Four Shades of Gray: The Amazon Kindle Platform

Rowberry, Simon Peter Book Four Shades of Gray: The Amazon Kindle Platform Platform Studies Provided in Cooperation with: The MIT Press Suggested Citation: Rowberry, Simon Peter (2022) : Four Shades of Gray: The Amazon Kindle Platform, Platform Studies, ISBN 978-0-262-36911-4, The MIT Press, Cambridge, MA, https://doi.org/10.7551/mitpress/11985.001.0001 This Version is available at: https://hdl.handle.net/10419/281349 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/4.0/ Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Four Shades of Gray Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Platform Studies Nick Montfort and Ian Bogost, editors Racing the Beam: The Atari Video Computer System, Nick Montfort and Ian Bogost, 2009 Codename Revolution: The Nintendo Wii Platform, Steven E. Jones and George K. Thiruvathukal, 2012 The Future Was Here: The Commodore Amiga, Jimmy Maher, 2012 Flash: Building the Interactive Web, Anastasia Salter and John Murray, 2014 I AM ERROR: The Nintendo Family Computer / Entertainment System Platform, Nathan Altice, 2015 Peripheral Vision: Bell Labs, the SC 4020, and the Origins of Computer Art, Zabet Patterson, 2015 Now the Chips Are Down: The BBC Micro, Alison Gazzard, 2016 Minitel: Welcome to the Internet, Julien Mailland and Kevin Driscoll, 2017 Super Power, Spoony Bards, and Silverware: The Super Nintendo Entertainment System, Dominic Arsenault, 2017 The Media Snatcher: PC/CORE/TURBO/ENGINE/GRAFX/16/CDROM2/SUPER/DUO/ ARCADE/RX, Carl Therrien, 2019 Who Are You? Nintendo’s Game Boy Advance Platform, Alex Custodio, 2020 Four Shades of Gray: The Amazon Kindle Platform, Simon Peter Rowberry, 2022 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Four Shades of Gray The Amazon Kindle Platform Simon Peter Rowberry The MIT Press Cambridge, Massachusetts London, England Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 © 2022 Simon Peter Rowberry All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher. The MIT Press would like to thank the anonymous peer reviewers who provided comments on drafts of this book. The generous work of academic experts is essential for establishing the authority and quality of our publications. We acknowledge with gratitude the contributions of these otherwise uncredited readers. This book was set in Filosofia OT by Jen Jackowitz. Printed and bound in the United States of America. Library of Congress CataloginginPublication Data Names: Rowberry, Simon Peter, author. Title: Four shades of gray : the Amazon kindle platform / Simon Peter Rowberry. Description: Cambridge, Massachusetts : The MIT Press, [2022] | Series: Platform studies | Includes bibliographical references and index. Identifiers: LCCN 2021013279 | ISBN 9780262543507 (paperback) Subjects: LCSH: Kindle (Electronic book reader) | Electronic book readers. | Electronic books. Classification: LCC Z286.E43 R689 2022 | DDC 004.1675— dc23 LC record available at https://lccn.loc.gov/2021013279 10 9 8 7 6 5 4 3 2 1 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Contents Series Foreword vii Acknowledgments ix A Note on Spelling xi Introduction xiii 1 Ebookness 1 2 The Amazon Infrastructure 27 3 Modeling the Book Trade 47 4 Format 61 5 The Everything Store? 81 6 Paratext Services 101 7 Social Reading 121 8 1984 Redux 143 9 The Future 155 Glossary 161 Appendix A: Methodology 167 Appendix B: Ebook Bibliography 177 Notes 185 Bibliography 217 Index 243 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Series Foreword How can someone create a breakthrough game for a mobile phone or a compelling work of art for an immersive 3D environment without understanding that the mobile phone and the 3D environment are different sorts of computing platforms? The best artists, writers, programmers, and designers are well aware of how certain platforms facilitate certain types of computational expression and innovation. Likewise, computer science and engineering have long considered how underlying computing systems can be analyzed and improved. As important as scientific and engineering approaches are, and as significant as work by creative artists has been, we also have much to learn from the sustained, intensive, humanistic study of digital media. We believe it is time for humanists to seriously consider the lowest level of computing systems and their relationship to culture and creativity. The Platform Studies series has been established to promote the investigation of underlying computing systems and of how they enable, constrain, shape, and support the creative work that is done on them. The series investigates the foundations of digital media— the computing systems, both hardware and software, that developers and users depend on for artistic, literary, and gaming development. Books in the series will certainly vary in their approaches, but they will all share certain features: • A focus on a single platform or a closely related family of platforms • Technical rigor and indepth investigation of how computing technologies work Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [xiv] irrevocably changed publishing. The device had a dramatic impact on digital workflows by encouraging publishers to adopt new processes that enhanced the quality of all publications, both physical and digital, since readers would now expect an ebook edition. Beyond pushing publishers to embrace digital forms, the Kindle empowered some previously marginalized readerships. The emergence of Kindle exclusives demonstrates the demand for genres not traditionally supported by publishers. Readers not catered to by printoriented publishing’s drive toward best sellers can now find an abundance of reading material. Ebooks also offer new physical accommodations: parents can hold a Kindle in one hand while attending to their child, but a print book may be preferable to read in bed. Ereaders require less space than physical books for people who cannot build a large library. Readers with visual or motor impairments may benefit from the ability to change font size or use texttospeech features, and the introduction of the OpenDyslexic font in 2015 catered to the needs of readers with learning differences. Despite these improvements, critics and publishers assume that digital publishing should integrate multimedia, virtual reality, and video game elements. Nourry’s comments reflect the antagonistic relationship between Amazon and the book trade stemming from ongoing concerns about Amazon’s sales strategies. Discussions of ebooks are often polemics around “disruption” or the superior materiality and smell of print. 5 Publishers have developed a niche market for luxury hardcover editions in response to readers who remain committed to print. Digital evangelism is pitted against print Ludditism, emphasizing extreme positions rather than considering the two media as complementary. In this book, I move beyond this antagonism to assess the impact of publishers, and the book industry at large, ceding the development of ebooks to a company that exemplifies the excesses of late capitalism and surveillance culture. Through this lens, I analyze how the Kindle has been a boon for publishing while simultaneously limiting the opportunities for developing an inclusive and forwardthinking digital platform. The name “Kindle” reflects the tensions that have plagued the platform. The name was met with negative reactions owing to potential links to Ray Bradbury’s novel Fahrenheit 451 and book burning. Bezos pleaded innocence, arguing the name was an attempt “to talk about the future of reading, but in a small, not braggadocio way,” to spark the imagination through reading. 6 He hoped the Kindle would become synonymous with ebooks as a verb akin to googling: “When I’m stuck in the airport or on line, I can Kindle my newspaper, favorite blogs or half a dozen books I’m reading.” 7 According to Brad Stone, a technology reporter for Businessweek, the Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Introduction [xv] platform’s development was driven by a desire “to thrive as a bookseller in a new digital age, [through owning] the ebook business in the same way that Apple controlled the music business.” 8 After the iMac’s success on Steve Jobs’s return to Apple, the company noticed a growing market for digital music. To capitalize on this trend, Apple released iTunes in January 2001, with the iPod following nine months later. The combination legitimatized digital music consumption at a time when Napster and other peertopeer networks dominated the public imagination. The iPod also removed barriers to entry through vertical integration of hardware, software, and services from purchasing music to listening on the go. This required substantial investment from Apple. Ebooks were in a similar position by 2007, with a loyal community of dedicated hobbyists but no single catalyst for broader uptake. The Kindle as Platform To subvert the antagonistic comparison of digital reading with print, Four Shades of Gray tackles ebooks as digital objects, focusing on the Kindle as the largest platform. While other ebook platforms exist, none have been as transformative as Amazon’s, owing to its unparalleled technical and social infrastructure. Nick Montfort and Ian Bogost proposed platform studies as a tool for the “serious and indepth consideration of circuits, chips, peripherals, and how they are integrated and used” for creative computing. 9 For example, my title Four Shades of Gray refers to the limited color palette of the firstgeneration Kindle that Amazon used to guide the aesthetic of the device’s screen savers, discussed in further detail in chapter 1. Platforms range from a video game console to an operating system or smartphone. Excavations of the Atari VCS 2600 or Adobe Flash are not mere technical excavations but considerations of the technology in its broader cultural and historical context. The Kindle was far from the first ereader to market, but Amazon integrated a range of services and offered a large catalog of titles to create a more printlike experience than its rivals. As Montfort and Bogost argue, “A platform in its purest form is an abstraction, a particular standard or specification before any particular implementation of it.” 10 A mature ebook standard could not just replicate the form of print but needed to capture the book’s functionality and replicate the networks of its creation, distribution, and reception. The Kindle must negotiate the inherent tensions between print and digital reading through existing simultaneously as a borndigital platform and a surrogate for print. Platform studies document the connections between computational architecture and creativity, but how does this work Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [xvi] when a platform is constrained by the norms and expectations of another medium? Audiences consume differently in print and digital form. For example, romance has enjoyed a resurgence digitally, while the full size of the market was never explored in print. Otherwise, the ebook acts conceptually a lot like its physical equivalent. Experimental forms of “electronic literature,” including Amaranth Borsuk and Brad Bouse’s Between Page and Screen, develop the expressive vocabulary of digital media. Borsuk and Bouse’s book is unreadable without the use of a webcamenabled computer, as the pages feature hieroglyphics that display poems rendered in three dimensions in front of the reader once presented to a camera. 11 These playful experiments have a shorter shelf life than print: Between Page and Screen relied on Adobe’s Flash Player, which was rendered obsolete by Apple’s decision not to support the web browser plugin for the iPhone. Further tensions persist between the longestablished traditions of publishing, where many of the largest companies were founded before the 1930s (Penguin is a latecomer in 1935), and the meteoric rise of Silicon Valley as a hub for digital innovation in the latter half of the twentieth century. Throughout the book, I assess the consequences of the book trade’s ambivalence toward digital media and how Amazon filled this demand with a technologyfirst approach. While this decision was mutually convenient for both parties initially, the longerterm impacts are still unraveling over a decade after the Kindle’s launch. The relationship between digital and print publishing challenges traditional notions of platforms as discrete entities. Tom Boellstorff and Braxton Soderman propose the term “transplatform” to describe the generative creativity encouraged by rivalries such as Intellivision and Atari or Sega versus Nintendo. 12 When placed in direct competition, technology companies innovate to outsell their rivals. The perceived antagonism between print and ebooks is a further example of a transplatform rivalry, but this is complicated by the internal crossplatform nature of the Kindle as a dedicated hardware range available for web browsers, personal computers, and mobile devices. The Kindle exists “in the weird liminal spaces that bridge one computational architecture to another.” 13 Thomas Apperley and Jussi Parikka argue that “the platform requires a degree of stability and consistency as a technical object.” 14 This is less clear here: What do we count as the Kindle platform? The dedicated ereaders? The content? The services? The Kindle further complicates the notion of platforms through its position at the intersection of two oftenconflicting definitions of digital platforms. This book is part of the Platform Studies series, which pioneered what Esther Weltevrede and Erik Borra term approaching “platforms as Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Introduction [xvii] architecture,” or analyses of the sociotechnical infrastructure of specific platforms and its influence on the production of creative content. 15 Since the publication of Racing the Beam, this technical approach to hardware’s materiality has been overtaken by Tarleton Gillespie’s more popular notion of “the politics of ‘platforms.’” 16 This definition of platform serves as the prefix for several neologisms, including “platform politics” and “platform capitalism,” which stem from the distinction between platforms, supposedly neutral sites for users to consume and share content, and publishers, the curators with direct responsibility for material they disseminate. The Kindle is a platform from the perspective of computational architecture, but within the broader Amazon ecosystem, it also functions as a political platform. By emphasizing the synergies between these two definitions in relation to the Kindle, I address the blind spots around how users engage with platforms in the architectural approaches and the technical limits of political economy analyses. In an interview with the journalist Steven Levy to promote the Kindle, Bezos stated, “Books are the last bastion of analog.... Music and video have been digital for a long time, and shortform reading has been digitized, beginning with the early Web. But longform reading really hasn’t.” 17 Resistance to extended reading onscreen dates back to at least the 1990s with publications such as Sven Birkerts’s The Gutenberg Elegies, which challenged electronic media’s ability to replicate the immersive experience of print. 18 The line of criticism has continued into the ebook era, with several studies documenting a gap in comprehension between students reading a text in print and digitally. 19 These projects often focus on reading to learn rather than as a leisurely pursuit, where accurate recall is less important. While the web revolutionized journalism, reference, and academic publishing, the strengths of trade publishers remained staples of print. The book trade predates other media industries, and practices from analog culture are deeply ingrained within the conventions of the industry. Since the advent of the personal computer, a cultural shift had occurred in publishing toward using digital tools to create and market print books. For example, the introduction of Standardized General Markup Language (SGML) in the 1980s allowed publishers to typeset digitally, a practice that became widespread by the early 2000s. Less attention was paid to the media shift of digitization, or converting books into digital publications. Print presented unique hurdles for digital consumption: five centuries’ worth of history to convert, the lack of an established format for reading, strong associations between the materiality of print and content. The problem was exacerbated by distracting speculative visions of digital publishing: Spotify for Books, blockchain for managing rights, datadriven Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [xviii] acquisitions. A large gap exists between publishers’ desire to step into an entirely new paradigm (without the investment required) and focusing on a more organic evolution of the current strengths of digital publishing. Ebook platforms are part of broader transformations of publishing in the early twentyfirst century, especially in digital publishing. 20 For example, Matthew Kirschenbaum’s “Book.Files” project reveals the challenges of publishers’ reliance on digital asset management (DAM) for the preservation of contemporary book publishing. 21 Four Shades of Gray builds on this work to demonstrate the impact of the publishing industry ceding control of digital distribution to a technology company at the expense of the industry’s own autonomy and the interests of readers. As a result, the market for ebooks has largely diverged from print, enabling new genres to thrive rather than supplementing existing revenue streams. Publishers’ stronger embrace of audiobooks shows that the industry has the will to adapt to digital distribution, so why did they not prioritize ebooks? Publishers and technology companies can be an uneasy mix. Obsolescence timescales differ substantially between the two. Publishers proudly market any Nobel Prize winners or New York Times Best Sellers on their lists decades after their initial publication date, while smartphones are called obsolete within a matter of years. Lisa Nakamura extends this logic, arguing that “older reading platforms like the firstgeneration Kindle may be worth studying because they were quickly obsolescent” 22 Her comments were published six years after the Kindle 1’s launch but failed to account for the longevity of the hardware, which received its last update in March 2016 to ensure it remained compatible with changes in Amazon’s wireless network. 23 The longterm support for hardware does not extend to services, however, as many functions available originally on the Kindle 1, and many later generations, are no longer available. The platform is constantly in flux, prioritizing shortterm policy and experimentation rather than ensuring permanence and archival consistency. The Kindle has stagnated, allowing for a moment of reflection on the state of the platform after its first decade. This book tackles the Kindle’s impact from the perspective of technology, texts, and uses. By analyzing the interconnection between these three areas, I argue that Amazon’s influence on publishing extends beyond “disruptive technology” to embedding itself in all aspects of the contemporary trade. This analysis can only be conducted through separating the publisher and technology company discourse from actual uses. The Kindle is a vast and complex platform with over five million ebooks, fifteen hardware launches, and forty million users. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Introduction [xix] A single book could never cover all facets of the platform, and my perspective is constrained to an Anglophone context, but I hope my work here will encourage further accounts of the Kindle’s influence in other contexts. While individual chapters will primarily appeal to readers interested in computational culture, bibliography, or reception studies, the three themes clearly overlap. Amazon is first and foremost a technology company, and this context is vital to understanding the company’s impact on a creative industry that prioritizes cultural influence and tradition over digital experimentation. Frederick Kilgour saw the “electronic book” as the “seventh punctuation” in the history of the book, an event on equal footing with clay tablets, the printing press, and offset printing. 24 It is still too early to judge the longterm impact of ebooks compared to these older technologies, but the shock waves of the Kindle continue to shape publishing. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 1Ebookness When Jeff Bezos revealed the Kindle on November 19, 2007, critics lauded the device’s combination of previous ebook successes while learning from other companies’ mistakes. 1 Amazon’s strategy was encapsulated in Bezos’s comment that the Kindle “isn’t a device, it’s a service.” 2 The Kindle is synonymous with ebooks as both hardware and consumable media because Amazon strived to emulate the book trade rather than focusing on translating print directly onto the screen. Criticisms of the Kindle often focus on visible design flaws or comparisons with the benefits of print, ignoring the platform’s complex back end. Ebooks are best defined according to their own media specificity and materiality rather than just their relationship to print. Resistance to ebook hardware stems from its disruption of the “natural” practices of reading print. Matt Hayler argues that the perception of print’s naturalness is the result of its normalization through centuries of use. The ebook’s artificiality can further be ascribed to “the sense that the text is somehow placed behind the object, that there is an additional layer of visceral insulation that must be fought through before it can be accessed.” 3 Ebook users must charge batteries and purchase content separate from the hardware. 4 Amazon’s marketing of the Kindle 1 emphasized the benefits of ebookness, including a screen that “looks and reads like real paper,” and a device “lighter and thinner than a typical paperback.” 5 The advertising also offered rebuttals to common worries about computers, including promises that the device was “simple to use: no computers, no cable, no syncing,” and “the screen never gets hot.” 6 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [2] The name “ebook” attempts to resolve the tension between print and digital consumption. As with “email” and “electronic signature,” the “electronic” book denotes only a conservative departure from its source material. Ebook hardware might extend features from print culture, but unlike an app or web page, an ebook will never challenge the fundamental concept of a “book.” Ebook standards including the Kindle and EPUB are primarily designed for trade publishers such as Penguin or Hachette, which publish material for the broadest possible audience. Trade publishers were skeptical of digital publishing’s benefits, so the ebook’s success relied on conservative changes (synchronizing progress across devices, changing font sizes) rather than experimenting with form. Amazon’s marketing and hardware design instilled ebookness in the public imagination. From Book to Ebook In an interview promoting the first Kindle, Bezos acknowledged the book’s technological advantage over ebooks: “You know, the physical book is so highly evolved and so elegantly suited to its purpose that it’s hard to improve on. It isn’t like some other artifact, some other object. It’s something very, very emotional and personal for people. But the book has a feature which I think is hard to notice, but it’s the book’s most important feature. And that is that it disappears.” 7 As Bezos notes, it is impossible to understand the Kindle’s success outside of the wider context of book history. Ample scholarship exists in the field, but four core aspects influenced the ebook’s development: discrete characters and words, mass reproduction, paratext, and sociality. 8 The invention of the alphabet as a set of discrete characters kickstarted the book’s development. The shift from scriptio continua (continuous writing) to spaces between words accelerated the process of abstraction in language, forming the discrete units of words, paragraphs, pages, and books. 9 This rigid codification of text as distinct from oral culture enhanced the prestige and authority of textual transmission. Gutenberg’s introduction of movable type to Europe in the midfifteenth century solidified the book’s importance by emphasizing mass reproduction. Print’s perceived fidelity through perfect replication elevated the medium as the gold standard for knowledge exchange. The publishing industry therefore resisted the challenges faced by the perception of fluidity in digital publishing. In the centuries after Gutenberg’s breakthrough, the growing market for books developed conventions for tackling the overwhelming volume of text. Publishers created what Gérard Genette terms “paratext,” including indexes, page numbers, and tables of contents to aid navigation. 10 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 1 Ebookness [3] Printbased paratext took advantage of randomaccess features of the codex, a collection of bound pages, to allow readers to quickly navigate through the text. The ebook could not replicate this essential affordance, leading to experimentation with alternative navigation systems. Sociality, the final characteristic of the book, was part of a shift that the librarian and bibliographer David McKitterick describes as a transition from the “public word, mediated on stone, metal or paint on buildings,” to “the privately mediated word.” 11 Although readers engaged with words in a private space, the book remained a social object through marginalia, sharing, and substantial alterations of texts. 12 A sustainable ebook platform needed to engage with the long history of print to establish how digital reading was compatible with conventional models of consumption. Because of this long evolution, the physical book endured as the “last bastion” of mass analog consumption despite several prophecies of the ebook’s imminent arrival. 13 Television shows such as Star Trek primed consumers for the ebook’s arrival. After the original series replicated the computational architecture of its time (large mainframes with whirring tape machines), The Next Generation envisioned speculative technology including the Holodeck and voice interfaces. 14 A place for linear reading still exists in the Star Trek universe, since crewmembers read on singleuse hardware, called Personal Access Display Devices (PADDs), which resemble modern tablet computers. 15 PADDs are signifiers of busyness, with Captain JeanLuc Picard’s desk frequently seen scattered with the devices. A scene from a 1996 episode of Deep Space Nine, “Homefront,” reveals the differences between Star Trek’s model of ebookness and the realities of ebooks in the mid2000s. In the episode, Captain Ben Sisko carries an attaché case containing several PADDs to hand out for a meeting. 16 PADDs were depicted as singleuse devices with no apparent mechanism for sharing content between them, a clear oversight when compared to the reallife ereaders that would come to market just two years later. Beyond Star Trek, Michael Hart’s Project Gutenberg was the most pervasive example of ebook culture before the Kindle. Hart aimed to create a digital “Library of Alexandria” from public domain texts. Project Gutenberg removed the text from its container. Slow internet connections made highquality facsimile photographs impractical for transmission, so the project instead extracted the text rather than replicating the original book’s format and layout. Progress was slow, however, as a small team digitized just ten titles between 1971 and 1989. Productivity increased with the uptake of the web in the early 1990s and the 2000 launch of Distributed Proofreaders, a tool designed to allow multiple users to proofread a text simultaneously. The project has digitized more than sixty thousand Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [10] reinvigorated ebook hardware market after Sony’s announcement of the Librie in 2006. In the subsequent final FCC filing for the Kindle in March 2007, all mentions of “Kindle” and “Amazon” are redacted in the user manual. 59 Nonetheless, the leaks built a yearlong period of anticipation among ebook enthusiasts. The external design of the firstgeneration Kindle was settled as early as 2005. Zehr’s physical prototype of Fiona, which now resides in the Computer History Museum’s Raiders of the Lost Ark– esque Milpitas warehouse, mirrors the final design’s form. 60 Without the final external plastic casing, the asymmetry of the keyboard, navigation buttons, and overall form is emphasized. The final product showed only minor changes, except for the removal of a builtin reading light peripheral that was designed and mentioned in the FCC filings. 61 The wedge shape and asymmetric navigation design were both already present, although some of the final aesthetic flourishes were still being debated. Lab126 wanted to acknowledge the long influence and evolution of print culture on the ebook with engraved letters on the device’s rear. Two prototypes from 2006 show the two final designs: one features exclusively cuneiform, the Sumerian writing system, arranged into a neat grid; the other features a waterfalllike arrangement starting with runes that slowly transform into the Latin alphabet. 62 The latter version was implemented in the final product and was used on 1.1 Fiona prototype unit housed at the Computer History Museum’s Milpitas warehouse. © Computer History Museum. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 1 Ebookness [11] the rubber rear cover and featured prominently in the Kindle 1’s booklike box packaging. While the hardware presented substantial design challenges, the rise of opensource software such as Linux afforded the Lab126 engineers the opportunity to adapt a preexisting operating system. 63 Linus Torvalds conceived of Linux in the early 1990s as a freely available and reusable version of the popular Unix operating system. Linux gained popularity, as it was designed to work with a variety of processors rather than remaining restricted to a single type of computer. As with other companies, Lab126 customized its version of Linux, derived from release 2.6.10, to remove superfluous hardware driver files and reduce the operating system’s overall size by 45 percent to 110 megabytes decompressed, an important consideration when the device’s overall storage space was limited to 256 megabytes. Hardware choices were made on similarly pragmatic grounds. Lab126 chose ARM processors, which had a track record with Palm OS devices, iPods, and iPhones owing to their low power consumption for mobile uses. Since the Kindle depended on extending battery life, all hardware needed to be power efficient. Amazon settled on Gumstix as the lowpower microcontroller, the main hardware interface between the processor and other chips, memory, and display. The choice would build on Gumstix’s architecture for E Ink’s displays rather than developing the interface between the architecture and operating system from scratch. 64 Other than reducing the overall size of the operating system by removing superfluous controller interfaces and introducing new power and display routines to suit the new technology, the Lab126 team made minimal changes to Linux 2.6.10 for the Kindle 1. Delays marred the Kindle’s launch. Some of the problems were logistical, as the shipping process for the electronic paper screens damaged the product because of atmospheric conditions during shipping. 65 Amazon was an unwitting bystander in takeovers and litigations that further delayed the device’s production. Intel’s announced exit from the mobile device processor market in June 2006 and sale of its XScale ARM series of processors, the Kindle’s core architecture, to Marvell created a complex transitional period, causing delays in receiving the microprocessors. 66 The wireless modem, Qualcomm’s AnyDATA, was also delayed after litigation by Broadcom threatened to stop the arrival of the modems in the United States. 67 Exacerbating these supply issues, Bezos encouraged Lab126 to constantly tinker with hardware specifications, which led to further delays in finalizing the device. 68 The early supply chain issues likely led to an understocking of the firstgeneration device, which sold out immediately Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [12] at launch and remained out of stock for months afterward, missing out on a lucrative holiday season. Since launching the firstgeneration Kindle in the United States in 2007, Amazon has unveiled a new generation each year except 2008, after substantial interest in the firstgeneration device in the United States led to troubles meeting demand. The Kindle has developed in four distinct phases (figure 1.2). The first two generations represent the experimental phase, where Amazon explored the device’s optimal design. For example, the Kindle 2 (2009) introduced a fourdirection navigation pad after the Kindle 1 trialed a onedimensional navigation bar. The Kindle 3 (2010) marked a move to the mature stage, as Lab126 refined the hardware to 1.2 Kindle genealogy Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 1 Ebookness [13] mainstream acceptance. The third phase was the longest sustained period without major innovation after the introduction of Touch (2011) and Paperwhite (2012) as further enhancements of the original device and reflecting the dominant trends in touch interface design encouraged by the emergence of the iPad. The latest phase, segmented, was marked by the simultaneous launch of the Kindle Basic and Voyage in 2014 as Amazon sought to create a market for both budget and luxury ebook readers after the failed attempt to create a luxury ebook reader market with the Kindle DX (2009), which had initially been designed with a “fully quilted surface” to denote luxury. 69 The Kindle 1 was never released outside the United States, but the Kindle 2 launched internationally in October 2009, eight months after its American debut, at an event during the Frankfurt Book Fair. 70 The launch was delayed by the complexity of negotiating roaming mobile data contracts across Europe, as well as Amazon’s insistence on having a selection of 280,000 books at launch in the United Kingdom, its primary European market. Genevieve Kunst, Amazon’s senior manager of digital media, addressed the Independent Publishers Guild UK conference in March 2009 to encourage independent publishers to digitize their back catalogs. 71 Six months later, Amazon’s Kindle selection on launch day included books from many of the largest publishers, although Oxford University Press and Macmillan did not initially sign up. Random House abstained too because of an ongoing payment dispute with Amazon. 72 Similar struggles between large trade publishers and Amazon have been persistent during the Kindle’s life span, but the international launch solidified the Kindle’s dominance of the ebook market. Ereaders, led by the Kindle, represented 25 percent of “reader device market share” in September 2009, second only to the desktop and printed PDFs with 42 percent share. 73 The Kindle DX launched in April 2009, three months after the Kindle 2, to break into the Portable Document Format (PDF) market with a 9.7inch screen, almost twice the size of other Kindle models. 74 The large screen allowed Amazon to target readers who wanted to keep the fidelity of PDFs, as well as readers who require larger font configurations. Rather than reflowing the page, the Kindle DX allowed users to view print media in facsimile. The device’s marketing relied on PDFs rather than any of Amazon’s proprietary ebook standards. Once native PDF reading was introduced on the DX, every subsequent hardware release featured the option. Specialized PDF ereaders have been a continuous desire for digital publishers, from EveryBook’s collaboration with Adobe, EB Dedicated Reader, in the late 1990s to Sony’s release of Digital Paper in 2014. The launch of the iPad took the market share instead. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [14] Amazon designed the DX with higher education in mind, as the integration of a native electronic paper PDF reader at near US letter size would offer a clear advantage for reading journal articles and textbooks onscreen. Evan Schnittman, a vice president at Oxford University Press, claimed the DX was “a new holy grail” for electronic textbooks. 75 Bezos announced the device at Reed College as part of a pilot study into Kindle use in the classroom at seven universities, including Reed, Princeton, Washington, and Virginia. 76 The experiment failed, with the American Council of the Blind and National Federation of the Blind suing Amazon over a lack of accessible navigation controls. 77 The DX was not doomed to obscurity, however, as a second version, the Graphite, was launched in 2010, and the two models are among the most desirable collectibles for ebook enthusiasts because of their large screen size and PDF compatibility, with devices selling regularly for over $100 and up to $300. 78 The Kindle 3, the beginning of the mature phase of Amazon’s dedicated ebook readers, offered a significant overhaul of its design in response to the success of the iPhone in July 2009. The quirky elements of the original two generations were replaced with a straight QWERTY keyboard and symmetrical design. Amazon integrated wellreceived features from the DX, as the Kindle 3 introduced the ability to view PDFs and added voiceguided menus to address the accessibility concerns raised by the DX lawsuit. Having iterated on design, the Kindle 3 marked the moment where the subsequent generation stopped radically departing from the previous device. Amazon’s continual support for developing ereading hardware was instrumental in establishing the viability of ebooks for publishers. After ereaders’ brief period of dominance, hardware is no longer central to any ebook platform’s business, as users are more likely to read ebooks on smartphones or multifunctional tablets, including the Amazon Fire, than on a dedicated device. The Kindle’s rise mirrored the smartphone’s ascent, fueled by the availability of wireless communication infrastructures and highresolution screens that are more convenient to hold in transit than a book. Apple sold 1.3 billion iPhones between 2007 and 2017, creating a larger potential user base for ebooks than dedicated hardware, and Amazon would have been foolish to ignore the opportunity. 79 The company addressed four core challenges facing ebooks in designing the Kindle: (1) ensuring the platform included enough content to build a user base; (2) developing displays sophisticated enough for reading onscreen for a sustained period; (3) navigation; and (4) connectivity. The last three challenges marked the beginning of a new form of ebookness that extended the core principles developed throughout the history of Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 1 Ebookness [15] the book and ensured a usable and efficient hybrid between digital and physical reading systems. Content John Maxwell argues that ebooks replicate simplistic web pages, but “the ebook privileges content where the web privileges connection. In doing so, the ebook adheres to an industryera scarcity model where content experiences are relatively few and must be acquired at some cost, whereas the web assumes sheer abundance.” 80 While ebooks do not emphasize the connectivity associated with websites, standards such as HTML and CSS allow the reader to change the formatting to their preferences. 81 Ebook formats are designed to be robust and encompass a wide range of genres from biography to poetry. Amazon prioritized content over connectivity as a core element of ebookness. Amazon’s ambitions required extensive digitization. Book conversion lags behind other creative industries, including film and music, as text is more laborious to digitize, requiring several stages from taking photographs of pages to optical character recognition (OCR), a process where an algorithm determines what text appears on a scanned page, and manual correction of errors. Transcribing a book manually can be more costeffective than automated processes. Due to text’s perceived simplicity but technical complexity, Van der Weel states that “paradoxically, text is both the first and the last of the medial modes that is to go digital.” 82 Amazon created Mechanical Turk (MTurk) in 2004 to accelerate digitization for “Search Inside the Book.” 83 As a corollary, the automation process offered a safeguard to prevent users from uploading pornography on the Kindle. 84 MTurk allowed Amazon employees, and later third parties, to set small tasks for completion in exchange for a nominal payment. MTurk’s name references an eighteenthcentury chessplaying “automaton” that actually featured a secret cabinet hiding an excellent human chess player. 85 Bezos described Mechanical Turk as the “guts of Amazon” alongside Amazon Web Services and framed the project as “artificial artificial intelligence” and “humansasaservice” for exploiting cheap labor to undertake menial tasks that one would assume that computation could easily automate, such as correcting scanned texts. 86 Mechanical Turk continued Amazon’s reliance on usergenerated content that started when its professional editorial staff was replaced by an influx of volunteercontributed reviews. Amazon’s business strategy depends on commodifying humans both as contingent labor for its warehouses, delivery network, and smallscale tasks and as data points Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [16] of consumption. Jeremy Antley describes our relationship to these platforms as “data serfdom,” where users produce value for the data lords with little return. 87 Amazon’s approach through MTurk is conspicuous owing to its framing. Lilly Irani argues that the service, and Amazon more broadly, rationalizes the contingent labor and data serfdom by “rendering it manageable through computing code.” 88 The dehumanizing rhetoric of the exotic “Turk” (a name Bezos approved despite the negative connotations, noting that “he personally would bear the responsibility for any backlash”) 89 and framing humans as “services” and “artificial intelligence” are emblematic of Amazon’s attempts to mask the company’s reliance on manual human labor. Other than Amazon Logistics, the company’s delivery service, the vast labor force of over a half million workers is hidden from sight, with an emphasis instead on automated grocery shopping, drone delivery, and “artificial artificial intelligence.” Display and Navigation Electronic paper was a significant factor separating the Kindle and Sony’s 2004 launch of the Librie from earlier ereading hardware. 90 Futurists had long predicted “smart” paper that replicated the dynamism of digital displays while maintaining the flexible materiality of print, but the technology’s history starts in the 1930s with innovations in photocopying. Chester Carlson’s invention of “xerography,” or “dry writing,” in 1938 not only led to Xerox’s dominance of the photocopying industry but also created the “direct positivetopositive operation” technology behind electronic paper. 91 Serious work to translate the principles of xerography to display technology began in the early 1970s, when researchers at both Xerox PARC in Silicon Valley and Matsushita in Japan developed electronic paper screens. 92 Neither research group was focusing on developing the technology for digital book publishing, as Nick Sheridon aimed to create Gyricon as a display screen for desktops coming from Xerox PARC, while Matsushita was interested in using the display for clocks. By 1997, “electric paper” reemerged as one of the top priorities for Xerox PARC to rebrand as the “Digital Document Company.” 93 Sheridon’s vision was realized in the late 1990s, when the MIT undergraduates J. D. Albert and Barrett Comiskey, along with their professor Joseph Jacobson and investors Jerome Rubin and Russ Wilcox, founded E Ink. The team positioned itself as a disruptive force within publishing. An early profile in Wired highlighted the company’s desire “to replace ink as we know it.” 94 The startup entered into an exclusive partnership with the Dutch company Philips Electronics in 2001 to create “radio paper” Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 1 Ebookness [17] (the combination of electronic paper with wireless connectivity), with ambitions to launch in 2005. 95 When the device failed to materialize, Prime View International (PVI), a Taiwanese company, purchased the license from Philips in 2005, before acquiring E Ink in 2009. 96 Once Philips canceled its exclusivity contract with E Ink, companies were able to build on technology that had not yet fulfilled its potential. The Kindle 1 used E Ink’s electrophoretic display (EPD), the cutting edge of electronic paper technology by the mid2000s. The screen renders text and graphics using “positively charged black particles and negatively charged white particles.” 97 Electrode layers placed on either side of the electronic paper capsules foreground a ratio of black or white particles according to the shade of gray required. Figure 1.3 visualizes how six pixels are rendered through the electrode layers. Grayscale is more viable than fullcolor display, which requires an assembly of capsules in red, green, and blue. Electronic paper extends battery life. Once a page is rendered on the screen, the text can be displayed indefinitely without requiring further power until a page turn alters the signals from the electrode layers. Additionally, the technology did not require back lighting, producing a more natural reading experience. The technology does involve some tradeoffs: electronic paper can display static grayscale images, but colors and the refresh rate required for videos remain beyond the capabilities of electronic paper. Nonetheless, electronic paper is reliable for static images and remains the preference for a niche group of readers who continue to use ereaders. E Ink has improved its screen technologies significantly since 2007. The first update came with the arrival of the Kindle DX, which used E Ink “Pearl,” reducing the image update time of text by 50 percent and images by 40 percent. 98 The Oasis, Amazon’s flagship ereader in 2017, used E Ink’s updated Carta screens, which refreshed the image on 1.3 Electrophoretic display. Source: Run!, “File:EInk.Svg,” Wikipedia, 2006, https:// en.wikipedia.org/wiki/File:E-ink.svg. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [18] a screen in 75 percent of the time of the previous generation, creating a smoother reading experience. E Ink Carta also increased the fidelity of images and text rendering to 300 pixels per inch (PPI), a measure for how visible individual pixels are on a display. This pixel density offered the equivalent of the first generation of Apple’s Retina technology, which was marketed as the pixel density exceeding the input capacity of the human eye. 99 This clearly lagged behind the vanguard of mobile screens and came with a premium price, which has led to stasis in one of the most visibly important technologies in the Kindle’s stack. Amazon needed to adapt the Linux libraries for the Kindle to ensure that users could use the hardware’s limited audiovisual capabilities through Advanced Linux Sound Architecture (ALSA) and the libraries for JPEG and PNG image formats. Audio and still images were peripheral to the firstgeneration device but enabled Amazon to offer an “experimental” MP3 player, and publishers could embed images at a low resolution and use up to four shades of gray for images. Despite its continual reliance on electronic paper, the dedicated Kindle ereader has yet to be commercially viable with more than sixteen shades of gray. The introduction of E Ink Pearl with the Kindle DX upgraded the color palette to sixteen shades of gray, allowing more complex images to appear on the device. As an illustration, we can compare a screen saver image of Jules Verne, author of Journey to the Center of the Earth, that appeared on both the Kindle 1 and DX (figure 1.4). The limited palette of the original Kindle required optimization to be effective, an additional cost for publishers. The necessity to display PDFs on the DX required higher fidelity. Electrophoretic displays remain the most common commercial electronic paper technologies despite continual investment in developing alternate technologies such as “electrowetting,” a fullcolor, flexible “videospeed” form of electronic paper. 100 Amazon invested heavily in the technology and filed over 250 patents containing the keyword “electrowetting,” but no commercial product has included the new displays. Companies have also attempted to introduce full color to electrophoretic displays, with E Ink’s research and development team demonstrating a prototype in 2016. 101 A paradigm shift to fullcolor electronic paper will revitalize the ereader market by offering more direct competition with tablets. The closure of Liquavista, Amazon’s electrowetting display spinoff, in early 2018 painted a bleak picture for the future of the display technology. To overcome the limitations of electronic paper during the production of the first Kindle, Lab126 engineers incorporated a narrow second Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 1 Ebookness [19] screen using a polymer network liquid crystal display (PNLCD), a display technology similar to electronic paper that was not backlit and limited battery consumption to counter the slow refresh rate of the electronic paper (annotation 108 on figure 1.5). 102 The strip updates more quickly than the electronic paper to create a navigation aid so that users knew which line or menu option was selected without refreshing the primary screen. Lab126 hard coded the strip to feature a maximum of two hundred lines, restricting the total number of lines viewable on the Kindle at one time. 103 The navigation bar required constant battery power and thus did not persist once the user switched the display to standby. While the PNLCD was the final compromise, the Lab126 engineers also investigated the potential of configurable keyboards where touch screen sensors were placed under a second electronic paper screen that could refresh different configurations according to the user’s current navigation needs (figure 1.6). 104 Once the reader has opened an ebook, the keyboard is wasted screen real estate, when navigation is paramount. This option would offer more customization but would not solve the problem of finding one’s position on the primary screen. The idea would eventually be implemented with the arrival of the Kindle Touch. Electronic paper offered new opportunities for screen savers. Earlier implementations of screen savers for cathoderay tube (CRT) a) b) 1.4 Comparison of E Ink displays of Jules Verne screen savers. Left: Kindle 1, four shades of gray. Right: Kindle DX, sixteen shades of gray. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 2The Amazon Infrastructure The Kindle occupies an awkward position in histories of mobile computing. It was not the first electronic paper device, and it was overshadowed by the launch of the iPhone seven months earlier in March 2007. The Kindle instead depended on Amazon’s infrastructure to attract users. The company rapidly changed from a web retailer to essential infrastructure for a variety of firstand thirdparty services in its first decade. The company prioritized technologies and infrastructure to transform industries rather than replicate preexisting models. Take web retail as an illustration: in the mid1990s, mail order was an established business model, but emerging web technologies offered vast improvements over the web’s slower and less dynamic predecessor. Amazon took this approach to all new projects. Werner Vogels, Amazon’s chief technology officer, argued that “Amazon is a technology company. We just happen to do retail. Everything Amazon does is driven by deep technology stacks.” 1 Danny Fortson and Simon Duke, journalists for The Times of London, describe Amazon’s infrastructure as “the fourth utility— after water, gas and electricity.” 2 The wires and protocols of the internet are far more important as a utility than Amazon (notwithstanding the company’s entry into laying undersea cables in 2018), but the hyperbolic comparison gives us a reasonable starting point for thinking about infrastructure. 3 Utilities, roads, fiber optics, and WiFi networks are all examples of infrastructure. Etymologically, the prefix “infra- ” comes from the Latin preposition for “beneath.” Susan Leigh Star and Martha Lampland extend this definition by noting that “infrastructure is something that other things ‘run Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [28] on,’ things that are substrate to events and movements.” 4 Electricity powers other tools rather than being an intrinsically valuable commodity. Others build on the infrastructure developed by utilities and internet services. Often the infrastructure is hidden in favor of the final service. For example, we are less concerned with how an email is transferred from California to Hong Kong than with its authorship and receipt, but a vast technical infrastructure of servers and protocols ensures the message’s crosscontinental delivery within minutes. Large technology companies develop modular infrastructure to enable cross compatibility for engineers to rapidly build new internal projects. 5 Amazon recognized the value of these services to a wider audience and elected to allow third parties to access the infrastructure. The company diversified its business interests to ensure the maximum number of third parties used its services. The footer for Amazon’s home page lists forty different enterprises ranging from Withoutabox, a service for independent filmmakers to submit work to film festivals, to Amazon Restaurants, a homedelivery service built on Amazon’s delivery network. The company has further business interests in cloud computing and storage (Amazon Web Services, Drive), publishing (DPReview, Kindle Direct Publishing), film (Box Office Mojo, IMDb), fashion (Fabric, 6pm, Zappos), web analytics (Alexa), and video games (Twitch, Lumberyard, Games Studio). These diverse services are connected by providing infrastructure for others. Amazon’s annual reports often highlight these successes. For example, in the 2015 letter to shareholders, Bezos states: “Used well, our scale enables us to build services for customers that we could otherwise never even contemplate.” 6 Infrastructure forms “a gateway, permitting other systems to interact with [Amazon] to form a seamlessly interactive network.” 7 Any technology company forming an integral part of the network will remain important as infrastructure even if its primary products or services diminish. Amazon’s infrastructure formed the basis of the company’s plan for the Kindle’s broad adoption. Sony and Cybook launched ebook readers with similar specifications to the Kindle in late 2007, but the competitors lacked Amazon’s sophisticated infrastructure. Sony offers a useful illustration of the benefits of mature infrastructure, since the company failed to create a sustainable ebook business despite its resources and experience building hardware compared to Amazon, which had never released a device before the Kindle. We can think about media formats as a form of infrastructure. The physical medium of a CD or tape cassette plays a subordinate structural role to the content stored on it. Sony has a patchy history of introducing new formats, with Betamax and MiniDisc as just Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 2 The Amazon Infrastructure [29] two of its highprofile failures. The company’s investment in proprietary media formats highlights its main limitation in developing infrastructure: it focuses less on cross compatibility, leading to higher development costs. Amazon instead followed Apple’s road map for iTunes and the firstgeneration iPod by integrating its core services— webbased retail, a large cache of usergenerated data, Amazon Web Services, Mechanical Turk— into the Kindle platform. Early commentary asked if this was the ebook’s “iPod moment,” although the Wall Street Journal counteracted this early hype by suggesting that the Kindle was having an “eighttrack moment” as a temporary phenomenon. 8 This initial pessimism stemmed from previous ebook failures and framed Amazon primarily as a retail company in the mid2000s. The popular press’s reporting on Amazon drives its public perception. Profiles of the company viewed its history through the perspective of retail, indicated by titles such as One Click and The Everything Store. 9 For example, Brad Stone frames Amazon Web Services in terms of “the everything store” that “stocked Amazon’s shelves with incongruous products like spot instances and storage terabytes.” 10 Even in 2020, when global COVID19 lockdowns boosted Amazon’s share prices from $1,900 in January to almost $3,500 in September, the focus remained on the growth in terms of home delivery, ignoring how Amazon Web Services also saw a dramatic increase in traffic owing to the shift to working from home. 11 Table 2.1 shows thematic words that appear frequently alongside “Amazon” in news reports indexed by Google News between 2010 and 2018. Journalistic accounts focus on the company’s technology but emphasize its retail operations over developments in media and other markets. This is not atypical, however, as General Electric is still associated with hardware and Microsoft with productivity software despite the companies’ substantial investment in other business models. Amazon is most Table 2.1 Common themes in top 100 collocates of Amazon in the Brigham Young University NOW corpus Theme Frequency Technology (e.g., “Google,” “Cloud”) 39,522 Retail (e.g., “Purchase,” “Retailer”) 30,185 Media (e.g., “Studios,” “Netflix”) 11,258 Meta (e.g., “Bezos,” “CEO”) 11,196 Other (e.g., “Peru,” River”) 5,221 See appendix A for full details. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [30] productively categorized as one of the five “technology giants” alongside Alphabet (Google’s parent company), Apple, Facebook, and Microsoft. 12 Disparate interests across the companies can be tied together by a common use of technology. Any attempt to summarize Amazon’s complex and diverse historical trajectory will be reductive by nature, and others have already written detailed histories of the company. 13 My interest here focuses specifically on how Amazon the technology company’s development of the Kindle shaped our popular understanding of ebooks. Nonetheless, several broader concerns are worth addressing first. Undercover investigations by newspapers including the New York Times revealed exploitative labor practices in Amazon’s warehouses, head offices, and hardwaremanufacturing collaborations with Foxconn in the Hunan Province of China, which, along with perceived tax evasion tactics, overshadow the company’s dominance within retail. 14 The drip of negative articles about Amazon turned into a steady flow by the late 2010s, challenging the company’s narrative. Brian Merchant has documented the problematic connections between Amazon and fossil fuel companies. 15 Caroline O’Donovan and Ken Bensinger revealed the extent of deaths connected to Amazon drivers. 16 It is no coincidence that these investigations emerged while Amazon consolidated its position as one of the largest technology infrastructure companies. As with similar platforms, including Cloudflare, a comparison to more traditional infrastructure such as roads allows these companies to plead neutrality when offering support for causes that would otherwise be deemed unsavory. These criticisms still frame Amazon as a retailer, ignoring expansion into other areas even if they are the primary focus of the article, such as a subhead for an article about streaming football matches describing the company as a “US online retailer.” 17 The focus on expansion over profitability encourages this narrative, as it hides Amazon’s interests in diversification and surprises pundits when, for example, the acquisition of Twitch did not fit into preconceptions of Amazon’s strengths. 18 Retail wisdom suggests that scale is essential, but Amazon understood that, in the digital age, advanced technology trumps physical inventory. Amazon was an underdog to established retailers such as Walmart, Toys“R”Us, and Barnes & Noble, which had yet to exploit the technological and logistical affordances of online retail. Amazon’s focus on technology and infrastructure for retail encouraged Toys“R”Us to partner with the company to boost its online presence in 1999. 19 Amazon’s origin story is well known: Bezos was working as an analyst for the Wall Street investment firm D. E. Shaw in May 1994 when he had a “eureka moment” about online commerce. 20 The web was changing from a Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 2 The Amazon Infrastructure [31] utopian space of experimentation to a commercial juggernaut. 21 Bezos had an epiphany after reading that web use was likely to increase by 3,200 percent per year as users quickly adapted to webbased retail. He identified books as a suitable foundation for establishing an online retail business. International Standard Book Numbers (ISBNs) were widely adopted in the book trade and were easy to convert into a catalog. The materiality of print was also amenable to mail operation: books were nonperishable and could withstand manhandling while shipped. 22 After a beta launch in 1994, the website opened to the public in 1995 and was an immediate success. In 1996 the Wall Street Journal called Amazon an “underground sensation for thousands of booklovers around the world, who spend hours perusing its vast electronic library, reading other customers’ amusing online reviews— and ordering piles of books.” 23 The novelty of Amazon’s range of services helped normalize online retail and laid the foundations for a more extensive infrastructure. Virtualization and AWS Amazon Web Services (AWS), launched in 2004, was the catalyst for the company’s transition from an online retailer to a cornerstone of internet infrastructure. The service was intended as a commerce application programming interface (API) for third parties to use the powerful infrastructure Amazon had built to date. 24 Instead Amazon built a “cloud computing” platform integral to the internet’s infrastructure and the second part of what Bezos described as Amazon’s “guts” in tandem with Mechanical Turk. 25 Its entry into cloud computing was a dramatic turnaround from Bezos’s acknowledgment in 1999 that the company did “not have backup systems or a formal disaster recovery plan.... Computer viruses, physical or electronic breakins and similar disruptions could cause system interruptions, delays, and loss of critical data.” 26 Many largescale web applications use AWS, most notably Netflix, a direct competitor of Prime Video. Brad Stone argues that this marked a turning point in the development of Amazon, as it expanded the scope of Amazon’s products beyond tangible goods to infrastructure and services. 27 The company was an early mover with cloud computing in 2006, which allowed it to build capacity at scale. In turn, this profit could be turned into further investment in research to create new services that benefited external customers and, more importantly, Amazon’s businesses. The mythical origins of AWS recall that the service developed through selling spare server capacity that would otherwise only be used during peak demand. Amazon’s need to survive the onslaught of purchases Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [32] between Black Friday and Christmas led to surplus servers for the rest of the year. Rather than lease extra servers at a premium cost for the busiest period of retail, Amazon could purchase capacity for peak times and then rent the surplus for the remaining ten months of the year. Werner Vogels dismissed this narrative, stating that AWS was developed to provide Amazon employees with a suite of prebuilt network tools to avoid having to develop this infrastructure from scratch before undertaking new projects. Once Amazon introduced this service for employees, it was a short step toward selling this infrastructural service during the early hype around cloud computing. 28 Since Amazon pivoted earlier than its competitors, the company established itself as a leader when others, including IBM and Microsoft, took longer to scale up. AWS’s physical presence is hidden literally as commercially sensitive information, and metaphorically by the cloud’s association with ephemerality and immateriality. 29 Amazon’s publicfacing customercentric ethos requires the environmental and labor costs to remain hidden to drive the narrative. These centers are the counterpart to Amazon’s distribution centers in enabling the Kindle’s online infrastructure to prioritize areas according to socioeconomic affordances and limitations. AWS is one of Amazon’s most profitable services. In 2019, AWS provided 12 percent of all net revenue, or $35.03 billion in total, for a profit of $9.2 billion. Conversely, despite revenue in excess of $170 billion, North American product and service sales returned a smaller profit margin of $7.03 billion, with estimates that costs from COVID19 will further reduce profitability of the company’s physical services during 2020. 30 Providing critical digital infrastructure is more profitable than retail, leading to further investment in cloud services at the expense of the perceived traditional strengths of Amazon, including bookselling. AWS was instrumental to the Kindle’s development and success because of its focus on virtualization and abstraction of the computational process. An IBM white paper identifies the virtual computer as “a logical representation of a computer in software [to provide] more operational flexibility.” 31 The Kindle offers a similar logical representation of the book in software, extending beyond facsimile privileged by formats such as PDF. As a result, ebooks model books, while other formats model text or print. This logical representation requires additional features to differentiate a book from other digital written genres such as journalism or blogging. For example, EPUB extends the base HTML specification by introducing new tags including “spine” and font obfuscation to simulate the affordances of the book removed by textoriented specifications. The Kindle was the first ebook platform to focus on virtualization rather than offering a flat Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 2 The Amazon Infrastructure [33] representation of print, complementing its precursor. The flexibility of cloud computing, which moved ebook infrastructure from ondevice to remote storage, also created opportunities for reading surveillance on an unprecedented scale. Virtualization services such as AWS are an essential component to the Kindle’s success, as they encapsulate both the product and the process. The practice enables the complex modeling of the book trade and a way of integrating the structure of the book trade into a relatively underpowered computer. The Rhetoric of “Innovation” Beyond AWS, Amazon has invested heavily in new technologies. The five largest technology companies (Alphabet, Amazon, Apple, Facebook, Microsoft), known as the “Big Five,” use the rhetoric of innovation to compete in the marketplace. This manifests in prizing startups and disruption over focusing on improving existing products and processes. Technology companies use research and development as a central strand of their marketing strategies. For example, in an interview on 60 Minutes and a promotional website in 2013, Bezos revealed plans for “Amazon Prime Air,” a drone delivery system, before the technology was approved for testing. 32 The announcement was a statement of intent designed to build excitement and increase share prices rather than demonstrate meaningful innovation. Seven years later, the company is still performing closed tests in Cambridge, England. Amazon faced greater barriers to entry into an increasingly restricted drone airspace, leading to further delays, but the reveal of speculative technology so far in advance of its implementation is highly unusual for a company like Amazon. Beyond this big marketing push, patent filings, acquisitions, and academic collaborations provide a more meaningful insight into how Amazon frames its role as a technology company, as such documentation provides evidence of unfinished and abandoned ideas. Alessandro Delfanti and Bronwyn Frey used Amazon’s patent filings to explore how the company explores the future of work in an example of how this gray literature helps illuminate corporate ideologies. 33 The company’s famed secrecy prohibits engineers from producing material for academic journals and conferences in line with its main rivals. 34 Academic collaborations are often local and outside the company’s main business interests. For example, Amazon maintains strong links with the University of Washington, endowing the computer science department with two chairs in machine learning, and running the general Amazon Catalyst research fund for any initiative looking for funding at Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [34] the university. 35 The first rounds of Catalyst provided funding for projects as diverse as health, 3D printing, renewable energy, and robotics. Amazon Web Services offers an open credit scheme for educational projects to train future engineers. 36 Despite this gesture toward openness in the broader scientific community, this segmented access separates Amazon’s core businesses from academic collaborations, unlike the open nature of Microsoft Research or Google and Facebook’s interactions with universities. The company’s collaboration with academics therefore offers little useful insight into Amazon’s infrastructure. Acquisitions are more fruitful for understanding Amazon’s position in the technology marketplace. The company’s first acquisitions in 1998 were PlanetAll, “a Webbased address book, calendar and reminder service,” and Junglee, “a leading provider of Webbased virtual database technology, which allows visitors to access a variety of products sold by other merchants.” 37 The two startups were acquired for infrastructure purposes instead of adding a flashy new service direct to customers. A year earlier, Bezos’s first letter to investors gestured at this interest: “The Company’s current strategy is to focus its development efforts on creating and enhancing the specialized, proprietary software that is unique to its business and to license commercially developed technology for other applications where available and appropriate.” 38 Acquiring businesses to license technology was the first shift in the company’s priorities from retail to providing infrastructure for third parties. Amazon’s patents are its most public statement of intent regarding technological and infrastructural innovation. The company was one of the top twenty most successful patent applicants in 2018 in a list dominated by hardware manufacturers including Samsung, Qualcomm, Intel, and Taiwan Semiconductor. 39 Lab126’s focus on hardware increased the rate of patent filings, but Amazon also continued to protect algorithms and software. This diversity reveals what kind of hardware the company is developing, as well as many of the processes underlying these developments. Patents therefore are useful markers of Amazon’s technological progress. Bezos’s early patent filings were controversial, particularly “Method and system for placing a purchase order via a communications network,” filed in September 1997 and granted exactly two years later. 40 The innocuous title belies the muchcontested patent’s claim that Amazon was first to develop the oneclick payment option for webbased retail. The technology allows users who have entered their personal details and credit card information to choose to buy something without requesting confirmation. Users would not have extra steps to confirm their purchase and reconsider the decision, thus replicating impulse purchases in brickandmortar Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 2 The Amazon Infrastructure [35] stores. The patent caused immediate concern within the technology community, drawing criticism from Tim O’Reilly, an influential computer guide publisher, and Richard Stallman, an early free software advocate, who viewed the patent as an implementation of preexisting HTTP cookies rather than a substantially new idea. 41 While cookies were originally designed for Netscape Navigator, David Kristol and Lou Montulli’s Internet Engineering Task Force Request for Comments (RFC) “HTTP State Management Mechanism” proposed an interoperable standard for articulating “stateful sessions with HTTP requests and responses.” 42 In other words, cookies enable servers to identify returning visitors along with limited information about their previous visits. The initial RFC mentioned web retail as a possible implementation alongside personalized journalism recommendations. Paul BartonDavis, an early Amazon programmer, opposed the patent, which he believed was contradictory to the principles of open technology driving the web’s early development as well as Amazon’s early ethos. 43 Bezos rejected the claims, stating that “the vast majority of our competitive advantage will continue to come not from patents, but raising the bar on things like services, prices, and selection,” but he agreed that software patents should be protected for a shorter time and agreed to fund a prior art database that never materialized. 44 The controversy did not slow Amazon’s patentfiling policy, as it has been granted over eleven thousand applications since 1997, revealing a gap between ideology and practice in the company’s technological development. For example, Lab126 filed seven patents related to the Kindle 1 simultaneously in March 2006, more than eighteen months before the device was launched. The seven patents feature the same context and diagram but diverge in the claims made, outlining separate processes and innovations from the shape of the device to its cloudbased delivery. As an illustration, “Handheld electronic book reader device having asymmetrical shape” reveals Amazon’s interest in ebooks in 2006 and how this was reshaped by the attrition of the application process in its final iteration in 2016. 45 The original application made seventyone claims, which were whittled down to twenty in the accepted application. The delays revolved around contentions over the similarities between the proposed wedgeshaped reading device and a patent filed by Nokia in 2002. 46 In response, Zehr and Whitehorn emphasized the media specificity of the invention by focusing on ebooks rather than the original claims for “electronic media,” as well as the specific hardware configuration of four screens in nonparallel structures and the wedge’s specific angle, which replicates a foldedback paperback. Zehr and others at Amazon sought to claim the design was unique among mobile computers, but this was narrowed to outline how Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [42] broader plans to turn retail customers into users of the full suite of Amazon services. Subscribers soon gained access to premium Twitch, music, and video services while receiving discounts for specialist retail operations including Pantry, a home delivery grocery service. In the 2016 report to shareholders, Bezos describes Prime as one of the three pillars of Amazon alongside Marketplace and AWS. 71 All three services form the blueprint for Amazon’s “walled garden,” where users can access the majority of services required on the internet. 72 Schemes like Prime invoke a fear of sunk cost, where customers are more likely to purchase something via Amazon to make the most of the membership fee even if the product is cheaper elsewhere. By encouraging users to see Amazon’s ecosystem as an interconnected suite of services, Prime has become central to the company’s businesstoconsumer offerings. In this sense, Prime functions as a service infrastructure, offering a common base for consuming Amazon services that discourages users from moving to another platform. The subscription service permeates all parts of Amazon, including the Kindle. Two subscription services are available for the ebook platform: Prime Reading and Kindle Unlimited. The latter offers a fuller library at an additional cost, but Prime members gain the benefits of the former. Both schemes aim to change readers’ consumption habits so that they remain Kindleonly readers. If many exclusive ebooks cost only a couple of dollars, and Unlimited costs $9.99 to access the same content, users are encouraged to read up to five books a month to make the most of their membership. Amazon invested in online retail infrastructure to build a reputation in an underdeveloped niche. The company was keen to shift to a serviceoriented business model early in its development. While Amazon struggled through the malaise of the dotcom bubble in 2000, Bezos suggested that the company might become “an incubator for ecommerce companies that can start companies at lower costs and more quickly than any other company in the world.” 73 The rhetoric was toned down in future press releases, but the ethos remained throughout the development of thirdparty retail services starting with zShops in 1999. The service provided the blueprint for today’s thirdparty marketplace by allowing merchants to “offer a vast array of popular or hardtofind items” on the web store. 74 The storefront merged with Amazon Auctions, a failed attempt to compete with eBay, to form the current marketplace. Traditional retailing wisdom suggests that allowing third parties to sell direct to consumers undercuts sales, but Amazon understood the importance of transaction logs as a commodity. Businesses have taken advantage of its infrastructure for thirdparty retail. For example, Thriftbooks Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 2 The Amazon Infrastructure [43] sells books for under a dollar and sells up to twelve million books a year via Amazon through rescuing “landfillbound books, sight unseen, for around 10 cents a pound.” 75 The most successful operations employ data scientists and financial analysts to find the most profitable books. Amazon benefits from the data to enhance its recommendation algorithms, identifying older titles to bring to the Kindle and pocket transaction fees without the need to warehouse items. In 2006 the company launched Fulfillment by Amazon (FBA) to allow third parties to use its warehouse and delivery infrastructure. 76 FBA reflects Bezos’s focus on scale: Amazon warehouses are larger than any thirdparty startups’ storage facilities, so both parties will mutually benefit by using this scale to market to a large user base. The scheme also allowed Amazon to recoup the shipping costs lost through Prime subscriptions, as third parties pay the company to store and deliver the items. The ability to target both end users and businesses as customers was replicated in the development of the Kindle, as Amazon could use its large customer base to encourage publishers to digitize books that remained out of print but interested users. FBA is just part of Amazon’s shift from service provider to service infrastructure, where scale and volume are more important than limiting direct competition. In his 2017 letter to shareholders, Bezos noted that over half of all products sold through Amazon were sold by third parties, cementing the company’s importance as web retail infrastructure. 77 Ben Thompson argues that Amazon is interested in developing an infrastructure for all retail activity and transactions rather than focusing on one business model, no matter how contradictory those motives might appear externally. 78 The introduction of Amazon Pay in 2007 to compete with PayPal expanded the company’s transaction log data beyond the confines of its own retail website to gather further evidence of consumer habits across the web. 79 Aiming to leverage the company’s reputation for secure payments, Pay never reached the same customer base as PayPal, despite Amazon’s substantial market share in retail. PayPal is a direct competitor, and its user base may support it instead of Amazon and may boycott the company. The socalled Everything Store is designed to be comprehensive, so loyal customers have little need to shop elsewhere. Amazon Pay shows that the company’s infrastructure has its boundaries and outside of the hidden use of AWS, it is less likely to succeed outside of Amazon’s core service infrastructure. The internal retail operation was far more successful. Bezos capitalized on the fact that brickandmortar bookshops would never be able to compete with Amazon’s ability to cater to the demand of the “Long Tail,” the model that 90 percent of all interest comes to the top 10 percent of Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [44] most popular items, while the final 10 percent of interest is spread thinly through the remaining 90 percent of items. 80 In bookselling, this is the distinction between best sellers— titles that sell large quantities in a relatively short period— and the backlist, older titles that sell steadily if left visible in a store. Brickandmortar bookstores traditionally chase best sellers rather than selling older titles, which can only feature limited stock, as booksellers must ensure all books will sell in a reasonable period. Brickandmortar retailers rely on “core stock” such as The Highway Code and The Hobbit, which are likely to sell consistently over time. Conversely, Amazon’s extensive warehouse network and online catalog allow the company to profit from obscure titles hidden from traditional bookshops’ customers. A store with over four million unique titles can be unruly unless tamed by a search bar. The runaway success of best sellers such as a new Harry Potter title allows Amazon to offer such books at a heavy discount, thus attracting more new customers. Amazon’s catalog of sixtyfive million other books props up discounting practices even if each only sells a single copy a year. 81 The company’s data collection practices allow it to profile networks of users’ interests in niche genres and provide recommendations from the end of the long tail. In 2016, Amazon transitioned from online to physical retail by opening its first bookstore in Seattle, followed by other branches across the United States. Ironically, it wished to avoid another company undercutting its physical retail and filed a patent in May 2012 describing a method for blocking customers’ access to competitors’ prices when physically in a store. 82 The company had previously benefited from this phenomenon, as browsers refer to Amazon’s app for price comparisons when in the competitions’ shops, and the patent is a defensive tactic for ensuring other retailers did not develop similar technology. Amazon Books’ aesthetic revolves around the importance of usergenerated data and replicating the technology of online retail in a physical space. For example, Prime customers received instore discounts, while those without the premium service pay full list price for instore purchases. 83 Curated displays drew on usergenerated data, with shelves dedicated to esoteric categories including “4.8 Stars and Above” and “MostWishedFor Books.” Rather than work with traditional product categories, Amazon wished to experiment with new models of discoverability, drawing attention to its data structures rather than broader trends within publishing. Amazon’s entry into physical grocery sales with the launch of Amazon Go in December 2016 and acquisition of Whole Foods in June 2017 marked a broader commitment to brickandmortar retail. The Whole Foods acquisition provided Amazon with a large and loyal grocery Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 2 The Amazon Infrastructure [45] customer base, as well as the preexisting infrastructure to intervene in grocery home delivery. Whole Foods’ distribution network would provide the foundation for Amazon to enter the notoriously difficult grocery market and pivot away from strengths in nonperishable goods. Amazon Go’s launch demonstrated the extent of Amazon’s ambitions in physical retail through a provocation about the future of grocery shopping with the emergence of the internet of things, or a new wave of technology with native network functionality. Amazon Go is marketed as “a new kind of store with no checkout required.” 84 The generic title belies the fact that the company’s first checkoutfree store focused on food rather than books. RFID (radiofrequency identification) technology is pervasive in libraries, which lessens the perception of checkoutless book shopping as “magic.” Amazon Go’s food retail operation mixes AWS, internet of things, and knowledge of users’ purchase history to create a digital experience in a physical retail space. A New York Times report on the first location’s launch noted the abundance of cameras present on the store’s ceiling, indicating the importance of facial and product recognition over a more convoluted internetofthings solution. 85 While Amazon has expanded far beyond its original bookstore, the company continues to explore the role of digital infrastructure in retail as both a service for other businesses and a method for data collection. The launch of AWS in 2006 marked a consolidation of Amazon’s philosophy for virtualizing infrastructure even if doing so undermines the company’s previously profitable ventures. The approach allowed Amazon to shift from being a retailer to a retail platform, letting third parties use its extensive warehouse and delivery infrastructure. The launch of the Kindle saw the equivalent shift in Amazon’s approach to books. Before 2007, Amazon was a major online retailer through its reputation for books, but this model was unsustainable. Since the Kindle had no major competition in the mid2000s, it offered the opportunity to create the infrastructure for digital book distribution, mitigating any potential drop in physical books sales from Amazon. After an initial push with hardware, a move toward emphasizing infrastructure has made the Kindle vital to the future of screen reading, reflecting its position in digital and offline cultural more broadly. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 3Modeling the Book Trade Within Amazon’s warehouses, there is no distinction between a cloth copy of Jami Curl’s Candy Is Magic, a variety box of Skittles and Starbursts, or a Peppa Pig plush doll. All three items are stored, organized, and shipped in the same fashion. 1 Despite this ambivalence, books remain central to Amazon’s retail operations. Books, and specifically ebooks, accumulate and return a greater volume of data than other products. Amazon uses both distribution and consumption to monitor the ebb and flow of the book trade to predict future trends, working with wellestablished industry standards including ISBN, Online Information Exchange (ONIX), and machinereadable cataloging (MARC) to create a conceptual model of the book and the publishing industry. The Kindle built on the core pillars of Amazon Web Services’ virtualization tools detailed in the previous chapter and the company’s commitment to model the book based on the wealth of internal and external data it had accumulated. Warehouses of Metadata Bezos faced an uphill battle to compete with Barnes & Noble but used the publishing industry’s longstanding collaboration to develop ISBNs to Amazon’s advantage. F. Gordon Foster, a professor at the London School of Economics, developed the British Standard Book Number (SBN) in 1966 to manage the needs of WH Smith at the request of the British Publishing Association. The standard was quickly adopted by a range of booksellers and publishers for cataloging and warehousing, with the International Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [48] Organization for Standardization (ISO) ratifying the standard as International Standard Book Numbers (ISBNs) in 1970. 2 Publishers now use ISBNs as the backbone of spreading metadata to distributors, retailers, and libraries. 3 The metadata standard provides information about a book’s publisher and region in a humanand machinereadable format. 4 This data structure also distinguishes between different editions of the same book. For example, the 1991 Picador paperback copy of Alastair Gray’s Lanark has an ISBN of 0330319655. The first number indicates the publication region, with 0 representing the Anglophone market. The number 330 refers to Picador as the publisher, with the following five digits noting the specific book in Picador’s list. Each ISBN concludes with a check digit to ensure the number was correctly entered through “an elegant and rather ingenious system since it guards not only against inaccurately recorded digits but also against the apparently more common error of transpositions.” 5 ISBNs can identify specific editions of books with complex bibliographic histories. In the case of Gray’s Lanark, my search on Amazon.co.uk in July 2015 returned two Canongate Classics editions published in 2002 and 2007, and the original hardback from 1982 on sale for £87.47. Conversely, Amazon.com records the two Canongate titles, the original, an American edition published by Harvest Books, and the Picador edition. Amazon’s search function privileges local and new editions, but as it integrates ISBNs into product page URLs, it is possible to bypass this navigation system to locate specific editions. Book collectors can use the system to ensure they receive the expected version of a book rather than purchasing the latest reprint. The World Wide Web of 1994 at Amazon’s launch was vastly different from the contemporary web, where all information is assumed to be online. Data about the book trade appeared in sources such as Bowker’s Books in Print, which was available to retailers in print or on CDROM and contained a catalog of ISBNs for books still available to purchase from publishers and wholesalers. Retailers used Books in Print to supplement their instore holdings and as a data source for preordering books. Customers could request titles from the database through the retailer, but the information was not publicly available without purchasing access. Information about forthcoming books was only available through trade sources, the popular press, or bookshops. Amazon challenged this informational monopoly by building its online catalog from the Books in Print data set to demonstrate the large catalog of titles they could order from wholesalers rather than what was currently in stock. Amazon used Books in Print to inflate the number of titles available on its website to make its selection appear larger than that of other retailers. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 3 Modeling the Book Trade [49] The boast that Amazon was the “world’s largest bookshop” drew a lawsuit from Barnes & Noble in 1997, which was settled later that year when the companies agreed “they would rather compete in the marketplace than in court.” 6 Amazon’s sleight of hand opened an industry data source for public consumption to allow users with niche interests to discover relevant books. Brickandmortar bookshops need to stock books they are confident will sell within a reasonable time frame to justify the shelf space at the expense of discovering more obscure titles. Ordering directly from distributors and warehouses at the point of customer order while displaying a large stock rewrote these rules. Readers who did not feel represented by the limitations of brickandmortar retailers saw the extensive database as catering to their tastes. Amazon reshaped ISBNs as webbased metadata that were discoverable rather than an identifying number for warehousing and distribution, turning an industry standard into a broader public good. Product pages turned ISBNs into searchable entities on the web, acting as unofficial URIs for the book trade. Book metadata were no longer obfuscated through standards such as ONIX or MARC. Others have built on this process, including Daniel Green, the cofounder of CamelCamelCamel, a website that tracks price fluctuations on Amazon, which launched in 2008. 7 This is vital in the world of algorithmically mediated pricing, since automated systems can rapidly increase the cost of a book beyond its actual value, as with Michael Eisen’s example of a $23,698,655.93 copy of Peter Lawrence’s The Making of a Fly. 8 External services including Google Books link to Amazon as a centralized repository of book metadata that would only exist in a more authoritative form on individual publishers’ websites. The company’s ubiquity across book trade platforms is equivalent to Facebook’s integration of “likes” across the web, establishing the company as bookish infrastructure. In April 1998, David Risher, then lead of product development for Amazon, launched an investigation into selling music, with the new product shop opening later that year. 9 The music industry had no equivalent for ISBNs, so Amazon created a broader cataloging standard. The company decided to issue every item a tendigit Amazon Standard Identification Number (ASIN), which uses the full range of alphanumeric characters to identify nonbook items. 10 Currently every nonbook ASIN (including ebooks) begins with B. This practice can catalog up to three thousand billion unique items. The development of ASIN runs counter to the widespread use of Universal Product Codes (UPC), a barcode standard featuring thirteen or eighteen numbers to identify items from various countries. 11 UPCs integrated a superset of ISBN called ISBN13, formally introduced Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [50] in January 2007, although the format had a longer history starting with the inclusion of barcodes on books in the 1980s. 12 ISBN13 contains no additional metadata, as the first three digits, usually 978, refer to the country, which in the case of books is represented by the fictitious “Bookland.” 13 Preexisting ISBNs could be used with the new prefix and a recalculated check digit. Bookland was also given the prefix 979, doubling the possible number of ISBNs. ISBNs prefixed with 979 are not used in Anglophone markets as of 2018 but have already been adapted by the French ISBN authority, L’AFNIL (Agence Francophone pour la Numérotation Internationale du Livre). 14 French books with ISBN13s starting with 979 receive an ASIN to identify the product page, and the ISBN is only visible in the “Détails sur le produit.” 15 While ISBNs and ASINs can identify unique editions, Amazon uses a higherlevel metadata structure to connect separate editions to reuse valuable data internally. Christopher Weight, a member of Kindle Special Ops, led research on “title sets,” which connect similar books based on a probabilistic model of both metadata and textual similarities. 16 Strategic reuse of data also links relevant metadata across regions regardless of the local publisher. For example, reviews are shared across editions to boost the number of voices commenting on a product, although this creates problems when criticism of a poorly edited version is carried over to a more carefully produced edition. The Englishlanguage edition of Emily St. John Mandel’s Station Eleven is available to purchase from nine Amazon regions, with shared data despite the novel’s publisher varying between the United States (Vintage / Penguin Random House), Canada (Harper Perennial / HarperCollins), and United Kingdom (Picador/Macmillan). Most often, blurbs and professional reviews will be linked, but if a lower threshold for reviews has not been met, reviews from a more popular edition will be pulled through. It is common practice in book retail to use data sources such as Nielsen BookData or Bowker’s Books in Print to populate online catalogs. Data will appear consistently across Waterstones, AbeBooks, and Bokus, but Amazon expanded this approach by extracting subsets of its data for different markets and formats. Although ISBN was an early international identification standard, the book trade struggled to agree on other metadata standards. Bad practice is still common within the trade, encouraging Karina Luke, executive director of Book Industry Communications (the British publishing supply chain organization), to publish a statement condemning superfluous title metadata such as the introduction of “Man Booker Prize winner” or “The explosive next book from...” in the subtitle of product pages for Amazon and other online retailers. 17 The disparities between identifying books Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 3 Modeling the Book Trade [51] and authors have remained an open problem, as author metadata can be inconsistent and frequently requires disambiguation for common names such as John Smith or generates multiple records attached to transliterated names such as Fyodor Dostoyevsky. Projects such as Open Researcher and Contributor ID (ORCID) attempt to create consistent standards for authors in academia but rely on contributors signing up, limiting its appeal as a universal standard. 18 Amazon extended the ASIN standard to include persistent identifiers for both authors and customers, flattening the hierarchical relationship between the objects. For example, George R. R. Martin has an author ASIN of B000APIGH4, whereas the Kindle edition of his novel A Storm of Swords is the product number B004PIJEWD. Amazon integrated authors, bibliographic objects, and consumers into a single conceptual model that favors relationships over hierarchies of production. Amazon prioritized connectivity over authority when constructing its computational model of the book trade. 19 When everything is treated as a relational entity, it can be difficult to find meaning. For example, Paul Ford struggled to find patterns in the complexity of Amazon’s reviews and product information, since he did not have access to the full relational data underpinning Amazon’s data collection and analysis processes. 20 Since Amazon alone has access to the full data structure, it can offer public access to individual elements without jeopardizing its complex data surveillance system. ASINs are flawed due to the lack of external oversight. The system privileges books as a central part of Amazon’s infrastructure, but third part ies might not use due diligence to correctly enter metadata. If third parties create an incomplete or faulty listing for a print book, additional records are tied to new ASINs. For example, a search for Ian Bogost, Simon Ferrari, and Bobby Schweizer’s Newsgames in July 2017 produced nine separate product pages for the title despite the need for only two official product pages relating to the cloth and paperback editions. 21 The remaining seven product pages feature incorrect metadata about authors, titles, and publication date, with some listing the seventeenth century, and the addition of the publisher to the book’s title. These product pages are created by automated vendors, which set prices according to other offerings. Most of these prices exceed the cost of purchasing directly as a consumer from the MIT Press, which suggests the merchants do not currently stock the title but will purchase at point of order. The introduction of ASIN as an allencompassing structure requires tight control on Amazon’s part to avoid such duplicitous efforts, but the evidence from faulty Newsgames metadata is replicated with other titles. The company does not collaborate with librarians or other information professionals who also Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [58] the book trade itself. The company’s dominance ensures some collapse between the two categories. Timothy Graham separates recommendations between “popular” in views and “bestselling” with the example of Veronica Roth’s Allegiant, which averaged three stars from 8,241 reviews while sitting at the fourth bestselling Kindle spot. 40 The recommendations network in table 3.1 reflects how popularity can be gamed with the goal of improving sales, although this tactic favors shortterm gain. Amazon uses its “p13n” (personalization) data structure to generate information on elements such as titles “frequently bought together.” These recommendations often come with restrictions: my account will only show items available in the United Kingdom and will prioritize products available directly through Amazon (tagged as A1F83G8C2ARO7P). Further conditions influence the display of recommendations: “frequently bought together” will only display a title if “has_seller_difference” and “has_ diff[erent]_avail[ability]” are both false, restricting recommendations to books in Amazon’s inventory rather than third parties. The focus on usergenerated reviews demonstrates the company’s commitment to producing Table 3.1 Frequently recommended titles ASIN Title Inbound links B005F9ZLD2 Mark Shepard, Simple Sourdough (Simple Productions, 2014) 6,831 B0057XK230 Jason Edwards, Will Allen and the Great Monster Detective (Rogue Bear Press, 2013) 4,212 B00CHTEMUG Aaron Shepard, The Legend of Lightning Larry (Skyhook Press, 2013) [no longer available] 3,940 B00CYPKEN2 Aaron Shepard, Pictures on Kindle (Shepard Publications, 2014) 3,672 B005FG163Y Aaron Shepard, From Word to Kindle (Shepard Publications, 2014) 3,220 B00BMHUDP2 J. S. Scott, The Billionaire’s Obsession (Kindle Direct Publishing, 2013) 3,122 B00APM2K5Q Earl Nightingale and Robert C. Worstell, How to Completely Change Your Life in 30 Seconds (Midwest Journal Press, 2012) 2,893 B00B56PP26 Steve Scott, How to Write Great Blog Posts That Engage Better Readers (Kindle Direct Publishing, 2014) 2,745 B00BSG4LXW Steve Scott, 61 Ways to Sell More Nonfiction Kindle Books (Kindle Direct Publishing, 2013) 2,651 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 3 Modeling the Book Trade [59] large volumes of data over traditionally authoritative sources. Bezos hired James Marcus and other reviewers to contribute expert opinions and suggested that the general philosophy of hiring reviewers and editors was “to seem smart and authoritative— to become not just a store but a destination.” 41 The editors also curated the home page to highlight titles in a similar manner to the tables at the front door of a brickandmortar bookshop. This editorial team was later dismissed as Amazon began to focus instead on usergenerated content. Bezos wanted reviews to reflect the honest opinions of customers rather than appear to be filtered, so he allowed both positive and negative reviews on the site. The current Amazon home page balances these two extremes by mixing personalized recommendations with generic advertisements. Editorial content still appears at the fringe of the front page through blogging insights such as Amazon Chart’s Week in Books. Amazon’s incursion into brickandmortar shops accelerated the trend through curating categories according to usergenerated data, including “Most Wished For Cookbooks” and “Highly Rated: 4.8 Stars & Above.” Unfortunately, Amazon’s popularity and the high stakes of getting good reviews on its websites led to an influx of fake or paid reviews, which Amazon is constantly battling to remove. 42 Amazon created mechanisms for user feedback to facilitate a greater curation of reviews. These include rating reviews if they are useful— which then affects a review’s visibility— and the ability to respond to reviews. This has become a potent battleground in recent years, with readers using the reviews to attack authors’ ideologies. For example, the release of Hillary Clinton’s What Happened garnered a flood of 1,500 reviews within twentyfour hours of release, with a high proportion of onestar reviews coming from unverified purchases, which suggested the users were reviewing the book out of spite rather than having read the content. 43 Daniel Allington argues that this plurality of opinions reflects “popular” rather than “literary” taste in titles like The Inheritance of Loss. 44 Users talked about the aesthetics of an edition or other tangents rather than focusing on the textual content. Occasionally the focus on the materiality of an edition, mixed with Amazon’s enthusiasm for reusing content, can have an adverse effect on a book’s product page, as criticisms of a poorly bound hardcover can translate to an additional poor score on the Kindle edition. Conversely, a poorly scanned text can have a negative impact on a luxury highproduction print copy. Despite these limitations, Ann Steiner argues that the Amazon review systems “enhance [users’] feeling of belonging to a global community of readers with similar daily problems and desires.” 45 This community spirit is not always apparent in review data. For instance, Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [60] only 9 of the 218 reviews of Mark Levinson’s The Box received more than ten recommendations. Downvote data are not publicly displayed but are evident from the position of the final two reviews, which are displayed outside the usual set of “verified purchases.” User engagement is low, as only twenty comments were left for nine reviews. The most popular reviews were 745 percent longer than the mean review length and were posted early in the book’s life cycle, authored by prolific reviewers in the “top 1,000” or otherwise mocked for stating “I though it would be about using Cargo Containers as homes but it was not I waisted money on this book.” The prioritization of reviews by verified purchases and “Vine voices” outweighs user ranking of reviews, and the algorithm also promotes longer reviews. It is unclear whether the recommendation algorithm drives low user engagement or if the community issues emerge from knowledge that unless one is part of the several elite clubs, a review is unlikely to be promoted. Amazon reviews do not encourage extended social interactions but archive changes in taste postpublication, as early reviews are supplanted by more critical reappraisals of material later. Individuals more interested in substantial engagement with peers use websites such as LibraryThing or Goodreads that encourage more extended interactions rather than the superficial callandresponse system of Amazon’s review service. The review data therefore reveal the biases of Amazon’s data collection and processing more than contemporary literary tastes. Although Amazon is the largest book retailer and clearly drives the culture, it still is insular, and much of the publicfacing infrastructure reflects the company’s practices rather than broader issues within publishing. The book trade exhibits an underlying distrust of the datadriven analysis of publishing trends emphasized in accounts such as Jodie Archer and Matthew Jockers’s The Bestseller Code in doubling down on the concept of taste as a “gut instinct.” 46 This gulf leads to exaggerated claims on both sides and conversations that are not productive for both parties, and as a result, the book trade is not geared for tackling the challenges presented by large technology companies. As Amazon has taken control of much of the bookish infrastructure and forced companies to adopt certain workflows, these new methods have created a degree of distrust. Mark Davis argues, “ebooks can be understood as an important strategic site in a wider corporate struggle over the digital commons, and for information ownership and control, taking place among large digital corporations, and as a corporate tool in that battle.” 47 Amazon has managed to increase the volume of data available for internal analysis while closing off the information from publishers, resulting in a divergent marketplace where ebooks designed for the Kindle Store succeed above those that have been developed for print distribution. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 4Format File formats are an essential part of digital media consumption. A software package requires instructions for how to present and interpret the data compiled in a media file. Despite their significance, formats often receive less attention than other aspects of media and platforms. Jonathan Sterne argues that format is too often conflated with medium, which only reveals itself during moments of competition: Betamax/VHS, Bluray / HD DVD, MiniDisc/MP3. Format encapsulates “a whole range of decisions that affect the look, feel, experience, and workings of a medium. It also names a set of rules according to which a technology can operate.” 1 Within the confines of contemporary computing, the file extension is the most prominent signifier of format and demonstrates the fluidity between software and format. A word processor document can have multiple extensions (.doc, .docx, .odt), which in turn can be opened by various software packages (Microsoft Word, OpenOffice, Google Docs). Despite this fluidity, each of the formats has different material limits and structures. Ebooks require an understanding of digital format theory and the substantial history and conventions of print bibliography, as a clear continuity exists between the two. In the study of the transmission of printed texts, the format refers to a work’s wrapping, which signifies a book’s role. For example, clothbound academic publications are a marker of prestige designed primarily for libraries, while the subsequent paperback reflects the economic or cultural success of the initial publication. Formats are ingrained in book history: from Aldus Manutius’s development Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [62] of a Renaissance “paperback” to the distinctions between folio, quarto, octavo, and other bindings. 2 These formats referred to the binding and were an important indicator of a title’s prestige: a cheap edition featured more sheets of paper stitched together in a single gathering, while the most important titles comprised many single folded sheets of paper stitched together into a large volume called a folio. The digitization of the book has reshaped our understanding of format. Matthew Kirschenbaum argues that, rather than existing as a single finished object, “a book is an assemblage of digital assets, consisting, in practice, of multiple files and formats collected in a digital asset management system, or DAM. These files contain artwork, fonts, style sheets, metadata, and of course the text. A book is thus also a network, since these digital assets must be orchestrated to interact with one another in structured and predictable ways in order to generate desired outputs, such as an EPUB or an XHTML file.” 3 Even socalled plain text, displayed without any formatting attributes through a command line or editor interface, is mediated through format by character maps and font families. Character maps translate hexadecimal numbers into the letter or symbol to display. The “map” in the name is literal: specifications include a grid with coordinates to locate specific characters. For example, the ASCII character map represents a capital A with the hexadecimal number 41 (column 4, row 1), and its lowercase equivalent is represented by 61. ASCII does not dictate how a character should be displayed, which is handled by font rendering systems such as TrueType and PostScript that take the number and convert it into a recognizable shape. Users can then choose a font family to display their text according to individual preferences. Text display therefore requires several processes before a single character is rendered as a series of pixels on a screen (table 4.1). 4 Table 4.1 Layers of text display Layer Format Output Style sheet h1 {color: gray;} a Markup <h1> a Bitmap a a Font renderer Times New Roman [instructions for drawing a lowercase a in Times New Roman] Hexadecimal character map 0 x 61 (column 6, row 1) a (lowercase) Binary 110001 110001 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 4 Format [63] Markup languages such as HTML enable further presentational differences. Software packages interpret extra text surrounding content to be displayed to present headings, bold text, and other visual differences. For example, in HTML, <h1> is the tag for displaying a heading with the highest priority (designated by the number, which notes a hierarchy down to <h6>, the smallest subheading). CSS determines how headings are displayed through additional markup: “h1 {color: gray;}.” In this instance, the heading will be in gray, but further markup may include font size and alignment. Markup only offers recommendations, however, since readers may choose to increase the font size or change the color to suit their own preferences. While HTML and CSS are powerful tools for tinkering with presentation, the additional text can greatly increase the size of an uncompressed ebook. Developing the Kindle Format In Memory Practices in the Sciences, Geoffrey Bowker compares the Earth to a vast geological archive; similarly, the Kindle is a palimpsest that reveals earlier iterations if you just scratch below the surface. 5 This approach to formats ensures consistency but also makes it difficult to change things. As Daniel Pargman and Jacob Palme argue, “The most common solution is to patch things and hobble along. Any deviation from that solution requires potentially large alterations of the existing information infrastructure, and such alterations are expensive.” 6 As a result, examining the origins of the Kindle file format is essential to understanding the platform’s current limitations. Amazon designed the Kindle during a period of relative stability within the ebook industry. In the early 2000s, an industry group initially called Open Ebook, spearheaded by Dick Brass of Microsoft, had agreed on a set of standards. 7 Over forty publishers, booksellers, and technology companies collaborated on the Open eBook Publication Structure (OEBPS), the precursor to EPUB, an interchange specification designed to allow publishers to easily convert files to various proprietary formats. 8 No reading system interpreted OEBPS directly, so technology companies could create proprietary formats derived from the specification. For example, Sony developed BroadBand eBook (BBEB) for its firstgeneration Librie in 2004, before committing to EPUB with the launch of the PRS505 hardware in 2007. Likewise, Amazon’s acquisition of Mobipocket, discussed in chapter 1, included the popular MOBI format, another derivative of OEBPS. The reliance on Mobipocket technology two decades later brings significant archaeological baggage to the platform. Amazon’s commitment Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [64] to backward compatibility (ensuring new titles are still readable on older devices) while ensuring new devices have enough flagship improvements and ensuring broad crosshardware support places a substantial strain on format specifications. Amazon’s cumulative approach to format created a series of geological layers visible within each ebook. Each successive upgrade to the file format moved further away from the original design but still accounted for its genesis as Palm Resource Code (PRC), a database language designed for Palm OS in 1996. Amazon’s commitment to backward compatibility ties the company to this lineage to avoid obsolescence update cycles that have defined Apple’s business model during the 2010s. Apple’s yearly upgrade cycle primes customers for depreciated apps with major operating system updates, but the realities for textoriented software are different. Without flagship hardware improvements or significant security vulnerabilities, it is more difficult to justify breaking access to content for users who continue to use older devices. Since static text appears to be simple to render, any changes will be met with greater resistance. The first Palm hardware was released in 1996, the same year the XML 1.0 specification was published. Rather than adapting an emerging open standard, Palm built PRC and Palm Database (PDB) for text reading in the popular C programming language. These formats have little formal published documentation but provide the foundations for the contemporary Kindle ebook. The technical limitations of Palm OS determined how PRC files were processed on derivative devices. 9 The Palm Pilot 1000, one of the first Palm devices, released in 1996, handled an upper limit of twelve megabytes of randomaccess memory (RAM) and four megabytes of readonly memory (ROM). As a result, all documents were precompiled before transfer to Palm devices to reduce storage and memory costs. This asymmetrical structure was attractive to publishers hoping to protect intellectual property, and Kindle formats never deviated from this model. Until the development of KF8, all new Kindle file formats were renamed MOBI files, and devices would still process PRC ebooks. Amazon’s file formats are based on archaic standards from 1996 rather than EPUB3, which derives from modern web specifications such as HTML5 and CSS3. Several of the Kindle’s quirks emerge from this historical divergence and Amazon’s later attempts to reconcile the expectations of users for contemporary ebook design. Newer file formats integrate features from EPUB3, but the geological layers of the original format remain. PDA users who wanted to read ebooks in the 1990s had different needs from ebook readers in 2007. PDAs featured small, lowresolution screens, so reflowable text, a central part of “ebookness,” was more userfriendly than attempting to create print facsimiles. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 4 Format [65] MOBI in turn was built on Memoware, an amateur ebook production community launched in 1996 after the release of the first Palm Pilot. Once established, the community changed focus to create an early ebook store. 10 Memoware’s name links to Palm’s “Memo” software, as users wished to create ebooks for their Palm Pilots before reading systems were available for Palm OS. Early adopters focused on creating computer manuals for Memoware, which influenced the design of the MOBI format specification, which includes basic linking and interactive features required for the genre. The initial transition from MOBI to AZW was straightforward. For security reasons, Amazon’s development team removed the ability to use JavaScript, extended the DRM protection, and changed the file format’s name from MOBI to AZW. This early rebranding allowed publishers working outside of Amazon’s service infrastructure to provide DRMfree Kindleready titles. If readers change an ebook’s extension from MOBI to AZW, reading systems will interpret the file identically. While newer formats deviate from this model, newer devices still read legacy formats. Amazon’s choice of MOBI in 2005 was pragmatic. EPUB was not established as a consumerfacing format, and all proprietary formats derived from the best practices of OEBPS, HTML, and CSS. EPUB developed traction simultaneously to the Kindle, and without the momentum from the Kindle’s highprofile launch, it is likely that EPUB would never have seen widespread adoption. Google Books focused on facsimiles; Apple would not launch iBooks until the reveal of the iPad in 2010. Other than Sony, the ebook market depended on a diffuse network of bookstores and had not gained critical mass since substantial investment in hardware earlier in the decade. Since 2007, Amazon has developed several format specifications to meet the conflicting needs of dedicated hardware and thirdparty software, summarized in table 4.2. These sociomaterial constraints result in several contested aspects of the Kindle format’s perceived pitfalls. Criticisms are often based on typographic issues that remain challenging for other ebook file formats. 11 If we move beyond these common ailments, the format ecosystem contains the geological layers that can be peeled back to reveal a broader history of displaying text, from precompiled databases to cloudassembled files. No public documentation for Kindle formats is available, but Amazon provides “best practice” advice for publishers preparing HTML or EPUB content for conversion to a Kindle ebook. 12 Instead, to understand the Kindle’s formats, I reverse engineered available source code and DRMfree publications. Publishers can use a command line tool, KindleGen, or a graphical user interface wrapper, Kindle Previewer. These tools offer an Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [66] approximation of what the final output will look like on a range of devices to avoid any errors that can occur during the complex conversion process. Selfpublished authors can use Kindle Create, which converts Microsoft Word documents into Kindle files, alleviating the requirement to understand markup languages to produce a basic ebook. 13 Amazon’s documentation shows disregard for publisher autonomy through comments such as “publishers do not need to define a start reading location because Amazon does this during the upload process.” 14 Amazon asserts control over the Kindle ecosystem by keeping the internal workings of its file formats secret and stripping unwanted HTML tags from the conversion process. This approach ensures Amazon’s dominance by creating an environment resembling Jonathan Zittrain’s metaphorical use of “walled gardens” to describe “sterile appliances tethered to a network of control.” 15 On the other hand, due to the World Wide Web Consortium’s dedication to open standards, EPUB is thoroughly documented. The industry standard forms the basis for many Kindle files at the production stage, but its cloudbased conversion process reshapes these data. Kirschenbaum warns that “programmatic computational environments [apply] some particular logic— a certain formal materiality— to the string of bits in question.” 16 The differences in the “formal materiality” of Kindle files and EPUB stem from divergent ideologies in the production and consumption of digital content. Both formats have the same building blocks Table 4.2 Summary of Kindle file formats File extension Year released Features .prc 1996 Palm Resource Code. Database files containing onscreen reading documents. .mobi 2000 An ebook format based on PRC and developed to display HTMLlike attributes. .azw 2007 The original Kindle format, based on MOBI. Speculated acronym: Amazon Word. .tpz 2008 Topaz. The primary format for Kindle files downloaded for iOS devices. .azw3/.kf8 2011 A transformation of AZW to bring it more in line with EPUB standards. Launched alongside the Fire tablet range. .azw4 2014 A wrapper for PDFs. .kfx 2015 Completely rewrote the format logic and rules. Marketing through “enhanced typography.” Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 4 Format [67] derived from open web standards including HTML, CSS, and XML. The standards depart at the point of processing the raw materials. The most important differences are compression and data storage. While EPUBs use ZIP compression, which maintains the original file structure, Kindle files are delivered in binary format, which obfuscates the structure of the original file. This shift ensures Kindle formats are more difficult to preserve. Amazon changed its datastorage practices with the introduction of KFX, which changed the underlying format from XML, the standard for EPUB, to JSON. For our purposes here, the two markup languages are nearly identical beyond syntax differences. Let’s take a name as an example: XML: <person> <firstName>John</firstName> <lastName>Smith</lastName> <person> JSON: {“person” : { “firstName” : “John”, “lastName” : “Smith”}} Beyond minor technical and syntactical differences, the choice between the two standards comes down to developer preference and what systems the content needs to interact with. Nonetheless, Amazon’s choice to move away from XML, the industry standard, introduces an additional conversion process that can increase corruptions while also reducing publisher autonomy. The shift away from XMLbased standards was part of a larger move toward a broader typology of ebooks. Amazon introduced multimedia ebooks with the launch of the Fire tablet range, leading to the development of KF8, an extension of AZW that incorporated elements of EPUB3 rather than resurrecting older aspects of the MOBI specification. While many of these new web technologies feature in the Mobipocket specification, Amazon intentionally restricted them when launching the initial Kindle to avoid the possibility of malicious JavaScript or SQL injections and to suit the restrictions of an electronic paper screen. The new format allowed for the development of new genres, including digital comics, that were not covered in previous iterations of AZW. KF8 introduced “appamznmagnify,” a JavaScript library that allows publishers to locate areas to magnify on images rather than relying on the pinchanddrag metaphors common on smartphones and tablets. These incremental improvements focused on balancing the needs of new genres with ensuring the security of the format. KF8 enabled the publication of PDFlike content without simulating the flexibility of print. For example, in Doug Dorst and J. J. Abrams’s S., Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [74] text as “K OBUN C HINO,” which affects the quality of its texttospeech and Word Runner conversion. Choosing Windows1252 by default and stripping tags without context may create a consistent presentation, but the changes to accessibility and semantic access can adversely affect readability. Amazon engineers have developed various technological solutions to what remains a social problem. For example, Anubhav Kushwaha, a software development manager at Amazon, experimented with compressing ebooks by using an index dictionary. Each unique word received an identifying number in a dictionary that is then recalled in text rather than repeatedly spelling out words (figure 4.3). 38 The method has a historical precedent in early ebooks implementations developed for the US Navy to increase the efficiency and speed of reading reference manuals. 39 Pertinent strings can also be compressed into dictionary entries, reducing commonly used markup such as <h1> for headings or navigational elements to a numerical string far more compact than the original markup. The technique was initially designed to circumvent the expense of storage memory in the 1980s, but Amazon recast the technique for obfuscation of content in the constant arms race with users who were reverse engineering the Kindle DRM to transfer ebooks to other platforms. Kushwaha’s work mirrored earlier research into characterlevel dictionary indexing. Amazon headhunted Eric Menninga, a typography veteran of nineteen years for Adobe, to work on improving the layout of Kindle ebooks and “solve the problem of authoring complex documents for display on devices with varying capabilities.” 40 Menninga worked with Lokesh Joshi, another typography expert, to expand the Kindle character map to be compatible with UTF8 while restricting access to exploits within the system. 41 Building on Kushwaha’s work, Menninga and Joshi focused on generating indexed dictionaries of characters in an ebook rather than words, allowing for a customized character map for each title. Each ebook features a compressed version of UTF8 rather than offering access to the complete character map, with a customized map that includes only the characters present in a particular title. 42 A new character map with just an index of characters present in the text further obfuscates the content and adds an extra layer of security to the DRM. Just like Amazon’s approach to converting EPUB to AZW through JSON, the character map conversion introduces an additional risk for creating unintended errors that can be difficult to troubleshoot given the obfuscation process. As figure 4.4 shows, this represents clear progress for better internationalization support, but the approach is constrained overall by Amazon’s extensive use of proprietary solutions based on available open technologies. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 4 Format [75] 4.3 The user dictionary approach to compressing ebooks 4.4 Comparison of how a Kindle 2 (left) and Kindle Touch (right) render the same MOBI file of Фёдор Достоевский (Fyodor Dostoyevsky), Преступление и наказание (Crime and Punishment). Source: Федор Михайлович Достоевский, Преступление и наказание (Russian Edition) [Kindle Touch 5.3.7.3], AZW3, B00KHRXSMI (Общественное достояние, 2014), loc. 31. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [76] The Limits of Backward Compatibility This approach ensures that a reader can access the same ebook title on a Kindle 1 or Voyage and it will be optimized for that hardware, not to mention the array of secondary platforms Amazon supports. The hardware and software will also create fluctuations in the naming conventions of a file. For example, the same title will be branded “AZW” on a Kindle 3 but appear as “PRC” on a nonAmazonbranded Android device. These minor deviations appear through the longitudinal development of platformspecific workflows that lead to different results even if the output is nearly identical. The formatting of several editions of Walter Isaacson’s Steve Jobs biography shows a consistency between file types, but also format differences in variants. The ebook is not a single file, however, as several services are sideloaded through separate files (table 4.3). Steve Jobs is a technically simple ebook with occasional images, but otherwise no complex formatting. Nonetheless, the biography is made up of many distinct files: outside of the main ebook, we find over thirty different files sideloaded with metadata (table 4.3). Many of these files are optional services available only on certain devices, but in several instances, the Kindle offers a unique interpretation of formatting choices owing to its archaic roots. The available sideloaded content can be grouped into four primary categories: 1. Bibliographic metadata about the length of the book, the author, and other aspects often found on the opening pages of a printed book 2. Usergenerated metadata pertaining to reading time and the collation of popular highlights 3. Generative paratext created through algorithms packaged with the ebook to interact with other elements of the Kindle platform, including page numbers, indexing, reading speed, and vocabulary building 4. Metrics and logs: local files compiling information about the user to be sent back to Amazon The ad hoc assembly of files serverside has clear consequences for the platform. While users have the flexibility to view the same title across a spectrum of hardware and software configurations, with a range of additional features, the structure of the text as a material object is permeable and prone to error. Each download is unique and offers a narrative of both its production and consumption through metadata traces. The platform focuses on scale and variety, which can adversely affect the quality of products that have not been optimized for the Kindle. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 4 Format [77] Table 4.3 Folder structure for various active Kindle versions of Walter Isaacson’s Steve Jobs File Description Touch Kindle 8 Android Fire Mac AZW Base ebook file in most advanced format available for each device ✓ AZW3 ✓ KFX ✓ ✓ ✓ SDR Folder containing relevant metadata ✓ ✓ AuthorProfile.ASC Popup boxes in JSON about the author, and to appear at the beginning and end of the ebook ✓ StartActions.ASC ✓ ✓ EndActions.ASC ✓ ✓ ✓ ✓ AZW3F Encrypted reading metrics data ✓ TICR ✓ ✓ YJF ✓ MBP JSON reading metrics data ✓ AZW3R Annotations / page number data ✓ APNX Page number algorithm ✓ PHL Popular highlights XML ✓ ✓ ✓ ✓ ✓ MF A log of server requests ✓ LanguageLayer.kll SQL database for Word Wise, Amazon’s vocabulary builder ✓ XRAY.db SQL database containing the index information for the XRay system ✓ ✓ XRAY.asc ✓ amazn1.drmvoucher.ast DRM information ✓ ✓ kfx.luci A series of 11 encrypted metadata files ✓ Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [78] Kindle ebooks are created on Amazon Web Services to optimize titles to work with new features while remaining backward compatible with older devices. Over the Kindle’s first decade, the company introduced new formats while maintaining support for older devices. The resulting infrastructure supports an ecosystem of formats rather than a single file format that has been developed alongside hardware and software innovations in an incremental fashion. The Kindle file format is not a stable object until a user downloads the file for a specific hardwaresoftware combination. The results are closer to the letterpress era, where print runs could be varied according to their production rather than a seamless and standardized process. Since 2012, Amazon has processed a stack of “source ebook data” including “text,” “graphics,” and “font data” through a server to produce “modified ebook data” personalized for individual users. 43 The shift from storing compiled ebooks on a server to generating personalized copies on the fly offers extra DRM protection through identifying specific copies circulating in the wild while also allowing publishers to provide a single “master copy” that is then converted to suit specific hardware configurations by stripping out or adding content where appropriate. The complexities of the Kindle’s formatting process are difficult to reconcile within the usual confines of format, as the Kindle exhibits a greater fluidity emblematic of Amazon’s algorithmic control of the process. No archetypal Kindle file format exists, and even though newer formats derive from older specifications and remain stable based on the hardwaresoftware configuration of a particular user, the notion of a prototypical Kindle format is difficult to formalize in practice. It is closer to the complex multiserver assemblage of Google Docs than the more straightforward singlefile PDF. The combination of sideloaded paratext and vastly different file formats for the primary content, as well as the generation of data and content algorithmically through first and subsequent use, makes the Kindle file format difficult to pin down as a single conceptual unit. Jiminy Panoz describes the equivalent EPUB phenomenon as a spectrum, since reading systems are not required or able to interpret all aspects of a common format. 44 This courtesy also extends to developers to adapt the standards to best suit their platforms. The common principles that emerge across format iterations are more important than the differences, as this is what defines the contemporary ebook. Amazon is unlikely to switch to the status quo of EPUB. The decision to implement JSON on top of keeping elements of PRC allows the company to maintain control over the features available to the title and protect publishers’ intellectual property. An asymmetrical process allows the company to only include whitelisted elements rather than experience Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 4 Format [79] unwanted behavior. As a consequence, the most accurate representation of a Kindle file exists not in publishers’ archives or on readers’ devices but in the databases containing JSON to be converted to a title when requested by a user. From the perspective of creators, it is impossible to ensure that the carefully designed ebook will be preserved through the upload and download processes according to the latest, or historic, software demands. Users— and researchers— will not necessarily have an identical experience on any two devices, and finding consistency is difficult. Through Amazon’s continual expansion of the platform, and its long history embedded in older computing traditions, the format has remained one of the most complex parts of the Kindle infrastructure. Amazon’s commitment to backward compatibility while simultaneously attempting to diversify into tablets and encourage a small number of readers to upgrade on a yearly obsolescence cycle has led to a complicated ecosystem of file formats that are distinct yet similar. As the Kindle platform has grown, quality control has become more difficult to maintain, leading to differences in the display of titles between a Kindle 1 and iPhone X. Simultaneously, each new iteration of the Kindle file format has introduced new proprietary and divergent rules and logics, ensuring that Amazon maintains control over what is allowed on the platform. While this initially came at the cost of accessibility and internationalization, newer devices have struck the balance between control and access. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 5The Everything Store? Once Amazon started to build its extensive warehouse network, Bezos envisioned “Project Fargo,” a reference to the 1996 Coen brothers’ film, where the company would have at least one of every product in stock. 1 Early Lab126 mission statements echoed the sentiment for ebooks, with the desire to have a copy of every book ever published available for download. 2 The US Kindle Store boasted over 5.5 million titles in September 2017, but Amazon faced an early hurdle to offer a more modest catalog of 88,000 titles at launch. The technical barriers of digitization were insignificant compared to the social challenges involved in persuading publishers to create ebooks. Publishers were comfortable with producing books digitally, but they remained skeptical of digital distribution before the Kindle’s launch. The gamble paid off, and Amazon boasted a large catalog in November 2007 “including more than 101 of 112 current New York Times Best Sellers.” 3 The combination of best sellers and a range of older titles built a critical mass for the Kindle’s launch and encouraged consumers that publishers were serious about ebooks. Ebook publishing divides into four main segments: frontlist, backlist, public domain, and digitalonly / selfpublished works. The frontlist and backlist are mainstays of legacy publishing. Publishers release new titles, called the frontlist, aiming to sell them in high quantities and create a best seller. After the peak window for sales, thought to be as short as twelve weeks for most books, titles move to the backlist, where they remain in print if they continue to sell enough copies to justify a reprint. For publishers with a rich history, such as Penguin, Faber & Faber, and Oxford Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [82] University Press, the backlist represents a valuable revenue stream, as readers continue to consume historically significant titles still protected by copyright. As of 2020, books published before 1924 are in the public domain in the United States and can be published without prior permission or for free on the Kindle. Ebook platforms’ infrastructure bypasses the workflows and timescales of legacy publishing. Startups have taken advantage of these new freedoms, including digitalonly publishers such as Canelo and self-published authors who choose to circumvent the gatekeeping function of publishers entirely. Recent journalism prophesying the demise of ebooks focuses on printoriented publishers’ difficulties in adapting to the demands of digital publishing with new titles, and the expenses of digitizing the backlist, while ignoring the rise of digitalonly publications owing to their association with vanity presses. 4 There are a few exceptions: Mills & Boon, a romance publisher, has pivoted toward digital publications to satiate its audience’s demands. Other publishers focus on print first with ebooks remaining an afterthought, outside of a few publishers who continue to hire digitalpublishing advocates. The longterm success of ebook platforms such as the Kindle depends on a fine balance of these four segments. Kindle Direct Publishing and the promotional value of Kindle Unlimited built new audiences for borndigital content, but this increased readership exists in a symbiotic state with more traditional publishing, which drives readers to the Kindle Store before discovering new genres and experiences. Despite initial optimism, printoriented publishers now contest any challenge to the orthodoxy. For example, in 2008 Frankfurter Buchmesse participants agreed that digital sales would overtake print by 2018, and Dick Brass, vice president of technology development for Microsoft, predicted that “the last paper edition of the New York Times will appear in 2018.” 5 Nonetheless, popular Kindle titles still correlate to the whims of print. From June 2010 to November 2011, Amazon maintained the Kindle Million Club, a list of authors who had sold over a million ebooks. 6 In the eighteen months this list was active, fourteen members were inducted. Stieg Larsson was the first inductee in July 2010, followed by other fiction heavyweights such as James Patterson, Lee Child, Suzanne Collins, and Michael Connelly in the following twelve months. Two Million Club members, John Locke and Amanda Hocking, were selfpublished, offering an early indication of the changes afoot. Contrary to the perceived phenomenon of “mommy porn” in the years since the release of Fifty Shades of Grey, this list reveals a convergence toward crime fiction and thrillers, with the occasional science fiction and young adult title. Many brandname authors only entered the club with the release of a film or television adaptation of Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 5 The Everything Store? [83] their works. For example, George R. R. Martin sold over a million copies of his works after the popularity of the first season of Game of Thrones in September 2011, and Stephenie Meyer’s appearance on the list in November 2011 coincides with the penultimate Twilight Saga’s release date in cinemas. The club has several notable absences, including heavyweight authors such as Stephen King and J. K. Rowling. Both decided against publishing ebooks owing to skepticism about the format. King’s hesitance was born out of the failed attempt to serialize The Plant in 2000, although his books had continued to be available for sale as ebooks, on Amazon no less, before the Kindle’s launch. The success of the firstgeneration device persuaded King to pen a promotional story, “UR,” released with a cover adorned with his custom pink Kindle 2, to mark the launch of the device. King joined Bezos for the device’s launch in New York City’s Morgan Library in February 2009, fully endorsing the device. Nonetheless, in the two years after King’s entrance to the Kindle market, he had not sold a million copies despite having forty novels in print at the time. Rowling was another highprofile abstainer and resisted the Kindle’s siren call until 2011, when she launched Pottermore to challenge Amazon’s ebook dominance. Pottermore was initially designed as an interactive version of the seven core Harry Potter books featuring several gamelike features but evolved into the exclusive ebook distributor for the series. Rowling elected to distribute ebooks directly via Kindle Unlimited in 2015 to promote the subscription service. Readers who wished to purchase the titles were instructed to visit Pottermore, allowing Rowling to sidestep the Amazon infrastructure and dictate terms of access. Rowling’s position was unique, as the Harry Potter series had greater cultural and economic clout than other authors, and she could dictate terms to Amazon: Nielsen BookScan data show that Rowling sold 1 percent of all books sold in the United Kingdom between 2001 and 2018. Rowling and King raised the potential for an alternative ebook ecosystem, disconnected from the whims of Amazon, but as the market has consolidated, even the refuseniks have fallen into line. Without a concerted effort from major publishers and highprofile authors, it is unlikely that Amazon’s grip on the ebook market will slip. The Frontlist and Best Sellers The best seller is an enigma in publishing studies, as it is one of the only metrics embraced by an industry that otherwise shuns quantification. Best sellers are editorialized and depend on algorithms with clear human biases. For example, when the New York Times publishes its bestseller Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [90] and publishers may not have granted Amazon the additional permissions to convert the book to a Kindle edition. “Look Inside” reproduces a facsimile rather than requiring the costlier step of extracting the text via optical character recognition to create a publishable ebook. Second, the genre of many books available exclusively via “Look Inside” reveals variations in digital distribution, as large academic publishers such as Springer Nature and Elsevier used the marketing potential of “Look Inside” but elected to not sell the books via the Kindle. Scholarly publishers invested in digital workflow technologies earlier than other publishers and gain more benefits from digitizing the rest of their back catalog owing to the additional merits of printondemand business models for the scale of print runs in academic publishing. The sector set up alternate infrastructures for disseminating ebooks on platforms such as ScienceDirect. As the library subscription model is the primary revenue stream for academic publishers, they rely less on the availability of titles for the Kindle. The historical and geographical specificity of publishing also affects the availability of ebooks from 1989. While readers still consume classic works, most publications remain forgotten. The Cold War, the HIV/AIDS crisis, and futurist speculations on the 1990s were popular topics in 1989 publishing. Some of these titles may remain appealing, but the majority are unlikely to generate sufficient sales beyond curiosity to be profitable enough to digitize. The long tail of publishing uncovered by analyzing a slice of all books published in a single year also reveals a body of localhistory titles that would not have a largeenough scale of interest to prompt a digitization campaign. In a talk at Ebookcraft 2016, Teresa Elsey, then digital managing editor at Houghton Mifflin Harcourt, contextualized the lack of progress with the backlist by demonstrating the pressures of maintaining a backlist that remains perpetually “in print.” Elsey calculated her group’s workload for 2015— a total of 1,400 ebooks— of which only 20 percent were new titles. The remaining titles were books that had never received an ebook edition (10 percent) and new editions of preexisting ebooks (70 percent). 28 As more titles are introduced as ebooks, the burden of updating becomes a large proportion of the workload, leaving the older titles less likely to be digitized. Elsey attributes this constant churn to two factors: (1) the initial digitization gold rush, and (2) the shifting demands of ebook production and the anachronistic connection between ebook titles and reading systems. The digitization gold rush in the mid2000s, facilitated by Google and Amazon, emphasized quantity over quality, and consequently an abundance of OCR errors, uncorrected printbased hyphenations, and other conversion artifacts remained in the process; in today’s mature Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 5 The Everything Store? [91] ebook marketplace, these errors are noticed by retailers and readers and need to be corrected. Elsey also notes the importance of the legal issues regarding permissions for digital editions that have emerged with a more substantial effort to digitize the backlist. The shifting demands of ebook production affect all backlist titles, including those born digital. Older titles need maintenance according to updated standards, and the conventions of ebook design shift, so even working titles may require updating to reflect the market. This is exacerbated by the anachronistic connection between titles and reading systems, which show no sign of aging, unlike their print counterparts. 29 Maintaining a backlist with the pressures of updatability remains an open challenge a decade after the Kindle’s launch, decreasing the resources available for publishers to digitize more of their backlists. The Emergence of DigitalOnly Publishers The nowinfamous Fifty Shades of Grey trilogy challenged orthodox conceptions of publishing success made possible through developments with ebooks. 30 E. L. James, the pen name of Erika Mitchell, presented a mythical “ragstoriches” narrative highlighting the potency of selfpublishing. The novel started as fan fiction of the Twilight Saga posted on fanfiction.net, published in weekly installments for a rapt audience. Once James found a large audience, she published a revised version of the fan fiction as Fifty Shades of Grey, using print on demand before Vintage Books acquired the rights. James initially found success without the support of a traditional publisher, but her biography reveals that this is not a straightforward narrative. Before the publication of Fifty Shades, James worked in television, and her husband, Niall Leonard, remains a screenwriter. James had appropriate connections to media industries and understood the nuances of contemporary publishing when choosing to selfpublish the work. Given the book’s origins as fan fiction, it could not be published in its original form. James’s mediasavvy marketing through fanfiction.net built an audience that an author taking the traditional approach would struggle to replicate. The blurring of the lines between best seller and selfpublished work that Fifty Shades of Grey pioneered reveals the blueprint for entrepreneurial authors willing to start with the Kindle. The ability to read “mommy porn” inconspicuously was immaterial compared to its use of selfpublishing, as the success of Fifty Shades depended equally on the print edition and its appearance in various gendered locations, including supermarkets. For example, in Norway, 96 percent of 177,000 total sales Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [92] for the Fifty Shades trilogy were print rather than digital. 31 Fifty Shades sold over eleven million paperback copies in the United Kingdom in its first five years, according to Nielsen BookScan, further solidifying the series’ bestseller status. The third title, Fifty Shades Freed, was the weakest seller but still sold twice as many copies as the fourth most purchased book from 2012 to 2017, Harry Potter and the Cursed Child. Fifty Shades was a phenomenon across formats that revealed the scale of the ebook market during its ascent. Regardless of the authenticity of James’s underdog narrative, Fifty Shades demonstrated the marketing potential of fan fiction forums and selfpublished books to build loyalty before embarking on a more traditional publishing model. The rise of selfpublished authors reverses twentiethcentury trends but fits into more historical trajectories of publishing. The publisher has not always been separated from the author; for example, Samuel Richardson was the author, publisher, and printer of Clarissa in the 1740s. It was only with the professionalization of the printing press and publishers that Richardson’s doityourself mentality became abnormal. Selfpublishing was tarnished as vanity publishing, since if a traditional publisher deemed a book unworthy, it was unlikely to be worth the money. Vanity publishing required authors to subsidize the cost of production and place advertisements at the end of the book to ensure a financial return for the printers. 32 Predatory vanity publishers, particularly in academic publishing, generated a stigma around selfpublishing that presented a hurdle for the legitimacy of services such as Kindle Direct Publishing. Kindle 1 promotional material heralded its extensive range of New York Times Best Sellers and backlist titles, but the launch of Digital Text Platform (the original name of Kindle Direct Publishing) marked a more substantial transformation. Amazon did not want to reveal its selfpublishing ebook platform before the launch of the hardware, but the Digital Text Platform page launched on the same day as the Kindle and encouraged users to join up with Amazon. The splash page was accompanied by an FAQ that sold Amazon’s ability to act as an intermediary in place of the publisher because “the Kindle Store provides the business, marketing, and vendor support you need to maximize your sales. Titles published on the Kindle Store are eligible for automated merchandising and personalization, Search Inside, Amazon.com’s excellent Customer Service and order fulfillment services, and much more.” 33 The FAQ, though not publicly released until the November launch date, was written in May 2007. KDP represented a strong backup plan for Amazon if print publishers were unwilling to publish digital publications. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 5 The Everything Store? [93] Despite the term “direct publishing,” authors using Kindle Direct Publishing benefited from association with the Amazon brand. Rather than releasing a MOBI file to a web page and hoping for the best, selfpublished titles were boasted by Amazon’s fullservice infrastructure and discoverability platform. While traditionally published and selfpublished works largely remained separate, authors could leverage the benefits of Amazon as de facto publisher in the process. Kindle Direct Publishing was not just another place to sell a few ebooks; it offered a new way for authors to make a living without relying on traditional publishing structures. The company’s role as publisher is clear from how it shapes pricing: while authors can charge whatever they like, in 2015, authors only enjoyed a 70 percent royalty if they priced their title between $3 and $10. 34 Incentivizing lower prices encouraged the production of material deemed viable at that cost. KDP is not selfsufficient, however, as it relies on the prestige of legacy publishing to ensure that readers use the platform. The size of the market attracts users and authors to the platform, but without the prestige of a publisher’s imprint, the Kindle platform would compete with startups like Wattpad and Archive of Our Own that offer similar services for free. These two divergent markets have been the subject of intense trade scrutiny, as official figures from the Publishers Association noted a 17 percent drop in sales of consumer ebooks in 2016, resulting in joyous articles about the downfall of ebooks. 35 The figures exclude two large segments of the market: digitalonly publications and the entirety of Kindle Direct Publishing. Early success stories shaped Kindle Direct Publishing’s impact. John Locke capitalized on his fame as the first selfpublished author to sell over a million books. Locke was fortunate to be the first author to hold the top two slots in mostpurchased Kindle titles. 36 His book covers reference the large volume of titles that have been sold, and a sense of bravado permeates his author brand. Locke signed a distribution deal with Simon & Schuster in mid2011, allowing him to continue selfpublishing new works. 37 This unusual deal marked a transition point where selfpublishing was now seen as a viable career option, shedding its historical association with vanity publishing. Locke was later implicated in paying reviewers to secure a fivestar rating for his titles and furthering the tactic by encouraging reviewers to purchase an Amazonverified copy of the book to both increase his sales ranking and game Amazon’s review algorithm. 38 The investigation raised important questions about the ethics and sustainability of selfpublished best sellers, which would often have to be priced at less than a dollar to be considered, compared to the higher price points Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [94] expected from printoriented publishers. After this negative publicity, and after the initial lucrative flurry of sales diminished, Locke stopped publishing new material in 2012. The Locke model for selfpublishing relied on the early Kindle marketplace, and as it has matured, so has the role of selfpublished and digitalonly publications. The arrival of Kindle Unlimited, and more recently Prime Reading, is bolstered by the presence of these types of publication. Select popular legacy publications like the Hunger Games trilogy and Harry Potter series pepper a catalog of over a million titles. The tactic worked for Suzanne Collins, as the Hunger Games books were revealed as three of the top five most purchased fiction titles in the Kindle’s first decade, alongside the first two Fifty Shades titles. 39 Collins outperforms Rowling because the latter entered the platform later and sold books via a third party. Licensing conditions for Kindle Unlimited require publishers to publish exclusively via the Kindle platform, optimizing the scheme for Kindleexclusive titles available via KDP while remaining a barrier for traditional publishers. 40 The subscription service offered a revenue kitty to be shared according to volume of pages read. In August 2017, the KDP Select Global Fund was $19.4 million, although the payout per page was 0.4 cents, indicating that subscribers viewed more than 4.6 billion “pages” per month. 41 In this competitive attention economy, authors moved from Locke’s scheme of paying for reviews to using underhanded tactics to encourage readers to skip to the end of the book or otherwise increase the visibility of a title. Locke’s early exploitation of Kindle Direct Publishing’s fragility has been usurped by more flagrant abuses in recent years, including accusations of money laundering through selling extremely expensive fake books using authenticated authors’ credentials. 42 David Gaughran, an author and advocate for selfpublishing, has documented a litany of algorithmic gaming practices and errant punishments to authors who follow the rules to crack down on unscrupulous behavior. The Kindle Unlimited royalty pot has led to a common fake book scam, where content is stolen and rendered in large 2,500page volumes, offered for free, and then added to Kindle Unlimited once its visibility is high to capture a larger proportion of the KDP royalties that month. 43 Authors and publishers of original content have also sunk to the same lows of click farming, or paying for third parties to download, “read,” and review a book to increase its visibility. 44 Amazon’s response has been to punish these unusual behaviors, but with the unintended side effect of harming authors who use thirdparty services such as BookBub to advertise their books. 45 The benefits of selfpublishing for groups underrepresented by printoriented publishers have been overshadowed by automatically generated Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 5 The Everything Store? [95] content that floods the Kindle Store. 46 Despite various schemes to undermine the credibility of selfpublishing via Amazon, several authors have set an example of the new careers available in the digital age. Amanda Hocking, the second selfpublished author on the Kindle Millions Club, was one of the forerunners of the second wave of selfpublishers. Hocking released her first book, My Blood Approves, in March 2010, and by the time she had sold a million books, she had published two series: My Blood Approves and The Hollows. The second book in The Hollows series was published a day before her induction into the Kindle Million Club, indicating a surge of sales related to this title. St. Martin’s Press signed Hocking in 2011, but before that point, she was selling an average of nine thousand books a day. 47 Hocking’s success demonstrates that selfpublishing ebooks could lead to sustained print sales, as her most recent book series, The Kanin Chronicles, was published in 2015 by St. Martin’s Press, four years after the start of the relationship. Selfpublishing platforms such as Kindle Direct Publishing allowed authors such as E. L. James and Amanda Hocking to launch their careers, but both authors chose a more traditional route once they had established track records. These success stories have encouraged others to selfpublish, with many authors electing to stick with selfpublishing because of the autonomy and royalties unmatched by legacy publishers. The rise of digitalonly publishers like Canelo, a team who drew on experience from print publishers including HarperCollins and Bloomsbury, shows a sustainable approach to the development of separate print and digital reading audiences. The ebook marketplace has matured around the expectations of value from a consumer perspective. While readers may spend up to $25 or £20 on a newly released hardback book, the equivalent ebook’s value proposition is far trickier. The perception that ebooks should be cheaper has led to more agile publishing models for digitalonly titles, where publishers and selfpublished authors use newly created spaces within the ebook market. For example, ebooks can be much shorter than print equivalents, which makes it viable to charge as little as a dollar for a single book. Canelo’s mission statement lists what it considers to be viable digital markets: crime, commercial women’s fiction, action fiction, and “what we call ‘geeky crossover,’ such as horror, or historical fantasy.” 48 The company focuses on genres it believes will thrive at a low and affordable price. Canelo expanded its offerings by opening Abandoned Bookshop as an imprint in 2016, with a mission statement to “uncover the best books we’ve forgotten about, lost sight of, or never even knew existed in the first place.” 49 Focusing on recovering forgotten titles— orphaned works, titles out of copyright, or simply books that were not paid attention by their Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [96] original publisher— approaches the backlist as an opportunity rather than a risk. Abandoned Bookshop leverages the credentials of publishers as curators to differentiate the venture from companies that have loaded portions of their backlist to the Kindle Store without a clear focus, and the mass of selfpublished work that defines digitalonly publishing today. Amazon’s role in the distinction between selfpublishing and digital only is complex. Since Amazon provides many services associated with the publishing process, can Kindle Direct Publishing be called selfpublishing rather than traditional publishing, albeit with little editorial intervention? Amazon published 1.6 million titles through Kindle Direct Publishing in 2016, which would make the company the largest publishing company in the world. 50 It has run various schemes that fit into legacy publishing models, but with a digital twist. In 1999, “the company quietly leased the rights to a defunct imprint called Weathervane, and repackaged... creatures from the black lagoon of the remainder table.” 51 The project was shortlived and erased from Amazon’s institutional memory. The company returned to the idea while gearing up for the launch of the Kindle with the 2005 launch of Amazon Shorts, later retitled Kindle Singles. The program offered authors the ability to publish “shortform work (2,000– 10,000 words, fiction or nonfiction)” that would be “available to customers via PDF, HTML, plaintext email, and they are stored in customers’ Digital Lockers forever.” 52 The unfortunate mention of permanence in the marketing of an obsolete product demonstrates the precarity of the Kindle within Amazon’s ecosystem. The company invested in the scheme in the signature Amazon style of commissioning exclusive content for the launch of the Kindle 2: Danielle Steel penned an exclusive autobiographical essay, “Candy for the Soul,” for the launch of Amazon Shorts. In a 2012 editorial for the New York Times, Dwight Garner praised the series as “probably the best reason to buy an ereader in the first place” as the home for “works of longform journalism that seek out that sweet spot between magazine articles and hardcover books.” 53 Enthusiasm for the format led to 1,248 publications in the series, but production slowed in recent years from a peak of 242 titles in 2015 to just 109 in 2017. Since the Kindle’s launch, Amazon launched a further fourteen imprints, ranging from Waterfall Press, a specialist Christian publisher, to Amazon Crossing, which specializes in translations. The imprints that focus on popular trade fiction, including most prominently Thomas & Mercer, a mystery and thriller imprint (911 titles since 2011), and Montlake Romance (1,378 titles since 2011), have had the greatest impact on publishing. Amazon publishes print copies of these texts as part of the promotion, although other retailers have hesitated to stock a book published by their Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 5 The Everything Store? [97] fiercest competitor. 54 Through all these imprints, Amazon has published eight hundred books per year since 2009, making the company one of the twenty largest publishers in the United States. Coda: Amazon Charts May 2017 saw the launch of Amazon Charts, a weekly update on the most read and purchased books across the US Amazon web store, the company’s latest attempt to disrupt trade publishing orthodoxies. The Washington Post— owned by Jeff Bezos, albeit independently of Amazon— included the data in its bestseller lists in January 2018 to provide the first list to account for the largest ebook retailer. 55 Amazon Charts data did not come with any figures but offered an inclusive definition of “book,” which encompassed print, ebooks, and Audible. The latter two categories provided observable data for the “Most Read” charts. Kindle Unlimited books were counted “once a customer has read a certain percentage— roughly the length of a free reading sample,” 56 offering a surprising degree of transparency regarding methodology compared to the New York Times Best Sellers. The list also includes other features, such as tidbits about the authors’ agents, that are unlikely to interest nonindustry members, as well as facts about the books based on the data, including tagging books as “unputdownable” if readers had generally finished the book in a single session, an honor that Sarah J. Maas’s A Court of Wings and Ruins received in the first week. 57 The separation of “Most Read” and “Most Sold” reveals the limitations of focusing on sales, which might not map over to what readers consume. I analyzed Amazon Charts’ first eighteen weeks between May and September 2017 to show how the two categories work in parallel. Unsurprisingly, both reading and purchasing habits are strongly tied to film and television adaptations: Margaret Atwood’s The Handmaid’s Tale was omnipresent in both lists due to the television show’s run from April to June and its availability on Kindle Unlimited. Negative reviews of a film have a chilling effect on Amazon users’ purchasing and reading habits. For example, Stephen King’s The Gunslinger appeared on the “Most Sold” list between May and July 2017 in anticipation of the film’s launch on August 20. The book appeared on the “Most Read” lists for four weeks between July 30 and its cinematic release, when interest immediately dissipated. The influence of Kindle Unlimited is stronger for fiction than nonfiction. For example, the Harry Potter titles frequently appear in the “Most Read” section, but as the series is exclusively available via Kindle Unlimited, this does not register as purchasing the title. The data demonstrate a distinct split between hype Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [98] cycles and marketing campaigns fueling purchases and what readers are actually interested in. Amazon Charts provide a useful counterpoint to the New York Times Best Sellers list, a barometer of taste and frequent reference point for Amazon during the development of the Kindle. Thirtynine percent of fiction and 43 percent of nonfiction that appeared on the New York Times Best Sellers list also appeared on Amazon Charts. More books appear on both the Times list and Amazon’s “Most Read” books (47 percent of fiction titles, 51 percent for nonfiction). Readers are drawn to the same big books— Atwood’s The Handmaid’s Tale, Neil deGrasse Tyson’s Astrophysics for People in a Hurry, Trevor Noah’s Born a Crime— but differences emerge in the titles that hover farther down the charts. Some of these variations may be a matter of sampling. For example, Stephen Covey’s The 7 Habits of Highly Effective People would be classified as a perennial seller by the New York Times and would be ignored even if it outsold other titles. Elsewhere the differences show how agency pricing has warped the market. Stuart Wood and Parnell Hall’s Barely Legal debuted at number 5 on the New York Times Best Sellers list on August 13, 2017, but it did not appear on Amazon Charts. In September 2017, the Kindle edition was priced at $17.54, a mere fifteen cents cheaper than the hardcover. Since consumers value digital goods less than the equivalent physical artifact, the lack of discount has boosted print sales at the expense of ebook growth in mainstream publishing. 58 Amazon released the bestselling Kindle title for each of the first ten years to celebrate its first decade (table 5.2). 59 The international launch of the Kindle in 2009 leans toward a convergence with the print marketplace, but the first two years are more intriguing. The 2008 best seller, The Complete User’s Guide to the Amazing Amazon Kindle, outsold Khaled Hosseini’s A Thousand Splendid Suns and Patricia Cornwell’s Book of the Dead as users wished to learn more about the new platform. The inclusion of Ken Follett’s The Pillars of the Earth, a novel first published in 1989, is more intriguing. Follett’s sequel, World without End, was published in October 2007 and was one of the featured $9.99 New York Times Best Sellers on the firstgeneration Kindle’s launch page. The prequel outsold the original, as it was both cheaper ($6.39) and longer (1,000 pages), demonstrating the volatility of ebook pricing from the start of the Kindle’s life span. 60 The ability to read a dense novel from the backlist for cheap without carrying it around marked a landmark moment for the early adoption of the device as users began to see its advantages over print. The Kindle’s first decade has seen a wave of experimentation and the divergence of print and digital marketplaces. This shift has not necessarily Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 5 The Everything Store? [99] democratized contemporary publishing, as predicted by early advocates of selfpublishing. As indicated by the presence of the legacy publishing heavyweights in the “Most Read” section of Amazon Charts, the industry still depends on hype cycles, perennially popular titles, and best sellers. The initial promise of digitization and the backlist did not materialize into substantial new business for legacy publishers. While legacy publishers continue to engage with digital publishing, albeit with the caveats of agency pricing or with other concessions from retailers such as Amazon, the selfpublishing and digitalonly models will continue to thrive; but if the scaffolding from legacy media companies is removed, there is less evidence to demonstrate that this community is selfsustainable, and much of the cultural record may disappear with the planned obsolescence cycles of digital media. The ebook marketplace exists in a precarious state, still reliant on printoriented publishers that resist committing to digital publishing. Amazon’s infrastructure is essential to maintaining a strong presence for ebooks, but the lack of trust between publishers and Amazon continues to erode public interest in ebooks. The Kindle Store requires both continued investment from printoriented publishers and the encouragement of digital publishing startups to become sustainable. Table 5.2 Kindle yearbyyear best sellers * Year Kindle best seller 2007 Ken Follett, The Pillars of the Earth (1989) 2008 Stephen Windwalker, The Complete User’s Guide to the Amazing Amazon Kindle (2008) 2009 Dan Brown, The Lost Symbol (2009) 2010 Stieg Larsson, The Girl with the Dragon Tattoo (2008) 2011 Kathryn Stockett, The Help (2009) 2012 E. L. James, Fifty Shades of Grey (2011) 2013 Dan Brown, Inferno (2013) 2014 John Green, The Fault in Our Stars (2012) 2015 Paula Hawkins, The Girl on the Train (2015) 2016 Paula Hawkins, The Girl on the Train (2015) 2017 Margaret Atwood, The Handmaid’s Tale (1985) *Wheat, “10 Years, 10 Books.” Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [106] for data transfer. An alwayson bookshop removed several obstacles for continual consumption in app. Amazon even began to sell books from within other titles, although readers can opt out. For example, when readers complete Suzanne Collins’s The Hunger Games, 16 a popup appears that encourages them to purchase Catching Fire, the second book in the trilogy. James Petts and colleagues working on the Personalization Platform developed an algorithm to further monetize reading through identifying references to books in a title and introducing links to their product page in the ebook. 17 The technology team never implemented the algorithm due to the sanctity of the page, but it transformed other parts of the Kindle app to include personalized advertising. Kindle ereaders’ home page libraries from the Kindle Touch onward feature a bar of recommended titles by default, with only three of the reader’s owned books appearing until the reader clicks through to their full library. Free titles are conspicuously absence in these recommendations, even if a reader primarily consumes this material. Daily deals and best sellers are prominently displayed and can fill up to a third of the screen on dedicated devices. In the default setup for the Kindle Touch 5.3.7.3, the advertised books outnumber the titles in a user’s library in volume, although proportionally, owned books take up most of the screen space. Nonetheless, users are constantly reminded that they are only a click away from purchasing new titles. The Kindle 5 extended the logic of alwayson consumption by integrating “special offers,” or advertisements on the sleep mode screen, in exchange for a small reduction in the cost of the hardware. Users could purchase the subsidized device for $10 cheaper than full price, with the opportunity to pay $30 to remove the advertisements from the device at a later date. Many advertisements delivered to my devices displayed offers exclusively related to Kindle accessories and purchasing books via the store, but Amazon provides the opportunity for third parties to advertise products (books or otherwise) on the platform. 18 The shift from culturally significant to commodified screen savers demonstrates that Amazon’s selfreported customercentric approach can instead be seen as an attempt to monetize all areas of the Kindle platform. User Labor Beyond the storefront, Amazon relies on usergenerated content to increase the appeal of its ebook catalog. This is most prominent in Kindle Popular Highlights, discussed in detail in the next chapter, but there are further examples of the company repurposing user labor to create Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 6 Paratext Services [107] metadata for books through Shelfari. Josh Hug and Kevin Beukelman, two former employees of RealNetworks, an early audio streaming service, founded Shelfari as a social network for bibliophiles in late 2006. Hug and Beukelman wanted to compete with LibraryThing, launched by Tim Spalding in 2005, by funding a social reading platform through Amazon affiliate links. 19 Amazon eventually purchased the startup in 2008 to form the foundations of the “book extras” function after investing $1 million in the company in 2007. 20 More recent changes in the Kindle platform suggest that this was a forerunner of the consolidation of readingrelated activities to Goodreads, an acquisition that was initially positioned as independent from Amazon’s core activities but has become more integrated into the platform with each new hardware launch. The sixyear period of Shelfari’s integration into the Kindle platform helps illuminate how these paratext services survive while Amazon considered them valuable and how the company retains the most useful aspects and deletes the less desirable. Shelfari data formed the basis of the Kindle’s “book extras” feature until the introduction of “About the Book” and integration of Goodreads in 2014. “Book extras” contained a range of trivia related to the title including a synopsis, awards, and brief descriptions of characters. A small disclaimer at the top of the popup started: “This content comes from Shelfari, the editable book encyclopedia.” The feature is only available in books downloaded before the introduction of “About the Book.” Newer titles integrate data from Goodreads, and any extra contextual information in the “XRay” sidebar appears directly from Wikipedia, removing any bookspecific information. While the switch removed several core features of Shelfari, it also decreased redundancy between features present on both platforms, including outlining characters and awards. Shelfari was initially kept separate from the main Amazon services before its integration into the Kindle reading system. This follows the example of both the Internet Movie Database (IMDb) and Twitch.tv, which were acquired by Amazon in 1998 and 2014 respectively, with press releases indicating that the services would remain separate from Amazon’s primary businesses. While both sites remain independent through mid2017, there has been an increased level of integration between primary and secondary businesses. In the case of IMDb, the vast cache of actors, trivia, and other filmrelated data has been used as part of “XRay for Movies and TV shows,” a service for Amazon Video that replicates the functionality of Kindle XRay for visual content. Twitch.tv’s independence has already been challenged by the launch of Twitch Prime, linked to Amazon Prime accounts. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [108] Shelfari remained semiautonomous from the Kindle, as directly editing content from the reading system was never possible. Users would need to sign in to the service via a web browser to update any incorrect information. The social network functioned primarily in a similar way to Goodreads, allowing users to create virtual bookshelves and review content in a social network dedicated to book reading. In late 2009, Amazon made a major alteration to the design of Shelfari pages through a beta editable book details page. 21 Users were now directed to produce information on specific topics such as “description,” “ridiculously simplified synopsis,” brief descriptions of characters, bibliographic information, quotations, themes, interpretation, and other paratext that might be expected in a critical edition. 22 Amazon changed Shelfari from a thirdparty marketing tool for affiliate sales to a labor force for an optional paratext feature. The user base was not large enough to undertake this work in any uniform fashion, leaving the “Book Extras” page uneven across ebooks. The integration of usergenerated content into the “authorized” space of the book masked the underlying exploitative wikibased labor practices on sites such as Shelfari. Users were encouraged with frequent references to editing pages or seeing their history. Even though Shelfari offered a direct login from Amazon accounts, no clear information indicated that any edits on the website would be scraped for inclusion in commercial ebooks. New pages were left empty, and users were expected to fill them in with a series of edits. As a consequence, the pages for popular books such as the Harry Potter series were replete with relevant information and minutiae, while many pages were left empty. The quality of the final product was variable and based on the free labor of volunteers. This was further complicated by the presence of “Shelfari librarians,” who monitored changes and often rejected changes made by Shelfari staff. 23 The community policing and affective labor behind Shelfari librarians ensured a baseline for quality, but without further investment from either publishers or Amazon, the resource was inconsistent and dependent on the whims of a book’s fan base. In 2017, Amazon closed the remaining Kindle Popular Highlights and Shelfari services to consolidate reading social networks. Numbers talk; Goodreads boasts over twenty million active users. After resisting integration, the service’s “To Read” lists and a feed of friends’ activity appeared on the Kindle home screen for users. 24 This new feature lacked the ability to talk about the book without exiting the primary reading interface or edit the data about a book, as is possible with Shelfari. Users can access the full range of Goodreads services through a link, but this remains at a distance to the Kindle content beyond the home page. The new social infrastructure lacked the granularity of the separate social networks available to readers Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 6 Paratext Services [109] beforehand. Users could no longer directly influence on the information “About the Book,” and Goodreads encourages them to comment on pages rather than specific quotations. The restructuring also created borders between the distinct parts of Amazon’s service paratext. Rather than gravitating around a centripetal force, users needed to leave the main book interface to access new features related to the book. For example, the Kindle for Fire OS 10.29, one of the most featurerich reading systems, separates the Kindle Store, Goodreads, Audible, Wish Lists, Word Runner, and Flashcards into discrete entities. Readers could access a wealth of extra information about a book or otherwise view or hear the book in a different form, but this needed to be a conscious decision rather than accidentally clicking on the wrong option. The separation of different paratext service indicates Amazon’s acknowledgment that the platform’s feature creep challenged the unity of a discrete book. Automation While its reliance on user labor for paratext services has diminished, Amazon continues to use automation as a remedy for the lack of platformexclusive content from publishers. Many of these services for the Kindle rely on text processing, which requires high competencies in natural language processing, from the ability to deduce the meaning of a word from context to disambiguation of words and correctly identifying parts of speech. 25 The algorithms and data structures underpinning these efforts remain proprietary, but errors in available tools show that Amazon is unlikely to have invested the resources into refining the automation to avoid incorrect markup for a corpus featuring billions of words, despite investment in these areas to develop the Alexa voice assistant ecosystem. Three major services— the dictionary, XRay indexing, and ebook searches– depend heavily on automation to provide users with adequate definitions and navigation. The dictionary function suffers from both presentational and semantic issues. The feature is emphasized by a popup box that appears with a definition if a user highlights a word on all but the Kindle 1. If a definition is too short, multiple words are displayed to fill the space. For example, when searching for the definition of “cofounder” in “EDWIN CATMULL. A cofounder of Pixar and later a Disney Executive,” 26 the twoline definition states: co•found•er n. a joint founder, cofound v, co•func•tion n. [MATHEMATICS] the trigonometric 27 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [110] This is followed by a prompt for users to view the definition with the dictionary ebook. Users will likely recognize “cofunction” as a separate definition rather than an extension of the lookedup word. Nonetheless, the presentational quirk could be avoided if Amazon loaded the dictionary as a database rather than pulling content from a regular ebook without checking for overlapping definitions. Since this would require a different content license and more investment, the lack of nuance for this feature demonstrates Amazon’s inability to develop features that would benefit the user unless they offer a substantial payoff in data collection or audience development. In an extensive analysis of the Kindle’s dictionary mechanism, two lexicographers, Theo Bothma and D. J. Prinsloo, document a smorgasbord of errors in the automated tagging system as the result of incorrect links between words and definitions. 28 Automated dictionaries require accurate tagging, particularly with identifying lemmas, or the “canonical” form of a word. For example, the word “men” does not have its own dictionary entry but should redirect to the lemma “man” with a note that “men” is the plural form. The researchers found several major parsing errors. For example, “who’s” correctly linked to “who,” but if the first letter was capitalized, the algorithm rendered the whole word in uppercase, leading to a definition of “World Health Organization” instead. Other errors are more inexplicable, including “but” linking to “and” as well as “lay” changing to “burn.” 29 Bothma and Prinsloo could not identify the cause of these errors, but their analysis points toward some significant fissures in the dictionary algorithm. Since 2013, Amazon has invested substantially in voice recognition for Alexa, which requires more nuanced natural language processing to respond correctly to questions. The Kindle’s dictionary errors point toward this infrastructure not being integrated into the ebook services, revealing its place in the Amazon hierarchy. If the dictionary does not contain a definition, which is often the case for proper nouns such as famous people or companies, Wikipedia is the fallback. Users can also search for phrases via Wikipedia where the dictionary only defines a single word. The website mixes Amazon’s predilection for both automation and user labor. Kindle integration endorses Wikipedia’s quality despite frequent controversies regarding its commitment to “neutral points of view” and uneven coverage of subjects according to users’ interests. 30 Where multiple instances of the same name exist, users are presented with the disambiguation page. For example, a search for “Bistritz” in Dracula links to a page that lists several locations across Romania and the Czech Republic, where readers unfamiliar with the novel’s context may not know that the top result in Transylvania is the likely destination. 31 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 6 Paratext Services [111] Despite these limitations, sideloading Wikipedia and the dictionaries is cheaper for both platforms and publishers, although without careful NLP and human auditing to ensure that the correct reference always appears, the resource can only be as valuable as the “I’m feeling lucky” result for each definition. 32 Paratext services that update can relieve the burden of updating an ebook’s metadata, but the system is only as reliable as the data and algorithms underpinning it. 33 The Kindle’s service paratexts are often superficial and therefore lead to errant responses. When a word might have multiple definitions, the algorithm does not flag the word for review to ensure the most accurate definition is returned, but the first result in the chosen dictionary appears by default. For example, on my Kindle 2, when I click on “Lee,” the software returns the first biographical result for the surname Lee in the dictionary (“Lee 1 Ann [1736– 84] U.S. religious leader...”) 34 rather than noting the word is a popular foreand surname. Proper nouns can be disambiguated through context, which could be provided by the publisher or otherwise deduced by the text, but Amazon’s efforts largely depend on the available external data. Despite the issues with developing sophisticated NLP techniques for dictionary identification, Amazon was confident in its service infrastructure to correctly identify definitions through a combination of machine learning and Mechanical Turk volunteers. 35 Daniel Rausch, a publisher who was hired by Amazon to lead Kindle product management, filed a patent in 2010 that used the dictionary extensively as a service. Rausch detailed Amazon’s ability to build vocabulary tests, diachronic analysis of meaning changes, and recommendations for titles using the word in the same sense (figure 6.3). When combined with demographic data including reading level, the service paratext would also alter the text to adjust to the literacy of the reader. 36 This intrusive service exists only in the theoretical realm of patents, not least because editing texts would present challenges with intellectual property. All evidence of languagebased automation also demonstrates that attempts to alter the text would likely lead to incorrectly changed material. Amazon’s marketing for XRay states that the feature “lets you explore the ‘bones of a book.’ You can also view more detailed information from Wikipedia and from Shelfari, Amazon’s communitypowered encyclopaedia for book lovers.” 37 XRay uses a “term frequency– inverse document frequency (TFIDF)” algorithm to determine which terms are significant enough to index rather than more nuanced or humancurated mechanisms. 38 As the name suggests, the Kindle’s TDIDF compares how frequently a word appears in a text with other Kindle titles. The algorithm can identify unusual terms but would not necessarily identify common terms Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [112] 6.3 Amazon’s prototype “vocabulary service” (adapted from patent). Source: Killalea and Hamaker, Disambiguation of term meaning. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 6 Paratext Services [113] such as “God” or “books” that might be central to a particular title. Amazon discourages the use of a print index in favor of XRay. Jan Wright questions the intellectual property of this arrangement: “Who owns the XRay file: the author, Amazon, the publisher?” 39 This question can be extended to multiple elements of the automated Kindle infrastructure. Elsey notes that the licensing mess associated with backlist titles— ensuring ebook edition rights for photographs, illustrations, and quotations— creates an extra burden. 40 Automated paratext exacerbates the challenges of both licensing and updatability, since it takes agency from the publisher, which may no longer control the overall quality of its project despite an association with all the material provided by the platforms. Wright concludes that XRay lacks the finer detail made possible through a carefully curated index. 41 For example, in Richard Wrangham’s Catching Fire, indexes contain relational information allowing a reader to identify passages on “fat” and more specifically on the relationship between “fat” and “energy.” 42 Despite this lack of complexity in mapping relationships, the XRay format is technically sophisticated compared to other paratext services after Amazon changed the storage medium from JSON to SQL, a database language that emphasizes relationships between entities. XRay files now construct relationships between excerpts, book locations, concepts, and definitions to ensure information is delivered consistently across reading systems. The data are still prepopulated, however, and users need to update the content manually or through an update pushed by Amazon and the publisher. Mistakes are common, as descriptions frequently come from Wikipedia, leading to awkward references to apple trees in place of Apple Inc., or Muhammad the Prophet rather than a character called Muhammad. Human intervention could catch these errors, although it would require understanding the context, and this would be an additional cost for publishers or Amazon. Of the 210 titles available for the Kindle that appeared on Amazon Charts lists between May and September 2017, around 70 percent use XRay. Fiction is more likely (80 percent) to feature the function than nonfiction (64 percent). Print versions of nonfiction titles often include an index, which is easier to convert to an ebook, while fiction may benefit from this conceptual map if automated even when the print edition does not require an index. No correlation exists between the publisher’s investment in a book and the appearance of an XRay index, as prominent publications such as Hillary Clinton’s What Happened launched without the feature. XRay is not universal; since it requires publishers to create a dedicated index for the Kindle, this is unsurprising. Amazon’s rejection Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [114] of broader cross compatibility reduces the number of features available compared to competitor platforms, which do not have the same user base, ensuring development is limited. This limits the potential of ebooks overall, as there is no one platform where innovation is profitable for publisher or platform. Amazon promoted XRay’s ability to visualize the book on a onedimensional plane, showing where characters, places, and themes appear across the novel. 43 Readers can use this information to ascertain how important a character or theme is to the book, although in fiction this led to spoilers if a character reappears after a lengthy absence from the narrative. 44 Since the process is automated, clustering can occur in interesting ways. For example, in Donna Tartt’s The Goldfinch, the relational database entries from excerpts can be 3,835 bytes long and contain as many as ten entities (three locations and seven characters). 45 If an entity is referenced multiple times in a single passage, the data are amalgamated into a single reference. For instance, Hobie appears in the excerpt but is continuously highlighted for a much larger region of the text. Likewise, the indexing catches the variations between “Andy,” “Andy’s,” and “Andy Barbour.” The index algorithm is greedy, matching the garbled phrase “Andy’s death, Death in general) if I thought a familiar person came to meet us at the door, because— writing this now, I’m close to tears— I think of poor Andy” as a single instance because of how Amazon’s search tool splits the blocks of text into the discrete chunks of text as locations. 46 The index therefore returns entries in proximity rather than revealing multiple instances in short passages. Since the algorithm iterates the overlap detection mechanism, results can encapsulate lengthy passages unintentionally. The intext search feature reveals the limitations of Amazon’s natural language processing facilities. 47 A search for “man” in Milton’s Paradise Lost returns all words in the text containing “man,” including “manner,” “commander,” “mans,” and “Adamantine,” even when punctuation or white space is included at the beginning or end of the word. 48 The only exception comes from “man!” which only matches exact results, including punctuation. While the patterns appear to suggest that the search results favor “man” as a prefix, returning “Adamantine” demonstrates that the pattern also matches the term in the middle of a word. If white space is added to both sides of the word to indicate searching for the word in isolation, the results are greedier, returning “command” and other words not present in the original search. “Man.” and “man?” follow the same patterns respectively. This comes at a cost, as any other search for “man” will not catch this in the results because the exclamation mark is inextricably bound to the word. Likewise, any instances of “man” at the Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 6 Paratext Services [115] end of a sentence are not included owing to the proximity of a punctuation mark. Due to this awkward mix of partofspeech tagging without associated documentation to explain discrepancies, the search function is highly volatile and cannot return reliable results. Automation has become a crutch for Amazon in the harsh reality of publishers’ hesitance to create content optimized for the Kindle. These extra features are useful marketing tools, but without further investment and quality control, the invisible biases and limitations of the tools hinder reader interpretation, thus running counter to their designed use. The natural language struggles are matched by further challenges in constructing new navigation idioms with the flat screen. In one instance, Amazon considered extending the XRay/spine metaphor of the book further through a new navigation tool described in the patent “Edge navigation user interface.” 49 Scott Dixon, a Kindle Software Development Kit (SDK) engineer, conceptualized XRay as the literal spine of threedimensional book interface (figure 6.4). The metaphorical use of threedimensional space to navigate the book was just one of several experiments, although it was the most grounded in physical metaphors. The patent contains several images of more complex navigational tropes that rely on users’ knowledge of multitouch interfaces. Augmenting the book is a central element of the patent’s background. Dixon suggests that “as electronic devices continue to evolve, there remains a need for improving a reader’s ability to interact with these devices.” 50 The XRay metaphor requires further visual elements to enhance the reader’s understanding of the ebook’s representation of a twodimensional object. This intervention sought to use the visualizations as a direct navigational tool. Unfortunately, this was never integrated and was superseded by the later addition of Page Flip as a twodimensional version of the same trope. Alongside other new navigation elements and counter to improving the automation of XRay, search, and other features, these new paratext features emphasize feature creep and innovation. The reading space of the Kindle has remained consistent over its first decade, but new paratext services serve to justify yearly updates on the marketing copy for the product page. Hardware innovations have been glacial, so minor software adjustments and new services represent the best opportunity for marketing. Synchronizing Media While automation is used primarily for metadata production, it has been most successful in linking media. This refashioning of the text as a malleable object to be reused in various contexts was one of the strengths of the Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [122] has formed shared reading on the platform by algorithmically filtering the billions of highlights stored on its servers. The scale of social reading on the Kindle invites closer scrutiny of how the service might shape our understanding of contemporary publishing. Users receive frequent prompts to share on the platform. The Kindle 1 featured “My Clippings,” a function “to capture your favorite quotes and share with others.” 2 The Reading Experience team experimented with other forms of user interaction, including prototyping inapp book clubs according to region, age, education level, or profession. 3 The idea was never implemented, but traces remain in the Kindle’s social infrastructure, such as the Popular Highlights website, where users were encouraged to follow readers with similar tastes to track what they read and highlighted. The most direct implementation of the book club came in the form of supplementary PDFs with questions for book clubs in 2012. 4 Amazon’s social reading strategy was solidified with the Kindle 2. The Reading Experience team listened to feedback about web access to personal “clippings” and was tasked with creating a website for organizing users’ highlights and notes. Deanna Glaze designed several prototypes for the platform to emphasize three main activities: “Read. Review. Remember.” 5 The home page featured a prominent “Daily Review” flash card to encourage users to review their previous highlights when entering the site. The website offered tools for exporting clippings and tracking reading progress by noting completion of books or desire to pick up new titles. The prototype downplayed sharing through adopting a broadcast model in early iterations. Users were encouraged to “publish” highlights while noting the number of their “subscribers” and only prompting subscriptions to books. Reciprocity was prioritized at launch, with users following one another with the ability to comment on highlights from other users. Shared highlights were optin, and users could individually select what would remain private or be seen by a wider audience. The Kindle Popular Highlights website offered private access to users’ personal clippings and displayed shared quotations from both individuals and in aggregate. The most frequently shared passages were prominently displayed on the home page, with the ability to click through to a dedicated highlights page linked to an ebook’s ASIN, leading to multiple individual pages for each digital edition of a title. The dedicated pages featured a selection of five popular highlights in aggregate and a live feed of recently shared quotations and notes. Access through the reading system would offer more granular data in the form of the ten most popular highlights for a title. Readers who follow others can see their shared annotations when reading the same book. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 7 Social Reading [123] Kindle Popular Highlights hints at the scale of Amazon’s data surveillance practices. Titles such as Victor Hugo’s Les Misérables and the English Standard Version Bible feature over a million highlights. The eight hundred ebooks discussed in this chapter have received a cumulative twentyfive million annotations. Many titles are not as popular as those discussed here, but overall, Amazon has a data set containing billions of shared highlights, with an even larger cache of saved “clippings” that users elected to keep private but store via Whispersync. The data asymmetry between Amazon and the public grew in mid2014, when the list of most shared passages was removed from the Kindle Popular Highlights website. 6 The list had been static since the last update in 2012, but it had become a “spine” for navigating the website and discovering popular Kindle titles. Searching via the site or using exact ASINs allowed users to find individual product pages, but this required preexisting knowledge of the book’s existence. The closure of Kindle Popular Highlights in 2017 marked a turning point in Amazon’s publicfacing services. While the company has blocked public access to the website, it continues to develop new methods for analyzing and using its large data sets to model consumption, including personalizing advertisements in printondemand books according to a user’s metadata profile to defray costs. 7 Understanding what readers are interested in and how far through a book they will get, data only available through profiling digital consumption, creates more accurate profiles for personalization. Zuboff argues that the term “personalization” under surveillance capitalism is contradictory, as “it defiles, ignores, overrides, and displaces everything about you and me that is personal.” 8 Since the data are aggregated, they do not reflect individual habits as much as a homogeneous “audience.” Publishers have the same level of access as the general public 7.1 An example of a popular highlight from Jane Austen’s Pride and Prejudice (2007). Screenshot from Kindle for PC 1.23.1. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [124] due to Amazon’s selfinterest in surveillance practices when completion rates and most highlighted passages might inform editorial and marketing developments. But what value do these data have? Can shared highlights provide insight into social reading? Previous answers to these questions avoid the issues of scale and the limitations within the Popular Highlights data set with smaller case studies of shared annotation patterns in Sherlock Holmes, public domain works, or the most popular titles. 9 These projects have largely ignored the integrity of the public data sources by focusing on manageable subsets to rein in the potential for chaos. The aggregated and individual highlights show Amazon’s mediation of shared reading and how this might influence users’ understanding of a book, just as with an annotated copy of a library book. With access to the reading patterns of individuals and the aggregated list of ten popular highlights, we can see the influence of previous readers’ interests in shaping new interpretations. Kindle Popular Highlights and associated metrics offer a form of virality akin to retweeting, where resharing amplifies the visibility of a particular passage’s sentiment. The algorithmically mediated selection of specific individual and aggregated highlights reflects Amazon’s marketing agenda. Authors can use the public slice of Kindle Popular Highlights to assess what material readers engage with and write more on that subject. The marketled focus is reflected in Bezos’s boast in a letter to shareholders that “over a thousand independent authors surpassed $100,000 in royalties in 2017 through Kindle Direct Publishing.” 10 The Reading Experience team collaborated with the Reading, Mining and Analytics team to explore how to integrate other social reading elements into the platform. The development of the @author scheme in 2011 allowed readers to post questions to twelve authors through their ebooks or an interface on the Amazon website. 11 A beta version launched in 2011, and the service was live for five years, despite never expanding beyond the initial twelve beta testers. 12 @author used Twitter’s @- mentioning system, notifying the author on Twitter whenever a user asked a question via the reading system. Responses were posted on the @author_posts Twitter account, with 1,256 recorded responses between August 2011 and November 2012. Traces of the project were removed in January 2016, although remnants still appear in the Kindle Touch’s system files, which launched alongside @author. The Touch’s firmware (5.0.0) handled sideloaded content through XML, and several templates were bundled with the device to render the metadata. A file called “stockauthorpage” contains the template for the “About the Book” sidebar and includes an “answersList” with templates for questions, answers, and tracking responses. @author’s Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 7 Social Reading [125] failure came from replicating Twitter’s functionality without adding any benefits. Users flocked to the social media platform to interact with celebrities, including authors, and Amazon’s service increased the number of steps required to post to Twitter, dooming the scheme. Looking at social reading platforms offers a healthy corrective to critics who dismiss digital publishing as inferior to print. 13 Matt Hayler notes that the relationship between print and digital media cannot be reduced to a simplistic binary, since the two media forms are deeply intertwined. 14 Studies of the relative merits of print and digital reading focus on the attention economy, pitting the constant push alerts of mobile devices against an idealized form of distractionfree reading. Environmental factors are necessarily more complicated, and sociality is an important part of the experience. Amazon controls the relationship between readers, ebooks, and other users through its interface and consequently shifts many copyspecific elements of reading— annotations, bookmarks, and so forth— into algorithmically mediated networked experiences. Such mechanisms are essential, since social reading requires annotations to appear across multiple copies of a publication rather than existing in one physical volume. The shared highlights infrastructure was designed to present different content according to the audience. Figure 7.2 distinguishes between users who have purchased a copy of the text and “prospective customers” who may be browsing Kindle Popular Highlights to identify a title of interest. Users could browse popular titles and discover what passages other readers were highlighting. Each page also featured a link to purchase the title on the Kindle Store. The Public Highlights website was only an important resource for as long as it was a useful marketing tool for the company. Once its marketing potential diminished, it was inevitable that Amazon’s desire for secrecy would kick in once more. This attitude toward potentially valuable data sets demonstrates the precarious nature of researching trends in contemporary digital reading. The extent of the data gap became apparent in 2015, when Amazon introduced “About the Book” as a popup that appears when a reader opens an ebook for the first time. The sidebar publicly revealed the scale of highlighting activity alongside information replicating the role of a print book’s cover, including a publisherprovided blurb, links to reviews and related titles, and an estimate of reading time. Amazon supplemented this popup menu with statistics about shared highlights. For example, an edition of Dracula featured “4,673 passages [which] have been highlighted 451,791 times.” 15 “About the Book” compiles at point of download and remains static until the user requests a new copy of the book. Its statistics reveal a lower bound of Amazon’s extensive surveillance practices, since it Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [126] is out of date once the reader receives the data. In figure 7.3, I compare the percentage of shared passages accessible through the Popular Highlights feature with the total reported number of annotations in “About the Book” in eight hundred titles. Readers can only see ten unique shared passages for each title. If an ebook features fewer than ten public annotations, nothing is displayed, ensuring no title features comprehensive information. 7.2 Patent figure showing Amazon’s model for shared annotations (adapted from patent). Source: Tom Killalea et al., Aggregation of highlights, US Patent 9,087,032, filed January 26, 2009, and issued July 21, 2015, sheet 1. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 7 Social Reading [127] Displaying ten popular highlights encourages readers to use the sharing features. Joseph Reagle Jr. argues that a stub, a nearempty page on Wikipedia, encourages users to add new content more effectively than a nonexistent page. 16 Likewise, Kindle users share prehighlighted passages as the interface reminds them that this is an option. Despite the influence of the visible annotations, even large samples from popular books can only account for, on average, 10 percent of highlighting activity. Public highlights reveal a hive mind shaped by corporate interests and algorithms 7.3 Proportion of visible unique and total highlights Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [128] rather than demonstrating the breadth of reading on the Kindle. In more popular books, these annotations perpetuate interest, as it is unlikely that new passages will be highlighted thousands of times without the support of appearing as a shared quotation by default. Amazon’s algorithmic mediation of the process further shapes the perception of the text, as the company can elect to remove the most frequently highlighted passages if they are deemed offensive or otherwise unsuitable from a marketing perspective. In fact, the size of the highlight is one of several factors determining the chance of an annotation appearing, hence the use of “popular” over “top” highlights. 17 The Kindle hardware encourages “alwayson” connectivity, but the Popular Highlights files, an XML document with a PHL extension, download asynchronously. Social reading services encourage users to connect to the internet, but the choice to sideload the PHL postdownload dissuades users from remaining connected if they see past the facade of the Kindle’s networked connectivity. Users who choose the “download and transfer via USB” option will receive just the main ebook without any sideloaded documents, which must be downloaded on the device itself. 18 Despite the static nature of PHL files, users can only access this content by synchronizing their ebook with Whispersync. Customers must trade their data to receive the service. Even if the data are not shared with other users, the highlights form yet another data point for Amazon’s extensive surveillance of contemporary reading. The Kindle for Mac version of Robin Sloan’s Mr. Penumbra’s 24Hour Bookstore PHL file shows how Amazon mediates the popular highlights process. The first shared annotation of the novel (“This is Mat’s secret weapon, his passport, his getoutofjailfree card: Mat makes things that are beautiful”) 19 is represented as “<annotation type=‘popular_highlight’ num_users=‘27’ end_pos=‘54499’ start_pos=‘54383’/>.” Annotations are mapped to byte registers rather than the more userfriendly page numbers (288 pages per the US Farrar, Straus and Giroux hardback edition, which forms the basis of the ebook in both the United Kingdom and United States) or the location register, which finishes at 4,158. 20 Byte registers (116) are not identical to character counts (110) owing to Amazon’s UTF8 compression techniques (discussed in chapter 4). Since the schema depends on a granular approach to identifying location, this introduces an additional challenge for Amazon’s engineers when updating books with even a single correction. A commitment to diversity and updatability leads to precarity in the case of ensuring consistent highlights. To illustrate these challenges, the Penumbra PHL file contains metadata about its creation, including a time stamp of September 2015 and an Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 7 Social Reading [129] update identification number: 105. When revisited in November 2017, the time stamp read October 31, 2017, but the version number had reverted to 14. The updated file introduced several new highlights with fewer shares than more popular sections from the previous copy of the file. Byte registers vary substantially, although some highlights remain persistent, indicating that the update number relates to a change in Amazon’s algorithm or a new edition of the text. Other PHL files further call into question the reliability of the figures. The copy of Pride and Prejudice that came bundled with the Kindle 3 in Britain received almost 600,000 highlights despite being “out of print” since 2011, but the PHL notes just fourteen updates, with the latest change arriving in January 2016. Asynchronous downloads and new versions of the text might account for the low number. The design of the popular highlights system challenges the marketing of the Kindle as an alwayson device. Instead we can see the Kindle as an alwaysupload device that does not necessarily benefit users. Users were offered updates at irregular intervals with the illusion of constant refresh. Lauren Cameron’s investigation of patterns in highlighting of The Adventures of Sherlock Holmes over the summer of 2011 captures the periodic refresh twice: first in the week commencing June 15, and later in the final week of August. 21 Cameron’s analysis is unfortunately based on a coincidence of data rather than mapping the chronological development of shared reading habits. Researchers are not the only casualties of the lack of transparency: Amazon does not offer reliable data for publishers, since the publicly available data are ephemeral and instable. Highlighting Patterns Regardless of the limitations of the public data, Kindle Popular Highlights shows clear patterns in Amazon’s processing and the substantial variation between individual and aggregated practices. As an illustration, Donna Tartt’s The Goldfinch has received substantial user interest with more than a quarter million shared highlights. The novel’s final paragraph, which begins “It is a glory and a privilege to love what Death doesn’t touch,” is the book’s most shared annotation and appeared on its Kindle Popular Highlights page 103 times out of 2,027 publicly available passages in eleven variations. While the version in the popular highlights stops at “it is a glory and a privilege to love what Death doesn’t touch,” 22 in the more extended varieties, readers continue highlighting through to the end of the paragraph and novel. Most users have shared less text than the original highlight, indicating the importance of the popular highlights’ framing in the reading system. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [130] The discrepancies between the individual and popular highlights data result from the aggregation process detailed in figure 7.4. 23 Users A, B, C, and D all share similar but unique sections of a title. The aggregation algorithm establishes overlap between annotations before removing any duplicates and combining the discrete passages into a single visible highlight that does not necessarily resemble the intention of any one user. The algorithm corrects perceived errors (for example, where the first word of a sentence has been chopped off), which removes creative elements of interpretation, such as clipping sentences to alter the meaning. Readers lose their agency as Amazon controls the types of highlights that remain in the system. The Kindle Popular Highlights displayed in reading systems show further evidence of truncation, as shared content is substantially shorter in reading applications than on the website, where the limit was determined by the maximum amount of text displayed on a single page. For example, on the Kindle Touch with the Caecilia Condensed font in the smallest available size, it is possible to display locations 385 to 399 of Isaacson’s Steve Jobs on a single screen, which could then be converted into a single highlight. 24 Journalists were quick to pick up on the potential of the Popular Highlights data set. 25 Jordan Ellenberg’s “The Summer’s Most Unread Book” playfully suggests that the average location of the popular highlights is an indicator of the likelihood of a reader finishing the book. 26 He coined the “Hawking Index” as a reference to Steven Hawking’s A Brief History of Time: a book that sells but is rarely finished. Ellenberg analyzed the popular highlight locations of nine titles including Sheryl Sandberg’s Lean In and Steven Hawking’s A Brief History of Time and found a stark contrast between the highlights from Thomas Piketty’s Capital in the TwentyFirst Century with a mean location of 2.4 percent, and Donna Tartt’s The Goldfinch’s 98.5 7.4 Collating highlights. Source: Killalea et al., Aggregation of highlights. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 7 Social Reading [131] percent average. From this, Ellenberg argues that readers were more likely to finish reading The Goldfinch than A Brief History of Time. Ellenberg’s smallscale experiment was largely supported by a report by Kobo on big data and publishing that revealed that only half of readers would ever finish a book, although Kobo noted a clear correlation between the brand recognition of the author and completion. 27 Overall, nonfiction, including A Brief History of Time, had lower competition rates that fiction because nonfiction titles include additional end matter and readers are less likely to complete a reference guide. If a book has a 50 percent attrition rate, it follows that most highlights will be shared closer to the beginning before most readers stop. In figure 7.5, I expand on Ellenberg’s method 7.5 Indexed highlight locations and volumes Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [234] Milliot, Jim. “The Nook Arrives.” Publishers Weekly, October 26, 2009, 6. Milroy, Rollin. About Agrippa. Vancouver: Heavenly Monkey, 2015. “MIT Press Announces a New Collaboration.” MIT Press, 2017. https:// mitpress .mit .edu /blog /mit -press -announces -new -collaboration. MobileRead. “Kindlet Index.” MobileRead Wiki, 2016. http:// wiki .mobileread .com /wiki /Kindlet _Index. Mobipocket. “OPF XMetadata Tags.” Mobipocket Developer Center, 2007. http:// web .archive .org /web /20090116182003 /ht tp:// www .mobipocket .com /dev /article .asp ?BaseFolder = prcgen & File = tagref _opfxmetadata .xml. Modern Language Association of America. MLA Handbook. 8th ed. New York: Modern Language Association of America, 2016. Montfort, Nick. “Continuous Paper: The Early Materiality and Workings of Electronic Literature,” 2005. http:// nickm .com /writing /essays /continuous _paper _mla .html. Montfort, Nick, and Ian Bogost. Racing the Beam: The Atari Video Computer System. Cambridge, MA: MIT Press, 2009. Morris, Jeremy Wade. “Curation by Code: Infomediaries and the Data Mining of Taste.” European Journal of Cultural Studies 18, nos. 4– 5 (2015): 446– 463. Murray, Simone. The Digital Literary Sphere. Baltimore, MD: Johns Hopkins University Press, 2018. Nakamura, Lisa. “‘Words with Friends’: Socially Networked Reading on Goodreads.” PMLA 128, no. 1 (2013): 238– 243. National Library of Scotland. “The Silent Corner / Dean Koontz.” National Library of Scotland, 2017. http:// main -cat .nls .uk /vwebv /holdingsInfo ?bibId = 9420635. Neate, Rupert. “Amazon’s Jeff Bezos Pays Out $38bn in Divorce Settlement.” The Guardian, June 30, 2019, Technology. https:// www .theguardian .com /technology /2019 /jun /30 /amazon -jeff -bezos -ex -wife -mackenzie -handed -38bn -in -divorce -settlement. Neilan, Catherine. “Amazon Urges Indies to Digitise.” The Bookseller, March 13, 2009, 10. Neilan, Catherine. “Digital Will Overtake Print in 2018, Says FBF Survey.” The Bookseller, 2008. https:// www .thebookseller .com /news /digital -will -overtake -print -2018 -says -fbf -survey. Neill, Graeme, and Catherine Neilan. “Amazon Launches Kindle Worldwide.” The Bookseller, October 12, 2009, 3. Nelson, Theodor Holm. Computer Lib / Dream Machines. Redmond, WA: Tempus, 1987. New York Times Book Review. “About the Best Sellers.” New York Times Book Review, 2017. https:// www .nytimes .com /books /best -sellers /methodology/. New York Times Editorial Desk. “King Closure.” New York Times, December 1, 2000, A36. Newman, Jared. “Amazon Settles Kindle ‘1984’ Lawsuit.” PC World, October 1, 2009. https:// www .pcworld .com /article /172953 /amazon _kindle _1984 _lawsuit .html. Noorda, Rachel, and Stevie Marsden. “TwentyFirst Century Book Studies: The State of the Discipline.” Book History 22, no. 1 (2019): 370– 397. Nosowitz, Dan. “A Penny for Your Books.” New York Times, October 26, 2015. http:// www .nytimes .com /2015 /10 /25 /magazine /a -penny -for -your -books .html. Nuovo, Frank. Portable communication device. US Patent Application Publication no. 2002/0050981, filed June 25, 2001, and issued May 2, 2002. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Bibliography [235] O’Donovan, Caroline, and Ken Bensinger. “The Cost of NextDay Delivery: How Amazon Escapes the Blame for Its Deadly Last Mile.” BuzzFeed News, September 6, 2019. https:// www .buzzfeednews .com /article /carolineodonovan /amazon -next -day -delivery -deaths. Ooba, Seiichi, Shingo Ooue, Hiroyuki Ueda, Hirotoshi Endo, Makoto Murakoshi, and Masashi Yanagida. Dictionary reading device. US Patent 3,612,676, filed November 22, 1968, and issued October 12, 1971. Open eBook Authoring Group. “Supporters.” Open eBook, October 4, 1999. http:// web .archive .org /web /19991004043319 /http:// openebook .org /who .htm. ORCID. “ORCID: Connecting Research and Researchers.” ORCID, 2018. https:// orcid .org/. O’Reilly, Tim. “An Open Letter to Jeff Bezos.” O’Reilly and Associates, 2000. http:// www .oreilly .com /amazon _patent /amazon _patent .comments .html. Ota, Isao. Electrophoretic display device. US Patent 3,668,106, filed April 9, 1970, and issued June 6, 1972. Ota, I., J. Ohnishi, and M. Yoshiyama. “Electrophoretic Image Display (EPID) Panel.” Proceedings of the IEEE 61, no. 7 (July 1973): 832– 836. OverDrive. “Enhanced Media Publishing.” OverDrive, 2017. https:// www .overdrive .com /publishers /enhanced -media -publishing. Overell, Simon, and William TunstallPedoe. Extracting structured knowledge from unstructured text. US Patent 9,110,882, filed May 12, 2011, and issued August 18, 2015. Packer, George. “Cheap Words.” New Yorker, February 10, 2014. https:// www .new yorker .com /magazine /2014 /02 /17 /cheap -words. Panoz, Jiminy. “Ebooks, Beta Testing, and the Apocalypse.” GitHub, 2017. https:// jaypanoz .github .io /ebookcraft2017 / #/. Panoz, Jiminy. “Re: Thoughts on the Future of EPUB 3.” [email protected], January 19, 2018. https:// lists .w3 .org /Archives /Public /public -epub3 /2018Jan /0012 .html. Pargman, Daniel, and Jacob Palme. “ASCII Imperialism.” In Standards and Their Stories: How Quantifying, Classifying, and Formalizing Practices Shape Everyday Life, edited by Martha Lampland and Susan Leigh Star, 177– 199. Ithaca, NY: Cornell University Press, 2009. Parkes, M. B. Pause and Effect: An Introduction to the History of Punctuation in the West. Aldershot: Scolar Press, 1992. Patankar, Rashmi Arun, Jeffrey Matthew Bilger, and Colin Ian Bodell. Providing opaque recommendations. US Patent 10,049,397, filed March 6, 2013, and issued August 14, 2018. Patel, Pankaj C., Arash Azadegan, and Lisa M. Ellram. “The Effects of Strategic and Structural Supply Chain Orientation on Operational and CustomerFocused Performance.” Decision Sciences 44, no. 4 (2013): 713– 753. Pearson. “Penguin Signs Up to ‘Search Inside!’ Programme with Amazon.co.uk.” Pearson, 2006. https:// www .pearson .com /corporate /news /media /news -announcements /2006 /10 /penguin -signs -up -to -search -inside! -programme -with -amazoncouk .html. Peckham, Matt. “Play Doom on a Printer— Thanks to a Serious Security Flaw.” Wired, September 23, 2014. http:// www .wired .com /2014 /09 /doom -printer/. Perez, Thomas E., Samuel R. Bagenstos, John L. Wodatch, Allison Nichol, and Kate Nicholson. “Letter of Resolution between the United States and Reed College.” Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [236] Information and Technical Assistance on the Americans with Disabilities Act, 2009. http:// www .ada .gov /reed _college .htm. Petts, James C., Aaron James Dykstra, Laura Ellen Grit, Lindsey Christina Fowler, Dennis H. Harding, George M. Ionkov, and Samuel A. Minter. Ebook citation enhancement. US Patent 9,639,877, filed July 6, 2012, and issued May 2, 2017. Phillips, Angus. “Have We Passed Peak Book? The Uncoupling of Book Sales from Economic Growth.” Publishing Research Quarterly 33, no. 3 (2017): 310– 327. Plantin, JeanChristophe, Carl Lagoze, Paul N. Edwards, and Christian Sandvig. “Infrastructure Studies Meet Platform Studies in the Age of Google and Facebook.” New Media and Society 20, no. 1 (2018): 293– 310. Platt, Charles. “Digital Ink.” Wired, May 1, 1997. https:// www .wired .com /1997 /05 /ff -digitalink/. Preda, Alex. “Sociotechnical Agency in Financial Markets: The Case of the Stock Ticker.” Social Studies of Science 36, no. 5 (2006): 753– 782. Pressman, Jessica. “The Aesthetics of Bookishness in TwentyFirstCentury Literature.” Michigan Quarterly Review 48, no. 4 (Fall 2009). Publishers Association. PA Publishing Yearbook 2016. London: Publishers Association, 2017. Publishers Weekly. “S&S to Distribute Kindle Bestseller John Locke.” Publishers Weekly, August 22, 2011. https:// www .publishersweekly .com /pw /by -topic /digital /content -and -e -books /article /48433 -s -s -to -distribute -kindle -best seller -john -locke .html. Purgathofer, Peter. DIY Kindle Scanner. Vimeo, 2013. http:// vimeo .com /73675285. Quincey, Andrew de. “KIF: An Infocom Text Adventure Interpreter for the Kindle.” Andrew de Quincey’s Livejournal (blog), October 9, 2010. http:// adq .livejournal .com /108011 .html. Quittner, Josh. “Jeff Bezos: Bio: An Eye on the Future.” Time, December 27, 1999. http:// content .time .com /time /magazine /article /0 ,9171 ,992928 ,00 .html. Radder, Hans. “Exploring Philosophical Issues in the Patenting of Scientific and Technological Inventions.” Philosophy and Technology 26, no. 3 (2013): 283– 300. Ragen, Brian Abel. “Reading Becomes Electric: The Amazon Kindle.” Papers on Language and Literature 44, no. 3 (Summer 2008): 328– 332. Rausch, Daniel B. Determining reading levels of electronic books. US Patent 8,744,855, filed August 9, 2010, and issued June 3, 2014. Ray Murray, Padmini, and Claire Squires. “The Digital Publishing Communications Circuit.” Book 2.0 3, no. 1 (2013): 3– 23. Reagle, Joseph Michael, Jr. Good Faith Collaboration: The Culture of Wikipedia. Cambridge, MA: MIT Press, 2010. Reid, Calvin. “Random House, Modern Library to Offer Ebooks.” Publishers Weekly 247, no. 32 (August 7, 2000): 10. Reid, Calvin, and Karen Holt. “Barnes & Noble.com Exits Ebook Market.” Publishers Weekly 250, no. 37 (September 15, 2003): 9. Reid, Calvin, and Steven Zeitchik. “Gemstar Ebooks Shutdown.” Publishers Weekly 250, no. 25 (June 23, 2003): 10. rodrigoccurvo. “Correct Mobi Location Formula (Maybe).” MobileRead Forums, 2011. https:// www .mobileread .com /forums /showthread .php ?t = 159357. Rose, Charlie. “Interview with Jeff Bezos,” November 19, 2007. https:// charlierose .com /videos /11791. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Bibliography [237] Rose, Charlie. “Interview with Jeff Bezos,” February 26, 2009. https:// charlierose .com /videos /22164. Roseman, Neil C., Paul Kotas, Jeffrey P. Bezos, Bruce C. Moore, Richard L. Dalzell, and Jeffrey M. Blackburn. User interfaces and methods for facilitating usertouser sales. US Patent 7,472,077, filed May 9, 2002, and issued December 30, 2008. Rosen, Rebecca. “The Missing 20th Century: How Copyright Protection Makes Book Vanish.” The Atlantic, March 30, 2012. http:// www .theatlantic .com /technology /archive /2012 /03 /the -missing -20th -century -how -copyright -makes -books -vanish /255282/. Rosenblatt, Bill. “Amazon.com Acquires Mobipocket.” Ebooklyn (blog), 2005. http:// web .archive .org /web /20160324100829 /ht tp:// www .ebooklyn .net /p /amazon com -acquires -mobipocket .html. Rowberry, Simon Peter. “Commonplacing the Public Domain.” Language and Literature 25, no. 2 (2016). Rowberry, Simon Peter. “The Limits of Big Data for Analyzing Reading.” Participations 16, no. 1 (May 2019): 237– 257. Rubincam, David P. Electronic book. US Patent 4,159,417, filed October 28, 1977, and issued June 26, 1979. Run! “File:Eink.svg.” Wikipedia, 2006. https:// en .wikipedia .org /wiki /File: E -ink .svg. Salvador, Stan Weidner, and Vlad Magdin. Predictive natural language processing models. US Patent 9,336,772, filed March 6, 2014, and issued May 10, 2016. Saroyan, Strawberry. “Amanda Hocking, Storyseller.” New York Times, June 17, 2011. ht tp:// www .nytimes .com /2011 /06 /19 /magazine /amanda -hocking -storyseller .html. Satariano, Adam, Karl Russell, Troy Griggs, Blacki Migliozzi, and Chang W. Lee. “How the Internet Travels across Oceans.” New York Times, March 10, 2019, Technology. ht tps:// www .nytimes .com /interactive /2019 /03 /10 /technology /internet -cables -oceans .html. Schaffert, R. M., and C. D. Oughton. “Xerography: A New Principle of Photography and Graphic Reproduction.” Journal of the Optical Society of America 38, no. 12 (December 1948): 991– 998. Schlesinger, Jacob M. “Walkman of Words: Sony’s Data Discman Can Squeeze an Encyclopedia into a Portable Electronic Book.” Wall Street Journal, October 21, 1991, r12. Schuessler, Jennifer. “A Tribute to the Printer Aldus Manutius, and the Roots of the Paperback.” New York Times, February 26, 2015. http:// www .nytimes .com /2015 /02 /27 /arts /design /a -grolier -club -tribute -to -the -printer -aldus -manutius .html. Seely Brown, John. “1997 Objectives,” 1997. Mark Weiser papers, ca. 1975– 1999. Series 5, box 41, folder 10. Shah, Mehal. Dynamic character biographies. US Patent 9,690,451, filed April 9, 2015, and issued June 27, 2017. Sharma, Sudeep. “Netflix and the Documentary Boom.” In The Netflix Effect: Technology and Entertainment in the 21st Century, edited by Kevin McDonald and Daniel SmithRowse, 143– 154. London: Bloomsbury, 2016. Shelfari. “About the Shelfari Librarians & Editors Group,” February 25, 2011. http:// web .archive .org /web /20110225161532 /http:// www .shelfari .com /groups /10713 /about. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [238] Shelfari. “Harry Potter and the Deathly Hallows (Harry Potter 7) by J. K. Rowling,” October 23, 2011. http:// web .archive .org /web /20111023055435 /http:// www .shelfari .com: 80 /books /1064883 /Harry -Potter -and -the -Deathly -Hallows. Shelfari. “Questions and Feedback,” November 13, 2009. http:// web .archive .org /web /20091113093818 /http:// www .shelfari .com: 80 /faq /29. Sheridon, N. K., and M. A. Berkovitz. “The Gyricon— a Twisting Ball Display.” Proceedings of the Society for Information Display 18, no. 34 (1977): 289– 293. Sherman, William H. Used Books: Marking Readers in Renaissance England. Philadelphia: University of Pennsylvania Press, 2008. Siegel, Hilliard B., Udi Manber, and Jonathan Leblang. Method and system for providing annotations of a digital work. US Patent 8,131,647, filed January 19, 2005, and issued March 6, 2012. Sims, Andrew. “The Most Moving Parts of ‘Cursed Child,’ According to Kindle.” Hypable (blog), August 12, 2016. http:// www .hypable .com /cursed -child -kindle -popular -highlights/. Skochinsky, Igor. “Hacking the Kindle Part 3: Root Shell and Runtime System.” Reversing Everything (blog), December 21, 2007. http:// igorsk .blogspot .com /2007 /12 /hacking -kindle -part -3 -root -shell -and .html. Sloan, Robin. “The Kindle Wink.” The Message (blog), April 23, 2014. https:// medium .com /message /the -kindle -wink -4f61cd5c84c5. Smith, Kelvin. The Publishing Business: From Pbooks to Ebooks. Lausanne: AVA Publishing, 2012. Snodgrass, Ryan J., James C. Slezak, Matthew E. Goldberg, Jeremie Leproust, Guillaume Jeulin, and Felix F. Antony. Ebook encryption using variable keys. US Patent 8,826,036, filed June 28, 2010, and issued September 2, 2014. Somers, James. “Torching the ModernDay Library of Alexandria.” The Atlantic, April 20, 2017. https:// www .theatlantic .com /technology /archive /2017 /04 /the -tragedy -of -google -books /523320/. Sorotokin, Peter, Garth Conboy, Brady Duga, John Rivlin, Don Beaver, Kevin Ballard, Alastair Fettes, and Daniel Weck. “EPUB Canonical Fragment Identifiers 1.1,” 2017. http:// www .idpf .org /epub /linking /cfi/. Squires, Claire. “Taste and/or Big Data? Postdigital Editorial Selection.” Critical Quarterly 59, no. 3 (2017): 24– 38. Stallman, Richard. “(Formerly) Boycott Amazon!” GNU Operating System. Accessed February 6, 2017. https:// www .gnu .org /philosophy /amazon .html. Standage, Tom. The Turk: The Life and Times of the Famous EighteenthCentury ChessPlaying Machine. New York: Berkley Trade, 2003. Star, Susan Leigh, and Martha Lampland. “Reckoning with Standards.” In Standards and Their Stories: How Quantifying, Classifying, and Formalizing Practices Shape Everyday Life, edited by Martha Lampland and Susan Leigh Star, 3– 24. Ithaca, NY: Cornell University Press, 2009. Starosielski, Nicole. The Undersea Network. Durham, NC: Duke University Press, 2015. Stefik, Mark. “The Digital Document Company.” Palo Alto, CA: Xerox PARC, July 29, 1995. Mark Weiser papers, ca. 1975– 1999. Series 2, box 8, folder 12. Steiner, Ann. “Private Criticism in the Public Space: Personal Writing on Literature in Readers’ Reviews on Amazon.” P@rticipations 5, no. 2 (2008). https:// lup .lub .lu .se /record /1288297. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Bibliography [239] Sterne, Jonathan. MP3: The Meaning of a Format. Durham, NC: Duke University Press, 2012. Stone, Brad. “Amazon Erases Orwell Books from Kindle Devices.” New York Times, July 17, 2009. http:// www .nytimes .com /2009 /07 /18 /technology /companies /18amazon .html. Stone, Brad. The Everything Store: Jeff Bezos and the Age of Amazon. London: Bantam, 2013. Streitfeld, David. “Book Reviewers for Hire Meet a Demand for Online Raves.” New York Times, August 25, 2012. http:// www .nytimes .com /2012 /08 /26 /business /book -reviewers -for -hire -meet -a -demand -for -online -raves .html. Striphas, Ted. The Late Age of Print: Everyday Book Culture from Consumerism to Control. New York: Columbia University Press, 2011. Swanson, Clare. “The Bestselling Books of 2014.” Publishers Weekly, January 2, 2015. ht tps:// www .publishersweekly .com /pw /by -topic /industry -news /bookselling /article /65171 -the -fault -in -our -stars -tops -print -and -digital .html. Sweney, Mark. “Amazon Breaks Premier League Hold of Sky and BT with Prime Streaming Deal.” The Guardian, June 7, 2018, Media. http:// www .theguardian .com /media /2018 /jun /07 /amazon -breaks -premier -league -hold -of -sky -and -bt -with -streaming -deal. Sweney, Mark. “‘Screen Fatigue’ Sees UK Ebook Sales Plunge 17% as Readers Return to Print.” The Guardian, April 27, 2017. https:// www .theguardian .com /books /2017 /apr /27 /screen -fatigue -sees -uk -ebook -sales -plunge -17 -as -readers -return -to -print. Tartt, Donna. The Goldfinch. London: Abacus, 2014. Telfer, Stephen J., and Michael D. McCreary. “A FullColor Electrophoretic Display.” SID Symposium Digest of Technical Papers 47, no. 1 (2016): 574– 577. Terry, Ana Arias. “Electronic Ink Technologies: Showing the Way to a Brighter Future.” Library Hi Tech 19, no. 4 (2001): 376– 389. Thompson, Ben. “Amazon’s New Customer.” Stratechery, June 19, 2017. https:// strat echery .com /2017 /amazons -new -customer/. Thompson, John B. Merchants of Culture: The Publishing Business in the TwentyFirst Century. 2nd ed. London: Polity, 2012. Thylstrup, Nanna Bonde. The Politics of Mass Digitization. Cambridge, MA: MIT Press, 2019. Tian, Xuemei, and Bill Martin. “Impacting Forces on eBook Business Models Development.” Publishing Research Quarterly 27, no. 3 (2011): 230– 246. Tillett, Barbara. “What Is FRBR? A Conceptual Model for the Bibliographic Universe.” Washington, DC: Library of Congress, 2004. http:// www .loc .gov /cds /downloads /FRBR .PDF. Tremblay, Geoffroy. “KindleBerry Pi.” Ponnuki— Electronic Media Art and Yoga (blog), 2012. http:// www .ponnuki .net /2012 /09 /kindleberry -pi/. Trettien, Whitney Anne. “A Deep History of Electronic Textuality: The Case of English Reprints Jhon Milton Areopagitica.” Digital Humanities Quarterly 7, no. 1 (2013). http:// www .digitalhumanities .org /dhq /vol /7 /1 /000150 /000150 .html. Tseng, Walter Manching, Adam J. Iser, Michel L. Goldstein, Ravi Shankar Thangavel, Janna S. Hamaker, and Ankur Jain. Generating a game related to a digital work. US Patent 9,449,526, filed March 19, 2012, and issued September 20, 2016. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [240] Tsukayama, Hayley. “Why Amazon Is Paying Nearly $1 Billion to Acquire Twitch.” Washington Post, August 25, 2014, The Switch. https:// www .washingtonpost .com /news /the -switch /wp /2014 /08 /25 /amazon -said -to -be -close -to -acquiring -twitch/. “U.P.C.s, Barcodes, & Prefixes.” GS1 US. Accessed July 3, 2017. https:// www .gs1us .org /upcs -barcodes -prefixes /overview. Usbourne, David. “Amazon’s Electronic Book Turns a New Page in the History of the Written Word.” The Independent, November 20, 2007. Van der Weel, Adriaan. Changing Our Textual Minds: Towards a Digital Order of Knowledge. Manchester: Manchester University Press, 2011. Vogels, Werner. “Amazon and the Lean Cloud.” HackFwd Build 0.7, posted to Vimeo, September 2011. https:// vimeo .com /29719577. Vogels, Werner. “How and Why Did Amazon Get into the Cloud Computing Business?” Quora, January 14, 2011. https:// www .quora .com /How -and -why -did -Amazon -get -into -the -cloud -computing -business. W3C. “W3C Publishing Working Group.” W3C, 2017. https:// www .w3 .org /publishing /groups /publ -wg/. Ward, Charles L., Donald Ryan Willhoit, Lars C. Ulness, Robert C. Borja Jr., James D. Mackraz, and Edward J. Gayles. Selecting contentenhanced applications. US Patent 9,268,734, filed March 14, 2011, and issued February 23, 2016. Ward, Miles J. Physical store online shopping control. US Patent 9,665,881, filed May 4, 2012, and issued May 30, 2017. Warner, M. “Can Amazon Be Saved?” Fortune 144, no. 11 (November 26, 2001): 156. Warren, John W. “Zen and the Art of Metadata Maintenance.” Journal of Electronic Publishing 18, no. 3 (2015). https:// doi .org /10 .3998 /3336451 .0018 .305. WashPostPR. “The Washington Post Launches Most Comprehensive Bestselling Books Lists.” Washington Post, February 8, 2018, WashPost PR Blog. https:// www .washingtonpost .com /pr /wp /2018 /02 /08 /the -washington -post -launches -most -comprehensive -bestselling -books -lists/. Weight, Christopher F., Andrew D. Birkett, Janna Hamaker, Tom Killalea, and Alexander William Robb Nelson. Identifying book title sets. US Patent 9,881,009, filed March 15, 2011, and issued January 30, 2018. Weight, Christopher F., Janna Hamaker, Tom Killalea, Bruno A. Posokhow, and Daniel B. Rausch. Book version mapping. US Patent 9,846,688, filed December 28, 2010, and issued December 19, 2017. Weingart, Scott B. “The Route of a Text Message, a Love Story.” Vice (blog), February 22, 2019. https:// www .vice .com /en _us /article /kzdn8n /the -route -of -a -text -message -a -love -story. Weiss, Steven K., Craig S. Griffin, John Lattyak, Lawrence Arnold LynchFreshner, and Thomas A. Ryan. Dynamic display dependent markup language interface. US Patent 8,453,051, filed March 31, 2008, and issued May 28, 2013. Weltevrede, Esther, and Erik Borra. “Platform Affordances and Data Practices: The Value of Dispute on Wikipedia.” Big Data and Society 3, no. 1 (June 1, 2016). https:// doi .org /10 .1177 /2053951716653418. Wheat, Alynda. “10 Years, 10 Books— a Look Back at Kindle Best Sellers.” This Week in Books, November 15, 2017. https:// www .amazon .com /article /twib /kindle -tenth -anniversary .html. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Bibliography [241] Wingfield, Nick. “Amazon Pushes Facial Recognition to Police. Critics See Surveillance Risk.” New York Times, May 24, 2018, Technology. https:// www .nytimes .com /2018 /05 /22 /technology /amazon -facial -recognition .html. Wingfield, Nick. “Inside Amazon Go, a Store of the Future.” New York Times, January 21, 2018, Technology. https:// www .nytimes .com /2018 /01 /21 /technology /inside -amazon -go -a -store -of -the -future .html. Wisher, Robert A., and J. Peter Kincaid. “Personal Electronic Aid for Maintenance: Final Summary Report.” Alexandria, VA: US Army Research Institute for the Behavioral and Social Sciences, March 1989. Wolf, Gary. “The Great Library of Amazonia.” Wired, December 1, 2003. http:// www .wired .com /wired /archive /11 .12 /amazon .html. Wrangham, Richard. Catching Fire: How Cooking Made Us Human. London: Profile Books, 2010. Wright, Jan. “The Devil Is in the Details: Indexes versus Amazon’s XRay.” The Indexer 30, no. 1 (2012): 11– 16. Wykes, Zoë. ISBN13 for Dummies, Special Edition. Hoboken, NJ: Wiley, 2005. Yamazaki, Shunpei, and Toshiji Hamatani. Paperless portable book. US Patent 5,339,091, filed October 14, 1992, and issued August 16, 1994. Zagorie, Aviram, Michael V. Rykov, Craig S. Griffin, and John Lattyak. Device specific presentation control for electronic book reader devices. US Patent 8,423,889, filed December 11, 2008, and issued April 16, 2013. Zehr, Gregg E., John E. Johnston, Jateen P. Parekh, Symon J. Whitehorn, and Thomas J. Hobbs. Page turner for handheld electronic book reader device. US Patent 8,018,431, issued September 13, 2011. Zehr, Gregg, and John Hollar. “Zehr, Gregg, Oral History,” May 14, 2014. http:// www .computerhistory .org /collections /catalog /102739920. Zehr, Gregg, and Symon J. Whitehorn. Handheld electronic book reader device having dual displays. US Patent 8,950,682 , filed March 29, 2006, and issued July 6, 2010. Zhou, Hanning, Jian Liang, and Sherif M. Yacoub. Ondemand generating ebook content with advertising. US Patent 9,892,427, filed September 2, 2014, and issued February 13, 2018. Zittrain, Jonathan. The Future of the Internet. London: Penguin, 2009. Zuboff, Shoshana. The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power. London: Profile Books, 2019. Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Index AbeBooks, 50 Accessibility, 7, 14, 25, 72– 75, 79, 104, 157 Adobe, 8, 13, 74 Flash, xv, xvi Reader, 148– 149 Advanced Linux Sound Architecture (ALSA), 18 Aegitas, 85 Agency pricing, 84, 98, 99 Allington, Daniel, 59 Altice, Nathan, 178 Amazon criticism of, xiii, 30, 35, 156 “Day One”, 148 Digital Products (team), 53 Echo Show, 37, 156 as infrastructure, xv, 6– 8, 24, 25, 27– 45, 49, 51– 53, 60, 65, 83, 93, 99, 104, 111, 119, 142, 148, 152, 154, 168 monopolistic tendencies, 149, 156 presentation as “customercentric,” 32, 106, 118 as technology company, xix, 27, 28, 33, 40, 87 as web retailer, 7, 27, 30, 31, 39, 45, 48, 60, 97, 159 Amazon Books, 44 Amazon Charts, 97– 99, 113, 171 Amazon Crossing, 96 Amazon Go, 44– 45 Amazon Marketplace, 52 Amazon Mechanical Turk (MTurk) exploitation, 152– 153 integration with Amazon infrastructure, 29, 31, 104, 111 origins, 15 role in digitizing books, 52 Amazon Pages, 8 Amazon Pay, 43 Amazon Prime, 41– 44, 107 Amazon Rekognition, 37 Amazon Shorts, 96 Amazon Standard Identification Number (ASIN), 49– 52, 54, 71, 122– 123, 148– 149, 170, 173, 181 Amazon Upgrade, 8 Amazon Web Services (AWS), 15, 28– 29, 31– 34, 36, 38, 42– 43, 45, 47, 52, 78, 104, 152– 153. See also Cloud computing Elastic Compute Cloud (EC2), 152 Amazon XRay, 40, 102, 104– 105, 107, 109, 111, 113– 115, 149, 153 American Association of Publishers, 5 American Council of the Blind, 14 AntConc, 167 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [250] Personal digital assistant (PDA), 64. See also Palm Personal Electronic Aid for Maintenance (PEAM), 4– 5 Personalization, 35, 58, 59, 78, 92, 106, 123 Personalization Platform (Amazon), 106 Petitjean, François, 85 Petts, James, 106 Philips, 16– 17 Picador, 48, 50 Picard, JeanLuc, 3 Piketty, Thomas, 130, 133 PlanetAll, 34 Polymer network liquid crystal display (PNLCD), 18– 19 Portable Network Graphics (PNG), 18, 69 PostScript, 62 Prime Reading, 42, 86, 94 Prime Video, 31, 36 Prime View International, 17 Printing press, xiv, 53, 92, 180 Print on demand, 85, 90, 91, 123, 136, 147, 152 Project Gutenberg, 4, 85, 86, 88, 148 Public domain, 3, 36, 52, 81, 82, 85, 86, 87, 88, 101, 124, 148, 173, 181. See also Copyright Publishers Association, 93 Publishing@W3C, 155– 156, 157. See also W3C; IDPF Purgathofer, Peter, 148 Qualcomm, 11, 34 R (programming language), 167, 174 Radiofrequency identification (RFID), 45 Ragen, Brian Abel, 72 Randomaccess, 3, 64 Random House, 13, 50, 84, 87 Raspberry Pi (computer), 145– 146 Rausch, Daniel, 111 Reading Experience team, 54, 104, 122, 124 Reading system, 8, 107, 108, 122, 124. 129, 130, 148, 151, 170, 172, 174 consistency, 113, 153, 157, 159 digital and physical, 15 featurerich, 109 flexibility, 78 interpreting formats, 63, 65, 71 limitations, 22, 90– 91, 119, 145 Reading, Mining and Analytics team, 121, 142 Reagle, Joseph, Jr., 127, 132 RealNetworks, 107 Recommendations, 35, 40, 43, 44, 56– 60, 63, 71, 106, 111, 170 Research and development, 18, 33, 37, 169 Reverse engineer, 65, 74, 152, 153 Reviews (book), 15, 31, 50– 51, 53, 55, 56, 58– 60, 93, 94, 97, 121, 125 RFID (radiofrequency identification), 45 Rice, Christopher, 178 Richardson, Samuel, 92 Risher, David, 49 Rocket eBook, 6, 21, 148 Romance (genre), xvi, 82, 135, 158, 173 Roseman, Neil, 40 Rosen, Rebecca, 88 Rosenblatt, Bill, 8 Rosetta Books, 87 Roth, Veronica, 58 Rowling, J. K., 52, 83, 94, 135 Harry Potter series, 44, 84, 86– 87, 92, 94, 97, 108, 134– 135, 174 Pottermore, 52, 83, 86 Royalties, 93– 95, 124, 149 S. (J. J. Abrams and Doug Dorst novel), 67– 68, 69, 164 Sandberg, Sheryl, 130 Sarem, Lani, 84 Scalable Vector Graphics (SVG), 69 Schmieg, Sebastian, 144 Schnittman, Evan, 14 ScienceDirect, 90 Screen saver, xv, 18, 19, 21, 106, 144, 145 Scriptio continua, 2 Sewell, Anna, 86 Shakespeare, William, 101 Shelfari, 102, 104, 107– 108, 141 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 Index [251] Shepard, Mark and Aaron, 57– 58 Sheridon, Nick, 16, 146. See also Electronic paper; Xerox PARC Sherlock Holmes, 124, 129 Sideload, 76, 111, 124, 128 Silicon Valley, xvi, 16 Simon & Schuster, 5, 93 60 Minutes, 33 Sloan, Robin, 128 Smartphone, xv, xviii, 14, 67, 68, 72, 146, 157 Société typographique de Neuchâtel (STN), 53, 152 Softbook, 5, 6 Sony BroadBank eBook, 63 Librie, 10, 16, 36, 63 MiniDisc, 28, 61 PRS500, 157 Xperia E1, 167 Speedreading, 4, 117 Spotify, xvii Springer Nature, 90 Stallman, Richard, 35 Standardized General Markup Language (SGML), xvii, 85 Star, Susan Leigh, 27 Start Reading Location (SRL), 66, 72 Startups Amazon, xiii, 41 Amazon’s acquisition of, 34, 107 deification of, 33 digital publishing startups, 82, 85, 93, 99, 159 E Ink, 16 thirdparties, 43 Statistically Improbable Phrases, 53 Steel, Danielle, 88, 96 Steiner, Ann, 59 Stephenson, Neal, 8– 9 St. Martin’s Press, 95 Stone, Brad, xiv, 29, 31 Structured Query Language (SQL), 67, 113 Styron, William, 87 Table of contents, 23, 72 Text Encoding Initiative (TEI), 117 Texttospeech, xiv, 73, 74, 116– 118 Thomas & Mercer, 96 Thompson, Ben, 43 Thriftbooks, 42 Thriller (genre), 82, 96 Ting, Nathalie, 72. See also Mobipocket Torvalds, Linus, 11. See also Linux Touchpress, 101 Touch screen, 19 Toys“R”Us, 30 TPZ (file extension), 181 Transport Layer Security (TLS), 24 Tremblay, Geoffroy, 146 Trettien, Whitney, 85 TrueType, 62 Tseng, Walter, 104 Turing, Alan, 9 Twitch.tv, 28, 30, 42, 107 Twitter, 103, 117, 121, 124– 125, 141 Two Lions, 86 Typography, 22, 53, 65, 68, 69, 72– 74, 86 Tyson, Neil deGrasse, 98 Unicode, 73 US Navy, 74. See also Personal Electronic Aid for Maintenance (PEAM) US Patent and Trademark Office (USPTO), 168– 169 Universal Product Code (UPC), 49. See also Barcodes; International Standard Book Number (ISBN) University of Sydney Library, 150 UTF8, 72– 74, 128 van der Weel, Adriaan, 15 Vanity publishing, 82, 92, 93 Video game, xiv, xv, 9, 28, 40 Vintage Books, 50, 91 Virtualization, 31– 33, 45, 47 Vogels, Werner, 27, 32. See also Amazon Web Services Vonnegut, Kurt, 87 Wallace, David Foster, 57 Wall Street Journal, 29, 31 Walmart, 30 Waterfall Press, 96 Waterstones, 50 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023 [252] Wattpad, 93 Weight, Christopher, 50 Whispernet, 24, 153 Whispersync, 25, 54, 119, 121, 123, 128, 142, 153 Whitehorn, Symon J., 35 Whole Foods, 44– 45 WH Smith, 47 WiFi, 24, 27, 105, 141, 146 Wikipedia, 103– 104, 107, 110– 111, 113, 127 Windows1252, 72– 74 Word Runner, 73– 74, 109, 117, 119, 120 Word Wise (Amazon), 103, 104, 149 World Wide Web, 40, 48, 49, 105, 122 early development, xvii, 3, privileging connection, 15 standards, 64, 66– 67, 70, 157, 159 (see also W3C) utopian visions of, 102 web archives, 150 web browser, xvi, 70, 108 web retail, 27, 30– 31, 34– 35, 40 World Wide Web Consortium (W3C), 70, 155– 157, 159 Wright, Jan, 113 Wylie, Andrew, 87 Xerox PARC, 16, 146 XML (Extensible Markup Language), 64, 67, 68, 124, 128, 151, 183 XScale, 11 XYML, 68 Young adult (genre), 82, 84, 141, 173 YouTube, 151 Zehr, Gregg, 6, 9, 10, 25, 35 Zombie media, 144 zShops, 42 Zuboff, Shoshana, 40, 56, 123, 152 Downloaded from http://direct.mit.edu/books/book-pdf/2087444/book_9780262369114.pdf by ZBW/DT ZENTRALBIBLIOTHEK FUR WIRTSCHAFTS user on 11 October 2023