Ulb ich , Lena; Egbe , Simon
A icle — Published Ve sion
In Palan i we us ? Regula ion o da a analysis pla o ms
in public secu i y
Big Da a & Socie y
P o ided in Coope a ion wi h:
WZB Be lin Social Science Cen e
Sugges ed Ci a ion: Ulb ich , Lena; Egbe , Simon (2024) : In Palan i we us ? Regula ion o da a
analysis pla o ms in public secu i y, Big Da a & Socie y, ISSN 2053-9517, Sage, Thousand Oaks, Vol.
11, Iss. 3, pp. 1-15,
h ps://doi.o g/10.1177/20539517241255108
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/310963
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h ps://c ea i ecommons.o g/licenses/by-nc/4.0/
In Palan i we us ? Regula ion o da a
analysis pla o ms in public secu i y
Lena Ulb ich
1,2
and Simon Egbe
3
Abs ac
O ganiza ions inc easingly ely on digi al echnologies o pe o m asks. To do so, hey ha e o in eg a e da a banks o
make he da a usable. We a gue ha he e is a g owing, academically unde explo ed ma ke consis ing o da a in eg a ion
and analysis pla o ms. We explain ha , especially in he public sec o , he egula o y implica ions o da a in eg a ion and
analysis mus be s udied because hey a ec ulne able ci izens and because i is no jus a ma e o s a e agencies o e -
seeing echnology companies bu also o he s a e o e seeing i sel . We p opose a pla o m- heo y-based concep ual
app oach ha di ec s ou a en ion owa ds he specific cha ac e is ics o pla o ms—such as da afica ion, modula i y,
and mul ila e ali y and he associa ed egula o y challenges. Due o a sca ci y o empi ical analyses abou how public sec-
o pla o ms a e egula ed, we unde ake an in-dep h case s udy o a da a in eg a ion and analysis pla o m ope a ed by
Palan i Technologies in he Ge man ede al s a e o Hesse. Ou analysis o he egula o y ac i i ies and conflic s unco e s
many obs acles o e ec i e pla o m egula ion. D awing on ecen ini ia i es o imp o e in e media y liabili y, we ul im-
a ely poin o addi ional pa hs o egula ing public sec o pla o ms. Ou findings also highligh he impo ance o poli ical
ac o s in pla o m egula ion-as-a-p ac ice. We conclude ha pla o m egula ion in he public sec o is no only abou
echnology-specific egula ion bu also abou gene al mechanisms o democ a ic con ol, such as he sepa a ion o powe ,
public anspa ency, and ci il igh s.
Keywo ds
Palan i echnologies, public secu i y, police da abanks, da a p o ec ion, pla o m egula ion, ci il igh s
Regula o y challenges o da a in eg a ion
and analysis pla o ms in public secu i y
In many coun ies, public agencies wan o exploi he ben-
efi s o big da a, and o se e al yea s, c i ical da a s udies
ha e conside ed his issue (e.g., boyd and C aw o d, 2012).
Wha has ecen ly changed is ha public agencies a e inc eas-
ingly elying on pla o ms ha allow hem o in eg a e and
analyze a ious da a sou ces (Bigo, 2020). This pla o miza-
ion o he s a e p omises o allow be e decision-making
due o la ge da abases, as da a analysis, and sophis ica ed
da a isualiza ion (B ayne, 2021; Fe guson, 2017; Ulb ich ,
2020). While much a en ion has been paid o he socie al
benefi s and isks o s a e-agency use o big da a analysis
o classi y ci izens (e.g., Dencik e al., 2018), he same
canno be said abou he egula ion o s a e au oma ion, espe-
cially when i comes o public sec o pla o ms (Bellano a
and De Goede, 2022).
1
Public agencies mus mee high s anda ds when using
ci izen da a; hence, hese p ac ices a e subjec o democ a ic
o e sigh . The e a e consequen ly many egula o y chal-
lenges associa ed wi h accessing la ge da a banks wi h
ci izen da a and using hem o au oma ed analysis.
2
Fo
his eason, egula ion has been a majo esea ch opic in
pla o m esea ch, which has explo ed how go e nmen s
y o hold echnology companies accoun able o he ac i -
i ies ha happen wi hin hei pla o ms (Bo ges, 2023;
Gillespie, 2010). This esea ch has no ye been connec ed
su ficien ly wi h he li e a u e on s a e au oma ion. Mo e
gene ally, s udies ha empi ically analyze pla o m egula-
ion a e s ill a e, especially when i comes o public sec o
echnology use and specifically o da a in eg a ion and
1
Poli ics o Digi aliza ion Depa men , WZB Be lin Social Science Cen e ,
Be lin, Ge many
2
Resea ch G oup Technology, Powe , and Domina ion, Weizenbaum
Ins i u e o he Ne wo ked Socie y, Be lin, Ge many
3
Facul y o Sociology, Biele eld Uni e si y, Biele eld, Ge many
Co esponding au ho :
Lena Ulb ich , WZB Be lin Social Science Cen e , Poli ics o Digi aliza ion
Depa men , Weizenbaum Ins i u e o he Ne wo ked Socie y, Resea ch
G oup Technology, Powe , and Domina ion, Ha denbe gs . 32, D-10623
Be lin, Ge many.
Email: [email p o ec ed]
C ea i e Commons Non Comme cial CC BY-NC: This a icle is dis ibu ed unde he e ms o he C ea i e Commons A ibu ion-
NonComme cial 4.0 License (h ps://c ea i ecommons.o g/licenses/by-nc/4.0/) which pe mi s non-comme cial use, ep oduc ion and
dis ibu ion o he wo k wi hou u he pe mission p o ided he o iginal wo k is a ibu ed as specified on he SAGE and Open Access page (h ps://us.
sagepub.com/en-us/nam/open-access-a -sage).
O iginal Resea ch A icle
Big Da a & Socie y
July–Sep embe : 1–15
© The Au ho (s) 2024
A icle euse guidelines:
sagepub.com/jou nals-pe missions
DOI: 10.1177/20539517241255108
jou nals.sagepub.com/home/bds
analysis pla o ms. These a e e y ele an because hey
a e he basis upon which any kind o decision-suppo
sys em elies.
Consequen ly, in ou pape , we answe he ollowing
esea ch ques ions: How does pla o m egula ion play
ou in public sec o au oma ion? And how do public agen-
cies ha engage in da a in eg a ion and analysis pla o m
p ojec s ensu e ha he pa icipa ing echnology compan-
ies espec egula ion?
We answe hese ques ions based on an empi ical case
s udy conce ning he implemen a ion o hessenDATA, a
modified Go ham pla o m om Palan i Technologies,
which is being used by he police o ce in he Ge man
ede al s a e o Hesse.
This s udy o he deploymen o Palan i so wa e by
he Ge man police o ce is ele an beyond Ge many. In
in e na ional compa isons, Ge many is ega ded as
ha ing high egula o y s anda ds o he use o digi al
echnologies in public a ai s. Mo eo e , i has a long
his o y o da a p o ec ion laws, and he Ge man go e n-
men has been s i ing o imp o e he digi al so e eign y
o s a e au ho i ies o some yea s now (Mölle s, 2021).
Palan i , in u n, has been c i icized as opaque due o i s
business model and i s unclea bu seemingly close ela-
ionship wi h he US s a e secu i y agencies (e.g.,
B igham, 2020). The e o e, hessenDATA is an in e es ing
case o s udy ega ding whe he and—i so, how— he
wo ldwide expansion o digi al secu i y pla o ms can be
b ough in line wi h egula o y equi emen s.
Wha makes Palan i ’s Go ham so wa e e en mo e ele-
an is ha i enables secu i y au ho i ies o conduc c oss-
da abase esea ch and da a analysis, which is cu en ly a
widely sha ed goal among public agencies a ound he
wo ld (Bigo, 2020; Egbe , 2019; Ga es, 2019; Wilson,
2021). Las bu no leas , Palan i is one o he mos powe -
ul da a analysis fi ms oday—no only in secu i y, bu also
in public heal h and po en ially many o he public se ices,
including social policy and educa ion (Taylo e al., 2020).
Su p isingly, s udies on academic analyses o he company
and i s socio echnical in as uc u e a e s ill a e (excep ions:
B ayne, 2017; Iliadis and Acke , 2022; Munn, 2018). The
ew exis ing s udies ocus p ima ily on Palan i ’s US ac i -
i ies, emphasizing he su eillance dimension o hei so -
wa e, bu hey do no add ess he ques ion (and di ficul y)
o egula ing such pla o ms.
Ou a gumen begins wi h a e iew o he academic dis-
cou se on digi al pla o ms wi h pa icula e e ence o
public secu i y pla o ms and pla o m egula ion. We
nex discuss he implemen a ion o Palan i so wa e in
Ge man police o ces and desc ibe he s udy’s me hodo-
logical app oach. Then we analyze he empi ically obse ed
egula o y ac i i ies and ela ed poli ical conflic s in he
case o hessenDATA. In ou conclusion, we summa ize
ou findings and call o a new egula o y app oach o
da a in eg a ion and analysis pla o ms in public secu i y
ha d aws on liabili y s uc u es as (now) known om
social media pla o ms.
Digi al pla o ms, pla o miza ion, and
egula ion
Unlike he many s udies abou da a-d i en policing ha
use he concep s o “su eillance”,“secu i iza ion”, o he
“indus ial-secu i y complex” o add ess he subjec o ana-
lysis and ha zoom in on he specifici ies o public secu i y
(A adau and Blanke, 2015; Fe guson, 2017; Ulb ich ,
2018), we ocus on cha ac e is ics ha a e ypical o
digi al pla o ms as we assume hey will engende pa icula
egula o y challenges and dynamics.
Digi al pla o ms
Digi al pla o ms a e one o he mos discussed phenom-
ena in cu en academic deba es on digi aliza ion.
Va ious s udies ha e paid a en ion o social media pla -
o ms like YouTube and Facebook and hei cu a ion
powe (Gillespie, 2010; Go wa, 2019) and o he eco-
nomic and wo k- ela ed epe cussions o pla o m com-
panies like Google, Mic oso , Apple, Ai bnb, and Ube
(e.g., Vallas and Scho , 2020). While he phenomena
ela ed o he e m “pla o m”a e e y di e se, hey s ill
sha e some basic ai s: a connec i i y-o ien ed in as uc-
u e ha aims o acili a e in e ac ions by a leas wo hi d
pa ies, a mode o unc ioning based on massi e and
di e se da a, and a modula a chi ec u e (e.g., Ande sson
Schwa z, 2017; Riede and Ho mann, 2020; an Dijck
e al., 2018). Pla o ms a e o en connec ed o no el ways
o mone izing su eillance, da a, and da a-d i en decision-
making (S nicek, 2016; Zubo , 2019). Hence, digi al pla -
o ms can be unde s ood as “in as uc u al a angemen s
ha si ua e digi al ope abili y on p op ie a y sys ems ha
a e, o some deg ee, p og ammable and/o cus omizable
by he sys em use s, making possible one- o mul i-sided
ma ke exchanges”(Ande sson Schwa z, 2017: 375).
Ande sson Schwa z adds ha , as “su aces on which
social ac ion akes place, digi al pla o ms media e—and
o a conside able ex en —dic a e economic ela ionships.”
(Ande sson Schwa z, 2017: 375). Re e ing o his media -
ing capaci y, mos au ho s ag ee ha , al hough many pla -
o m companies y o con ince he public o he wise
(Gillespie, 2010), pla o ms ha e a pe o ma i e dimen-
sion, as hey “do no simply connec social and economic
ac o s bu undamen ally s ee how hey connec wi h
each o he ”( an Dijck e al., 2018: 24, emphasis emo ed).
This is why “a pla o m is a media o a he han an in e -
media y”( an Dijck, 2013: 29).
In summa y, he pla o m li e a u e ag ees ha pla o ms
a e cha ac e ized by h ee p ope ies: da afica ion, modula -
i y, and mul ila e ali y. In addi ion, pla o ms de elop in
2Big Da a & Socie y
ma ke s ha a e o en ansna ional and oligopolis ic and
hei public image e eals a ension wi hin pla o ms as
neu al in e media ies e sus influen ial media o s.
Digi al pla o ms o public secu i y
Gi en his ension wi hin pla o ms as in e media ies
and media o s, i is c ucial o ake a close look a digi al
pla o ms in he public sec o .
3
As media o s, hey gene a e
s ee ing knowledge o s a e agen s, whose decision-
making p ac ices a e mos ly add essed o ci izens and
o en a ec hei ci il igh s. Because s a e agencies a e
able o make decisions ha impac ci izens’li es, in
mode n democ acies hei powe s need o be legi ima e
and subjec o many legal and e hical no ms (Bigo e al.,
2011). A new le el o isk a ises when p i a e ac o s a e
b ough on boa d in he decision-making o decision-
suppo p ocess. This may occu in he o m o public-
p i a e pa ne ships, which a e commonly used in digi al
echnology p ojec s (Bossong and Wagne , 2017). He e,
p i a e co po a e ac o s ge access o da a ha should ypic-
ally be he sole p ese e o s a e ac o s and is hence subjec
o s ic ules (Ans is, 2021: 9). Mo eo e , when complex
echnologies a e he basis o a p i a e company’s se ice,
an in o ma ion asymme y may a ise be ween he
company and he s a e agency (Ans is, 2021: 9). Da a in e-
g a ion and analysis pla o ms also b ing abou epis emic
shi s in public agencies ha need o be c i ically sc u i-
nized. In he public secu i y field, such epis emic shi s
include he use o big da a, uns uc u ed da a, and heu is ics
ha end o ely on co ela ions and good p edic o s ins ead
o causal heo y (e.g., Amoo e, 2013). These aspec s ha e
been hea edly discussed o se e al yea s, as hey ha e
mani old implica ions o he ule o law in mode n democ-
acies (e.g., Fe guson, 2017). They a e especially ele an
o da a in eg a ion and analysis pla o ms, as hei key
aim is o enhance he in e ope abili y be ween da abases
o “sma analysis”, he “so-called sma way o connec
he do s and o a oid con inuing o wo k in ‘silos’wi h
segmen ed in o ma ion ne wo ks”(Bigo, 2020: 410). As a
consequence, he egula ion o public secu i y pla o ms
dese es mo e academic a en ion. Pla o m egula ion is
he e o e a ui ul esea ch opic.
Pla o m egula ion
Rele an esea ch has poin ed o he many isks associa ed
wi h he ad en o digi al pla o ms, o example, he p o-
li e a ion o illegal o undesi ed online con en and he
disc imina ion, manipula ion, and exploi a ion di ec ed
agains use s and wo ke s alike. O he p oblems o pla -
o ms include isks linked o da a p o ec ion, p i acy,
disc imina ion, and social so ing due o in asi e and
la ge-scale su eillance; he limi ed anspa ency, obse a-
bili y, and accoun abili y o da a- ela ed p ac ices o la ge
online pla o ms; and he lack o cla i y on go e nmen s’
da a so e eign y in collabo a ions wi h (o en o eign) ech-
nology companies (Helbe ge , 2020; an Dijk e al., 2018).
These isks and p oblems ha e led o a ious calls o s ic-
e egula ion o pla o m-based companies and se ices
(Hildeb and , 2020; Pasquale, 2015).
Much engagemen wi h he ques ion o pla o m egula-
ion has occu ed in ecen yea s; he e ha e been many
egula o y ini ia i es by legisla i e bodies such as he
Eu opean Commission,
4
inc eased judicia y ac i i y,
5
and
in ensi e academic esea ch. The o e all impe us is o
make pla o ms accoun able o p ac ices ha ake place
wi hin hei s uc u es and o o ce hem o ake measu es
o p e en ha m (Go wa, 2019). Typical egula o y ins u-
men s associa ed wi h his in e media y liabili y a e com-
panies’ epo ing du ies owa ds he public o owa ds
o e sigh au ho i ies, epo ing and flagging sys ems o
use s, no ice-and- akedown p o ocols, and sanc ions o
noncompliance (Eu opean Commission, n.d.).
Howe e , egula ion is no easy: Schola s ha e no ed
many obs acles o pla o m egula ion and poin ed ou
ha egula ion is o en inadequa e o ine ec i e (Pasquale,
2015; Zubo , 2019). One se o obs acles o pla o m
egula ion conce ns he p ope ies o digi al echnologies,
o example, he ac ha many digi al se ices a e highly
opaque, making i ha d o ace which da a a e collec ed,
s o ed, and used by whom and o wha pu pose. This
opaci y can be oo ed in he complexi y o digi al p ac ices,
which may ely on semi- o unsupe ised machine lea ning.
Opaci y, especially when i comes o secu i y p ac ices, can
also o igina e om business o s a e sec ecy, as companies
and s a e agencies o en p o ec hei da a- ela ed p ac ices
om public sc u iny and he e o e om legal and demo-
c a ic con ol (Bloch-Wehba, 2021). Limi ed accoun abil-
i y is a amilia p oblem o public-p i a e pa ne ships
(Bo ai d, 2004). The o he se o obs acles o pla o m
egula ion s ems om he socie al condi ions in which
digi al pla o ms a ise: he monopolis ic ea u es o digi al
ma ke s p e en igh s-p ese ing business models om
flou ishing (Colangelo and Maggiolino, 2018) and o e -
sigh agencies a e unde s a ed such ha egula ion is no
p ope ly implemen ed (Jo i, 2015).
Bellano a and de Goede igh ly s a e ha (2022: 107),
“conce ning secu i y algo i hms as a a ge o egula ion,
he e is [s ill] a need o g ea e unde s anding o how he
ope a ionaliza ion and legal en o cemen o alues (…)
ake place in and h ough da a a chi ec u es.”Fu he o
his, we mus add ha analyses ha s udy conc e e p ac ices
and no jus o mal no ms a e needed o unde s and pla -
o m egula ion. A p esen , mos esea ch abou pla o m
egula ion ocuses on he o mal no ms o egula ion,
such as exis ing and p oposed legisla ion (Hildeb and ,
2020) and he e ms, condi ions, and codes o conduc
issued by echnology companies (Go wa e al., 2020;
Pasquale, 2015).
Ulb ich and Egbe 3
S udies ha ha e sc u inized he ac ual p ac ices o pla -
o m egula ion, including legisla i e p ocesses, day- o-day
ule implemen a ion, and moni o ing, a e e y a e. A ew
empi ical s udies ha e indica ed ha echnology companies
o en ha e a la ge leeway o define whe he and how hey
comply wi h egula ion and ha hey a e lenien owa ds
hemsel es; his is e iden wi h ega d o he sel -
moni o ing o ule en o cemen (Gille e al., 2022), and
wi h ega d o assessing and p e en ing acism (Siape a
and Viejo-O e o, 2021). O he s udies ha e analyzed legis-
la ion in esponse o pla o ms—pe aining o sha ing
economy companies such as Ube and Ai bnb (Aguile a
e al., 2021)—a he na ional and subna ional le el,
finding ha he s ic ness o s a e egula ion depends upon
he o e all egula o y s ic ness in a gi en policy a ea,
and on domes ic powe cons ella ions (Go wa, 2021;
Lau e and Seidl, 2021). Pla o m companies s ongly
oppose egula ion and use a ious lobbying s a egies o
de ea i : Fo example, hey lobby agains new egula ion
(Mazu and Se afin, 2023) and in a o o co po a e sel -
egula ion (Medzini, 2022), hey e use o collabo a e in
implemen ing egula ions (Colomb and Mo ei a de Souza,
2023), hey engage in discu si e legi ima ion o hei ac i -
i ies (Chan and Kwok, 2022), and hey mobilize use s and
cus ome s in he co po a e in e es (Ya es, 2023).
Socie al ac o s ou side he company a e usually no
in ol ed in egula o y implemen a ion (Bloch-Wehba,
2022).
6
The abo e-men ioned s udies ela e o s a e con ol
o e echnology companies; e y ew s udies add ess cases
whe e s a e ac o s make use o echnology se ices and pla -
o ms. The e is one excep ion, namely a s udy abou digi al
go e nance in he sma ci y o Shenzhen (China), which
showed ha g owing da a cen aliza ion educes he
con ol ha public se an s a lowe adminis a i e le els
ha e o e ci izen da a. The s udy also showed ha he in e-
g a ion o ci izen da a a ac s echnology companies
(G oße-Bley and Kos ka, 2021). The de ails o da a go e n-
ance and egula ion we e, howe e , no comp ehensi ely
add essed in he s udy.
We complemen hese s udies by analyzing he ules and
p ac ices o egula ion in a specific case o a public pla -
o m, mo e conc e ely in he case o police pla o miza ion.
In his con ibu ion, we unde s and egula ion as s a e
con ol—o e companies, bu also o e public agencies,
and o he socie al ac o s (Koop and Lodge, 2017). The dis-
inc i e ea u e o public sec o pla o m egula ion is ha i
implies wo le els o con ol: fi s , s a e agencies con ol-
ling hei co po a e con ac o s, and second, s a e agencies
( hose in cha ge o egula ion and o e sigh ) con olling
o he s a e agencies ( hose who a e in cha ge o he echnol-
ogy p ojec ). In p inciple, public pla o ms he e o e imply
bo h o ms o egula ion: adi ional egula ion be ween
s a e agen s and companies; and s a e sel - egula ion, wi h
i s many p oblema ic implica ions. S udying pla o m egu-
la ion wi h ega d o public sec o pla o ms is he e o e
concep ually in e es ing because i gi es insigh in o how
his sel - egula ion plays ou wi hin he s a e. Simila
laissez ai e mechanisms may be e iden he e as in co po -
a e sel - egula ion bu he e may be en i ely di e en ones a
play. In p inciple, in democ acies, many ins i u ional
mechanisms a e aimed a cons aining he powe o s a e
agencies and es ablishing democ a ic con ol. These
include a ious o ms o powe sepa a ion, public anspa -
ency obliga ions, and cons i u ional and ci il igh s.
We now u n o analyzing a case whe e public agencies
decided o closely coope a e wi h a co po a e pla o m and
had o find ways o comply wi h egula o y needs ha
esul ed in hei u iliza ion o a da a in eg a ion and analysis
pla o m.
Pla o mized public secu i y in Ge many
Ou s udy ocuses on he da a in eg a ion and analysis pla -
o m Go ham by US-based Palan i Technologies and i s
u iliza ion by police o ces, wi h special emphasis on
he case o hessenDATA in he Ge man ede al s a e o
Hesse. In his sec ion, we will con ex ualize he case o
hessenDATA in ligh o o e all da afica ion ends in
Ge man policing.
Me hodological app oach
To unde s and how he hessenDATA p ojec was imple-
men ed, wha egula o y p oblems a ose, wha egula o y
ac i i ies ook place, and wha discou ses we e ele an ,
we esea ched and analyzed a numbe o publicly a ailable
documen s. The comple e lis includes 85 documen s,
mainly pa liamen a y and go e nmen documen s, bu
also media a icles wi h ele an quo es by pe sons in ol ed
in he egula o y p ocess (see Annex). In addi ion, o alid-
a e ou obse a ions and fill he gaps in he publicly a ail-
able ma e ial, we conduc ed a semi-s uc u ed in e iew
wi h he p ojec eam o hessenDATA, which included a
demons a ion o he pla o m, and h ee guided in e iews
wi h membe s o he pa liamen a y opposi ion who we e
pa o he pa liamen a y commi ee ha in es iga ed he
ende ing p ocess (see he in e iew lis in he Annex).
These in e iews especially ocused on ep esen a i es o
opposi ion pa ies, as we assumed ha hei pe spec i e
would no be su ficien ly ep esen ed in he o ficial docu-
men s, which e y closely eflec ed he go e nmen ’s
aming. We combined he a ious ypes o da a o ob ain
a comp ehensi e pic u e o he implemen a ion p ocess o
hessenDATA and he (poli ical) discussion ha ook place
a ound i .
The documen s and he ansc ip s o he in e iews we e
subjec ed o a quali a i e con en analysis by bo h au ho s
and coded un il code sa u a ion was assumed o be
achie ed. The coding p ocess was based on he con en ana-
lysis me hod acco ding o Kucka z and Rädike (2023),
4Big Da a & Socie y
which combines bo h induc i e and deduc i e coding s a -
egies. This mean ha he analysis was su ficien ly ocused
on he opic a hand bu was, a he same ime, su ficien ly
open o unexpec ed insigh s o eme ge om he empi ical
da a. Hence, despi e i s hema ic ocus, he analysis was
able o shi he esea che s’p econcep ions and gene a e
new knowledge abou he subjec . Fo ins ance, he ocus
on egula o y challenges and conflic s was no an icipa ed
a he beginning o he analy ical p ocess; i eme ged in
he cou se o he mo e de ailed analysis o he da a, as
hese sugges ed ha he hessenDATA pla o m is pa icu-
la ly di ficul o egula e. This, we will show, has o do
wi h i s pla o m cha ac e .
Palan i Go ham in Ge man policing
The mos p ominen cu en end in police da afica ion
in Ge many is he inc easing use o da a in eg a ion and
analysis pla o ms by police o ces. Hesse was he fi s o
use he se ices o US-based Palan i Technologies and
i s Go ham so wa e. The Hessian police ha e been using
Palan i Go ham since he summe o 2017 (Be e ungen,
2021). Soon, o he Ge man egions ollowed: s a ing
in 2021, police in No h Rhine-Wes phalia pilo ed a
“sys em o c oss-da abase analysis and esea ch”(called
Da enbankübe g ei ende Analyse und Reche che in Ge man,
o DAR o sho ), also based on he Palan i Go ham
pla o m (Landesk iminalam No d hein-Wes alen, 2020),
which has been in ope a ion since sp ing 2022. The
Ba a ian police ha e a p ojec called he “C oss-P ocedu al
Sea ch and Analysis Pla o m”(Ve ah ensübe g ei ende
Reche che und Analyse in Ge man, o VeRA o sho ),
which led o a con ac wi h Palan i in sp ing 2022
(Baye isches Landesk iminalam , 2022). Al hough he o i-
ginal plan was o make VeRa (as “VeRa Bund“)a ailable
o all o he ede al s a es in Ge many ia a amewo k
ag eemen , coo dina ed by he Fede al Minis y o he
In e io , he Fede al Minis e o he In e io ecen ly
decided o s op his plan o now and o ins ead aim o
an in-house sys em (Zie e e al., 2023).
Founded in 2004, Palan i is a digi al echnology
company commonly ega ded as ema kably sec e i e and
publicly con o e sial, no leas due o i s delibe a ely
opaque appea ance and he con o e sial image o he com-
pany’s in es o Pe e Thiel
7
(e.g., Cha kin, 2021). The main
se ice Palan i o e s o secu i y agencies is i s da a in e-
g a ion and analysis so wa e Go ham. The company
desc ibes he scope o i s so wa e as ollows:
Ou p oduc s se e as he connec i e issue be ween an
o ganiza ion’s da a, i s analy ics capabili ies, and ope -
a ional execu ion. Palan i ’s pla o ms ie hese oge he
by b inging he igh da a o he people who need i , allow-
ing hem o make da a-d i en decisions, conduc sophis i-
ca ed analy ics, and efine ope a ions h ough eedback
(Palan i , 2020b). As we will illus a e in he ollowing
sec ion, Palan i Go ham elies upon da afica ion, modula -
i y, and mul ila e ali y— he eby quali ying as a pla o m.
Like he ope a o s o he digi al pla o ms desc ibed in
he li e a u e (Gillespie, 2010), he p o ide s o da a in e-
g a ion and analysis pla o ms end o cha ac e ize hei
so wa e as agnos ic and neu al, o en denying he media o
s a us o hei pla o ms. In his ein, Palan i s a es ha
hey “a e no a da a company”(Palan i , 2020b), since
hey do no use da a om hei clien s o hei pu poses.
Ins ead, hey p esen hemsel es as a “so wa e company”
ha is “build(ing) digi al in as uc u e o da a-d i en
ope a ions and decision-making.”Ou analysis will show
he dis inc i e egula ion- ela ed challenges connec ed o
his o m o pla o m policing.
A pla o miza ion pionee in Ge many:
hessenDATA
Since he summe o 2017, he police in he s a e o Hesse
ha e been using Palan i ’s Go ham so wa e, unde he
name hessenDATA, o gene a e ime-c i ical in o ma ion
by means o c oss-da abase esea ch, o access he e oge-
neous sou ces and co ela ion-based con ex analyses, and
o implemen his in o ma ion di ec ly in police ope a ions
and s a egies (Hessisches Minis e ium ü Inne es und
Spo , 2018: 59). The main goal is o make he policing
o “ e o is s and se ious c iminals”mo e e ec i e
(Hessische Land ag, 2019a: 1).
The p og am was implemen ed compa a i ely quickly
since he ele an au ho i ies publicly claimed ha he e
was a su ficien ly conc e e isk o e o is a acks in
Hesse a ha ime and a need o quickly make he
p og am ope a ional. Ins ead o issuing a call o ende ,
as legally equi ed, he con ac was awa ded di ec ly o
Palan i o a limi ed ime pe iod on he g ounds o
special u gency; i was la e g an ed pe manen s a us
(Hessische Land ag, 2019b: 20).
The goal o hessenDATA is o enable human analys s o
find associa ions be ween en i ies (e.g., people, spaces,
objec s) wi h a iew o p e en ing e o is a acks o disco-
e ing o ganized c ime ne wo ks. The sales a gumen was
ha an in es iga o in e es ed in a suspec could ei he
spend all hei ime ollowing ha pe son, o hey could
se up an ale in he so wa e. Acco ding o he 2018
annual epo o he Hessian Minis y o he In e io and
Spo , be o e hessenDATA, much da a p ocessing was
done manually and equi ed he aw da a o be o wa ded
o IT specialis se ices a he Hessian S a e C iminal
Police O fice, whe e he da a we e p ocessed wi hin a
week o so (Hessisches Minis e ium ü Inne es und
Spo , 2018). The use o hessenDATA pu po edly made
i possible o cu ou hese ime-consuming manual s eps
and ob ain a comple e and s uc u ed o e iew o he da a
in minu es (In e iew Police Hesse).
Ulb ich and Egbe 5
The hessenDATA pla o m wo ks as a “d agne ,”as i
allows he ela ionships be ween people, objec s, and
places o be ep esen ed in he o m o a ne wo k (see
also B ayne, 2017). In es iga o s can use a sea ch in e ace
(see Figu e 1) o analyze his web o ela ionships om any
node. Fo example, in es iga o s ecei e a message when
ce ain en i ies (names o people o places, keywo ds, and
objec s) appea in a su eilled pe son’s elecommunica-
ions. Palan i o e s addi ional modules ha can be added
o Go ham, such as A a, a se o AI unc ions, and
Dossie , a ool o in e -o ganiza ional collabo a ion. In
Hesse, addi ional unc ions ha e been added a he
police’s eques , such as he in eg a ion o p ofile pic u es
and mugsho s. A mobile e sion has also been es ablished
(In e iew Police Hesse).
Significan ly, his ype o analysis pla o m is cha ac e -
ized by he “desiloza ion”o da abases, ha is, he b eaking
up o “da a silos,”allowing o fice s o access di e en da a
banks wi hin one pla o m om one cen al i ual place.
Associa ed wi h his, he e is he po en ial o link nume ous
sou ces o da a (in e nal police da a and ex e nal da a),
which enables analyses a high speed. Re e encing
hessenDATA’s so wa e a chi ec u e, some commen a o s
ha e ap ly spoken o a “Robocop-Google”(B ühl, 2018)
(see also Figu e 1). Fo police ope a o s, hessenDATA’s
desiloza ion unc ion is o u mos impo ance since i
makes i much easie o access da a om di e en da a
banks. In doing so, i enhances he in e ope abili y o di e -
en da a banks and sys ems.
The ollowing police and nonpolice da a sou ces a e
included in hessenDATA analyses: POLAS, CRIME,
and ComVo . POLAS (POLizeiAuskun sSys em [Police
In o ma ion Sys em]) s o es law en o cemen da a. CRIME
(C iminal Resea ch In es iga ion Managemen So wa e)
is a p e en i e da abase, a case-p ocessing sys em whe e
da a necessa y o in es iga i e p ocedu es in he u u e a e
s o ed. ComVo (Compu e ges ü z e Vo gangsba bei ung)
is he s anda d case-p ocessing sys em, whe e all police
p ocedu es a e kep and whe e pa ol o fice s documen
hei ac ions. Fu he mo e, hessenDATA can also access
a fic da a om elecommunica ions su eillance and da a
om elecommunica ions p o ide s. In addi ion, da a om
seized cell phones can be in eg a ed ( o ensic ex ac s). O
pa icula in e es a e he elephone numbe s and communi-
ca ion da a, such as he ime o calls, he leng h o calls, and
he people called (who can hen be analyzed wi h he ocus on
associa ions). Police elexes, an in e nal police email sys em
o o mal communica ion, a e also included in
hessenDATA; hese a e especially used o in es iga e c im-
inal cases o sha e impo an news, such as he a es o a
high-le el suspec . Finally, hessenDATA can also access
da a om social media si es. This includes bo h public and
nonpublic da a— o example, in o ma ion ob ained by
Facebook h ough a mu ual legal assis ance eques o he
Uni ed S a es. Howe e , hessenDATA only has access o
hese ne wo ks in indi idual cases, a e a cou o de .
Summing up, hessenDATA is cha ac e ized by da afica-
ion, modula i y, and mul ila e ali y. Mul ila e ali y esul s
Figu e 1. S a sc een hessenDATA (Sou ce: Au ho pho o).
6Big Da a & Socie y
om he linkage o da abases and, hence, da a. Palan i
Go ham also links di e en police o ces: O fice s om
di e se uni s main ain and e alua e da a, some imes beyond
he egion o Hesse. One consequence o his is ha analys s
in he back o fice a e gi en g ea e powe han police o fice s
in he field, whose supe io knowledge abou he local con ex
and g ea e field expe ience is ul ima ely de alued (B ayne,
2021: 77; Wilson, 2017: 118). The linking o da abases is
also a egula o y p oblem in iew o ea ma king p inciples
and he di e en legal unde pinnings o he da abases, as
will be sc u inized in he ollowing sec ion.
Analysis: egula o y conflic s and
s a egies in he case o hessenDATA
Gi en ha new and in eg a ed pla o ms o en aise new
ques ions when hey a e implemen ed, i is no su p ising
ha he es ablishmen o hessenDATA c ea ed a ious
egula o y conflic s. Agains his backd op, in ou empi ical
analysis, we will engage wi h hese egula o y conflic s in
mo e de ail, especially wi h he conce ns aised by he
implemen a ion o hessenDATA. The conce ns we e linked
o he h ee defining ea u es o pla o ms—da afica ion, mul-
ila e ali y, and modula i y— igge ing conflic s abou da a
p o ec ion, ci il igh s p o ec ion, and he o e all anspa -
ency o he p ojec . Ano he conce n was linked o who
would ake esponsibili y o he ac i i ies wi hin he pla o m
and whe he Palan i could be conside ed an in e media y;
he hi d se o conflic s was abou ai ma ke compe i ion
and he p ocu emen p ocess ha led o he con ac wi h
Palan i .
The de ende s o he hessenDATA p ojec we e mos ly
ep esen a i es o he Minis y o he In e io , he Hessian
police, and elec ed membe s o pa liamen o he majo i a -
ian Ch is ian Democ a ic Union (CDU). C i icisms o he
p ojec we e mainly oiced by he opposi ion pa ies in
he Hessian pa liamen and by ci il igh s o ganiza ions.
Da afica ion, mul ila e ali y, modula i y,
and conflic s abou da a p o ec ion and
ci il igh s
Da afica ion and conflic s abou da a p o ec ion. The main
egula o y challenge o he HessenDATA p ojec emana ed
om da afica ion and he ela ed da a p o ec ion p oblems.
One o he majo conce ns o he pa liamen a y opposi ion
was ha hessenDATA could os e po en ially unlaw ul
o ms o s a e su eillance. The op ion o combine
a ious da a sou ces p omp ed pa icula ea s ha any
ci izen could become a a ge o su eillance and suspicion
(Hessische Land ag, 2019b: 2). In addi ion, he po en ial
o in eg a ing social media da a aised conce ns ha all
kinds o in o ma ion could be deemed ele an o c iminal
in es iga ions and lead o a huge inc ease in he numbe s o
innocen ci izens unde su eillance (Hessische Land ag,
2019a: 2). To add ess hese conce ns, he go e nmen in e-
g a ed a new pa ag aph in o he exis ing public secu i y law
o explici ly allow au oma ed da a analysis o pe sonal da a
in cases o public in e es .
8
The aim was o c ea e a legal
basis o hessenDATA, because da a p o ec ion law gene -
ally o bids da a mining o sensi i e pe sonal in o ma ion
(Hessische Land ag, 2019b: 2). Da a mining, as defined
by he Ge man Fede al Cons i u ional Cou , is dis inc
om simple da a analysis because i combines massi e
and mul iple da a sou ces o c ea e unexpec ed insigh s,
he eby c ea ing “new knowledge”. Acco ding o he
Fede al Cons i u ional Cou , in i s in e p e a ion o he
Ge man Fede al Da a P o ec ion Ac , da a mining poses a
h ea o a ious ci il igh s and he e o e mus only be
used o achie e impo an aims, o example, o deal wi h
an imminen h ea , and mee specific equi emen s. The
Fede al Da a P o ec ion Ac hus es ablishes mo e s ingen
c i e ia o da a mining han he Eu opean Da a P o ec ion
Law En o cemen Di ec i e (EU) 2016/680, which i imple-
men s.
9
Acco dingly, he new pa ag aph o he Hessian Law
on Public Secu i y and O de allows o au oma ed analysis
o connec ed da a o p e en se ious c imes (such as co up-
ion, child po nog aphy, mu de , e c.) and o wa d o exis -
en ial h ea s o he na ion. Howe e , a collec i e o lawye s
and ci il igh s o ganiza ions submi ed a cons i u ional
complain agains he new legal pa ag aph in 2019,
a guing, among o he hings, ha he condi ions o da a
access and analysis we e oo b oad. In ea ly 2023, six yea s
a e he fi s implemen a ion o hessenDATA, he Fede al
Cons i u ional Cou uled mos ly in a o o he
plain i s (Bundes e assungsge ich , 2023). Consequen ly,
hessenDATA is cu en ly ope a ing on he basis o an uncon-
s i u ional law, and he Hessian go e nmen has s a ed he
p ocess o modi ying he ele an passages (Voig s, 2023).
Mul ila e ali y: da a-d i en policing and ci il igh s p o ec ion.
The ac ha hessenDATA is mul ila e al and connec s
da a (banks), bu also a ious uni s wi hin he Hessian
police and po en ially links he Hessian police wi h o he
Ge man police uni s—a ec ing ci izens and socie y as a
whole—has igge ed conce ns abou how well ci izen
igh s would be p o ec ed wi h ega d o da a p o ec ion,
he p esump ion o innocence, he eedom o mo emen ,
and p o ec ion agains disc imina ion. While he pa liamen-
a y opposi ion and (c i ical) jou nalis s pain ed a pic u e o
a socie y cha ac e ized by pe asi e da a-d i en policing,
enzied p osecu ions, and au ho i a ian aspi a ions
(Hessische Land ag, 2019b: 19), he go e nmen jus ified
hessenDATA by unde sco ing i s use ulness o p osecu -
ing Islamic e o is s and p e en ing u u e e o is
a acks. In hese jus ifica ions, i e e ed o he publicly
no ed lack o collabo a ion be ween police uni s in
Ge many in he wake o e o is a acks like he one pe pe-
a ed in Be lin in Decembe 2016.
Ulb ich and Egbe 7
Modula i y: anspa ency p oblems and modula da a
p o ec ion. The modula i y o hessenDATA made i p ob-
lema ic o he public and he pa liamen a y opposi ion o
keep ack o he p ojec . The unclea scope o da a collec-
ion and use and hei socie al e ec s made hessenDATA a
opic o public con o e sy, and he modula s uc u e o
Palan i ’s se ices was pe cei ed by he pa liamen a y
opposi ion as an obs acle o financial anspa ency. They
c i icized he ailu e o publish he o e all p ojec budge
and suspec ed an o e un o he budge ini ially s ipula ed
in he p ocu emen p ocess as addi ional se ices we e
added o he ini ial se ice desc ip ion. A he same ime,
he modula s uc u e o Palan i ’s se ices was also used
as an a gumen in a o o he p ojec , mainly as an
elemen o he da a p o ec ion s a egy: The go e nmen
s essed ha Palan i ’s o e o ailo a modula da a p o ec-
ion and da a secu i y sys em o Hessian police’s needs had
been one o he main easons o choose Palan i (In e iew
Police Hesse). Palan i o e s o de elop da a p o ec ion
concep s o i s cus ome s, i has an in-house P i acy and
Ci il Libe ies Enginee ing Team, and i o e s da a
con ol nudges and checkpoin s in i s so wa e (Palan i ,
2020a). Acco ding o he Hessian police, hessenDATA
has a ious s uc u al and p ocedu al ea u es o ensu e
da a p o ec ion and in o ma ion secu i y in daily da a-
in eg a ion and da a-analysis p ac ices. One elemen con-
ce ns he s uc u e o he da a bank: hessenDATA limi s
da a flows om police da a o hessenDATA, hus making
su e ha he o iginal police da a banks emain unchanged
(In e iew Police Hesse). In addi ion, he da a used by
hessenDATA is s ill held on police se e s and supposedly
does no go h ough Palan i se e s; hessenDATA is no
connec ed o he in e ne bu only o he police in ane
(“Sonde ne z”) (Hessische Land ag, 2019b: 19); in add-
i ion, he da a ed in o hessenDATA a e dele ed a e wo
yea s (In e iew Police Hesse). Ano he da a p o ec ion
module acili a es he con ol o use s by a sys em o
access ules and con ol p ocedu es (based on a “ ole-based
access con ol”model). Any police analys who wan s o
use he sys em has o eques access and p o ide easons
ha jus i y he use o hessenDATA, he eques mus be
p edica ed on in es iga ing o p e en ing some so o
se ious c ime, and hessenDATA canno be used o o he
in es iga ions. Access is g an ed o a limi ed pe iod o
ime and can be es ic ed o subse s o he da a. Use s
ecei e a wo-day aining cou se p io o hei fi s use o
he sys em. A logging sys em eco ds all ac i i y in he
sys em, and andom checks o app op ia e use ake place
(In e iew Police Hesse).
In e media y esponsibili y
Did he Hessen go e nmen hold Palan i accoun able o
possible isks o ha ms wi h ega d o he pla o m? The
da a p o ec ion and ci il igh s conce ns led o a deba e
on whe he he go e nmen and Palan i we e exhibi ing
su ficien esponsibili y o he p ojec and we e accoun -
able o he public. One aspec ha a ac ed a en ion was
he lack o cla i y on whe he hessenDATA was connec ed
o he in e ne . A hessenDATA ope a o es ified ha i is
indeed possible o in eg a e da a om an in e ne sea ch
in o hessenDATA (In e iew Police Hesse). The 2018
annual epo o he Hessian Minis y o he In e io also
s a ed in a sec ion on hessenDATA ha ma ching wi h
in o ma ion openly iewable on he in e ne , such as om
social ne wo ks, was possible (Hessisches Minis e ium ü
Inne es und Spo , 2018: 58). These s a emen s emain
ambiguous, as no consis en di e en ia ion is made
be ween publicly iewable and non-publicly- iewable
in o ma ion om social ne wo ks and o he sou ces. Also,
he s a emen s do no explain how publicly iewable da a
se s a e disco e ed by he so wa e and how hey a e
added o hessenDATA o whe he his is pa o Palan i
Beagle’s scope, which is desc ibed by hessenDATA o fi-
cials as analysis so wa e o he public a eas o social ne -
wo ks (Hessisches Minis e ium ü Inne es und Spo ,
2018: 58). Beagle unc ions as an in eg a ed solu ion o
sea ches in social ne wo ks and in o he open sou ces ig-
ge ed by indi idual cases. Ano he ma e o public conce n
was whe he Palan i could access police da a in o de o
imp o e i s p oduc s o e en o sha e i wi h US in elligence
agencies. In esponse, he Hessian go e nmen said i had
es ablished a sys em o egula ion and o e sigh , bu i s
de ails a e no accessible o he public. The go e nmen
has, o ins ance, s essed ha he ele an da a p o ec ion
and in o ma ion secu i y ules a e o malized in guidelines
(In e iew Police Hesse); howe e , hese ha e no been
sha ed wi h he public o wi h he pa liamen a y opposi ion
(In e iew opposi ion pa y 1, 2, 3).
Wi h ega d o o e sigh , hessenDATA is egula ly
e iewed by he c iminal police’s in e nal da a p o ec ion
o fice and by Hesse’s independen da a p o ec ion au ho -
i y (In e iew Police Hesse), bu , again, he ele an
e iew epo s ha e no been made public. We con ac ed
he da a p o ec ion au ho i y, bu i did no make anyone
a ailable o alk o us abou i s ongoing assessmen o
hessenDATA; indeed, he au ho i y did no make any
public s a emen s on he ma e h oughou he whole
p ocess, wi h he excep ion o hose made in he o al hea -
ings be o e he Fede al Cons i u ional Cou , whe e he
da a p o ec ion au ho i y was summoned as an expe
wi ness (De Hessische Beau ag e ü Da enschu z und
In o ma ions eihei , 2022). This a i ude con as s wi h
ha o he independen da a p o ec ion au ho i y in
No h Rhine-Wes phalia, which c i icized he use o he
Go ham-based DAR a an ea ly s age—on he g ounds
ha i ep esen ed da a mining and ha i allowed he
use o da a collec ed o ano he pu pose, which would
equi e i s own legal basis (Land ag No d hein-Wes alen,
2021). The opposi ion pa ies in Hesse s a ed ha he
8Big Da a & Socie y
Vallas S and Scho JB (2020) Wha do pla o ms do?
Unde s anding he gig economy. Annual Re iew o Sociology
46(1): 273–294.
an Dijck J (2013) The Cul u e o Connec i i y: A C i ical His o y
O Social Media. Ox o d ; New Yo k: Ox o d Uni e si y P ess.
an Dijck J, Poell T and Waal Md (2018) The Pla o m Socie y:
Public Values in a Connec i e Wo ld. New Yo k: Ox o d
Uni e si y P ess Inc.
an Dijk N, Tanas A, Romme ei K, e al. (2018) Righ enginee -
ing? The edesign o p i acy and pe sonal da a p o ec ion.
In e na ional Re iew o Law, Compu e s & Technology
32(2–3): 230–256.
Voig s H (2023) Hessen: Eine Gese zes e o m ü die Wei e nu zung
on „Palan i “.F ank u e Rundschau, 21 Feb ua y. A ailable
a : h ps://www. .de/ hein-main/landespoli ik/hessen-eine-gese
zes e o m- ue -die-wei e nu zung- on-palan i -92101651.h ml.
Wilson D (2017) Algo i hmic pa ol. The u u es o p edic i e
policing. In: Za šnik A (ed) Big Da a, C ime and Social
Con ol. 1s ed. Cham, Swi ze land: Rou ledge, 108–127.
A ailable a : h ps://www. aylo ancis.com/books/
9781315395777 (accessed 9 May 2020).
Wilson D (2021) The new pla o m policing. In: Za šnik A and
Badalic
V (eds) Au oma ing C ime P e en ion, Su eillance,
and Mili a y Ope a ions. Cham, Swi ze land: Sp inge
In e na ional Publishing, 47–68. A ailable a : h ps://link.
sp inge .com/10.1007/978-3-030-73276-9_3 (accessed 18
No embe 2022).
Ya es L (2023) How pla o m businesses mobilize hei use s and
allies: Co po a e g ass oo s lobbying and he ai bnb ‘mo e-
men ’ o de egula ion. Socio-Economic Re iew 21(4): 1917–
1943.
Zie e M, Ka heuse B, Schö el R, e al. (2023) Bund ück on
So wa e Palan i ab. agesschau.de. A ailable a : h ps://www.
agesschau.de/in es iga i /b - eche che/palan i -so wa e-analyse-
polizei-100.h ml.
Zubo S (2019) The Age o Su eillance Capi alism: The Figh o
a Human Fu u e a he New F on ie o Powe . New Yo k:
PublicA ai s.
Ulb ich and Egbe 15
