ArticlePDF Available

Caught inside the black box: Criminalization, opaque technology, and the New York subway MetroCard


Abstract and Figures

This article investigates how an account of hidden, internal properties of an everyday technology became a framework to interpret human action as a serious crime. Using a case study situated in the New York subway system, I examine the criminalization of a practice of New York’s poor known as “selling swipes” performed by so-called “swipers”. A high court came to classify the practice as felony forgery, interpreting it though an expert-witness account of how objects physically manipulated by swipers interact with a secretive, proprietary digital information system. Thousands of felony arrests – overwhelmingly of nonwhite men – have been legitimated under this theory, in which the crime occurs on a plane of technical interactions to which swipers have no access. Through close examination of the underlying technology (known as MetroCard), however, I show considerable problems in the authorities’ understanding of the technology, illustrating the hazards of interpreting human action through proprietary or complex systems, especially as they are represented solely through expert accounts. The case demonstrates fresh connections between technology and unequal outcomes in the U.S. criminal justice system, and suggests an emerging form of social vulnerability, to interpretations of our actions through the logic of technologies black-boxed to us.
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
The Information Society
An International Journal
ISSN: 0197-2243 (Print) 1087-6537 (Online) Journal homepage:
Caught inside the black box: Criminalization,
opaque technology, and the New York subway
Noah McClain
To cite this article: Noah McClain (2019): Caught inside the black box: Criminalization,
opaque technology, and the New York subway MetroCard, The Information Society, DOI:
To link to this article:
Published online: 25 Jul 2019.
Submit your article to this journal
View Crossmark data
Caught inside the black box: Criminalization, opaque technology,
and the New York subway MetroCard
Noah McClain
Department of Social Sciences, Illinois Institute of Technology, Chicago, Illinois, USA
This article investigates how an account of hidden, internal properties of an everyday
technology became a framework to interpret human action as a serious crime. Using a case
study situated in the New York subway system, I examine the criminalization of a practice
of New Yorks poor known as selling swipesperformed by so-called swipers. A high court
came to classify the practice as felony forgery, interpreting it though an expert-witness
account of how objects physically manipulated by swipers interact with a secretive, propri-
etary digital information system. Thousands of felony arrests overwhelmingly of nonwhite
men have been legitimated under this theory, in which the crime occurs on a plane of
technical interactions to which swipers have no access. Through close examination of the
underlying technology (known as MetroCard), however, I show considerable problems in
the authoritiesunderstanding of the technology, illustrating the hazards of interpreting
human action through proprietary or complex systems, especially as they are represented
solely through expert accounts. The case demonstrates fresh connections between
technology and unequal outcomes in the U.S. criminal justice system, and suggests
an emerging form of social vulnerability, to interpretations of our actions through the logic
of technologies black-boxed to us.
Received 10 April 2019
Accepted 5 July 2019
Criminal prosecution;
criminalization; inequality;
informal economy; law;
magnetic stripe; MetroCard;
New York subway
In the early 1990s, the New York City subway intro-
duced a new fare system based on a magnetic farecard
(called MetroCards
) to replace its aging revenue
system reliant on metal tokens. By the end of that
decade, a growing number of metrocard users had
learned that an extra, unbought ride could be
extracted from a spent metrocard by bending the card
in a specific spot on its magnetic strip before
swipingit at a subway turnstile. This trick became
central to a hustle of the urban poor in which
swipers(as they came to be known) stand at subway
entrances with a stock of bent metrocards, selling
entry to the subway at some discount below the offi-
cial fare.
Swipers were subject to significant policing,
but it was unclear what laws swipers were violating.
The state soon passed a law to make the practice
a misdemeanor, carrying a potential penalty of three
months of incarceration, but police and prosecutors
also sought more significant charges, and began
to classify swipersbent metrocards as forged
instruments. When charged as a misdemeanor,
making or possessing a forgery can carry a one-year
Yet in one jurisdiction the borough of
Manhattan police began, in the early 2000s, to arrest
swipers with felony forgery, a charge which can carry
a much higher penalty. According to police data, over
thirteen years, more than eleven thousand accused
swipers were arrested under felony forgery charges;
92 percent of these arrestees were black or Hispanic,
and 97 percent were male.
When this policing practice emerged, anyone look-
ing for an explanation of just how bending a magnetic
farecard can produce a forgery would have discovered
that the technology underlying the metrocard system
is proprietary and kept secret from the public.
inner workings of the metrocard technology, however,
rose to the forefront of a legal case before the highest
court in New York State in which the court affirmed
the theory that a bent metrocard can be a forgery.
Interestingly, the court did not derive its rationale
from what is plainly observable: a bent-metrocard
cause and unpaid subway ride effect. Instead, it found
its rationale inside the technology, or at least in an
CONTACT Noah McClain Department of Social Sciences, Illinois Institute of Technology, Seigal Hall #116, 3301 S. Dearborn
Street, Chicago, IL 60616, USA.
Published with license by Taylor & Francis ßNoah McClain
expert-witness description of the inner workings of
the metrocard system and how those inner workings
interact with a swipers bent metrocard.
Although metrocards are ubiquitous in the pockets
of subway users, the metrocard system is a black box,
generally known only in the simplified terms of what
it does, not how it does it (MacKenzie 2005; Latour
1999; Pinch and Bijker 1984). In effect, swipers were
accused of a specific crime by dint of technical fea-
tures they could neither directly inspect nor challenge
with an alternative interpretation of the technology.
Moreover, swipers came to be defined as felons for
what they technically do, that is, for their actions
framed and interpreted through a context of the tech-
nology. To analyze this scheme of things the term
technical agency,which has been used elsewhere to
describe the agency of technology (see Neff, Jordan,
McVeigh-Schultz, and Gillespie 2012), will be re-pur-
posed here as a term of convenience for: human
action interpreted in strict reference to a relevant
technology, independent of the actors awareness
or intentions.
The criminalization of technical agency, as we see
in the case of swipersmanipulation and possession of
bent metrocards, raises a number of compelling ques-
tions: What are the politics of public systems which
operate on proprietary and secret underlying technol-
ogy (see Levine 2007)? Can human criminality
become an artifact of technical design? What is the
role of expertise in the joint interpretation of technol-
ogy and criminality? Just how do mundane technolo-
gies intersect with problems of race and social
The criminalization of selling swipes offers a case
in which a subsistence practice of the urban poor has
been subject to a punitive interpretation through the
frame of a proprietary technology. Defendants have
had no direct access to the technology, or to inde-
pendent expertise, making existing claims about how
the technology works more or less irrefutable, and by
extension, making claims about how a crime is com-
mitted through that technology more or less irrefut-
able as well. Asymmetrical distributions of technical
knowledge can be especially impactful when, as this
case shows, the prevailing account of the technology
is flawed, but still serves as the basis for the elabor-
ation of criminal law. More broadly, the case illus-
trates how understandings of technology may become
a fresh vector through which existing inequalities
travel, when claims about complex, proprietary tech-
nology are leveraged against certain groups and prac-
tices. The complexity or secrecy of technology can
itself be impactful when claims about human behavior
are built upon them.
This article, then, investigates the emergence of
swipers as a policing and legal controversy which was
eventually settled on the technical groundsof met-
rocard technology; that is, on dynamics hidden in the
black box of the metrocard system. To understand
and situate that outcome, this article digs deep into
the metrocard technology, precisely where a high
court placed social stakes, and upon which it affirmed
an aggressive pattern of policing and prosecution.
However, we find that there are a host of ambiguities
in how the metrocard technology was understood by
the court, and how it might relate to forgery statutes
yet they were not identified in the court case (or
heretofore in any venue). These ambiguities suggest
alternate interpretations of technical action that
defendants and their advocates have been precluded
from proposing in their defense.
Opening the black box of the metrocard can show
how the workings of a technological system and
how that system is understood, or not can have
heretofore-unacknowledged relevance for sociological
explanation. Our case allows us to see how a claim
about how a technology really worksaffirms and
stabilizes policing strategy firmly leveraged against
minority communities. It is also a step towards
addressing a dual disciplinary myopia articulated by
niz (2016): the reluctance of Science
and Technology Studies (STS) scholarship to address
issues of race so central to sociology, and the parallel
reluctance of sociologists to address the role of the
material and technical in the making of the social.
niz suggests to bridge the divide
through engagement with the sociotechnical material-
ity of race and racial domination(p. 214) the
actual stuff which helps produce stratified racial
orders, from technologies of knowledge production to
medical devices (see Law 1991; Braun 2014). This case
study advances that agenda by showing that, while the
inequalities delivered through policing and criminal
justice practices may hinge on issues long explored by
sociologists and socio-legal scholars, they can hinge
also on accounts of a technology to which criminal
justice actors point to justify their practices.
This inquiry sits at the intersection of important
themes in existing literature and policy debates.
Policing at the subway turnstile has recently emerged
as a focal point in the politics of city policing, but has
long been at the center of broken windowsstrategies
deployed in New York since the 1990s now the
archetype for national transformations in urban
policing. The nature of expert scientific interpretation,
and the technology and methods of forensic analysis
in the U.S. legal system are important questions in
both socio-legal studies and STS the criminalization
of selling swipes relies upon an expert account of
swiperscrime and also an entire plane of action
inside the technology where the crime allegedly
occurs. Furthermore, scholars and commentators of
many stripes increasingly point to the emergence of
technology as an axis and vector of social inequality;
this article shows how claims about technology, and
related interpretations of human action, is an emerg-
ing terrain on which vulnerable groups become still
more vulnerable. Before discussing methods and data,
I briefly situate how this case is framed by these
debates, and outline how this case links them.
Criminalization at the subway turnstile
In a vast city where fewer than half of households
own cars and subway fares continue to rise faster than
real wages, the subway turnstile has not only been a
gateway into a major public good, it has also been a
gateway into the criminal justice system. The adoption
of broken windowspolicing strategy in the early
1990s led to an especial focus on fare-beating at sub-
way entrances, where a small fare-beating transgres-
sion offered police a pretense to conduct frisks,
warrant checks, and develop justifications for searches
and arrests (Harcourt 2001; Kohler-Hausmann 2018).
The best available data on the policing of fare eva-
sion in the subway have long suggested that policing
fare evasion contributes to a broader criminalization
of everyday life for poor nonwhite urbanites (see Rios
2011; Stuart 2016). Recent research has relied on tri-
angulation techniques to study racial and ethnic bias
in the policing of fare evasion by comparing frequen-
cies of arrests with neighborhood demographics where
the arrests occur, adjusted for the volume of subway
passengers (Flagg and Nerbovig 2018; see also Stolper
and Jones 2017). In 2018 concern for the bias in
police scrutiny led to a new law requiring the New
York Police Department (NYPD) to report data on
fare-evasion policing,
which made new fare-evasion
arrest data available by race, gender, and age for an
18 month period. While the data (NYPD 2019) is
hardly granular (Lancman 2018; Winston 2018), the
broad strokes are telling: over that period, ninety-two
percent of Theft of Service (TOS) arrests involved
black or Hispanic arrestees. Even these hard numbers
likely obscure the true textures of subway fare
enforcement thanks to how the NYPD records and
reports arrest statistics, by the single most-serious
charge. A police encounter that begins as a stop for
fare-beating might, for example, become an arrest
which also includes a charge for drug possession, and
be counted as an arrest for drugs rather than for TOS.
When confronting fare-evaders, New York police offi-
cers have discretion to make a misdemeanor arrest, issue
a summons returnable to criminal court, or (most pref-
erable to the accused) issue a civil Transit Adjudication
Bureau Notice of Violation(TAB/NOV). While consti-
tuting 92 percent of official arrests, blacks and Hispanics
received 69 percent of TAB/NOVs over that same eight-
een-month period (NYPD 2019a). By official policy, a
fare-beating suspect may be disqualified from receiving a
including not having a verifiable address, or fitting into
the category transit recidivistThis category includes
more than two TAB/NOVs in the prior two years
(which can be issued for things like laying down on a
subway bench; see NYPD 2019b). This policy thereby
compounds bias in policing by ratcheting up the conse-
quences of the next encounter with the police for some-
one with prior minor offenses, or who is homeless. We
cannot tell, however, what proportion of arrestees for
TOS were disqualified from receiving a TAB/NOV, or in
what proportion of these arrests police might have issued
TOS policing is in many ways still mysterious.
The policing of swipers is even more elusive to
decipher. Swipers have been arrested under an array of
charges, including TOS, loitering, and petit larceny, but
there is no ready way to discern from statistical aggre-
gates which of the arrests stem from selling swipes. Even
though powerful local political actors have expressed
concern over the apparent disproportionate policing of
swipers in minority neighborhoods (Stewart 2018), their
efforts to confront the issue have been stymied by lack
of data to even demonstrate the pattern.
The focus on
the deployment of police scrutiny is only part of the pic-
ture, however; the criminalization of swipersstock-in-
trade, and the basis for felony-level charges, rests not
only on police strategy and uses of discretion; it also
rests on an interpretation of the metrocard developed
through expert testimony, framed by prosecutors, and
stabilized by a high court.
Technology, expert interpretation, and criminal
legal process
The STS literature has explored the ways in which sci-
entific and technical knowledge is deployed and
contested before judges and juries possibly ill-
equipped to evaluate it. Litigation such as toxic torts
when liability may hinge on scientific evaluation often
encompasses broader issues in the relationship between
science and society, such as how expert authority is
established as credible, how complex, messy procedures
of analysis are represented to courts, and how certain
claims prevail and are accepted as true by nonscientific
audiences (Jasanoff 1997).
With significant exceptions (Jasanoff 1998; Lynch
1998), criminal cases do not generally entail contests
of scientific expertise between well-matched adversa-
ries. Rather, criminal defenses generally have signifi-
cantly less access to expertise than their prosecuting
counterparts (Giannelli 2004; Murphy 2007). Limited
financial resources are one obvious reason, but others
are noted by scholars (Hoeffel 1990; Lynch and Cole
2005; Murphy 2015; Giannelli 2018; Saks and
Faigman 2008). Prosecutors are the key customers of
private crime labs and forensic experts, creating disin-
centives for them to report irregularities or error, or
express doubt. Moreover, physical evidence ends up
in the hands of the state, and, in the case of DNA
analysis, the prosecutions tests may consume all test-
able material, leaving none for defenses. In effect,
expert witnesses are prone to overstate the credibility
of forensic techniques while courts have often been
seduced by superficially-plausible explanations of tech-
nique or method(Mnookin 2008 p. 343). Even the
development of independent expertise is hampered by
high costs of entry and other obstacles, as in the case
of forensic techniques like GPS tracking, data mining
and DNA typing, which entail a complex interpretive
machineryand often require access to privately-held
computerized records and databases not readily avail-
able to independent parties (Murphy 2007).
Unequal access to expertise conjoins with formal
restrictions on access to technologies in ways that
may be pivotal for criminal justice outcomes.
Technologies of criminal justice decision-making are
opaque and biased; smartproducts profile certain
identities, places, and situations as somehow suspect
according to inscrutable calculations (Crawford and
Schultz 2014; Joh 2017b). Criminal sentencing and
parole decisions may be informed by an algorithm
which has been observed to rank minorities as riskier
than white counterparts; yet their outputs even
when based on outright inaccurate input data have
resisted challenge because aggrieved people and their
advocates cannot prove just how the proprietary algo-
rithm weighs input variables (Angwin, Larson, Mattu,
and Kirchner 2016; Wexler 2018). Moreover, courts
often protect technologies at the core of forensic evi-
dence gathering (e.g. DNA kits) from adversarial scru-
tiny (Hoeffel 1990; Mellon 2001). In a telling New
York case, a judge ruled that a defendant accused of
drunk driving was not entitled to examine the source
code of the breathalyzer device, reasoning that the
state does not possessthe source code sealed within
the device; that the proprietary code would likely be
unreadable, anyway; and that breathalyzer device
should be trusted by the court because it is trusted by
the state health authority.
In other words, when the
code is black boxed to the state, it can also be forcibly
black-boxed for defendants.
The metrocard technology has long been veiled in
secrecy by the Metropolitan Transportation Authority
(MTA), the agency which operates the subway. Expert
examination of bent metrocards used in prosecutions
of swipers can be feasibly performed by only the
smallest set of insiders in the MTA, who have exclu-
sive access to equipment necessary to evaluate if any
specific bent metrocard has the proper data configur-
ation to count as a forged instrument. More funda-
mentally, the criminalization of selling swipes relies
on an expert account of an entire technological plane
of action in which a bent metrocard can be seen as a
forged instrument at all. Charles Goodwin (1994)
offers the example of how a specialized way of seeing
can reframe action in a trial setting: how a person
(shown in a video) prostrate on the ground receiving
blows from a police baton is framed in a courtroom
as displaying aggression warranting the blows accord-
ing to an expert police perspective. Because common
sensehas no credentialed experts, such an expert
claim can (and often does) prevail without even being
contested. While swipers are denied access to the met-
rocard technology and the plane of action it putatively
constitutes, they are nonetheless vulnerable to a crim-
inal interpretation of their stock-in-trade which is
more or less impossible to challenge without that
access, and a competing expert interpretation.
Opaque technology and social vulnerability
Scholars of digital inequality have noted how a lack of
access to, and related sophistication with, Internet-
connected devices become incapacities to act effect-
ively in an increasingly online world (Dimaggio,
Hargittai, Celeste, and Shafer 2004; Robinson et al.
2015). The point also applies to decision-making sys-
tems in domains of consumer credit, insurance, hous-
ing, employment, and receipt of public benefits, which
produce consequential outputs through processes
concealed from, but pivotal for, the identities they
process, rate and score (Barocas and Selbst 2016;
Eubanks 2018). The rise of big data poses the add-
itional challenge of so-called franken-algorithms
which are ongoingly self-modifying, and increasingly
ungovernable even by programmers who set them in
motion (Smith 2018).
Other technologies also grow in prevalence in
everyday life without parallel growth of public under-
standing, much less direct engagement (Pasquale
2015). The criminalization of selling swipes illustrates
consequences from a lack of access to the black-boxed
interiors of a public technology ubiquitous in one
major city. It poses the problem that whatever lies
beneath a user interface can come to define action
taken through a technology, even while these internals
are not inspectable, not under democratic control, and
might even be subject to modification beneath an
unchanging human interface. Swipers, without access
to even a description of the inner workings of the
metrocard system, became vulnerable to the accus-
ation that their bent-by-hand metrocards are forgeries
on an electronic plane according, at least, to one
expert account of electronic interactions. But how?
Background and data
This article has roots in a series of technical and
sociological puzzles: In the mid-2000s, local New York
City press reported that swipers then proliferating
in subway stations were increasingly being arrested
under forgery laws, seeming to imply that swipers
manipulation somehow altered a metrocards value.
Could imprecise human hands actually modify digital
data? How did police and prosecutors come to under-
stand bent metrocards as forgeries? Why was the issue
confronted with policing instead of with re-engineer-
ing a publicly-commissioned technology?
Obstacles I encountered in efforts to understand the
technology underlying selling swipes became important
clues towards the obscured nature of the technology as
itself an integral part of the research object. The fram-
ing of human acts through an opaque technology
opened up fresh questions about the distinction
between reflexive action on one hand and technical
action on the other, and just who has access to tech-
nical modes of interpretation and how those modes are
deployed in the criminal justice system.
In interviews with subway station employees who
sell metrocards (twenty in total; some also observed in
), informants were unable to explain the
technical mechanics of swipersmethod. My
examination of their computer terminals, coding devi-
ces, and observation of operational and troubleshoot-
ing procedures illustrated how these interfaces
betrayed little about the underlying technology. These
workersinstructional materials
also hovered at the
surface of the work interface. Grey literature offered
important details of the fare systems overall architec-
ture, but failed to clarify the precise technical proper-
ties of the system or of selling swipes.
The first and only official account of how a bent
metrocard interacts with the MTAsfareprocessingsys-
tem made available for public dissemination was con-
tained in a criminal case, People v. Mattocks, argued
before the Court of Appeals in 2009 (see Chan 2009).
However, the technical account of selling swipes in the
Mattocks case seemed inadequate in two ways. First,
swipers as I observed them in fieldwork inevitably
encounter a pattern of error messages from the turnstile
computer as they execute their procedure, but those
error messages were unmentioned and unexplained in
the Mattocks account of the process. Second, the account
appeared at odds with a technical principle described as
essential to a metrocard-like fare system in a patent filed
by the metrocards developer, Cubic Automatic Revenue
Collection Group (Cubic, hereafter).
Questions about the Mattocks account, and about
the actual operation of the metrocard system were
eventually satisfied through the cultivation of a key
informant, one of few individuals with extensive direct
knowledge of the metrocard system. The informant, a
retired MTA official, had been the official keeper of
the metrocard transaction database and tasked with
monitoring records for anomalies indicative of fraud
or other problems from the systems inception until
the late 2000s, and was one of a small coterie of MTA
personnel dispatched to give expert-witness testimony
in the trials of accused swipers. Our several interviews
and correspondence (spanning 2013 through 2016)
were instrumental to building the understanding of
the technology I share in these pages, which reflects
the fare control system as it was more or less stable
from about the year 2000 through at least the mid-
2010s (although it may have also been subject to tem-
porary changes not explored here).
My understanding of the metrocard system is
cross-validated against the work of self-identified
computer hackers who took up the challenge that the
metrocard cannot be read by commercial card-readers
(see Michael and Michael 2009), and worked in the
1990s and 2000s to reverse-engineer the metrocard
and its data (Balaclava 1994; blueski-mask and the
wrapper 1997; Redbird 2006; see also Harmon 1997).
Their work culminated in the discovery that metrocard
data can be decoded when treated as sound, by passing
its magnetic strip across a tape-recorder head interfaced
with a PC sound card to produce musicconvertible to
binary (Redbird 2005a; b). The organization and mean-
one-to-one comparisons of metrocards with certain
known properties, such as time or location of its most
recent use, date of purchase, and so on.
The conceptual illustrations of the fare-processing
technology and the way a bent metrocard interacts with
it in Figures 1 and 3employ the black box as a visual
metaphor to also indicate important technical processes
that are hidden from users. Figure 2 extends that meta-
phor in blacking out processes which were neglected in
the Mattocks account, illustrating the understanding of
swipersmanipulation now embedded in law, and what
is missing from that understanding.
While I formed a critique of the Mattocks account
of how swipers deceive the fare processing system
with great help from my key informant, I must note
that my key informant was also the prosecutions
expert witness in the Mattocks case. This may appear
paradoxical, but the two accounts emerged in two
very different contexts. Both cross-examination and a
sociological interview lead to interactively-produced
knowledge. Yet the procedural conventions of court
examination do not give witnesses much capacity to
offer answers outside of the scope of the questions
posed to them, or to otherwise shape interactional
agenda (Molotch and Boden 1985). As I describe
below, when the expert witness was questioned in the
Mattocks trial, he made a surface-level reference to an
interaction between a swipers manipulated metrocard
and a turnstile computer necessary to obtain an
unbought ride. The experts cross-examiners did not
seek and likely did not think to seek a deeper
understanding of that interaction, which remained
glossed even through the highest level of appeal in the
Mattocks case. My research interviews, by contrast,
did seek to understand that interaction, and found
that it complicates the prevailing understanding of
swiperstransgression at the same technical level at
which the Mattocks case was finally decided.
Conversations with staffpersons in the Clerks of
Court office in four subway-served boroughs of New
York City strongly suggested that felony forgery
arrests of swipers were predominantly processed in
Manhattan, and might be explained by the prosecutor-
ial philosophy of the District Attorneys office in that
In turn, I interviewed the individual who
led the criminal court division of that office in the
early 2000s, and was involved in implementing new
policy governing the prosecution of swipers during
his tenure.
Official data on the arrests of swipers under forgery
charges come from the New York City Police
Department via a Freedom of Information request.
Other statistical data was furnished by the New York
State Department of Criminal Justice Services, and the
Manhattan D.A.s office through direct correspondence.
The metrocard and the emergence of swipers
The MTA began to explore a transition to electronic
farecards in the early 1980s as a solution to chronic
turnstile jumping and widespread use of slugs
instead of an official token.
The MTA has long
treated every instance of turnstile-jumping or slug-use
as missing revenue, making fare evasion salient to any
budgetary shortfall. The agency had hoped that mag-
netic, refillable stored valuefarecards would under-
mine the market for slugs, cut transaction costs by
encouraging riders to buy multiple fares at once, and
allow the MTA to replace station clerks with farecard
vending machines.
The metrocard was phased-in over several years
beginning in the early 1990s, alongside new subway
turnstiles designed to make turnstile-jumping more diffi-
cult (Weidner 1996). Each turnstile houses a computer
for processing metrocards users swipe their metro-
cards one or more times until their fares are successfully
processed and they are allowed entry. As this suite of
devices was phased in, official MTA statistics showed a
steep drop in the rate of fare evasion (Reddy, Kuhls,
and Lu 2011). Soon after, the MTA undertook to reduce
or eliminate stations staffing (Pierre-Pierre 1996), a goal
the agency has pursued ever since.
The MTAs apparent success was threatened by the
end of the decade, however, when riders discovered a
major vulnerability in the metrocard technology: A
metrocard with no remaining value, scratched in a
precise spot, could be used over and over again to
enter the subway an unlimited number of times. The
vulnerability was soon broadcast citywide in the local
press (Coleman 1998; Rohde 1998). The press did not
discover, and the MTA did not reveal, that the vulner-
ability stemmed from a data remnant which remained
on the metrocard after all purchased fares have been
spent; the scratch obscured other data which indicated
the metrocard no long held valid fares. In response,
the MTA worked with the metrocards developer to
modify how metrocards are processed by turnstile
computers (Rutenberg 1998a). One modification had
turnstiles reject metrocards after two failed attempts
to read their data (Newman 1998), but the fix was
soon abandoned it was not uncommon for users to
experience multiple failed reads with a perfectly valid
metrocard, and the modified turnstile software would
reject them, too (Rutenberg 1998b).
A subsequent revision of the turnstile fare-process-
ing software (the version at the center of this analysis)
Figure 1. How a turnstile computer processes metrocard fares, and the routine errors it encounters, produces, and corrects.
Notes. The metrocards magnetic stripe has two value-fields. The turnstile computer treats the lowest readable fare-total between
these two fields as the current value of the metrocard. Our examples begin with a metrocard worth three fares. The metrocard is
processed as it is passed through a slot in a turnstile containing a scan head, a write head and another scan head in physical suc-
cession. Swipe # 1 is ideal: The first scanner (A) determines the lowest readable fare total as the current value; then, a write head
(B) records a new, reduced fare total of two fares to the adjacent value field, then (C) the second scanner reads the freshly-written
data to verify the data written in process B. The computer will now allow the passenger entry, and the value field with the fare
total of three becomes obsolete but is left intact. Swipe # 2 leads to an error in which the computer cannot read a fare total on
either value field, and requires the user to try again. In Swipe # 3 the fare totals on both value fields are successfully read, and
the computer attempts to write a new fare total but the new fare total cannot be verified (C), possibly because either because the
verification scan failed, or the write(B) was corrupt, or both. The computer instructs the user to swipe again at that turnstile. The
turnstile computer then defaults to a mode, seen in Swipe #4, in which it simply re-attempts the write (B) and verify (C) processes;
re-starting with a read (A) risks deleting a users fare. Swipe # 5 is ideal, like Swipe #1. If the metrocard is swiped again (as in
swipe #6), it will be rejected because the lowest readable value field is now zero.
was stabilized before the turn of the century.
The update indeed halted the acceptance of scratched
metrocards, but was vulnerable to a more elaborate
procedure of bending, swiping, unbending, and swip-
ing to produce an unpaid subway ride (Daly 1999).
The procedure still takes advantage of the data rem-
nant on a spent metrocard, but that remnant ends up
deleted, so the procedure will work only once per
spent metrocard.
Swipers became a common presence in many sub-
way stations in the early 2000s. Near many entrances,
swipers would gather discarded, spent metrocards,
skillfully bend them in a highly specific way, recruit
a customer, and swipe a metrocard through the turn-
stile reader several times until a message prompted
the rider to enter. In some instances, the MTAs
de-staffing campaign left swipers as the only human
fare-sellers at a subway entrance, and swipers some-
times secured a total monopoly at the entrance by
jamming paper into adjacent metrocard vending
machines, disabling them temporarily (Luo 2004).
Each transaction earns the swiper a dollar or so. In
the argot of subway users, a swipe refers to any
attempt to use a metrocard at a turnstile, but only a
swiper will sell swipes. Swiperstechnique is only
viable at subway turnstiles; on busses, where a mech-
anical apparatus (not a human hand) guides a metro-
card through its processing, a bent metrocard will be
rejected as either unreadable or valueless. (As we will
see, this is an important clue: Why would the
manipulation work at the subway turnstile but not at
the bus farebox if bending a spent metrocard, on its
own, makes it appear valid to an electronic reader?)
Why could not the MTA just check every
metrocard against master records and have turnstile
computers reject any metrocard which ought to be
depleted of fares? The reason is that turnstile com-
puters have no way of telling just what a given metro-
cards value ought to be, because they process each
metrocard based solely on data on the card, without
Figure 2. The Mattocks account of bent metrocards as forged instruments.
Notes. Selling swipes begins with a metrocard in which the lowest of the two value-fields shows zero fares remaining, which will
always be rejected by the turnstile computer (illustrated in swipe #0). Cross-examination of an expert-witness in the 2005 People
v. Mattocks criminal trial characterized a bent metrocard (as in swipe #1) as capable of producing free entry into the subway
because the fare processing system gives the benefit of the doubtin case a metrocard had been somehow damaged. Testimony
also made reference to the existence, but not the substance, of subsequent steps necessary to achieve that outcome. Those steps
not revealed in this account suggest ambiguity in the legal classification of a bent metrocard as a forged instrument.
first checking records of that metrocards value in a
centralized database. In an arrangement designed long
before the present era of stable real-time networks,
each turnstile computer forwards records of its trans-
actions in batches to a station computer, which then
forwards batches of records on to a central main-
If a node or link in this network fails or is
merely sluggish, transaction records can continue to
accumulate (and fares can continue to be processed)
until the records can be forwarded on to the next
strata, allowing ongoing fare processing in spite of
that failure. One consequence of this loosely-coupled
arrangement, however, is that a rider might add or
use fares anywhere, any time in the vast subway and
bus network, but the record of each transaction may
not reach the mainframe for hours after each
occurred, and not necessarily in order of their occur-
rence, either (Reddy et al. 2011). Data written to the
metrocard itself is therefore treated as authoritative by
the turnstile computer for the purpose of providing or
denying entry.
In turn, a manipulated metrocard
without valid fares only comes into administrative
focus hours or even days after a turnstile computer
has already provided entry.
Without an obvious software solution, the MTA
turned to lobbying police, prosecutors and politicians
for a sterner stance towards swipers. The MTA also
initiated a poster campaign advising riders that selling
Figure 3. The hidden technical logic of swiperspractice.
Notes. Swipe # 0 involves a metrocard that no longer has any value. Before Swipe #1, the card is bent in a very specific way, mak-
ing concave the portion of its magnetic strip containing the value field with a fare total of zero. When it is swiped (Swipe #1),
that portion of magnetic strip is out of reach of the scanner head (A), preventing a successful read of the value field containing a
fare total of zero. However, the computer tenuously accepts that one fare it can read, and attempts to write a reduced fare total
(B) to the opposite value field, but is unable to verify if the fresh data has been correctly recorded (C) because the kink drew the
value field out of reach of the write and verify heads. The swiper is told to try again. Immediately after, the swiper un-kinks
the metrocard and holds it flat as it guides it through the mechanism (Swipe # 2). The computer, defaulted to only re-attempt the
write and verify scan steps, indeed writes (B) and verifies (C) that a fare total of zero is recorded to the formerly-unreachable value
field, completing essential steps to process the one fare the computer read in Swipe # 1. The pattern of Swipes #1 and # 2 will
require the user to swipe the metrocard once more (Swipe # 3) to zero out the remaining field indicating one fare, leaving two
value-fields with fare totals of zero, before the user is granted entry.
swipes is illegal. It turned out, however, there was no
consensus over why it is illegal.
Policing and prosecuting swipers
Precisely what law did selling swipes violate? Charges
for trespassing, loitering, and criminal mischief were
each leveled against swipers, but these were indirect at
best. Theft of Service was a bad fit because swipers are
not generally the users of the stolen service. One major
arrest initiative accused swipers of petit larceny, a Class-
A misdemeanor, under the theory that their earnings
were stolen from the MTA. In one of the two instances
where the highest court in the state heard arguments
over appropriate charges for swipers, the Court of
Appeals rejected petit larceny. Partly concerned of set-
ting a precedent that would criminalize unfair or illegal
business competition, the court reasoned that property
was not the case for swipersearnings.
In 2004, the state legislature responded to MTA
lobbying with a new law prohibiting Unauthorized
Sale of Certain Transportation Services(USCTS), a
Class B misdemeanor that carries up to three months
In the months after the introduction
of the law, local media reported that police used the
charge agaisnt swipers only infrequently (Donohue
2005). However, USCTS was still used in more than
six thousand arrests through year 2018. Blacks and
Hispanics make up 93 percent of these arrestees (65
and 28 percent, respectively). Members of those
groups account for 97 percent of those convicted
under the law through 2012 (the most recent year for
which I have conviction data).
Accused swipers have faced a more common and
more serious arrest charges in the form of Forgery
in the 2
degree and Criminal Possession of a Forged
Instrument in the 2
degree, both Class Dfelonies,
carrying sentences of incarceration up to seven years.
According to NYPD data, its Transit Bureau made
more than eleven thousand such felony arrests from
2002 through 2014. As can be seen in Table 1, these
felony arrests also follow a powerfully-racialized skew:
Ninety two percent of these arrestees were black or
Hispanic, as well as overwhelmingly male and adult.
Almost two-thirds of the citywide felony arrests of
swipers made by the NYPD Transit Bureau were
processed in Manhattan. According to Rick Castello,
who led the Criminal Court Division of the
Manhattan District Attorneys Office when felony
arrests for swipers first occurred, these charges were
welcomed for defendants with prior records. When
asked what occasioned the policy to allow prosecutions
of swipers under felony forgery laws, he responded:
Idont remember if this is one of those [charges] that
complaint room with them, or if they bombarded, and
it eventually filtered up. Imprettysureitwasthe
police, that they came up with this potential charge and
then our office decided how to handle it and agreed
that it was a clever but legally-sufficient charge.
In such instances, according to Castello, his office
would seek internal consistency by making a
command decisionon the acceptability of a charge,
and then instruct junior prosecutors who otherwise
draft complaints more or less independently. The
approach was, according to Castello, consistent with
the Manhattan District Attorneys general expectation
that the NYPD would deliver arrests under most-ser-
ious charges. Prosecutors, in turn, might exercise dis-
cretion to reduce the charges, or offer a reduced
charge as part of a plea deal. Castello recalled, Most
of them [swipers with a felony arrest] would plead
guilty to something [ ] If you had a recidivist and
you really wanted to get him six months in jail, and
you charged him with a felony and they thought you
might be serious about indicting that case, theyd take
the six months.
Table 1. Arrests of accused swipers under two forgery-related
felony charges by New York Police Department Transit
Bureau, 20022014.
Arrests Arrestees
Total (%) Total (%)
All arrests 11,142 (100)
Arrest Charge Race or Ethnicity
§170.254,865 (44) Asian/Pacific Islander 261 (2)
6,277 (56) Black 6,715 (60)
Hispanic 3,571 (32)
Borough of Arrest White 541 (5)
Bronx 1,363 (12) Native American 13 (0)
Brooklyn 2,191 (20) Unknown 41 (0)
Manhattan 7,112 (64)
Queens 474 (4) Sex
Staten Island 2 (0) Female 388 (3)
Male 10,754 (97)
Year of Arrest
2002 467 (4) Age Distribution
2003 644 (6) 10-17 1,139 (10)
2004 1,887 (17) 18-24 3,218 (29)
2005 1,815 (16) 25-40 3,936 (35)
2006 1,504 (14) 41-59 2,765 (25)
2007 1,092 (10) 60þ84 (1)
2008 692 (6)
2009 479 (4)
2010 512 (5)
2011 369 (3)
2012 388 (3)
2013 516 (5)
2014 777 (7)
Criminal Possession of a Forged Instrument - 2nd Degree
Forgery - 2nd Degree
Some swipers were indeed tried under felony
The Manhattan District Attorneys office
reports that it prosecuted 174 bent metrocardcases
as felony possession of a forged instrument between
2001 and 2018, with half of these in the span of 2004-
Instances include one swiper convicted of a
felony count for each of the 96 bent metrocards found
in his pockets, but allowed to serve 96 sentences of
two-to-four years concurrently.
Another defendant,
convicted of just two felony counts for his two bent
metrocards, was sentenced to serve two consecutive
prison terms of 2-6 years each, although a higher
court later halved the penalty, noting that defendants
diagnosis of schizophrenia had been ignored at
The relative small proportion of felony prosecu-
tions suggests that the main purpose of the felony
charge is (as Castello puts it) the great leverageit
offers in plea bargaining leading many accused
swipers to plead guilty to lesser charges rather than
risk a felony prosecution. Even if no felony conviction
or prison time resulted from such arrests, they are
nonetheless consequential for arrestees. A full discus-
sion is not possible in these pages, but we can defer
to Kohler-Hausmanns(2013,2018) analyses of the
ways New Yorks criminal justice system asserts con-
trol even without conviction by managingdefend-
Elements include periods in which defendants
are markedfor added consequences in any future
criminal justice encounter (through designations like
transit recidivist), and marked with potential
obstacles to jobs and housing through searchable pub-
lic data on open criminal cases. The courts further
entangle defendants by mandating performance of a
program, of service or therapeutic activity, to move
their cases towards dismissal, all the while requiring
defendants to re-orient their lives around the vagaries
of court schedules and procedures, and inevit-
able delays.
Criminalizing technical agency inside the
metrocard system
Criminal forgery allegations against swipers often did
little to specify how a bent metrocard is a forgery, for
example, by saying only that a defendant possessed a
metrocard bent in a manner that would allow illegal
entry into the subway system.However, more
evolved language located the criminality of the bent
metrocard in its relationship with a computer, on a
technical plane not observable by the accused, such as
by alleging that the bend obliterates the encoded data
and alters the fare of the card as read by the turnstile
As we will see, such claims oversimplify
the operations of the fare-processing system and how
a bent metrocard can interact with it (explained in
Figures 1 and 3). Yet legal proceedings held to adjudi-
cate just what swipers doin a technical and criminal
sense embedded that oversimplification in the fabric
of law, stabilizing the legal status of bent metrocards,
along with the social impacts of that status.
Catching swipers in the black box in People
v. Mattocks
The treatment of bent metrocards as forgeries was
rejected by at least one court which reasoned that
because the MTA does not sell visibly-bent metro-
cards, a bent metrocard is not falsely alteredbecause
(as that term is defined in a key statute
) a bent one
does not purport to be in all respects an authentic
This way of thinking about technology
atechnology frame(Orlikowski and Gash 1994)
reflects what is observable to human actors. Its appli-
cation to the metrocard did not long survive, however,
and was soon replaced, through People v. Mattocks, by
a technology frame empathetic to what the turnstile
computer purportedly perceives.
In a trial, which began in 2005, a Manhattan jury
convicted defendant Jonathan Mattocks of Criminal
Possession of a Forged Instrument 2
this criminal trial, the MTA Director of Metrocard
Security, James Eastman, was designated an expert in
metrocard security. He was asked to explain how a
bent metrocard interacts with a turnstile computer
and to indicate which of the defendants had the
proper data configuration to enter the subway without
paying, based on his examinations performed through
MTA equipment. The testimony and its interpretation
later became the authoritative account of what turn-
stile computers see when presented with bent metro-
cards, and thereby was key to stabilizing bent
metrocards as forgeries under the law.
The expert witness account
In Mattocks2005 criminal trial, Eastman revealed
fundamental elements of the metrocard technology
(consistent with Figure 1): The metrocards magnetic
stripe contains not one but two data fields designated
to hold fare data. The totals written to these fields are
not equal one total is always one fare higher than
the other. When swiped at a turnstile, its computer
always treats the lowest fare total as the correct,
current value, and writes the new, reduced fare total
to the other value field (see Figure 1). The two phys-
ical value fields thus take turns possessing the current
fare total of the metrocard. Following this alternating
logic, a metrocard used until no fares remain will
have a value-field that indicates a fare total of zero
and will be rejected at the turnstile. Yet, the other
value-field will still show a fare total of 1. That linger-
ing sum is what makes swipersmanipula-
tion possible.
Questions posed to Eastman produced the follow-
ing account of how a swipers bent metrocard inter-
acts with the fare processing system (summarized in
Figure 2), which we must take care to examine closely:
A swipers bend to a spent metrocard damagesthe
value-field with a fare total of zero. When swiped at a
turnstile, its computer can then only read the opposite
field, which still shows one fare, and will accept that
one fare as valid, and allow the customer entry.
Eastman described the acceptance of just one readable
fare total as a backupfeature which gives the benefit
of the doubtin case the card had been damaged.
When Eastman was asked if certain of the defend-
ants bent metrocards would workto obtain unpaid
rides if swiped at a turnstile, Eastman stipulated yes,
but youd have to swipe them twice. This was a crit-
ical answer but it passed without notice; if any party
had interrogated why this second step was necessary
and what is accomplished through it, a more nuanced
account of the metrocard system might have emerged,
suggesting a more complex relationship between a
bent metrocard and forgery laws.
The Court of Appeals interprets bent metrocards
Attorneys working on Mattocksbehalf challenged the
treatment of a bent metrocard as a forged instrument,
as well as the use of felony forgery laws to police and
prosecute an activity already prohibited under the
misdemeanor law specifically and directly passed
against it (Unauthorized Sale of Certain
Transportation Services). The argument eventually
won a hearing before the Court of Appeals, the high-
est court in the state. Arguments on matters of law
were to be based on existing transcripts and materials
produced for Mattockscriminal case; the existing
technical account of selling swipes thus became stabi-
lized and undisputable for the purposes of the appeal.
The Court eventually rendered its decision drawing
from that expert account, and the ways the account
was framed in the initial criminal trial and intermedi-
ate proceedings. What is most remarkable about the
Court of Appeals case is that it sought to specify the
legal status of a bent metrocard within a technological
context neither the litigants nor the court actually
understood clearly, grasping for metaphors perhaps
appropriate to their understandings of the metrocard
system, but inapt for its actual nuances.
According to New York State law, forgery is com-
mitted when someone falsely altersa written instru-
ment so that it then appears or purports to be in all
respects an authentic creation or fully authorized by
its ostensible maker or drawer(emphasis added).
In oral arguments before the Court of Appeals,
whether or not a swipersmetrocard appears in all
respects an authentic creation”–was the central ques-
tion. Mattocksattorney (from the Office of the
Appellate Defender) offered a simplifying metaphor: a
metrocard with only one readable value-field is akin
to a check with the signature line ripped off, and is
thus incomplete and invalid.
Just because the MTA
programmed its turnstiles to accept incomplete metro-
cards, he argued, does not mean that a metrocard
with missing data is a forgery.
An Assistant District Attorney (ADA) from
Manhattan advocated for the felony forgery charge.
Contradicting Mattocksattorney, she claimed that its
absolutely not true the turnstile reader accepts can
detect that the card is invalid, actually detects that its
invalid or unreadable and allows someone to enter any-
way. Thatsnottrue.
As she argued, even if the fare-
processing systems capacity to detect a false instrument
is inadequate, any unrecognized false instrument is by
definition completeto the system and is thus a for-
gery. Exemplifying the technological frame sympathetic
to an imagined perspective of technology, the ADA
claimed that to rule against the forgery charge because
MTA computers cannot detect the invalidity of a bent
metrocard would be blaming the victim.
The Court of Appeals sided with the Manhattan
District Attorney in ruling that a bent metrocard is
indeed a forged instrument to the eye of the relevant
audience, that of a turnstile computer. Yet the court
did not find grounds for forgery charges in the undis-
puted and observable fact that a properly bent metro-
card can be used to produce unpaid subway entry.
Rather, the court looked inside the technology, based
on a reading of the 2005 expert-witness testimony.
The Courts decision described how, according to
that expert account, a precise bend can obliterate
that information [zero fare total - au], so that, when
swiped, the computer is unable to detect that the met-
rocard is worthless the turnstile computer will
read the other field containing the backup
information, which gives the user of the card free
entry into the subway.The decision elaborated:
[T]he damage to the MetroCards thwarted the usual
computer processing of the information contained on
the magnetic strips. By bending the MetroCards,
defendant successfully destroyed the zero-fare
information encoded on one of the fields in the
magnetic strips and was able to acquire free rides on
what were worthless MetroCards. Thus, defendants
misused the benefit of the doubtsystem by
intentionally making the valueless MetroCards
purport to be authentic instruments.
The Mattocks ruling remains stable case law. Yet,
when we examine how bent metrocards interact with
the fare-processing system at a technical level, we can
find considerable ambiguity in the legal status of a bent
metrocard. Moreover, we discover how technology can
be misunderstood in ways that may wrongly attribute
criminality to actions taken with and through them.
How a bent metrocard interacts with the fare-
processing system
As the Court of Appeals articulated the process, if a
turnstile computer can only read one value-field (con-
taining one fare) on a metrocard, the computer will
accept that fare and allow the passenger to enter. This
is not correct. A turnstile computer will not, under
any circumstance, process any metrocard as valid for
a fare unless it has successfully: (1) read a fare on one
value field, (2) written a zero to the other value field,
and (3) read the freshly-written zero. Paradoxically,
the second and third of these steps cannot be com-
pleted with a bent metrocard, however, because the
portion of the magnetic strip containing the relevant
value field is not in full, flush contact with the write
or verify heads set in the turnstile. Thus, the process
must be accomplished across multiple steps: In the
first, the turnstile computer reads one fare. In the
second, when the swiper unkinks the metrocard and
holds it rigid, the turnstile computer writes a zero to
the other value field before then reading it for verifi-
cation. The data that the turnstile computer writes
and then verifies in the second and third steps makes
it treat the data it read in the first as legitimate as
having been valid in retrospect. The fare-processing
system will record that a fare has been used at
this point.
Understanding this process requires a brief tour of
the metrocards engineering. The slot set into subway
turnstile for processing metrocards contains a scan
head, a write head, and a second scan head (for veri-
fying fresh data) in linear succession so each can
perform its function as the metrocard strip is trans-
ported across it in one swipe through the slot. When
the MTA began to consider a farecard loaded with
multiple fares in the 1980s, existing products entailed
farecards where a new fare total was written over (and
thus deleted) a prior fare total each time the card was
used. A motorized mechanism would accept the card
and transport the card across the scan and write heads
properly, at correct speed, alignment and uninter-
rupted contact. These conditions are necessary to
allow the scan head to read the strips magnetic par-
ticles as coherent data, and for the write head to prop-
erly distribute those particles on the strip in a way
that can be read as data once again. While these
mechanical transport systems helped ensure those
conditions, they often jammed and trapped contami-
nants, requiring labor-intensive interventions.
Moreover, they slowed passenger-throughput below
levels the MTA deemed necessary for subway entran-
The MTA thus sought a system in which the
user would swipe the card through its processing by
hand, but this posed a new challenge: if the stripe is
not swiped across the write head under proper condi-
tions, what is written will not be readable as a data
once again it is corrupt, and the users value has
been deleted entirely (Cekander 1994).
The metrocards developer, Cubic, did not find a
way to prevent the chronic corruption of fare totals
due to bad writes, but did find a way to tolerate
that corruption. Cubics solution (outlined in a patent;
see Aubrey 1992) involved using magnetic strips with
two value-fields rather than one, and turnstile com-
puter software which would always treat the lowest
fare total between those fields as the current, proper
value of the metrocard. Each time the turnstile proc-
esses a fare, the current fare total is left untouched.
That total simply becomes obsolete once a lower value
is written to (and verified on) the opposite value-field.
Corruption would still be endemic to this arrange-
ment, but the errors are correctable by making users
swipe again until a proper write is verified; users
would be made to compensate for a technical short-
coming. The metrocard system was designed to toler-
ate metrocards with just one readable value-field
because the metrocard system so often creates them.
As illustrated in Fig. 3, swipersbend is placed right
where the value field holding a fare total of zero sits,
making it impossible to read from or write to that
region of the strip because it is concave and beyond
reach of the heads.
The turnstile computer will treat
the one fare it can read as valid, but only in retrospect
if it is subsequently able to write and then verify a
new, reduced fare-total of zero to the other value-
field. However, that verification fails because (as we
humans know) the value-field is unreachable by the
write head and scan head, meaning expected data will
not be found by the verification scanner.
A failed metrocard verification is normal and
endemic to the system, and can be a result of a cor-
rupt write, a misread for verification, or both (e.g.
Figure 1, swipe 3, process C). The turnstile computer
responds by prompting the user to swipe again here.
When the user complies, the computer defaults to a
mode in which it only re-attempts the write and verify
steps. This configuration seeks to avoid a major pitfall:
If the computer repeated the initial scan step (e.g.
Figure 1, swipe 1, process A), it might end up treating
a fresh, properly-written fare total it failed to verify as
the current total, and then debit a second fare without
having provided entry for the first.
As the swiper passes the card through the slot a
second time, they now hold the metrocard rigid,
which counteracts the kink in the magnetic strip.
Once flat, the turnstile computer can now write and
then verify that fare total of zero on the relevant
value-field. Verifying that zero closes the loop of steps
necessary to process a fare, and it is at this point that
the metrocard system records that a fare has been
processed. The passenger will be allowed entry as
soon as the user completes a final step, to swipe yet
again so the computer can destroy the one fare once
again lingering on a value field, thus leaving the met-
rocard without any fare totals at all.
Finding ambiguities in the black box of
the metrocard
Even if we accept that defendants are criminally culp-
able for the hidden machinations of the technologies
they use, we can find ambiguities in the interpretation
of bent metrocards as forgeries that have been beyond
reach of defendants and have never been presented
to courts.
State forgery statutes describe a forged document
as one that has been purposefully altered and which
then purports to be in all respects authorized by its
While swipersbend makes it impossible for
the turnstile computer to read a portion of the data,
the data is itself unaltered (and would be legible again
if the card were unkinked). If the turnstile computer
seesthrough contact, it seems apt to say that bend-
ing the metrocard is more like turning out the lights
when a cashier is examining our currency than it is
like handing over a fake dollar bill. The circumstances
of perception are modified, but the instrument is not.
The multi-step and iterative nature of swipers
manipulation is a second site of ambiguity. To achieve
unpaid entry, a bent metrocard must be read in three
different states (first, as only partly readable, and
second, as containing a fresh zero in the formerly-
unreadable field, and lastly, as containing no fares on
either value-field). The validity of the first fare-total
hinges on the successful recording of the second;
without the second step, even a properly bent metro-
card will always be recognized as invalid and be
rejected by the turnstile computer. The interaction
will not even generate a record. It is only when the
turnstile computer writes and verifies a fare-total of
zero to the opposite field that the prior total of one is
treated as legitimate, and a passenger is given entry.
However, the data configuration which permits
unpaid entry exists for only a split second, just before
the swiper is required to pass the card through the
turnstile slot a final time, when any lingering repre-
sentation of value is deleted before the passenger is
allowed entry (see Figure 3). Moreover, it is not a
bent metrocard which is able to produce this out-
come, but an unbent one.
The incapacity of a turnstile computer to accept a
bent metrocard, and the issue of unbending, were not
mentioned at any stage in the Mattocks trial. It is highly
unlikely that any prosecutions of swipers have taken
into account these problems. The evidence held against
swipers are bent metrocards with the potential to pro-
duce unpaid entry but which cannot under any circum-
stance produce unpaid entry or be considered complete
by the turnstile computer without further steps.
It is not clear if these ambiguities rooted in an
improved understanding of metrocard technology
could have resulted in a different conclusion in
Mattocks, much less halted the search for severe conse-
quences for selling swipes. It is certain, however, that
accused swipers were denied the potential to develop
important arguments based on these ambiguities.
Discussion and conclusion: Interpreting
technology, interpreting criminality
When the MTA discovered that users could ride free
using spent metrocards, the agency did make some
internal attempts to halt the practice, including a soft-
ware modification which limited the number of times
the manipulation could be performed on a given met-
rocard. Yet, as selling swipes began to proliferate,
swipersbehavior was framed as the source of the
problem, rather than a flawed or misbehaving technol-
ogy. Notwithstanding initial tinkering, the metrocard
technology was taken as an ungovernable given, like
the weather, over which there is no human control,
and about which there can be no politics, oversight or
public intervention (see Jasanoff 2016). Indeed, the
prosecutor advocating that bent metrocards are for-
geries argued convincingly to the New York Court of
Appeals to take the design of the metrocard verifica-
tion system for granted, and treat the (putative)
parameters of computer perception as given, even
when those alleged parameters can lead to a human
action being given a felonious status. Presumably, we
could all be forgers if we inserted a $1 bill into a
vending machine marked to only accept $5 bills, but
the vending machine treated our one dollar as five
dollars upon verifying that yes, the currency is green.
If criminality can be an artifact of technological
design, and the black box of a technical artifact masks
the very human decisions entailed in that design, we
discover that, through technology, human decisions can
have un-confronted, autonomous impact this is one
foundational insight of STS research. The engineers of
the metrocard system are not unique in failing to
anticipate alternative uses of technology. Indeed, in a
second basic observation in STS, they are
interpretively flexible(e.g. Leonardi 2009;Oudshoorn
and Pinch 2003). However, notwithstanding examples
like consumer-safety regulations, we lack clear political,
legal or other mechanisms to problematize the engin-
eering of things, much less to intervene in it. The opa-
city if not proprietary secrecy of many technologies
presents a barrier: what can we contest or regulate
when we do not know what to contest or regulate?
Artifacts may have politics, as we have long known
(Winner 1986) but they need to be noticed in order for
an intentional human politics to emerge over them.
The stealth work of opaque infrastructure stands to
be more insidious because infrastructure is inescapable
without also giving up all to which it is central (Star
1999). Opting out of AC electric power means also
giving up toasters and television, just as opting out of
the fare-processing system of your citys mass transit
means giving up a vital mode of mobility. Another
layer is added to this picture in the case of technolo-
gies deployed by arms of the state in which, as Joh
(2017a) notes of the engineering of police body cam-
eras, design is policy. The mechanics of the technol-
ogy itself has the impacts of public policy but is too-
often granted the neutrality of ungovernable and
undisputable things.
The criminalization of bent metrocards offers a fur-
ther twist on this already-fraught problem. Even in a
court proceeding which sought to examine a putative
relationship between a technology and a criminal act,
an inadequate (and quite arguably wrong) account of
that technology was made central to a legal decision.
It is not only artifacts that have politics; accounts of
those artifacts have politics as well.
Accounts of technology rise in significance when
the opacity of a system is not merely the outcome of
cognitive or intellectual neglect, but of active secrecy.
Secrecy grants monopolies of knowledge and access to
those with interests in specific accounts of that tech-
nology in a way that is difficult to dispute. As we saw
in the Mattocks case, a singular account from an
expert witness was glossed and construed to describe
not only the status of a bent metrocard, but also to
conjure the technological context in which that bent
card could have that status. The 2004 misdemeanor
law which targeted the selling swipes, against the
unauthorized sale of transportation services, at least
described the activity in a way accessible to swipers
who might intuit that they were indeed selling trans-
port services without authorization. The use of felony
forgery charges, by contrast, alleges a crime its perpe-
trators themselves cannot witness because it unfolds
within a domain to which they have no legitimate
access. While certain allegations against an accused
swiper might be contestable, the fact of a bent metro-
card being a case of forgery is hardly contestable,
thanks to the secrecy of the metrocard system, and
domination of related expertise and even basic infor-
mation by those invested in that secrecy.
Surely, forms of bias are revealed in the social selec-
tion of selling swipes as a problem warranting highly
punitive arrest policies, which yielded arrests of (almost
exclusively) African American and Hispanic men. They
are also revealed in efforts by prosecutors to defend
these arrests and prosecutions up to the highest levels.
However, those efforts found traction in the putative
interiors of technology. Labeling technical agency may
be a new vector of inequality in criminal justice, espe-
cially when contesting a criminal interpretation of tech-
nical agency requires exactly the resources marginalized
defendants tend to lack, and the expertise and institu-
tional access almost everybody lacks.
Within every black box of technology is many more.
We must worry that each might contain or be said to
contain a technical rationale to reframe human action.
Those who cannot anticipate those re-framings are espe-
cially vulnerable to them. Swipers offer one example. A
contrasting case is that of the computer hackers whose
capable probing uncovered both how to duplicate valid
metrocards, and how these duplicates would be discov-
ered by MTA computers in fairly short order.
sophistication with technology helped them to anticipate
duplications as a risky if not futile forgery.
value of this kind of sophistication is likely to grow
when the internal operations of technical systems are
cited to frame just what users do (in a technical sense)
when they do this or that (in a way that is self-aware).
While the practice of selling swipes is much in
the Mattocks precedent stands ready to gov-
ern analogous practices with other digital media. The
linkage between a bent metrocard and forgery can now
be invoked without being revisited, a black box of its
own. The flawed understanding on which that linkage
is based, however, illustrates the problem of opaque
technology deployed to frame the actions of unwitting
users. No data suggests that any actor relevant to the
Mattocks case put forth what they knew to be a flawed
understanding of the technology, but the incomplete
account of metrocard technology which emerged in the
case is perhaps an outcome of a kind of structural
secrecy(Vaughan 1996), achieved when an expert wit-
ness without legal training answered questions posed
by lawyers unable to identify ambiguities in the picture
of the technology which emerged from his answers. Yet
even if the Mattocks ruling were based on a rigorous
understanding of the metrocard system, now that the
legal status of bent metrocards has been stabilized,
there is no nexus to update that legal status concordant
with updates of the internal, hidden aspects of the tech-
nology. A hackers 2016 attempt to decode the metro-
card, for example, suggests some modifications to the
way data was previously written to the metrocard,
while usersexperience has been unchanged.
job would it be to inform courts, defendants, and pros-
ecutors that the metrocard now works differently, and
the criminal status of bent metrocards warrants recon-
sideration? With no institutionalized means of evaluat-
ing and updating existing accounts of technology,
definitions of human action can be trapped within
those accounts, with little chance of escape.
1. Hereafter, I avoid the capitalization of MetroCardto
diminish distraction and to reject treatment of public
infrastructure as a private or branded product.
2. This paper describes selling swipes of what are known
as pay-per-ridemetrocards. Another less-common
type entails selling the use of unlimited ride
metrocards, generally policed under civil rules of
mass-transit conduct (see 21 CRR-NY 1050.4). The
latter practice requires no manipulation, but needs an
expensive unlimited ride card, which entails a delay
between uses.
3. A 2002 juvenile case, In the Matter of D.U. [Family
Court, Queens County, Jamaica, New York, 192 Misc.
2d 601 (N.Y. Misc. 2002).
the-matter-of-du] appears to mark the first
prosecution of an accused swiper under forgery laws.
4. The secrecy survived two public inquiries into chronic
errors and double debitsexperienced by users. See
reports by New York Citys Public Advocate
(Sheppard, Chernobilsky and Mintz-Roth 2005), and
an oversight committee of the New York City Council
(Feerick and Gupta 1998), both on file with author.
5. New York City Administrative Code § 14-172 requires
the NYPD to post data on enforcement of Penal Law
§ 165.15(3) and on TAB/NOV summonses. See New
York City Council (2018).
6. In 2017, the Manhattan District Attorneys Office
announced it would no longer prosecute standalone
Theft of Service arrests (McKinley 2017) the only of
five city jurisdictions to adopt this policy. However, the
policy does not apply to defendants deemed threats to
public safety. Moreover, the NYPD continues to make
standalone TOS arrests in the Manhattan jurisdiction
(1,611 in 2018, according to NYC OpenData, 2019),
stemming from enforcement which may lead to the
discovery of additional charges, being taken into police
custody, and other consequences.
7. Phone discussion, Jarret Hova, Director of Policy, and
Counsel, Office of the Public Advocate, City of New
York, November 1, 2018.
8. See People v. Robinson.
9. Interviews stem from a broader project which also
included extensive ethnographic observation in the
subway system.
10. E.g. MTA New York City Transit (1999), on file with
the author.
11. Specifically, a farecard system patent assigned to Cubic
(Aubrey 1992) operates on the principle that a
turnstile computer would never, in any one swipe,
read a fare total from, and write a new fare total to,
the same value-field. The Mattocks account did not
explain how these processes were accomplished
separately, glossing the steps between a readand a
metrocard no longer able to provide an unpaid entry.
12. Each of the five counties of New York City has an
elected District Attorney, leading to variation across
13. Freedom of Information Law request to New York
Police Department #2013-0212, issued December 2012,
rejected twice, reopened twice on appeal, and partly
fulfilled in November 2015 as statistics on Transit
Bureau arrests of swipers for six forgery-related
charges from 2002 through 2014.
14. One industrial-scale slug-making enterprise was broken
up following a police raid on a workshop containing
hundreds of thousands of fake tokens. Its leader, a
former graduate student in sociology at the New School
(Dwyer 2004; Sullivan 1991) was charged with Criminal
Possession of a Forged Instrument in the 2
Degree, the
same felony leveraged against swipers years later.
15. See Prichard (1991), on file with author.
16. E.g. Metropolitan Transportation Authority (2011), on
file with author.
17. While the MTA is able to instruct turnstile computers
to reject certain metrocards (for example, ones that
were reported as stolen and were traceable through a
purchase record) by serial numbers, this bad listis
updated on each turnstile computer just once per day,
and is not usable for a real-time intervention.
18. See People v. Hightower. Oral arguments of November
14, 2011 available as video at https://www.nycourts.
(accessed on March 13, 2019); see comments of Hon.
R. Smith at 12:20.
19. Penal Law §165.16. Also see the 2004 Executive
Memorandum in support of the bill describing its
purpose, excerpted in Part III of the uncorrected
Mattocks (2009) decision, at
nyctap/I09_0062.htm (accessed on March 13, 2019).
20. Data refer to arrests in which Penal Law §165.16 is the
most serious, or topcharge. Data come from New York
State Division of Criminal Justice Services Computerized
Criminal History Database, on file with author.
21. Interview conducted via telephone, January 19, 2013.
22. These cases were generally charged under Criminal
Possession of a Forged Instrument in the 2
rather than Forgery in the 2
Degree, very likely
because bent metrocards act as physical evidence, and a
defendant cannot be convicted of both crimes for the
same instrument (see New York Penal Law § 170.35).
23. Email correspondence between the author and Danny
Frost, Director of Communication, Manhattan D.A.s
Office, Jan 22 and Jan 29, 2019.
24. People v. Curry.
25. People v. Solomon.
26. While Kohler-Hausmanns(2013,2018) research
follows trajectories of misdemeanor arrests, those
trajectories illustrate, if in broad outlines, felony
arrests which do not result in felony charges.
27. Compare People v. Richardson with People v. Santos.
28. NY Penal Law § 170.00(5, 6).
29. People v. Lopez, cited in Goell (2006/7).
30. People v. Mattocks 2005.
31. New York State Penal Law § 170.00 (6); see
also 170.10(3).
32. People v. Mattocks 2009. Oral arguments available
as video via request from Albany Law School
Government Law Center,
ctapps/forms/VideoForm.pdf (accessed on March 22,
2019), and on file with author.
33. Mattocksattorney erred. The computer establishes the
presumptive value of a metrocard based on the lowest
fare total even when there is only one legible fare total
on the card.
34. As we shall see, the ADAs claim was substantially
wrong. The turnstile computer always detects an
instance in which it cannot read a value field.
35. See Jeffreys (1984), on file with author.
36. To process a fare, the turnstile computer must also be able
to read the cardserial number, and other data, which
remains legible to the scanner despite the swipersbend.
37. This is precisely the reason why when users, after
experiencing errors at one turnstile try their luck at
another and sometimes find that their metrocards
have somehow lost value (see Blair 1998).
38. New York forgery laws define a falsely alteredinstrument
as one changed by means of erasure, obliteration, deletion,
insertion of new matter, transposition of matter or any
other manner(§170.090(6)).
39. Middle-class populations are better insulated from
such a radical labeling of mundane malfeasance.
Compare, for example, the case of selling swipes with
the use of plastic covers that obscure automotive license
plates to overhead cameras used to bill motorists for
tolls. Reportedly popular on the personal vehicles of
police officers (Cohen and Marsh 2017), their use
remains in the realm of civil violation. While they
prevent the collection of tolls, the perspective of the
deceived camera (or some similar technical reading) has
not been made to matter as a version of fraud or forgery.
40. The hackers noted that multiple metrocards bearing
the same serial number would lead to great imbalances
between fares used and fares purchased. Once noticed
by the central mainframe, that serial number will be
placed on a bad listupdated to every turnstile
computer each day. See video-recorded presentation of
Red Balaclava (1997) especially at 15:30-25:00.
41. In another example, when sophisticated media file
sharers migrated to file swarming (or torrents) in
which users download small fragments of a media file
from many sources instead of whole files from just
one, the distributed nature of the sources and
recipients made the legal interpretation of their
activity more ambiguous, more difficult to investigate,
and largely shielded those users from mass litigation at
least for a period of years (Caraway 2011).
42. For example, we learn that sending an email to a
neighbor becomes overseas communicationand can
therefore be read by the U.S. National Security
Administration when our message is, for whatever
reason, routed through overseas servers (Goldberg 2017).
43. The market for swipersservices was diminished in the
late 2009 when new emergency gates at subway
entrances offered nonpaying passengers a ready way in
(Lysiak and Goldiner 2009). Soon after, used
metrocards, the fundamental stock-in-trade of swipers,
became scarce thanks to a $1 surcharge added to the
purchase of new metrocards.
44. Hacker W. Woodruff found his (2016) reading of the
metrocard consistent with the prior discoveries of
Redbird (2005a), but could not discern the same
sentinels which mark the beginning and end of
data sequences.
The author expresses sincere gratitude to Harvey Molotch,
Ashley Mears, Owen Whooley and Christena Nippert-Eng
for critical feedback across the duration of this project. The
author is thankful for the insight or help of Daniel Bodah,
Diana Graizbord, Marie Hicks, Harry G. Levine, Ben
Merriman, participants in the Northwestern Ethnography
Workshop, and co-panelists and audiences at the meetings
of the Eastern Sociological Association, the American
Sociological Association and the Society for Social Studies
of Science. Daniya Kamran-Morley designed the Figures
with the authors input and is responsible for all of their
virtues and none of their flaws.
This work was partly supported under grants from the
National Science Foundation (SES-0542777) and the Harry
Frank Guggenheim Foundation.
Angwin, J., J. Larson, S. Mattu, and L. Kirchner. 2016.
Machine bias: Risk assessments in criminal sentencing.
ProPublica, May 23. Available at: https://www.propublica.
org/ (accessed on June 6, 2018).
Aubrey, K. R. 1992. Fare card read-writer which overwrites
oldest or invalid data. U.S. Patent 5,298,726 filed January
7, 1992, and issued March 29, 1994. Available at: https:// (accessed on June
13, 2019).
Balaclava, R. 1994. The New York City Metrocard. Paper
presented at HOPE: Hackers on Planet Earth conference.
New York, August. Video available at: (accessed on June 7,
Balaclava, R. 1997. Metrocard. Presentation at Beyond
HOPE conference, New York, August. Video available at: (accessed
on March 18, 2019).
Barocas., S, and A. D. Selbst. 2016. Big datas disperate
impact. California Law Review 104 (3):671732.
Blair, J. 1998. Vallone cites double-charging in metrocards.
New York Times, July 2. Available at: https://www.
charging-in-metrocards.html (accessed on June 13, 2019).
blueski-mask, and the wrapper. 1997. The ins and outs of
metrocard gold. 2600: A Hacker Quarterly 15:4448.
Available at:
14-2_Page_23.jpg (accessed on June 12, 2019).
Braun, L. 2014. Breathing race into the machine: The supris-
ing career of the spirometer from plantation to genetics.
Minneapolis, MN: University of Minnesota Press.
Caraway, B. 2011. Resistance commons: File-sharing litiga-
tion and the social system of commoning. Doctoral dis-
sertation, University of Texas, Austin.
Cekander, E. J. 1994. Design of a computer controlled turn-
stile. In Proceedings of WESCON 94, 444-448. New York:
Chan, S. 2009. Appeal fails in forgery of metrocard. New
York Times, April 30. Available at: https://www.nytimes.
ResultPosition=4 (accessed on June 14, 2019).
Cohen, S., and J. Marsh. 2017. NYPD cops are using illegal
license plate covers to thwart traffic cameras. New York
Post, February 9. Available at:
thwart-traffic-cameras/ (accessed on June 6, 2019).
Coleman, C. 1998. Hundreds swiping rides reusing expired
cards on subway. New York Daily News, January 8, 23.
Crawford, K., and J. Schultz. 2014. Big data and due pro-
cess: Toward a framework to redress predictive privacy
harms. Boston College Law Review 55 (1):129.
Daly, M. 1999. Swiping a subway ride. New York Daily
News. December 8, 7.
Dimaggio, P., E. Hargittai, C. Celeste, and S. Shafer. 2004.
Digital inequality: From unequal access to differentiated
use. In Social inequality, ed. K. M. Neckerman, 355400.
New York: Russell Sage Foundation.
Donohue, P. 2005. Swipe guys get free ride: MetroCard law
still snarled. in New York Daily News. August 8, 8.
Dwyer, J. 2004. 19042004; Fare-beater, Inc. New York Times,
March 28. Available at:
28/nyregion/1904-2004-fare-beater-inc.html (accessed on
June 12, 2019).
Eubanks, V. 2018. Automating inequality: How high-tech
tools profile, police and punish the poor. New York: St.
Martins Press.
Feerick, D., and A. Gupta. 1998. Please swipe again, and
again, and again, and again A survey of Metrocard
users. New York: New York City Council Office of
Oversight and Investigation.
Flagg, A., and A. Nerbovig. 2018. Subway policing in New
York City still has a race problem. The Marshall Project,
Septermber 12, 2018.
race-problem (accessed on June 12, 2019).
Giannelli, P. C. 2004. AKE v. Oaklahoma: The right to
expert asssitance in a post-Daubert, post-DNA world.
Cornell Law Review 89 (6):13051419.
Giannelli, P. C. 2018. Forensic science: Daubertsfailure.
Case Western Reserve Law Review 68 (3):869941.
Goell, G. 2006/7. People v. Lopez.New York Law School
Law Review 51 (3):72129.
Goldberg, S. 2017. Surveillance without borders: The traffic
shapingloophole and why it matters. The Century
Foundation, June 22. Available at:
hole-and-why-it-matters/ (accessed on June 12, 2019).
Goodwin, C. 1994. Professional vision. American
Anthropologist 96 (3):606633. doi: 10.1525/aa.1994.96.3.
Harcourt, B. E. 2001. Illusion of order: The false promise of
broken windows policing. Cambridge, MA: Harvard
University Press.
Harmon, A. 1997. What galls a hacker most? The
Metrocard. New York Times, August 11. Available at:
galls-a-hacker-most-the-metrocard.html (accessed on June
11, 2019).
Hoeffel, J. C. 1990. The dark side of DNA profiling:
Unreliable scientific evidence meets the criminal defend-
ant. Stanford Law Review 42 (2):465538. doi: 10.2307/
Jasanoff, S. 1997. Science at the bar: Law, science, and tech-
nology in America. Cambridge, MA: Harvard University
Jasanoff, S. 1998. The eye of everyman: Witnessing DNA in
the Simpson trial. Social Studies of Science 28 (5/6):
713740. doi: 10.1177/030631298028005003.
Jasanoff, S. 2016. The ethics of invention. New York: W.W.
Jeffreys, D. C. 1984. Evaluation of automatic fare collection
technology: Report to the MTA, volume II: Technology
assessment. Cambridge, MA: Arthur D. Little, Inc.
Joh, E. E. 2017a. The undue influence of surveillance tech-
nology companies on policing. New York University Law
Review 91:101130.
Joh, E. E. 2017b. Feeding the machine: Policing, crime data
& algorithms. William & Mary Bill of Rights Journal 26
Kohler-Hausmann, I. 2013. Misdemeanor justice: control
without conviction. American Journal of Sociology 119
(2):351393. doi: 10.1086/674743.
Kohler-Hausmann, I. 2018. Misdemeanorland: Criminal
courts and social control in an age of broken windows
policing. Princeton, NJ: Princeton University Press.
Lancman, R. I. 2018. NYPD disclosure of enforcement data
fails to comply with fare evasion reporting law. Council.
NYC.Gov, October 3. Available at:
(accessed on March 6, 2019).
Latour, B. 1999. Pandoras hope. Cambridge, MA: Harvard
University Press.
Law, J. 1991. Introduction: Monsters, machines and socio-
technical relations. In A sociology of monsters: Essays on
power, technology and domination, ed. J. Law, 123.
London: Routledge.
Leonardi, P. M. 2009. Crossing the implemenation line: The
mutual constitution of technology and organizing across
development and use activities. Communication Theory
19 (3):278310. doi: 10.1111/j.1468-2885.2009.01344.x.
Levine, D. S. 2007. Secrecy and unnacountability: Trade
secrets in our public infrastructure. Florida Law Review
59 (1):135193.
Luo, M. 2004. Subway headaches: Metrocard devices often
need repairs. The New York Times, February 3. Available
(accessed on June 13, 2019).
Lynch, M. 1998. The discursive production of uncertainty:
The O.J. Simpson dream teamand the sociology of
knowledge machine. Social Studies of Science 28(5/6):
829868. doi: 10.1177/030631298028005007.
Lynch, M., and S. Cole. 2005. Science and technology stud-
ies on trial: Dilemmas of expertise. Social Studies of
Science 35 (2):269311. doi: 10.1177/0306312705048715.
Lysiak, M., and D. Goldiner. 2009. Exit gates as MTAs
freeway, fare-beaters use busy stops to advantage. New
York Daily News. April 30. Available at: https://www.
busy-stops-advantage-article-1.408168 (accessed on June
12, 2019).
MacKenzie, D. 2005. Opening the black boxes of global
finance. Review of International Political Economy 12 (4):
555576. doi: 10.1080/09692290500240222.
McKinley, J. 2017. For Manhattan fare beaters, one-way
ticket to court may be over. New York Times, June 30.
Available at:
gion/subway-fare-beating-new-york.html (accessed on
June 13, 2019).
Mellon, J. N. 2001. Manufacturing convictions: Why
defendants are entitled to the data underlying forensic
DNA kits. Duke Law Journal 51 (3):1097137. doi: 10.
Metropolitan Transportation Authority. 2011. Concept of
operations for MTA new fare payment system: Bus and
subway deployment phase, Vol. 1.2. New York:
Metropolitan Transportation Authority.
Michael, K., and M.G. Michael. 2009. Innovative automatic
indentification and location-based services: From bar codes
to chip implants. New York: Information Science
Mnookin, J. L. 2008. Of black boxes, instruments and
experts: Testing the validity of forensic science. Episteme
5 (3):343358. doi: 10.3366/E1742360008000440.
Molotch, H, and D. Boden. 1985. Talking social-structure:
Discourse, domination and the Watergate Hearings.
American Sociological Review 50 (3):27388. doi: 10.2307/
MTA New York City Transit. 1999. MetroCard operations
training course: Station agent inductee participants
guide. San Diego, CA: Cubic Automatic Revenue
Collection Group.
Murphy, E. 2007. The new forensics: Criminal justice, false
certainty, and the second generation of scientific evi-
dence. Californa Law Review 95 (3):72197.
Murphy, E. 2015. Inside the cell: The dark side of forensic
DNA testing. New York: Nation Books.
Neff, G., T. Jordan, J. McVeigh-Schultz, and T. Gillespie.
2012. Affordances, technical agency, and the politics of
technologies of cultural production. Journal of
Broadcasting & Electronic Media 56 (2):299313. doi: 10.
New York City Council. 2018. A local law to amend the
administrative code of the City of New York, in relation
to requiring the police department to report on the num-
ber of arrests and summonses retrunable to the Transit
Adjudication Bureau for subway fare evasion. New York:
City of New York. Available at:
3106792 (accessed on June 18, 2019).
NYPD. 2019a. Subway fare evasion reports, 4
Quarter, 2019. New York: New York Police
Department. Available at:
(accessed on March 4, 2019).
NYPD. 2019b. Procedure 209-39: Personal service of Transit
Adjudication Bureau notice of violation (TAB/NOV) in
lieu of arrest for a qualified individual. In NYPD Patrol
Guide (loose-leaf). New York: New York Police
Department. Available at:
209-39-transit.pdf (accessed on March 30, 2019).
Newman, A. 1998. Transit agency planning to reprogram
turnstiles. New York Times, January 9. Available at:
planning-to-reprogram-turnstiles.html (accessed on June 13,
NYC OpenData. 2019. NYPD arrest data, updated March 8,
2019. New York: City of New York. Available at: https://
to-Date-/uip8-fykc (accessed on March 18, 2019.)
Orlikowski, W., and D. Gash. 1994. Technology frames:
Making sense of information technology in organizations.
ACM Transactions on Information Systems 12 (2):
174207. doi: 10.1145/196734.196745.
Oudshoorn, N., and T. Pinch. 2003. How users matter: The
co-construction of users and technologies. Cambridge, MA:
MIT Press.
Pasquale, F. 2015. The black box society: The secret algo-
rithms that control money and information. Cambridge,
MA: Harvard University Press.
People of the State of New York v. Cornell Curry., 71 AD3d
418 (App. Div. 2010).
People of the State of New York v. Joseph Hightower., 18
NY3d 249 (N.Y. 2011).
People of the State of New York v. Radames Lopez., 797
N.Y.S.2d 893 (Crim. Ct. Kings County 2005).
People of the State of New York v. Jonathan Mattocks.,WL
6392095 (N.Y.Sup. 2005).
People of the State of New York v. Jonathan Mattocks., 12
NY3d 326 (N.Y. 2009).
People of the State of New York v. Donald Robinson., 860
N.Y.S.2d 159 (N.Y. App. Div. 2008).
People of the State of New York v. Vincent Richardson.,18
Misc 3d 89 (App. Term 2007).
People of the State of New York v. Randolph Santos.,17
Misc 3d 520 (Crim. Ct., New York County 2007).
People of the State of New York v. Bernard Solomon.,78
AD3d 521 (N.Y. App. Div 2010).
Pierre-Pierre, G. 1996. Transit Authority is planning to cut
800 more token clerks. New York Times, March 12.
Available at:
clerks.html (accessed on June 12, 2019).
Pinch, T., and W.E. Bijker. 1984. The social construction of
facts and artifacts. Social Studies of Science 14 (3):
399431. doi: 10.1177/030631284014003004.
Pritchard, J. 1991. Review of the MTA capital program.
New York: Office of the Inspector General, Metropolitan
Transportation Authority.
Redbird [J. Battaglia]. 2005a. New York Citys MTA
exposed! 2600: A Hacker Quarterly, Spring:713.
Redbird [J. Battaglia]. 2005b. Magnetic stripe reading. 2600:
A Hacker Quarterly, Spring:28-49.
Redbird [J. Battaglia]. 2006. Magnetic stripe technology and
the New York City Metrocard. Presentation at HOPE
Number 6 conference. New York. Video available at:
(accessed on June 13, 2019).
Reddy, A., A. Lu, S. Kumar, V. Bashmakov, and S.
Rudenko. 2011. Entry-only automated fare collection
(AFC) system data used to infer ridership, rider
destinations, unlinked trips, and passenger miles.
Transportation Research Record 2110 (1):128136. doi: 10.
Reddy, A., J. Kuhls, and A. Lu. 2011. Measuring and con-
trolling subway fare evasion. Transportation Research
Record 2216 (1):8599. doi: 10.3141/2216-10.
Rios, V. M. 2011. Punished: Policing the lives of Black and
Latino boys. New York: New York University Press.
Robinson, L., S. R. Cotten, H. Ono, A. Quan-Haase, G.
Mesch, W. Chen, J. Schulz, T. M. Hale, and M. J. Stern.
2015. Digital inequalities and why they matter.
Information, Communication & Society 18 (5):569582.
doi: 10.1080/1369118X.2015.1012532.
niz, M. 2016. Bridgework: STS, sociology, and
the dark mattersof race. Engaging Science, Technology,
and Society 2:22426. doi: 10.17351/ests2016.74.
Rohde, D. 1998. Two are charged in a scheme to ride free
on metrocards. New York Times, January 8. Available at:
charged-in-a-scheme-to-ride-free-on-metrocards.html (accessed
on June 12, 2019).
Rutenberg, J. 1998a. Metrocard scam derailed by MTA. Pp.
30 in New York Daily News, January 16, 30.
Rutenberg, J. 1998b. TA seeks card fix that isnt bad deal.
New York Daily News, January 10, 4.
Saks, M. J., and David F. 2008. Failed forensics: How foren-
sic science lost its way and how it might yet find it.
Annual Review of Law and Social Science 4:149171. doi:
Sheppard, J.E., Y. Chernobilsky, and J. Mintz-Roth. 2005.
Stuck at the turnstiles: Failed swipes slow down subway
riders (A report by Public Advocate Betsy Gotbaum).
New York: Office of the Public Advocate, City of New
York. Available at:
govpub/2723stuck_at_the_turnstile.pdf (accessed on July
4, 2019).
Smith, A. 2018. Franken-algorithms: The deadly consequen-
ces of unpredictable code. The Guardian, August 30.
Available at:
(accessed on June 12, 2019).
Star, S. L. 1999. The ethnography of infrastructure.
American Behavioral Scientist 43 (3):377391. doi: 10.
Stewart, N. 2018. Selling Metrocard swipes remains
illegal, and a way of life. New York Times, June 17.
metrocard-swipes-arrests.html (accessed on June 19,
Stolper, H, and J Jones. 2017. The crime of being short
$2.75: Policing communities of color at the turnstile.
New York: Community Service Society. Available at:
short-2.75 (accessed on June 12, 2019).
Stuart, F. 2016. Becoming copwise: Policing, culture, and
the collateral consequences of street-level criminalization.
Law & Social Inquiry 50 (2):279313. doi: 10.1111/lasr.
Sullivan, R. 1991. Raid cuts off slugs for subway cheats.
New York Times, June 25. Available at; https://www.
subway-cheats.html (accessed on June 12, 2019).
Vaughan, D. 1996. The Challenger launch decision: Risky
technology, culture, and deviance at NASA. Chicago:
University of Chicago Press.
Weidner, R. Target-hardening at a New York City subway
station: Decreased fare-evasion, at what price? In
Preventing mass-transit crime, ed. R.V. Clarke, 117132.
Monsey, NY: Criminal Justice Press.
Wexler, R. 2018. Life, liberty and trade secrets: Intellectual
property in the criminal justice system. Stanford Law
Review 70 (5):13431429.
Winner, L. 1986. The whale and the reactor. Chicago:
University of Chicago Press.
Winston, Ali. 2018. Whatsatstakeinfightforsubwayfare-
beating data. New York Times September 19. Available at:
way-turnstiles-data.html (accessed on September 20, 2018).
Woodruff, W. 2016. Decoding the metrocard, part 2.5:
Early results. Enosuchblog (blog). October 7. Available at:
Part-2.5 (accessed on March 14, 2019).
... This is especially troubling as algorithms are increasingly used to make consequential decisions with high stakes implicationscredit and loan qualification, child protection, hiring and promotion in jobs, healthcare and insurance, and even prison sentencing. The potential built-in racial, gender, class, and other biases in big data and algorithms have been shown to be dangerous and destructive across many life realms including prison sentencing (Barocas and Selbst, 2016;Noble, 2018;McClain, 2019). ...
... Software for assessing an offender's risk of reoffending is routinely used in courtrooms in the U.S. (Angwin, et al., 2016), and data from body cameras and gunshot detection devices are frequently treated as objective evidence of sound police work (Merrill, 2017). Yet, studies of the efficacy of such technologies have uncovered low success rates (Dror and Mnookin, 2010;Merrill, 2017), as well as biases based on race, ethnicity, gender, and socioeconomic status (Angwin, et al., 2016;McClain, 2019;Noble, 2018;Nunn, 2001). Using facial imagery and DNAbased technologies to uncover crime patterns, for example, has disproportionately focused on people of color, thereby reinforcing the criminalization of minority group members and increasing their risk of stigmatization (Machado and Granja, 2020;Skinner, 2020). ...
Full-text available
Marking the 25th anniversary of the “digital divide,” we continue our metaphor of the digital inequality stack by mapping out the rapidly evolving nature of digital inequality using a broad lens. We tackle complex, and often unseen, inequalities spawned by the platform economy, automation, big data, algorithms, cybercrime, cybersafety, gaming, emotional well-being, assistive technologies, civic engagement, and mobility. These inequalities are woven throughout the digital inequality stack in many ways including differentiated access, use, consumption, literacies, skills, and production. While many users are competent prosumers who nimbly work within different layers of the stack, very few individuals are “full stack engineers” able to create or recreate digital devices, networks, and software platforms as pure producers. This new frontier of digital inequalities further differentiates digitally skilled creators from mere users. Therefore, we document emergent forms of inequality that radically diminish individuals’ agency and augment the power of technology creators, big tech, and other already powerful social actors whose dominance is increasing.
... It is a well built and updated resource with detailed information revealing no personal data and open for public use. Two datasets, NYPD Arrest Data [31] comprising 1,03,000 rows and 19 columns and Motor Vehicle Collisions-Crashes ( comprising 1.73 M rows and 29 columns, were studied thoroughly and combined, keeping only the columns that were desired attributes like latitude, longitude, accidents occurred, number of cyclist/motorist/pedestrians injured and killed, crime type comprising categories like dowry deaths, honor killings, exploitation by husband/relatives etc. and dropped columns like arrest key, jurisdiction code, law code and many more which were describing the points directly not linked to the danger index and insignificant in its calculation. ...
Full-text available
The study aims to formulate a solution for identifying the safest route between any two inputted Geographical locations. Using the New York City dataset, which provides us with location tagged crime statistics; we are implementing different clustering algorithms and analysed the results comparatively to discover the best-suited one. The results unveil the fact that the K-Means algorithm best suits for our needs and delivered the best results. Moreover, a comparative analysis has been performed among various clustering techniques to obtain best results. we compared all the achieved results and using the conclusions we have developed a user-friendly application to provide safe route to users. The successful implementation would hopefully aid us to curb the ever-increasing crime rates; as it aims to provide the user with a beforehand knowledge of the route they are about to take. A warning that the path is marked high on danger index would convey the basic hint for the user to decide which path to prefer. Thus, addressing a social problem which needs to be eradicated from our modern era.
... Judicial decisions in criminal prosecutions can render the police uses of new surveillance techniques, as well as arguments supporting their necessity and legitimacy, more visible. However, complex expert evidence can also render everyday technological processes opaque, which can reinforce unfairness in criminal trials (McClain 2019). We argue that the legal sanctioning of intrusive surveillance practices deployed by SIGINT agencies for foreign intelligence and national security objectives is extending into routine criminal investigations on both the surface and dark webs. ...
Full-text available
This paper examines Lisa Austin’s (2015) concept of lawful illegality, which interrogates the legal foundations for potentially unlawful surveillance practices by United States (US) signals intelligence (SIGINT) agencies. Lawful illegality involves the technically lawful operation of surveillance powers that might be considered unlawful when examined through a rule of law framework. We argue lawful illegality is expanding into domestic policing through judicial decisions that sanction complex and technically sophisticated forms of remote online surveillance, such as the use of malware, remote hacking, or Network Investigative Techniques (NITs). Operation Pacifier targeted and dismantled the Playpen dark web site, which was used for distributing child exploitation material (CEM), and has generated many judicial rulings examining the legality of remote surveillance by the FBI. We have selected two contrasting cases that demonstrate how US domestic courts have employed distinct logics to determine the admissibility of evidence collected through the NIT deployed in Operation Pacifier. The first case, United States v. Carlson (2017 US Dist. LEXIS 67991), offers a critical view of the use of NITs by the FBI, with physical geography constraining the legality of this form of surveillance in US criminal procedure. The second case, United States v. Gaver (2017 US Dist. LEXIS 44757), authorizes the use of NITs because the need to control crime is believed to justify suspending the geographic limits on police surveillance to identify people involved in the creation and dissemination of CEM. We argue this crime control emphasis expands the reach of US police surveillance while undermining due process of law by removing the protective function of geography. We conclude by suggesting the permissive geographic scope of police surveillance reflected in United States v. Gaver (2017 US Dist. LEXIS 44757), and many other Playpen cases, erodes due process for all crime suspects, but is particularly acute for people located outside the US, and suggest a neutral transnational arbiter could help limit contentious forms of remote extraterritorial police surveillance.
This chapter examines various levels of human and non-human-mediated forms of bias in place-based predictive policing that may lead to discriminatory outcomes. Informed by science and technology studies (STS), and drawing on empirical data on the implementation and utilization of crime prediction software in Germany and Switzerland, we grasp predictive policing as a socio-technical assemblage, encompassing not only the technical predictions themselves but also the enactment of the predictions on the street level by police—which can also have serious ramifications including discrimination. Further, we consider the broader socio-political and historical contexts of these predictive technologies and dispute technocentric accounts of discrimination in predictive policing. Rather, we argue for greater attention to be paid to the socio-political and historical contexts from which such technologies and practices emerge. Hence, we contend that STS approaches are vital for an appropriate understanding of the discriminatory potency of (place-based) predictive policing, as we need to decenter technology in accounts of discrimination in predictive policing. Our argument unfolds across four steps. First, we present a short introduction to predictive policing and highlight its socio-technical nature. Second, we analyze the main dimensions of discrimination sources in place-based predictive policing. Third, we argue for the need to decenter technology and call for an appreciation of broader socio-political and historical perspectives in the analysis of predictive policing. Fourth and finally, we argue for the need to incorporate science and technology studies (STS) into the analysis of the discriminatory potential of predictive policing.
Full-text available
Privilege afforded to a certain urban social milieu is an overlooked counterpoint to the aggressive policing and prosecution of the urban poor. This article investigates the divergent policing of two very similar activities which both involve manipulating information-containing objects to deceive computers deployed to read them. In the first case, subway farecards manipulated by the urban poor to enter the subway without paying are treated as forgeries, and nine in ten arrestees for possessing these forgeries under felony laws are Black or Hispanic. In the latter case, motorists obscure their license plates to avoid bridge and tunnel tolls but risk no more than a civil summons. The practice is strongly associated with police personnel themselves and a related milieu which commands "street-level privilege" which tends to exempt them from certain types of regulation. In spite of great formal similarities, we find that one practice is criminalized through a "technical logic" which frames human action in relation to a technology, while the other practice has not been subject to that same-or any other-criminalizing logic at all. The comparison foregrounds "under-policing" of some as an incisive contrast to the over-policing of others.
Full-text available
p>This short essay reflects on intellectual bridges that scholars have built, are building, and could build to connect critical sociologies of race and STS. Whereas much work in these respective fields have rarely intersected, greater exchange could help scholars better account for ways in which race shapes and stratifies contemporary societies. To this end, the essay begins with a recent example of bridgework—research on race and genetics. Next, I use my own research on ethnoracial statistics to describe how bridgework happening elsewhere can indirectly create openings for connections across the divide. Finally, I propose that research on the broader sociotechnical materiality of race and racial domination represents an important site for further bridgework.</p
The criminal justice system is becoming automated. At every stage, from policing to evidence to parole, machine learning and other computer systems guide outcomes. Widespread debates over the pros and cons of these technologies have overlooked a crucial issue: ownership. Developers often claim that details about how their tools work are trade secrets and refuse to disclose that information to criminal defendants or their attorneys. The introduction of intellectual property claims into the criminal justice system raises undertheorized tensions between life, liberty, and property interests. This Article offers the first wide-ranging account of trade secret evidence in criminal cases and develops a framework to address the problems that result. In sharp contrast to the general view among trial courts, legislatures, and scholars alike, this Article argues that trade secrets should not be privileged in criminal proceedings. A criminal trade secret privilege is ahistorical, harmful to defendants, and unnecessary to protect the interests of the secret holder. Meanwhile, compared to substantive trade secret law, the privilege overprotects intellectual property. Further, privileging trade secrets in criminal proceedings fails to serve the theoretical purposes behind either trade secret law or privilege law. The trade secret inquiry sheds new light on how evidence rules do, and should, function differently in civil and criminal cases.