The Carceral Archive: Documentary Records, Narrative Construction, and Predictive Risk Assessment

Over the past three decades, scholarship in sociology, criminology, law, public policy, history, science and technology studies (STS), and other allied fields has addressed various areas of concern at the intersections of race and correctional supervision in the United States. Some scholars have argued that the modern day prison industrial complex is a transparent extension of chattel slavery, while others have asserted that a ubiquity of carceral ideologies has contributed to a broader social inequality—one so affecting as to prevent people from oppressed and marginalized communities from full participation in American life. This scholarship unambiguously attests to ongoing disparities in rates of incarceration calculated along the indices of race and ethnicity; with data cultures expanding—evidenced by growth in sectors such as big data, data science, and data analytics—in the sociocultural landscapes and sociotechnical environments of the United States, more recent scholarship has focused on the ways that technology also contributes to—and often exacerbates—these disparities.

Over the past three decades, scholarship in sociology, criminology, law, public policy, history, science and technology studies (STS), and other allied fields has addressed various areas of concern at the intersections of race and correctional supervision in the United States. Some scholars have argued that the modern day prison industrial complex is a transparent extension of chattel slavery, while others have asserted that a ubiquity of carceral ideologies has contributed to a broader social inequality-one so affecting as to prevent people from oppressed and marginalized communities from full participation in American life. 12 This scholarship unambiguously attests to ongoing disparities in rates of incarceration calculated along the indices of race and ethnicity; with data cultures expanding-evidenced by growth in sectors such as big data, data science, and data analytics-in the sociocultural landscapes and sociotechnical environments of the United States, more recent scholarship has focused on the ways that technology also contributes to-and often exacerbates-these disparities. 1 Just as scholarship on correctional supervision was shifting to reckon with data and its attendant technologies, archivists-those who appraise, manage, preserve, and provide access to data-and archival studies scholars were naturally also embracing technological concerns. More recently, however, a body of archival scholarship has also emerged that speaks to concerns about race, representation, and the evidential value of documentary records. 2 Yet, despite the many overlapping interests of scholars, archivists, and activists working on data justice concerns, these two particular areas of inquiry-that which investigates the relationships among race, data, and law enforcement and that which sits at the nexus of race, data, and the documentary record-have remained distinct. This paper interweaves these discussions, examining how law enforcement data practices in the United States ( e.g. surveillant data mining, algorithmic primacy in offender risk assessment) both draw upon and generate documentary records and risk narratives that propagate a carceral archive, and how this carceral archive in turn perpetuates discriminatory practices in the criminal justice system. I begin by introducing and defining the carceral archive, then argue that law enforcement data practices create and sustain the carceral archive by appropriating and exploiting social media data cultures. Next, I interrogate how the carceral archive draws upon, contributes to, and perpetuates the risk narratives that are used by the state to justify the need for compulsory confinement. Connecting the dots between our current system of mass incarceration and earlier forms of social control, I argue that data cultures in contemporary policing have not been inaugurated by new technologies; instead, these ongoing practices are built on the violent history of chattel slavery and appeals to white panic constructed around notions of order and "safety. " Finally, I offer a vision for the liberatory potential Cultural Analytics The Carceral Archive of archives-and the archive-in contraposition to the carceral archive.
To conceptualize the carceral archive specifically as it relates to data justice, I engage critical archival studies, digital culture studies, and critical race theory as analytical frameworks. To frame my discussion of carceral archives and "the carceral archive, " I employ critical archival studies. Per archival studies scholar Michelle Caswell, critical archival studies: "(1) explains what is wrong with the current state of archival and recordkeeping practice and research and identifies who can change it and how; (2) posits achievable goals for how archives and recordkeeping practices and research in archival studies can and should change; and (3) provides norms and strategies and mechanisms for forming such critique. " 3 Caswell further asserts that, "Critical archival studies, like critical theory in general, is unapologetically emancipatory in nature. " 4 In critical archival studies, the transformation of archival practice and research is the ultimate goal. To frame an understanding of the social media landscape in which the surveillance of data cultures thrive, I employ digital culture studies. Digital culture studies asks how new technologies reflect the wider social world offline, how they create new cultural interactions, and how those new interactions reshape the real (non-virtual) world. Finally, throughout my analysis I employ critical race theory (CRT). CRT is a broad theoretical framework stemming from a synthesis of scholarly work that challenges dominant understandings of race and the law. For the purposes of this article, CRT provides the framework for a critique of the role of race and racism in the criminal justice system; CRT also aids in creating space for thinking about more liberatory data and archival praxes.

Cultural Analytics
The Carceral Archive allowing the collections and items to speak for themselves. In prison institutions, the archivist serves only the interpretations of an oppressive higher power. 9 Applying Benjamin's carceral imaginary to archival studies, Jimerson's prison institutions might be likewise designated"carceral archives. " Carceral archives are analogous to Jimerson's prison institutions in that they are those archiveswherein archives are simply and elegantly defined as "collections of records, the institutions that steward them, and the processes that designated them archival"-that fundamentally serve the interpretations of oppressive state powers. 10 A full analysis of state-sponsored archives as carceral institutions requires lengthy discussion, and warrants full treatment in a separate work. However, because the focus of this work is the carceral archive-in which carceral archives play an important role-I will briefly outline three examples of archives as carceral, paying specific attention to collections, institutions, and processes.
The defining characteristic of carceral archives is that they comprise recordsor collections of records-held by institutions, and that one or both work in the service of oppressive state powers. To offer a simple but important example, state records often provide the evidentiary basis to legitimize the power to punish. 11 As I will discuss in more detail later, both the policing and prison endeavors in the US use state records in risk assessment; recidivism and sentencing determinations; and pretrial decisions. These records are components of a carceral system that both generates and draws upon carceral archives. Similarly, recent archival studies scholarship has criticized American archives that decline to collect and make available evidence of human rights abuses that would support restorative and transitional justice, particularly for Black and Indigenous people. Archivist Lae'l Hughes-Watkins argues that these carceral archives, in their refusal to collect in the service of justice, are effectively "coconsipirator[s] in the violence against Black bodies. " 12 9 Randall C. Jimerson, Archives Power: Memory, Accountability, and Social Justice, (Chicago: Society of American Archivists, 2009). 10 For a clear and thoughtful delineation between "the archive" and "archives" please see Michelle Caswell, " 'The Archive' is Not An Archives: On Acknowledging the Intellectual Contributions of Archival Studies, " Reconstruction 16, no. 1 (2016): 1-21. Specifically, Caswell notes that "'the archive' denotes a hypothetical wonderland, the 'first law of what can be said, the system that governs the appearance of statements as unique events, ' according to Foucault, or a curious materialization of the death drive and pleasure principle according to Derrida. … [A]rchives … 11 By invoking the idea of "state, " archives here, I am referring to the larger body politic, rather than archives maintained by individual US states. What is perhaps the most important impact of carceral archives is the justifications for oppression that are codified, reinscribed, and reinforced in these spaces, from carceral records and recordkeeping practices to the narratives that emerge from them. Both as institutions and as physical places, archives frequently read as carceral spaces. Characterized by walls, gates, locked doors, surveillance cameras, and security checkpoints, this material carcerality rests upon appeals to societal ideals of "safety" and "risk, " wherein archives must protect humanity from itself. Archival repositories are therefore constructed as closed spaces, and are so frequently experienced as such by users that theory and practice related to archival access-from concerns about archival custody and the ownership of evidentiary records, to closed stacks practices and cultures of surveillance in repository reading rooms-is a subject of some disciplinary critique. 13 Records are more than data or information, they are also evidence of an activity or decision. Determining the institutional, cultural, and historical value of records based on their evidentiary, documentary, and informational value is part and parcel of the archival endeavor, and one of many processes that render records archival. Another such process is archival representation (the arrangement and description of archival records), in which classification and narrative are key components. Archivists and the users of archives co-construct archival narratives through processes of archival description and archival storytelling. Archival description, the process describing the materials in the archivist's care, is a highly political act. The narratives that are used to describe archival materials help to define the records themselves; description also plays a role in determining how the records are used and by whom. The users of archives also construct narratives. Indeed, from the archives emerge many of humankind's stories. The stories that emerge specifically from carceral archives can be thought of as carceral narratives.
Narrative construction-as well as the construction of counternarratives or counterstories-is a key methodological component of critical race theory. Similarly, the use of counterstories in archives challenge the cultural dominance of the status quo, aid in community building, and are indicators of the need to reallocate power. 14 Thinking critically about narrative construction, incorporating Benelischolar.library.yale.edu/jcas/vol5/iss1/6. See also Tonia Sutherland, "Archival Amnesty 14 Anthony Dunbar, "Introducing Critical Race Theory to Archival Discourse: Getting The Con-

Cultural Analytics
The Carceral Archive jamin's thinking on carceral imaginaries, and extending Jimerson's metaphor of prison institutions (carceral archives), I offer a concept related to-but distinct from-carceral archives: the carceral archive.
As is articulated above, archives are comprised of records, institutions, and processes. "The archive, " on the other hand, extends beyond the documentary record to "enclose in one place all times, all epochs, all forms, all tastes … a place of all times that is itself outside of time. " 15 The archive, then, is not temporally or materially bound; it encompasses all things and all times. The carceral archive can be conceived of as a sociocultural imaginary-a vision of the archive as a kind of "temporal penitentiary in which oppressed people are locked in to a dystopic present. " 16 The carceral archive represents the history and memory of human existence that has been formed by and/or bound to captivity, ownership, domination, control, imperialism, colonialism, hegemony, forced conformity, and white supremacy.
The carceral archive is one in which oppressed people cannot escape the historical memory and historical trauma that is codified, reinforced, reinscribed, and reified in the documentary record as part of the work of maintaining dominant cultural narratives. For oppressed and marginalized people, the carceral archive is an embodied carceral reality-a lived experience that stands in opposition to liberation and reinforces historical notions of inferiority and disposability in the present. The carceral archive comprises all of human experience that can be defined as carceral, but here I will focus on archival data and the narratives that emerge from them.

Carceral Data
It is important that a discussion of carceral archives begin by naming antiBlackness and other forms of racism as direct challenges to the evidentiary value of state records. One central expression of neoliberal culture is an emphasis on the state's carceral functions, what Noah de Lissovoy has termed "the carceral turn. " 17 I contend that the use of archives the United States has taken a sharp carceral turn, one in which the actual-and disproportionate-incarceration of marginalized people is justified by invoking the documentary record: primary source data (specifically, for this discussion, social media data) that has been appropriated for use by law enforcement and the criminal justice system. Government agencies (including state and national archives) capture, produce, and maintain a multitude of records and artifacts, including such things as founding documents, treaties, proclamations, records of conflict, and increasingly vast amounts of primary source data on individual citizens that comprise the documentary record. When primary source data, regardless of form or format, is deemed public, it becomes part of the public record. Evidence of the concept of public records dates as far into history as ancient Babylon (where vital records such as reports of births, deaths, and marriages were etched in cuneiform into clay tablets), and is also associated with Native Hawaiian, Inca, and ancient Chinese cultures (in which records were kept as quipas, or talking knots). In the US, records maintained by the state fall into several access categories, including classified, confidential, and public. While access to and use of classified and confidential records typically requires permission and/or security clearance, public records comprise an official reporting of facts that are accessible to any member of the community.
US law and policy both dictate what types of information is considered public, where public information can be stored, how public information can be distributed, and, importantly what categories of information are included in the public record. Although public records laws vary in scope and application, most government entities define public records similarly, using broad classes that encompass all related information regardless of physical form or characteristics. 18 These classification schemas were designed to remain relevant as the form and format of communication changed over time, and are inclusive of everything from parchment to paper to digital media. In most states, public records include documents, papers, letters, maps, books, tapes, photographs, film or sound recordings, information stored or maintained electronically, and data-processed or image-processed documents. Importantly, in recent years, social media data has also come to be considered part of the public record. In modern history, public records have been generally understood to represent "the official reporting of facts;" however, there has been a significant epistemic shift in how we understand the concept of public records-and notions of fact, or authenticity-where social media data is concerned.Whereas public records were historically just that-records of public activity-social media data exists to serve the private corporate and commercial 18 Many government entities have formally revised the language of their statutes to offer specific guidance on social media and public records.

Cultural Analytics
The Carceral Archive interests of the company. Although they might be considered shareholders, the user is not the customer on social media platforms. They are the indirect, rather than the direct source of revenue; rather, those that purchase user data and advertising space are the customers.
Social media can be beneficial in a multitude of ways. It allows people to build professional and personal networks; exposes new ideas; amplifies voices; allows for greater advocacy; and has the potential to act as an equalizer, creating space for multiple voices. On the other hand, rather than solely being a mode of interpersonal communication, social media is part of the state record. 19 Many social media users have a (false) expectation of privacy and conduct themselves on social media platforms as though they are in the private sphere; the reality is, however, that social media users gain access to the means of data production (the social media platform) by surrendering rights of ownership to their data, and by extension, their privacy. Research has shown that in the private sphere of social media, people construct identities that may not correspond to their more public lived realities. 20 Whether presenting idealized versions of themselves, or versions that are larger-than-life, social media data are not necessarily reliable evidence of a person's actual beliefs, attitudes, or even conduct. This tension between the private and public spheres is critical to inquiries about the evidentiary value of social media and its attendant data and metadata as public records, particularly in light of law enforcement's ongoing use of these data for criminal justice purposes. 21 In addition to the tensions between the public and the private, the nuances of the interplay between data and metadata and between narrative and counternarrative in social media data also raise questions about these data as reliable evidentiary records. Archivists are concerned with issues of provenance and the continuous chain of responsible custody; with preservation over interpretation; and records as of evidence of decisions, activities, and policies. This is true of all records and data, but thus far, the demonstrated use of social media data by law enforcement agencies lacks the nuanced reading of data as archival evidence that would make data justice possible. 19 State records, as I am using the term here, are analogous to and inclusive of public records and government records. Increasingly used with predictive technologies to assess potential risk, social media data feeds the mechanisms that reinforce the carceral state. For example, social media data is used in presentence investigation reports and by Homeland Security in the context of border surveillance and control. This data is routinely monitored, collected, and used by the FBI and other law enforcement agencies as part of their justification for incarcerating immigrants at the US border. 22 For those who are already vulnerable, social media data-shared digital files; Twitter, Instagram, Facebook, Reddit, Yelp! and other posts; uploads and downloads; participation in groups or other online communities; as well as the metadata such as date, location, and time stamps generated by each of these activities-is carceral data, evidence of activity open to interpretation by the state. Thus, every interaction on social media also becomes part of a person's carceral archives, part of the record that can be easily deployed to construct the narratives that are used to justify the need for compulsory containment. By 2014, police were already using social media to apprehend criminals; today, largely as a result of increasingly surveillant data cultures, these practices have evolved and expanded. 23 In her 2018 article "Critical Surveillance Literacy in Social Media: Interrogating Black Death and Dying Online, " critical race and digital studies scholar Safiya Umoja Noble advocates for a developing a critical surveillance literacy for social media, arguing that "increasing resources have been allocated in service of new forms of record keeping by the state, as new federal and municipal commitments to surveillance … [have escalated.]" 24 Noble further emphasizes two carceral concerns-commodification and political economy-in social media surveillance, arguing that "collecting and using social media records and digital artifacts contributes to the persistent domination of Black Americans. " 25 What follows the construction of carceral archives by way of surveillance and data mining in the modern criminal justice system is a process of narrative construction and storytelling: the transformation of evidentiary information, data, and metadata into legal evidence, and the furnishing of that evidence in hearings and trials in support of carceral outcomes. These narratives, fashioned from the records of state powers, easily lend themselves to the social and cultural construction of the carceral archive, signaling a need to compose counternarratives.  David Lyon, Director of the Surveillance Studies Centre at Queen's University, and credited with spearheading the field of surveillance studies, argues that an appropriate focus in surveillance studies is to center attention on "sites of surveillance, " and names government administration, policing and crime control, and consumer activity as three such sites. 27 Centering social media as a "site of surveillance" exposes the social media data being used by law enforcement and the criminal justice system as the deeply problematic product of a carceral archive that rests upon highly racialized cultures of surveillance (including surveillance technologies). The history of policing as a site of surveillance also offers important insights into the archive of contemporary US antiBlack risk narratives and the carceral archives that are used to construct them. In the United States, carcerality is a complex and racialized premise. The development of policing in the United States is closely aligned with the development of policing in England. In the early American colonies, informal and communal policing was known as "The Watch, " and private-for-profit policing, known as "The Big Stick. " 29 The watch system was a volunteer system in which the role of watchmen, who often slept or drank on duty, was to warn of impending danger. Augmenting the watch system was a system of constables, official law enforcement officers, paid on commission for warrants served. It was not until the 1830s that the idea of a centralized municipal police department first emerged in the United States, but by the 1880s all major U.S. cities had municipal police forces in place. 30 To articulate a critical race and critical archival counter-narrative of policing and crime control in the US is to acknowledge that these practices were-and continue to be-inextricably tied to the violent history of chattel slavery.

Carceral Narratives
In the southern United States the origin of modern policing organizations was the slave patrol. Slave patrols had three main functions: to apprehend and return enslaved people who had taken flight; to provide a form of organized terror in an effort to deter uprising; and to execute extrajudicial discipline (lynching and other forms of vigilante justice) to enslaved people deemed to be in violation of plantation rules. During Reconstruction, these vigilante organizations evolved into Southern police departments, existing primarily to control freed Black people and to enforce Jim Crow segregation laws. In the days after the Civil War, former enslaved people and their descendants were often arrested for minor violations, burdened with imposing fines, then incarcerated until the fines were paid. In a system penal labor known as convict leasing, the only way by which the fines could be paid was to either toil on the farms and plantations from which they had just been freed, to toil in the deathly depths of coal mines, or to labor in prisons that had been converted to work farms. 31 Because convicts were too meagerly compensated to pay their fines, this labor system effectively amounted to (re)enslavement.
The resubjugation of Black men and boys was a saleable enterprise that white Southerners could powerfully leverage in commercial endeavors. This "new" Southern economy demanded the criminalization of Black life, which in turn led to a long-term pattern of increased policing projects and new financial markets in security services. Incarcerating Black men became an exceptionally prof-

Cultural Analytics
The Carceral Archive itable financial enterprise of its own. At the same time that slave patrols and convict leasing were fiscally sound enterprises, they also proffered strict social control, maintaining what Michelle Alexander terms the American "racial caste system. " 32 Present-day threats to this political and financial economy only intensify state efforts to uphold policing (and subsequent incarceration) as a means of mitigating risk, which is ultimately seen as a public good. 33 Risk can mean threat, the possibility of loss or injury, the suggestion of hazard, the chance of peril. Patients might be at risk of infection, an investor risks the chance that a commodity will lose value. Safety, on the other hand, implies protection from danger or injury. Protection from risk. The racialized and colonial history of the concepts of risk and safety in America dates to the earliest days of settler colonialism and the idea of the savage native. White European colonizers met their demand for land by conquering the sacred ancestral lands of Indigenous tribes. Indigenous people were seen as a threat to European progress, risking the safety of the colonial project. Images of Indigenous people in books and newspapers painted them as savages to be feared, risks to be averted. As more Africans, and then enslaved Africans, were brought to the eastern shores, Black men were depicted as innately destructive, animalistic, and criminal. The Black "brute" was a sociopathic and anti-social menace, a predator whose primary victim was white women. Sociologist David Pilgrim writes that The "terrible crime" most often mentioned in connection with the Black brute was rape, specifically the rape of a white woman. At the beginning of the twentieth century, much of the virulent, anti-black propaganda that found its way into scientific journals, local newspapers, and best-selling novels focused on the stereotype of the Black rapist. The claim that Black brutes were, in epidemic numbers, raping white women became the public rationalization for the lynching of [Black men]. 34 Slave patrols were not only deployed in the service of returning enslaved people who had run toward freedom, they were expressly deployed to maintain the safety, security, and purity of white women. As the rhetoric of law and order was mobilized in the American South in the 1950s, systematically linking Civil Rights to crime-still embedded with notions of native savages, Black brutes, and the purity of white women-little effort was made to disguise the racial motivations behind rhetorics of safety and risk. 35 Oakland, 2015). 35 For further discussion of the history of the racialized rhetoric of law and order, see Michelle safety rhetoric to underpin contemporary thinking about risk analysis as a law enforcement tool is to understand that the omniveillance and mass incarceration of Black and brown communities is part and parcel of a ritual and deeply racialized performance a carceral archive, constructed around narratives of law and order, safety and risk. The racist tropes of the native savage and the Black brute and the racialized rhetoric of safety and risk contribute to a national carceral archive, with long-lasting and potentially devastating human consequences.
More than crime, modern police forces in the United States emerged as a response to perceptions of disorder. In Dark Matters, scholar SimoneBrowne argues that oppressed people have long been imagined as out of place and that technologies of control and order were developed in order to diagnose, map, and correct these conditions. 36 In the 19th century, social order was determined by the interests of the mercantile class, who were more interested in social control than crime control. In creating a police force, these mercantiles were interested in protecting their own financial enterprises, and in making those costs the responsibility of the state. 37 State-sponsored policing in the name of order and safety has, from the beginning, claimed a demand for surveillance and social control. That this history of policing is tied to commerce and chattel slavery is evident in modern-day law enforcement. Statistically, the number of incarcerated Americans from communities of color in the United States continues to rise disproportionately, and resulting concerns about the human and social consequences of mass incarceration at the current scale have led to calls for criminal justice reform and the abolishment of prisons.
The fiscal landscape of the criminal justice system (alternately dubbed the "prison industrial complex") in the US has also shifted, not away from profit, but toward even more complex and corrupt revenue streams. The United States government holds many lucrative contracts with the private prison industry. Private prisons receive a stipend from the government, the amount of which in most cases is determined based on the number of inmates housed in the penal complex. As with most businesses, shareholder primacy demands perpetual growth, which means that earnings for each fiscal year must supersede the year preceding. In order to stay in business, private prisons need a nearly constant stream of incoming inmates to replace those that have completed their sentences. In simple terms, this means that laws have to be enforced, and in some cases more strictly enforced. media outlets reported multiple instances of judges accepting monetary incentives from private prisons for subjecting criminal defendants to stricter-and longer-sentences in corporate, for-profit prison environments. 38 In March 2018 the Prison Policy Institute reported that the American criminal justice system holds almost 2.3 million people in state and federal prisons, juvenile correctional facilities, local jails, and Indian Country jails as well as in military prisons, immigration detention facilities, civil commitment centers, state psychiatric hospitals, and prisons in the U.S. territories. Another 3.7 million people are on probation and 840,000 are on parole. Black, Indigenous, and other people of color are dramatically overrepresented in the nation's prisons and jails. These racial disparities are particularly stark for Black people, who comprise 40% of those incarcerated despite representing only 13% of U.S population. Gender and age disparities matter as well: rates of incarceration have grown even faster for women and youth than for men. 39 Because the prison profit model depends on the continuity of incarceration, and because of America's structural, systemic, and institutional racism, critics of the justice system have advocated taking sentencing, bail, parole, and recidivism decisions out of the hands of fallible human judges in favor of "neutral" algorithmic tools known as Risk and Needs Assessment (RNA) algorithms.
Risk assessments have existed in various forms for a century, and courts have used actuarial tools in the service of offender risk assessment for at least the past 30 years. Over the past two decades, however, these tools have become ubiquitous in the American justice system, driven in part by advances in the social and computer sciences, and in part by the rapid uptake in human incarceration. Risk and Needs Assessments (RNAs) are proprietary algorithmic tools used to inform decisions at every stage of the U.S. criminal justice system from bail to parole. RNAs try to predict the risk of recidivism using statistical probabilities that are based on data extracted from public and other state records-a practice that I have already argued is deeply flawed, contributes to and draws upon a carceral archive, and that does not adjust or account for antiBlackness and racialized cultures of surveillance. Notions of risk and safety, as I discuss here, are also flawed and racialized concepts in the United States. Umoja Noble's Algorithms of Oppression: How Search Engines Reinforce Racism identifies data discrimination as a pressing social problem. In her book Breathing Race into the Machine historian Lundy Braun argues that cultural notions of race are embedded in the architecture of ordinary instruments. 40 Similarly, in her book Automating Inequality: How High-Tech Tools Profile, Police, and Punish the Poor, political scientist Virginia Eubanks investigates the impacts of data mining, policy algorithms, and predictive risk models on economic inequality in the United States. Others have discussed how the interview protocol that accompanies the RNA algorithmic tool also reinforce existing class-and race-based injustices and perpetuate cycles of incarceration. The data sets that feed the algorithms that determine RNA scores, as data scientist Cathy O'Neil argues in Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy, often includes data about nuisance crimes, like panhandling and vagrancy, that are most common in poverty-stricken areas. 41 Including this crime data skews the algorithmic analysis; once the data is included in a predictive model, neighborhoods with high levels of nuisance (rather than violent) crime are subject to increased policing which in turn leads to more arrests. Once a person has been arrested once, the likelihood of subsequent arrest rises, and much like convict leasing during Reconstruction, people who can't afford to pay court fines, legal fees, and other debt associated with the penal system, are re-incarcerated for failure to pay. As the organizers of the October 207 Data For Black Lives conference stressed in writing about the event, Tools like statistical modeling, data visualization, and crowd-sourcing, in the right hands, are powerful instruments for fighting bias, building progressive movements, and promoting civic engagement. But history tells a different story-one in which data is too often wielded as an instrument of oppression, reinforcing inequality and perpetuating injustice … Today, discrimination is a high-tech enterprise. 42 It is not surprising, given the racialized history of law and order in the US, that when data wielded as an instrument of oppression, it is frequently cloaked in the rhetoric of safety and risk. Carceral data, classed and sorted, classified and confidential, public and private is the cornerstone of contemporary omniveillance campaigns-one that is closely tied to slavery era risk management and disorder narratives. 40

Cultural Analytics
The Carceral Archive Criminal risk assessments are used in sentencing, parole, and bail. Tools typically consist of a set of questions that guide face-to-face interviews with offenders, probing behaviors and attitudes that are believed to be related to criminal reoffending, and as previously discussed, the questionnaire is supplemented with an official records check, of public and other state records. Responses to the questionnaire are statistically weighted, based how strongly each item is thought to correlate with recidivism. The tool then calculates an overall score that classifies an individual's risk of reoffending. This risk level and accompanying information about an offender's unique needs then inform decisions about bail, sentencing, probation, parole, detention facilities, and other considerations. RNA tools can be customized for use by different agencies at various decision points in the sentencing and corrections process.
There are typically at least seven dynamic risk factors closely associated with criminal conduct that are assessed: antisocial personality pattern, pro-criminal attitudes, social supports for crime, substance abuse, poor family/marital relationships, school/work failure, and lack of pro-social recreational activities. 43 A number of risk assessment instruments incorporate data about a defendant's employment status, income, education, and job skills. In addition to other forms of risk, these instruments calculate stated risks such as a defendant's risk of not showing up for court and risk of re-arrest. Arguably, there are many reasons that a defendant might miss a court appearance-work, childcare, transportationall of which have financial and racial correlates, meaning the risks being measured are automatically higher for some. As evidenced by a sample RNA used by the Correctional Offender Management Profile for Alternative Sanctions (COM-PAS), and the risk factors listed above, the interview protocols that accompany RNA algorithms ask questions that are proxies for race and class, such as questions about gang affiliation, belief that the criminal justice system is rigged, welfare status, and parental incarceration. These questions create a feedback loop wherein one's place of birth or one's current address might lead to a longer sentence, reduced chances for probation or parole, and assumptions about recidivism. Importantly, for the arguments set forth in this paper, the most widely known source of data used to train RNA algorithms is state or government data, specifically that which is deemed to be part of the public record. As previously discussed, this data includes everything from vital statistics and demographic information to social media posts and neighborhood crime statistics. There have been several pieces of proposed RNA legislation introduced before Congress in recent years. For example, Senate Bill 2123 (S.2123), the Sentencing Reform and Corrections Act of 2015, stipulates the development of a Post-Sentencing Risk and Needs Assessment System designed to determine recidivism risk, determine the risk of violence, collocate low-risk offenders, assign prisoners activities based on risk level, determine the specific criminogenic needs of each prisoner, determine a set schedule for reassessment, and provide best practices to lower the risk of recidivism. In developing the Assessment System, and determining what records it will include, the Attorney General is encouraged to consult with academics and criminal justice experts as deemed appropriate. 44 Strikingly, this proposed legislation recommends that the Bureau of Prisons incorporate existing Inmate Classification Systems into federal RNAs. Inmate Classification Systems records, or Objective Jail Classification Systems, are designed specifically to reduce escapes and escape attempts, suicides and suicide attempts, and inmate assaults. 45 Nothing in these systems or in proposed legislation speaks to rehabilitation, individual and community health and growth, or the potential for discriminatory design to embed racist beliefs and ideologiesinto the very architecture of safety and risk technologies such as RNAs. The ramifications are critical because proposed legislation would not allow for the system to be challenged, leaving those incarcerated with no means of appeal: Subject to any constitutional limitations, there shall be no right of review, right of appeal, cognizable property interest, or cause of action, either administrative or judicial, arising from any determination or classification made by any Federal

Cultural Analytics
The Carceral Archive agency or employee while implementing or administering the Assessment System, or any rules or regulations promulgated under this section. 46 Although we cannot know for sure how much, there is evidence to suggest that law enforcement databases also include robust data from the Department of Homeland Security. In May 2010, the Secretary of Homeland Security established a Policy for Integrated Risk Management (IRM), also known as the 2011 Homeland Security Risk Management Doctrine. Central to this policy were the ideas that risk management should be collaborative, and that risk management systems should be integrated with and sustained by federal, state, local, tribal, territorial, nongovernmental, and private sector homeland security partners, meaning that data could and should be extracted from all available sources. The Doctrine stipulates that the Department of Homeland Security should play an essential role in leading a unified effort to manage risks to the United States from a "diverse and complex set of hazards. " 47 Some of the most commonly used data sources for Homeland Security risk analysis include archival and other historical records (carceral archives); models and simulations; and interviews with subject matter experts (typically done when historical records do not exist or are not deemed to be "appropriate"). The Doctrine further stresses that "structured techniques, such as value focused thinking (emphasis mine), can help the analyst determine which aspects of [the consequences of a breach in security] should be included in the methodology" and that "the gathered data and evidence should be carefully studied and compared to previous work" as part of validation and verification, further noting that, "In a broad assessment, the decision maker will often have specific areas they are particularly interested in, and will ask the analysts to focus in on those areas. " 48 As has already been demonstrated, the "values, " "previous work, " "specific areas [of interest]" that fall under the purview of law enforcement are highly racialized and frequently lead to further oppression and marginalization of black, Indigenous, and other people of color.
The Department of Homeland Security U.S. Customs and Border Protection (CBP) also operates the Automated Targeting System (ATS), a "decision support" tool that compares traveler, cargo, and conveyance information against law enforcement, intelligence, and other enforcement data using risk-based scenarios and assessments. ATS compares information about individuals (identified as passengers, travelers, crewmembers, or persons appearing on documents supporting the movement of cargo) against the Terrorist Screening Database as well as against available data on outstanding wants and warrants. 49 ATS maintains an official record of license plate data; Department of Motor Vehicle registration data; biographical data; law enforcement and/or intelligence data; reports, and projects developed by CBP analysts that may include public source information; and information obtained through memoranda of understanding or other arrangements.
Finally, ATS maintains copies of or access to key elements of certain databases including, but not limited to: border crossing information, special protected classes data; student exchange and visitor information; the Enforcement Integrated Database, which includes-but is again not limited to-criminal arrest records (including RNA analyses) and immigration enforcement records; secure flight passenger data and master crew list data from the Transportation Security Administration; the Department of Justice's National Crime Information Center and Federal Bureau of Investigation's Interstate Identification Index; the Social Security Administration (SSA) Death Master File; nonimmigrant and immigrant visa data from the US Department of State; the National Insurance Crime Bureau's private database of stolen vehicles; e3 Biometrics System data; DHS Automated Biometric Identification System; and, notably, commercial data aggregators. A person flagged by the Automated Targeting System for any reason stands to be incarcerated and/or deported. Because of increased cultures of data sharing, and continuing rhetorics of safety and risk, access to these databases for any criminal justice or law enforcement purpose is likely to be granted.
Given the opacity of RNAs; their proprietary nature; the capitalist and corporate needs they serve; the carceral and racialized data practices that feed the algorithmic tools; and the racialized histories, surveillance cultures, and technologies that produce them, the inability to challenge the results of risk and needs assessments raises serious concerns about modern-day correctional supervision. Also given the history of policing from slave patrols and convict leasing to the enforcement of racist Jim Crow segregation laws, one would be remiss to not center concerns about the discriminatory aspects of risk and needs assessments, risk management, and algorithmic design. An algorithm is simply a set of guidelines that describe how to perform a task. Just as, historically, those guidelines for performing the task of policing were built on racist and oppressive beliefs and practices, so too can today's algorithms and computer models show bias against 49 Department of Homeland Security, "Privacy Impact Assessment Update for the Automated Targeting System, " January 2017, https://www.dhs.gov/sites/default/files/publications/privacy-pia-cbp006e-ats-april2017.pdf. oppressed and marginalized people. Too frequently the very data used to train those algorithms reflect existing cultural biases.
As I have already suggested, the for-profit prison industrial complex depends on a constant stream of offenders to maintain profit margins. Notions of carcerality, deeply imbedded in American history and culture, must continually be reinscribed and reinforced to serve the fiscal needs of the penal project and to maintain the racialized norms of an oppressive and carceral society. This is the function of carceral archives and, I would argue, what it means to live within the carceral archive.

Toward Liberation
In this paper I have introduced and defined two new and interrelated terms: carceral archives and "the carceral archive. " I have argued that mass incarceration, risk assessments, recidivism, sentencing determinations, and pretrial decisions are all components of a single carceral system that generates and draws upon the carceral archive, and I have argued that the state has created and sustained carceral archives by appropriating and exploiting data cultures, particularly social media data culture. I have interrogated how carceral archives contribute to and perpetuate risk narratives that are used by the state to justify incarceration. I have also argued that contemporary state data and algorithmic practices are not new, but are instead constructed around racialized, carceral narratives around order, risk, and safety.
Here, I will close by arguing that to take up a liberatory and decolonial position is to 1) insist that racism and colonialism no longer factor into sustaining and reinforcing the intersecting data and algorithmic practices of the criminal justice system, and 2) work to elucidate how sociocultural and sociotechnical bias in data mining, algorithmic practices, and data production contribute to the national epidemic of physical and socio-psychological incarceration of oppressed and marginalized people. I would like to suggest that if there is a carceral archive-and there are carceral archives-then we must also construct liberatory archives and attend to the archive of liberation. An archive of liberation, or the liberatory archive requires understanding how people are positioned by data systems and data science. The state records and data practices that support and enable algorithmic risk assessment are enacted by knowledge creators who imbue new types of technology with old social harms. These sociocultural biases, historically imbedded into unjust "justice" systems and fortified by capital enterprise,