Racial Innocence: Law, Social Science, and the Unknowing of Racism in the US Carceral State

Racial innocence is the practice of securing blamelessness for the death-dealing realities of racial capitalism. This article reviews the legal, social scientific, and reformist mechanisms that maintain the racial innocence of one particular site—the US carceral state. With its routine dehumanization, violence, and stunning levels of racial disparity, the carceral state should be a hard test case for the willful unknowing of obvious devastation. Nonetheless, the law presumes “no racism,” condones racial profiling, and interprets racial disparity in policing and imprisonment as evidence of true racial difference in criminality, not discrimination. Prominent social science research too often mimics these practices, producing research that aids in the collective erasure of racism.


INTRODUCTION
"It is not permissible that the authors of devastation should also be innocent. It is the innocence which constitutes the crime." So wrote James Baldwin in 1963, as the black freedom struggle forced a national reckoning with racial innocence [Baldwin (1963, p. 334]. Racial innocence, practiced blamelessness for the death-dealing realities of racial capitalism, is maintained through willful ignorance, blame displacement, and liberal reforms. Ignorance here is no passive lack of information but an active discipline of denial armored in the "American Creed" and the triumphal insistence that we have overcome (Balfour 1999, Golub 2018, HoSang 2010, Mills 1997, Taylor 2015.
This article reviews the legal, social scientific, and reformist mechanisms that maintain the racial innocence of one particular site: the US carceral state. Racism is best understood as "the state-sanctioned and/or extralegal production and exploitation of group-differentiated vulnerability to premature death" (Gilmore 2007, p. 247). By this definition, there is abundant evidence of racism in the carceral state. Police kill roughly 1,000 people a year, with blacks more likely to be killed even though they are less likely to be armed with a gun (Zimring 2017, Correia andWall 2018). The Black Lives Matter Movement confronted the spectacular violence of police killings as but one manifestation of the carceral structures of slow death-police stops, arrests, fees and fines, imprisonment-all of which produce and exploit group-differentiated vulnerability to premature death (Movement for Black Lives 2016, Ransby 2018. Each year the police make 18 million traffic stops; approximately 12% of all drivers are stopped, but fully 24% of all black drivers are stopped by police each year (Epp et al. 2014). Even without an "incident" of police brutality, normal police arrests are harmful. Even a low-level misdemeanor arrest record decreases the callback rate for entry-level jobs, and hence disproportionately high black arrest rates ultimately compound racialized wealth inequality (Uggen et al. 2014, Natapoff 2015, Kohler-Hausmann 2018. Approximately 6.8 million people are held behind bars or on probation or parole; of these, roughly 34% are black. Despite the Black Lives Matter national wake-up call in the summer of 2014, however, public support for police has since rebounded to preprotest highs (Butler 2017), and a modest majority of 2016 white voters supported Trump's antiblack, anti-immigrant, anti-Muslim law-and-order presidential campaign (Brown 2017, Strolovitch et al. 2017.
This article attempts to identify the overarching mechanisms that redefine, minimize, and legitimate the carceral state's production and exploitation of group-differentiated vulnerability to premature death. The next section introduces racial innocence and what I see as its three core mechanisms: willful ignorance, the invention of "the Negro problem," and liberal reformism. Each mechanism enables our collective unwitnessing of carceral devastation, as I explore in each of the three following sections. First, willful ignorance, enshrined in the law and replicated in prominent social science research, is an epistemology that begins with the presumption of no racism, searches for racism in narrow terms, and predictably concludes that there is meager support for "the discrimination hypothesis." Second, the invention of "the Negro problem," specifically the conflation of blackness with criminality, is an intellectual architecture that makes criminalization and caging seem necessary and even inevitable. Third, liberal reformism absorbs protest and affirms the false hope that well-intentioned carceral humanism can end the devastation.

RACIAL INNOCENCE: AN OVERVIEW
Racial innocence is the dominant US epistemology, a way of knowing fueled by the desire for unknowing. "This is the crime of which I accuse my country and my countrymen, and for which neither I nor time nor history will ever forgive them, that they have destroyed and are destroying hundreds of thousands of lives and do not know it and do not want to know it" [Baldwin (1963, p. 334, emphasis added]. Baldwin faults something beyond ignorance and prejudice, racial liberalism's preferred and misleading synonyms for racism. These terms distort twice over, once by reducing oppressive systems to personal educational deficits and once again by redirecting responsibility from the ruling class to the "uneducated" white poor (Guinier 2004Taylor 2015, Taylor 2016, Haltom & McCann 2004. Ignorance is no absence of knowledge; it is, rather, the cultivation of institutions, ideologies, and rhetorical mazes that unwitness racism, sometimes fully inverting evidence of racial inequality into national stories of equality, individualism, and progress (Bonilla-Silva 2006, p. 25;Golub 2018, p. 54;Taylor 2015, p. 56).

Willful Ignorance
Baldwin's idea of innocence entails "willful ignorance" (Balfour 2000, p. 27), and contemporary scholarship develops this form of unknowing under the moniker of "colorblind racism." Colorblind racism explains contemporary racial inequality as the by-product of ostensibly nonracist forces-market dynamics, naturally occurring tastes and preferences, and "cultural limitations" imputed to African Americans (Bonilla-Silva 2006, p. 2;Bonilla-Silva 2001). Colorblind racism's affective register is not necessarily boiling animus and frothing hatred but, as Williams (1997, p. 27) puts it, "a profoundly invested disingenuousness, an innocence that amounts to the transgressive refusal to know." Consider the "profoundly invested disingenuousness" of Chief Justice Roberts in striking down public school integration measures in Seattle and Louisville: "The way to stop discrimination on the basis of race is to stop discriminating on the basis of race" [Parents Involved in Community Schools v. Seattle (2007)]. In the legal infrastructure and practiced fantasies of racial innocence, to ignore race is to end racism.

Inventing "the Negro Problem"
Twinned with the fastidious unknowing of racism is the facile knowing-or rather, inventionof "the Negro." The invention of "the Negro" and "the Negro problem," as Baldwin explains, projects onto blacks the un-American vices of laziness, criminality, and sexual aggression, thereby "safeguard[ing] the purity" of the US political and economic order [Baldwin (1972Walker 2002, p. 226]. The invention serves racial capitalism by rewriting capitalism's formal attributes of expropriation, exploitation, and ceaseless accumulation into narratives of racial "difference" (Gilmore 2007, Taylor 2016.

Liberal Reformism
Liberal reforms secure illusions of racial progress without paying the price of the ticket. Baldwin deemed liberal reformism "the white fantasy of getting racial justice cheap," in Turner's (2012, p. 97) apt words. To the extent that liberal reformism absorbs radical protest, it secures redemptive narratives of "mission accomplished" for leveling the playing field, thereby fortifying messages that black people have only themselves to blame (Perry 2011, Taylor 2016. Reformism reposes on anemic definitions of racism, constrained to produce "meaning in the service of power" (Bonilla-Silva 2006, p. 25). With racism reduced to a psychological problem, framed as prejudice, irrationality, hatred, or even personal sickness, its remedy requires no redistribution of resources (Frymer 2005, Guinier 2004, Singh 2004.

WILLFUL IGNORANCE
This section sketches three ways of maintaining willful ignorance toward carceral devastation. First, antidiscrimination law refuses to witness racial disparity statistics as proof of discrimination. Instead, antidiscrimination law formalizes the presumption of racial innocence, tasking plaintiffs with the burden of identifying a perpetrator guilty of intentional discrimination (Freeman 1978, Crenshaw 1988, Selmi 1997, Haney-López 2000, Spade 2015, Lovell et al. 2016. Second, many law enforcement agencies have indeed banned racial profiling, but in name only. The official condemnation of prejudice imparts plausible deniability; in fact, the law condones racial profiling. Third, prominent social science research replicates antidiscrimination law's null hypothesis of no racism, searching for discrimination by the law's constrained standards of intentional, directly caused racial harm.

Knowing and Unknowing Racial Disparity
The carceral state produces and exploits group-differentiated vulnerability to premature death. This racism is often demonstrated through racial disparity statistics, especially in the incarceration rate. Measured as the proportion of the population held in state and federal prisons plus local jails, the US incarceration rate nearly quintupled from 1972 (161 per 100,000) to its peak in 2007 (760 per 100,000). Despite declines since 2008, the incarceration rate of 670 per 100,000 in 2016 leaves the United States the world's lead per-capita incarcerator, topping Russia (475 per 100,000), South Africa (294 per 100,000), and Brazil (274 per 100,000) (Beckett 2018, Kaeble & Cowhig 2018, Schoenfeld 2018. Blacks comprise 13% of the US population but roughly 35% of the imprisoned population, 42 % of death row prisoners, and 56 % of those sentenced to life without possibility of parole (Platt 2018;Kim et al. 2018).
The injuries of criminalization both precede and extend beyond "mass incarceration," which represents only "the tail end of the criminal process" (Beckett 2018, Beckett & Murakawa 2012. Many have therefore adopted a more capacious if clunky term: the carceral state, which refers to the networks of laws, institutions, and administrators that enable and enforce punishment, including but not limited to criminal law and procedure, police, criminal courts, criminal records databases, legal financial obligations, probation and parole, brick-and-mortar incarceration and e-carceration with electronic shackles (Beckett 2018, Gottschalk 2012, Hernandez et al. 2015, Lerman & Weaver 2014, Schoenfeld 2018. Approximately 20 million people are estimated to have a felony conviction, one-third of whom are black (Miller & Stuart 2017, Shannon et al. 2017. Eight percent of all US adults have a felony conviction, and a full 33% of adult African American men have a felony conviction (Shannon et al. 2017).
Black-to-white male disparities receive significant attention, in part because historically black men's struggles are framed as "representative" of black progress (Cohen 1999, Strolovitch 2007. Studies that disaggregate police stops by gender and race, however, find that women's black-to-white disparities are comparable to men's (Crenshaw & Ritchie 2015, Epp et al. 2014, Lundman & Kaufman 2003. Highlighting the surveillance and criminalization of black women and girls illuminates the structural relationship between the carceral state and black communities; by contrast, excluding them facilitates faulty analysis that lends itself to particularized tropes that black men and boys need "fixing" through programs such as My Brother's Keeper (Crenshaw & Ritchie 2015, Butler 2017, Crenshaw 2012. Carceral state violence against women and girls of color and trans and gender-nonconforming people is often unseen, unreported, or "privatized" through sexual violence that exploits and silences. For example, a survey by the Los Angeles organization Bienestar found that fully one-quarter of Latinx immigrant trans women, the majority of whom were undocumented, reported sexual assault by law enforcement officers (Ritchie 2017). Women of color also incur carceral injuries and costs through "proximal contact" with the carceral state (Walker & García-Castañon 2017, Gilmore 2007, Richie 2012, Ritchie 2017. In short, racial disparity statistics can be telling, but the numbers do not speak for themselves (Muhammad 2010, p. 277). In terms of compelling legal redress and support for punitive deescalation, why do these numbers sum to zero?

Legal defanging of racial disparity.
Antidiscrimination law holds that disparity is not proof of discrimination. As enunciated in McCleskey v. Kemp (1987), statistics are insufficient to meet the burden of proving, first, intent ("the existence of purposeful discrimination") and, second, causation ("that the purposeful discrimination 'had a discriminatory effect' on him") (p. 293). In enunciating intent and causation, the McCleskey court required evidence that Georgia "enacted or maintained the death penalty statute because of an anticipated racially discriminatory effect." The court drew on Washington v. Davis (1976), which held that discriminatory impact alone is insufficient to establish an equal protection violation, and Personnel Administrator of Massachusetts v. Feeney (1979), which established a "because of, not in spite of" intent standard that essentially requires evidence of racist malice (LaChance 2016, Lovell et al. 2016, Spade 2015. When confronted with statistical evidence of racially disparate policing and criminal sentencing, lower courts echo McCleskey's reluctance to interpret statistics alone as evidence of discrimination [Alexander 2010, Goel et al. 2016, Johnson 2010, Lynch 2013; see for example, United States v. Armstrong (1996).
Given these legal standards, racial disparity statistics must be coupled with nonstatistical evidence of intentional racial animus. Nonstatistical evidence that would meet McCleskey's requirements-proof of discriminatory intent and direct harm in a particular case or set of cases-is essentially a moment when "officers admit to racial animus" (Goel et al. 2016), a "stroke of luck" when plaintiffs fall upon the "miraculous happenstance of testimonial or documentary evidence of bias" (Huq 2017(Huq , p. 2454. Plaintiffs in Floyd v. City of New York (2013, p. 603; Fagan 2017) achieved the rarest of victories-a successful equal protection challenge to police stop-and-frisk practices-by combining statistics with "smoking gun" evidence that high-ranking New York Police Department officers named "young black and Hispanic youths [aged] 14 to 20" as the target population.
Without evidence of intent, antidiscrimination law witnesses racial disparity as a benign manifestation of two forces: criminal justice discretionary power and true differences between groups. Through the 1980s, conservatives challenged the validity of statistical evidence of discrimination in many arenas, including employment sex discrimination [EEOC v. Sears, Roebuck & Co. (1986)], minority workers' rights [Wards Cove v. Atonio (1989)], and affirmative action for minority-owned businesses [City of Richmond v. J.A. Croson Co. (1989)]. In its attack on race-and gender-conscious affirmative action, for example, Reagan's Office of Legal Policy (1987, p. i) contended that "naturally occurring statistical disparities between groups are inevitable in a heterogeneous society such as the United States." McCleskey's indifference to high rates of black execution reflects the suspicion that black criminality is another "naturally occurring" disparity that warrants harsh punishment (Seigel 2018, Donnelly 2017.

Whites' adverse reaction to racial disparity statistics.
Critics of the carceral state marshal racial disparity statistics as a persuasive tactic, but research from social psychologists suggests that the invocation of extreme racial disparity might diminish white support for reform. Stanford psychologists experimented in telling white Californians and New Yorkers different levels of prison racial disparity-one "realistic" black-to-white disparity and another "extreme" disparity with even higher percentages of black prisoners. When prison populations were represented as more black, whites were less likely to oppose three-strikes laws and stop-and-frisk practices (Hetey & Eberhardt 2014). Hetey & Eberhardt (2014) identified fear of crime as a mediating factor, with higher institutional blackness triggering crime concern. In short, knowledge of extreme racial disparity may perpetuate support for the very policies that produce disparities.
Additional social psychological studies connect perceptions of more blackness to greater punitiveness, whether the degree of blackness is within institutions or individuals. Blacks who are perceived as more stereotypically black in appearance (e.g., broad nose, thick lips, dark skin) are more likely to be perceived as criminal (Eberhardt et al. 2004) and more likely to be deemed deathworthy in capital cases involving a white victim (Eberhardt et al. 2006). One study found that, after controlling for differences in criminal history, defendants with stereotypically black facial features served up to eight months longer in prison than defendants who had the least stereotypically black features (Blair et al. 2004). Indeed, blackness signals criminal threat so deeply that psychological studies show that white subjects, when shown photos of black faces for less than one second, experience a lighting up in the "fear center" of the brain (Phelps et al. 2000).

Racial Profiling: Legally Prohibited, Legally Condoned
Racial innocence is, in Shulman's (2008, p. 134) words, the "denial of the reality of others and a disclaiming of this refusal." Observe this disclaimed denial in the legal lattice around racial profiling. Many law enforcement agencies do indeed prohibit racial profiling, and a recent survey of 36 large and mid-sized police departments found that all prohibited profiling (Ritchie & Jones-Brown 2017). In a federalist patchwork without a single policy, however, some agencies merely define the problem away, requiring proof that an officer used race as the sole factor of suspicion (Ritchie & Jones- Brown 2017, Glaser 2015. The same survey found that only onequarter of law enforcement agencies (10 out of 36) prohibited profiling based on sexual orientation and gender, and only a handful (5 out of 36) prohibited profiling based on gender identity and expression (Ritchie & Jones-Brown 2017). Such racial profiling bans therefore allow loopholes for police suspicion toward women and LGBTQ (lesbian, gay, bisexual, transgender, and queer/questioning) people of color, who are routinely profiled for carrying illegal drugs, prostitution-related offenses, and offenses related to child welfare (Kunzel 2008, Mogul et al. 2011, Ritchie 2017, Ritchie & Jones-Brown 2017.
Although racial profiling is banned in name, and although Fourth Amendment search standards ostensibly require "individualized suspicion," in effect courts sanction group-based suspicion and police actuarialism (Fagan & Geller 2015, Harcourt & Meares 2011, Feeley & Simon 1992, leading many scholars to conclude that courts affirmatively condone racial profiling (Butler 2017(Butler , 2018Glaser 2015;Johnson 2010). The Supreme Court has licensed police investigatory stops as permissible under the Fourth Amendment, upholding pedestrian stop-and-frisks in Terry v. Ohio (1968) and vehicle investigatory stops in Whren v. United States (1996). After Terry lowered stop standards from "probable cause" to "reasonable suspicion," courts validated police proxies for race with "location plus evasion" standards; that is, "reasonable suspicion" was satisfied by the conditions of policing in a "high crime area" and observing someone move away from police [see Illinois v. Wardlow (2000); Harris 1994; Lerman & Weaver 2014]. Whren upheld police practices of using minor traffic violations as pretext to investigate more serious violations, acknowledging the likelihood that pretextual stops risked racial profiling. Whren rearticulated the perpetrator model of antidiscrimination law-i.e., the burden of proof is on the suspect to show that an officer intentionally used race as the primary reason for the stop-but the Whren court simultaneously stated that an officer's subjective intent is irrelevant. No officer motivation could invalidate a legal stop. With this, Whren immunized police investigatory stops against equal protection claims (Butler 2017, Epp et al. 2014, Johnson 2010. The investigatory stop is a legal, institutionalized practice that routinizes racial profiling. The International Association of Chiefs of Police advertised Whren as a post hoc legal validation of already widespread practices of using traffic stops as pretext to uncover criminality. Although police departments are locally controlled, federal organizations and police professional networks institutionalized investigatory stops as a nationwide crime-fighting tactic. For example, a Drug Enforcement Agency initiative called Operation Pipeline trained state and local police in pretextual traffic stops; over the 1980s and 1990s, the DEA trained 27,000 officers, many of whom went on to train others (Epp et al. 2016, Seigel 2017. Policing training manuals, hot-spot policing, and the National Highway Traffic Safety Administration's programs further institutionalized the practice. With the investigatory stop as a normal institutional practice, individual officers and department leadership choose where to deploy these stops on the basis of (unstated) suspicions of black criminality (Epp et al. 2014(Epp et al. , 2016Beckett et al. 2006;Lynch et al. 2013;Fagan & Geller 2015;Seigel 2017).

The Penology of Racial Innocence
"What we know" through social science sometimes contributes to unknowing racism. Van Cleve & Mayes (2015) fault dominant research approaches for adopting "colorblind lenses," consequently minimizing or dismissing carceral racism. In a similar vein, Murakawa & Beckett (2010) fault criminal justice research for adopting the narrow standards of antidiscrimination law. In this "penology of racial innocence," researchers obscure racism by searching for invidious racial intent in discrete moments of racial causation. Just as antidiscrimination law puts the burden of proof on the plaintiff to prove discrimination, so too does social science set the null hypothesis as no racism (Murakawa & Beckett 2010).
Widely cited research from quantitative criminologists finds that racial disparity is largely attributable to legitimate legal factors, not racial discrimination. Consider two influential schools of research. The so-called Blumstein method compares changes in racial composition across different stages of criminal processing, from arrest to incarceration, per type of crime (Blumstein 1982). Various applications of this method attribute most of prisons' racial disproportion to black criminality. Blacks' "differential involvement" in crime accounted for 80% of prisons' racial differences in 1979 (Blumstein 1982), 76% in 1991 (Blumstein 1993), and between 70 to 75% from 2004 to 2008 (Beck & Blumstein 2018). Other studies produced lower estimates of "warranted" racial disparity-e.g., 61% in 2004 (Tonry & Melewski 2008)-but Beck & Blumstein (2018) challenge low estimates as an artefact of mis-categorizing Hispanics. Without categorically denying the possibility of discrimination, Blumstein (1993, p. 750) concludes that "the bulk" of prison racial disproportionality "is attributable to differential involvement in arrest, and probably in crime, in those most serious offenses that tend to lead to imprisonment." Another school of research examines sentencing outcomes and puts forward the "no discrimination thesis," contending that unwarranted racial disparities all but disappear when models properly account for legally relevant factors (Wilbanks 1987). Direct "race effects" tend to decline with the inclusion of precise measures for criminal history, offense severity, pretrial release status, pleading guilty, hiring an attorney, and providing evidence or testimony; notably, these legally relevant factors are interpreted as nonracial. Even with this strict dichotomy between racial and nonracial factors, many sentencing studies find "some discrimination, some of the time, in some places" (Sampson & Lauritsen 1997, p. 348, emphasis in original;Kramer & Steffensmeier 1993). Sentencing studies tend to find the strongest race effect when race, sex, and age are measured not independently but interactively, as judges exact a high cost for being a young black man (Steffensmeier et al. 1998, Spohn & Holleran 2000, Wooldredge et al. 2015. Many challenges to the "no discrimination thesis" suggest that race may operate indirectly, interactively, and cumulatively (Mitchell 2005, Spohn 2015b).
If controversies over "apparent race effects" seem like inconsequential academic disagreements, consider the upscaling of these logics in the applied form of risk-assessment tools. Risk-assessment tools magnify the penology of racial innocence. Their development and defense depend on faith in the possibility of separating racial from nonracial factors. As social protest and budgetary constraints have created the perfect storm of reform pressures, many have elevated evidence-based risk assessments as new race-free solutions, sometimes repackaging old actuarial sentencing as new algorithmic justice (Benjamin 2019, Browne 2015, Eaglin 2017, Eubanks 2017, Ferguson 2017, Gottschalk 2015, McLeod 2015, Muñoz et al. 2016. By 2017, when the American Law Institute endorsed the incorporation of actuarial risk tools into state sentencing structures, at least 13 states required risk assessment in sentencing and at least 20 states did so without statutory requirement (Eaglin 2017, Starr 2014. Many promote riskassessments as tools to reduce racial bias and the expense of mass incarceration (e.g., Flores et al. 2016, Skeem & Lowenkamp 2016, Muñoz et al. 2016. With racism conceived of as flawed individual judgment, it follows that less discretion promises more fairness. Yet risk-assessment tools are created from data sets of the already criminalized population, raising concerns that discrimination is woven into predictions (Eaglin 2017, Ferguson 2017, Goddard & Myers 2017, Hannah-Moffat 2013, Harcourt 2007, Schwalbe et al. 2007. Proprietary data means there are few studies testing this claim, but one recent study found substantial racial disparity in use of Northpointe Inc.'s popular tool known as COMPAS (Correctional Offender Management Profiling for Alternative Sanctions) . As used in Florida's pretrial sentencing, COMPAS assigned a higher mean risk score for black defendants. Two years after risk score assignment, the false-positive ratemeaning the proportion of defendants labeled high risk but not rearrested-was 45% for black defendants and 24% for white defendants. By contrast, the false-negative rate-meaning the proportion of defendants labeled low risk but subsequently rearrested-was 28% for black defendants and 48% for white defendants .
In a remarkable reassertion of racial innocence, risk-assessment defenders cited McCleskey, black recidivism rates, and standardized testing norms that "subgroup differences do not in and of themselves indicate lack of fairness." Instead, racial-group score differences reflected "true differences" in recidivism risk (Skeem & Lowenkamp 2016, p. 685). Flores et al. (2016) maintain that, with rearrest rates of 39% for white defendants and 52% for black defendants, the test is accurately calibrated and satisfies standards for predictive parity. In short, risk assessments materially disadvantage black defendants-an uncontested finding-but legal standards and proper statistical accounting of black criminality make racial injury illegible.
The penology of racial innocence has not gone unchallenged, however. Critiques-here limited to critiques delivered within a positivist perspective-highlight the following.

Race-laden benchmarks and controls.
Because researchers reject general population racial proportions as a valid benchmark against which to evaluate disparity, studies depend on baseline measures of group-differentiated criminality (differential involvement) and controls for individual-level criminality (legally relevant factors). These benchmarks and controls reflect racially differentiated policing and prosecution, all reflected in arrest records, offense history, and offense seriousness (Beckett et al. 2006, Lynch & Omori 2018, Van Cleve & Mayes 2015. Like many sentencing studies, risk assessments essentially "launder" racial inequalities by declaring criminal history a nonracial and legally relevant predictive factor (Goddard and Myers 2017). With criminal history a proxy for race, risk-assessment tools are likely to exacerbate but sanitize existing racial inequality (Harcourt 2007(Harcourt , 2015.

Cumulative racial disadvantage.
Many studies test for racism in a discrete moment, examining a snapshot of contact with the carceral state. This method washes out the cumulative effect of racism (Murakawa & Beckett 2010, Omori 2018, Van Cleve & Mayes 2015. Modeling "cumulative disadvantage" in case processing, recent sentencing studies find that black felony defendants face a greater likelihood of prison sentences owing to accrued discrimination across criminal justice decision points (Sutton 2013;Wooldredge et al. 2015). Including racial disadvantage accumulated through bail, pretrial detention, and adjudication as a felony, Stolzenberg et al. (2013) find that the odds of receiving a severe sentence are roughly 42% higher for black defendants even after controlling for so-called legally relevant factors. Kutateladze et al. (2014) find that black and Latinx defendants in New York City receive the most disadvantageous combination of outcomesdetainment, case retention, and finally incarceration (see Spohn 2015a,b). Still, this approach identifies "cumulative discrimination" only within individual case processing.

Organizational discrimination.
Sentencing studies that search for individual-level prejudice often overlook organizational and structural contexts that teach, normalize, and reinforce discrimination (Lynch 2013, Lynch et al. 2013, Haney-López 2000. Lynch & Omori (2018) found that US attorney's offices that aggressively charge crack cocaine (relative to other drugs) demonstrate higher black-white inequality in conviction rates across their entire caseload. At the level of the organization, general black criminalization followed from specific prosecution of notoriously harsh antiblack crack penalties and politics (Alexander 2010, Beckett 1997, Provine 2007, Tonry 1995.

The race effect versus the racism effect.
Regression analyses measure discrimination through the independent variable of defendants' race; this specification, however, measures the race effect, not the racism effect. This is a conceptual elision that effectively sets one's racial identity as a stand-alone self-triggering cause, forgetting that racism is enacted through institutional mechanisms, social relations, and political processes. In addition, estimating the race effect along with covariates like income, education, and criminal history means that these allegedly nonracial independent variables will absorb explanatory weight, reducing the race effect , Fields & Fields 2014, Holland 2008, Lynch & Omori 2018, Stewart 2008, Zuberi 2001. Finally, testing the race effect forecloses an understanding of the recursive dynamics between race and the carceral state. That is, crime policy and carceral institutions change with changing racial populations, which in turn may alter racial categories (Van Cleve & Mayes 2015).
This section illuminated the legal, psychological, and social scientific practices of dismissing racial disparity, eliding racial profiling, and affirming the no-discrimination thesis. But these deconstructions of racism are twinned with something more-constructions of black criminality, black dysfunction, and black blameworthiness.

INVENTING "THE NEGRO PROBLEM"
This section sketches the invention of "the Negro problem" as another mechanism for maintaining racial innocence. The fabrication of western civilization, the European race, and whiteness-all constructed as internally unified and intrinsically virtuous-required the corollary inventions of the Orient, indigenous peoples of the Americas, and Africa and the Negro (Robinson 1983, Kelley 1999, Crenshaw 1988. Black feminist intellectual traditions have long recognized criminalization as integral to these inventions, as notions of innocence and guilt were affixed to racialized gender binaries. Analyses from Ida B. Wells, for example, examined lynching and rape together to expose the interdependent construction of pristine white womanhood and the sexually primitive black slave. In a gendered racial triangle, white women's purity was exceptionalized vis-à-vis black women's impurity and imperiled vis-à-vis black men's voraciousness (Haley 2016, Ritchie 2017). Haley's (2016) study of Georgia's convict leasing and prison system identifies carceral development as integral to stabilizing gender identity for white women while enforcing the radical otherness of the black female subject. In short, criminalization does more than disproportionately harm people of color; it stabilizes categories of gendered racial identity, defining whiteness through hierarchical opposition to deviant racial others (Crenshaw 1988).
Racial criminalization, as historian Muhammad (2010) explains in his magisterial The Condemnation of Blackness, is the conflation of blackness with criminality and the simultaneous minimizing of white criminality as the transgression of white individuals, not a reflection on the white race. This intellectual architecture-the criminalization of blackness and decriminalization of whiteness-is maintained by diverse and recursive forces: positivist social science and criminology (Hawkins 1995, Ward 2015, Van Cleve & Mayes 2015, Brown & Schept 2017, Correia & Wall 2018; messaging from mass media and political elites (Beckett 1997, Weaver 2007, Murakawa 2014, Hinton 2016; policing and the carceral state (Alexander 2010, Muhammad 2011, Lynch 2013, McCorkel 2013, Butler 2017, Forman 2017; and surveillance beyond the criminal legal system (Browne 2015, Pager 2007, Rios 2011, Shedd 2015.

Mainstreaming Statistical Rhetoric of "Negro Criminality"
Progressive-era actuaries like Frederick Hoffman pioneered positivist validation of black criminality, securing "the statistical rhetoric of the 'Negro criminal'" as a respectable "proxy for a national discourse on black inferiority" (Muhammad 2010, p. 8). Through the twentieth century, prominent social scientists explained racially disparate crime rates through notions of black subculture, family structure, and social disorganization, perhaps with "sympathy" toward black disadvantage, but nonetheless stressing the pathological features of black life (Hawkins 1995). Popularized theories of black criminality, laden with statistical measures and social scientific terminology, structured a mainstream discourse that ensconced "the crime problem" as a subset of "the Negro problem" (Murakawa 2014, p. 13;Beckett 1997, Muhammad 2010, Weaver 2007. Punitive crime policy is littered with references to Moynihan's pathological black family, dysfunctional ghetto subculture, crack babies, super predators, and a nihilistic underclass (Kelley 1997, Hinton 2016, Camp 2016, Forman 2017. "Culturally, blackness signifies the realm of the always known, as well as the not worth knowing," writes Williams (1997, p. 74). References to at-risk youth, thugs, gangs, broken homes, welfare, and ex-felons, even when not explicitly marked, represent narratives that weave blackness with criminogenic upbringing, poverty, and family dysfunction (Perry 2011). These shorthand references to "Negro criminality" are already known and not worth knowing.

White Exemptions
By contrast, theories of white criminality have been left in the dustbin of history. Ward (2015) reminds us of one such discarded theory. Having observed white youths attacking black people and their property through the 1950s, psychologist Clark (1959) hypothesized that white juvenile delinquency is an underdetected by-product of the white youth's confused cynicism about proclaimed democratic egalitarianism alongside obvious white supremacy. Historically, there are classes of missing perpetrators whose whiteness reduces their likelihood of being reported, policed, or punished (Ward 2015). In the early-twentieth century urban North, whites were extended the benefits of rehabilitation and decriminalization that were withheld from African Americans, as "the black male criminal" supplanted white hoodlums and ethnic gangsters as the dominant signifier of urban criminality (Muhammad 2011). Official measures of crime also largely exclude white "resistance" to desegregation from the 1940s through the 1970s; such "resistance" might have been measured in Uniform Crime Report index crimes like aggravated assault, arson, and murder. Such omissions exaggerate racial gaps in measures of the crime rate (Ward 2015).
Present-day white exemptions include, to name just a few, not having school discipline farmed out to police, not being stopped or frisked, and being permitted to engage in unpoliced disorderly conduct (Lassiter 2015, Ward 2015. With "real criminality" fastened to blackness, prosecutors and judges are more likely to "divert" whites into drug courts, pretrial diversion, or other programming (Eaglin 2016, Schlesinger 2013. Contemporary practices are consistent with longer histories of creating and reserving rehabilitation programs, and social insurance and welfare in general, for whites (Alexander 2010, Muhammad 2010. By contrast, present-day returns to so-called therapeutic programming have reconfigured rehabilitation as responsibilization. Working with large proportions of black, Latinx, and Native American people, contemporary programs "rehabilitate" by counseling people to lower their expectations, control their anger, correct their defective selves and assume individual responsibility (Cox 2017, McCorkel 2013, Miller & Alexander 2016, Miller & Stuart 2017.

Routine, Cumulative, Compounding Racial Inequality
Racial criminalization is routine and ubiquitous, not a single "event" performed by a carceral administrator (Rios 2011, Shedd 2015. Given the pervasiveness of being watched with suspicion, young people of color experience "hypercriminalization," the process by which "everyday behaviors and styles become ubiquitously treated as deviant, risky, threatening, or criminal, across social contexts" (Rios 2011, p. xiv). Brame et al. (2014) estimated that 30% of black boys have been arrested by age 18 (versus 22% for white boys). This figure increases to 49% by age 23; i.e., half of all black men hit young adulthood having been arrested at least once (versus 38% for white men).
Disadvantages of carceral contact accrue over the life course, although racial criminalization means that blacks are marked even without an official criminal record. Sociologist Pager's (2007) groundbreaking racial audit studies, which assess callback rates for entry-level jobs, found that any kind of criminal record, even a low-level felony conviction without evidence of incarceration, reduces callback rates for white and black applicants. However, white applicants with a criminal record were three times as likely to receive a callback as equally qualified blacks with a criminal record (17% versus 5% called back). Further still, white applicants with a criminal record were just as likely to receive a callback as black applicants without any criminal history (17% versus 14% called back; the difference is not statistically significant) (Pager 2007, pp. 90-91;Shedd 2015). Internet search results are perhaps a more mundane example of reinforcing presumptions of black criminality. Compared with Google searches of whitesounding names (e.g., Geoffrey or Emma), searches of black-identifying names (e.g., DeShawn or Jermaine) are more likely to deliver results suggestive of an arrest record (Sweeney 2013). Results from a potential employer's Google search may in effect compound racialized wealth inequality, even if based on misinformation.
Beyond the legally marked individual, the costs of racial criminalization ripple across families, neighborhoods, and generations. The cumulative risk of experiencing paternal imprisonment for black children (25.1%) and white children (3.6%) exacerbates racial inequality in children's well-being, especially with regard to child homelessness and foster care placement, poverty, and food insecurity (Sykes & Pettit 2015, Wakefield & Wildeman 2014. Of all black children born in 1990 to fathers without a high school diploma, 50.5% would see their father imprisoned, up 130% from 1978. The risk of paternal imprisonment for black children of college-educated fathers (13.8%) is nearly twice the risk of paternal imprisonment for white children of fathers without a high school diploma (7.2%) (Wakefield & Wildeman 2014). The collateral consequences of punishment further entrench racial inequality in wealth, health, and well-being.

Statistical Illusions of Progress
Recall that "the Negro problem" safeguards fantasies of a just national order. This dynamic is well demonstrated in research finding that high black incarceration rates have the perverse effect of creating statistical illusions of racial progress. Commonly used surveys sample only individuals living in households, not those living behind bars. These surveys include the Current Population Survey, the National Health Interview Survey, the National Survey on Drug Use and Health, and the Survey of Income and Program Participation. Studies that rely on these data actually underestimate the magnitude of racial inequality, creating what Pettit (2012) calls "illusions of progress" through "siphoning effects." For example, Current Population Survey data suggest that men's black-white gap in high school completion narrowed from 13.6 to 6.3 percentage points between 1980. When Pettit (2012 counted the imprisoned, however, this narrowing of racial disparity all but disappeared. The racial gap in high school completion remained mostly stable, hovering at roughly 11 percentage points from 1988 to 2008 (Pettit 2012, Ewert et al. 2014. Similarly, conventional estimates of men's black-white employment and wage gaps exaggerate narratives of racial progress because the most marginalized black people are siphoned off from the count (Western & Beckett 1999, Western & Pettit 2005.

LIBERAL REFORMISM
This section considers liberal reformism as another mechanism for neutralizing radical protest and reaffirming the racial innocence of the carceral state. "Racial innocence is the alchemy by which Americans turn enduring and otherwise visible inequality into redemptive stories of rights, equal protection, individualism, and progress," as Taylor (2015, p. 56) explains. This alchemyof spinning virtue from hoarding and domination-relies on a narrative of progress and liberal reformism. Racial innocence is adaptive, and the "American Creed" that celebrates procedural equality enables and constrains political reform (Crenshaw 1988, HoSang 2010, Taylor 2015. Liberal reform reflects the recognition that movement demands are powerful and threatening, and therefore preservation of power relations is best served by adopting some movement demands after reducing them to a suitably moderate form (Omi & Winant 1986, Schept 2015.

Decarcerating the Relatively Innocent
Recent reforms have focused on relatively sympathetic figures like the "non, non, nons": the nonserious, nonrepeat, and nonviolent offenders (Beckett 2018, Gilmore 2015, Gottschalk 2015. This strategy, however, sidesteps the two most consequential macroshifts in sentencing policy since the 1970s: the increased likelihood that a felony arrest would result in prison admission and increased time served in prison (Beckett 2018, Tonry 2013, Travis et al. 2014. Confronting these equilibrium shifts would require massive sentencing reform, including but not limited to repealing mandatory minimums and three-strikes provisions, abolishing life sentences and virtual life sentences of greater than 40 years, replacing sentencing's high floors with low ceilings, reinstating good-time early-release credits, and reducing parole length and terms of revocation (Beckett 2018;Gottschalk 2015;Kim et al. 2018;Tonry 2013Tonry , 2016Mauer & Nellis 2018). In short, advocating for the relatively innocent is inadequate. Note, for example, that incarceration rates still would have quadrupled over the past 35 years even if all drug convictions were eliminated (Beckett 2018, Forman 2012, Gottschalk 2015. Advocating for the relatively innocent legitimizes the idea that the relatively guilty deserve what they get. Based on content analysis of legislative reforms and media language from 2000 to 2012, Beckett et al. (2016) found that reforms pertaining to nonviolent offenses were often justified as freeing resources to incarcerate violent offenders for even longer sentences. That is, reformers demand less prison in the name of more prison (Beckett et al. 2016;Gilmore 2015Gilmore , 2017. This reform tactic also hardens status divides between nonviolent and violent criminals, categories that are neither natural nor self-evident. Rather, police and district attorneys actively produce "serious" or "violent" felony charges, and "their use is part of a racial apparatus for determining 'dangerousness'" (Gilmore 2015).

Carceral Feminism
"Carceral feminism" critically names efforts to criminalize away the routine violence of rape, domestic violence, and heteropatriarchy (Burnstein 2007, Davis 2016, Thuma 2019). As a major "lever of legitimacy for expanding the carceral state," the Violence Against Women Act (VAWA) delivered unprecedented federal funding for sexual and domestic violence prosecution. VAWA was nested in the omnibus Violent Crime Control and Law Enforcement Act of 1994, which allocated nearly $10 billion for state prison construction and subsidized local hiring of more police officers (Thuma 2019). Other manifestations of carceral-feminist expansion include mandatory arrest policies and pro-prosecution protocols. Women who do not "fit the traditional image of the innocent victim"-that is, black women, women of color, poor women, sex workers, lesbians, and trans women-are likely to be criminalized rather than protected through such policies (Richie 2012, p. 123;Gottschalk 2006, Gruber 2018, Ritchie 2017. Much like "hate crime" legislation, carceral feminism builds punitive capacity while enabling lawmakers to stand in disavowal of state-sanctioned racial heteropatriarchy (Reddy 2011, Schept 2015, Spade 2015. In short, critics of carceral feminism highlight the mutually reinforcing dynamics of carceral expansion, coopted resistance, and falsely affirmed state legitimacy.

Training and "Police Humanitarianism"
Like the Black Lives Matter Movement, liberation movements and uprisings throughout the 1960s identified police as frontline enforcers of racial hierarchy (Taylor 2016, Platt 2018.
Reformers in both moments turned to police training, promoting what some critically call "police humanitarianism" and "carceral humanism" (Gilmore 2017, Gilmore & Gilmore 2016, Kilgore 2014, Schept 2015. In classic terms, The Iron Fist and the Velvet Glove identified the rise of "community policing" as complementary to the rise of aggressive paramilitary policing (Center for Research on Criminal Justice 1975). Indeed, funding for community policing rose with SWAT teams and paramilitary units (Camp 2016, DeMichele & Kraska 2001, Hinton 2016. Procedural justice and implicit bias dominate the current landscape of police training. Against the chorus of support, some scholars fear that procedural-justice training, as a standalone remedy, risks sustaining saturated police presence in communities of color, essentially teaching police to continue the same patterns but with superficial niceties (Bell 2017, Butler 2017, Epp et al. 2014, Vitale 2017. Some similar concerns apply to implicit-bias training, as companies like Fair & Impartial Policing ® instruct police to cultivate the skill of acknowledging their cognitive bias. Framed as a pervasive yet individual-level phenomenon, implicit bias is presented as mitigatable through education (Petersen 2018). This vision of training calls to mind HoSang's (2010, p. 31) very definition of racial innocence: it is the notion that "the self-possessed individual" can decide "to free himself or herself from the narrow trappings of prejudice" and "no longer be 'guilty' of racism." Experimental tests of debiasing interventions demonstrate some success in very short-term reductions of implicit biases, but there is limited and conflicting evidence of whether reductions endure past two or three months (Smith 2015). Observations of implicit-bias police training paint a disturbing picture of racial criminalization reaffirmed, not deconstructed; one popular training module assures officers that implicit associations between blackness and criminality are partially justified by reality (Petersen 2018). Critical race theorists have long relied on implicit bias research to combat legal definitions of discrimination as intentional harm, but recent pushback against the implicit-bias preoccupation is illuminating in the context of police training. Marshalling long citation chains of experimentally confirmed implicit bias research is a satisfying "hard evidence" refutation of the intent standard, but not all racial harm can or should be attributed to cognitive bias in the unwitting racist's interior (Seigel 2018, Selmi 2016. The habit of blaming implicit bias-a subset of the overarching habit of equating racism with psychological error-risks exonerating institutional arrangements that incentivize and legitimize racist harm (Butler 2017, Frymer 2005, Vitale 2017, Obasogie & Newman 2018.

CONCLUSION: REJECTING INNOCENCE
Law and social science are implicated in constructing the carceral state, of course, but this article illuminated the legal and social scientific tools used to obscure the carceral state, to blunt our perception of its intrinsic dehumanization and its routine racism. Criminal law condones racial profiling, caging, and killing, while antidiscrimination law formalizes the presumption of racial innocence by tasking plaintiffs with the burden of proving intentional discrimination. Positivist social scientists practice the same willful ignorance by setting a null hypothesis of no racism, fastidiously separating racial from presumably nonracial factors, and then measuring "the race effect" instead of racism's effects. Social science has furnished a respectable statistical rhetoric of black criminality since the Progressive era, and this articulation of "the Negro problem" justifies carceral development and its racial disparity. In short, willful ignorance, invention of "the Negro problem," and liberal reformism secure the racial innocence of the US carceral state.
Witnessing carceral devastation requires rejecting innocence. For scholars of the carceral state, this entails rejecting professional investments in race-free crime statistics and nonracial legal variables; measures and methods are freighted to an unfolding history in which race matters (Muhammad 2010, Thompson 2010, Ward 2015, Zuberi 2001. More broadly, witnessing carceral devastation requires rejecting the fantasy that social scientists can or should be innocent of normative commitments. The dehumanization of the carceral state exceeds what can be seen through racial disparity analysis, which is too often the stand-in for moral analysis. When Baldwin [ (1960, p. 210] challenged "urban renewal," he staked an historically-informed moral claim about race, place, and abolition, writing: "The people in Harlem know they are living there because white people do not think they are good enough to live anywhere else. No amount of 'improvement' can sweeten this fact… A ghetto can be improved in one way only: out of existence." The same should be said of prisons. Prisons normalize the practice of creating separate and subordinate spatial, legal, and political universes for whole categories of people (Gottschalk 2008, Simon 2007, Lerman & Weaver 2014, Miller & Stuart 2017. Further still, the carceral state reaffirms the foundational fictions of racial capitalism-fictions of freedom through the sanctification of property, fictions of earned opulence and deserved deprivation (Gilmore 2007(Gilmore , 2017Davis 2016;McLeod 2015;Camp 2016;Brown & Schept 2017;Vegh Weis 2018;Turner 2012). For these very reasons, abolitionist scholars insist that no amount of improvement can secure the innocence of the carceral state. Indeed, it is the pursuit of innocence that constitutes the crime.

DISCLOSURE STATEMENT
The author is not aware of any affiliations, memberships, funding, or financial holdings that might be perceived as affecting the objectivity of this review.