The Law‐Before: Legacies and Gaps in Penal Reform

Despite vast expenditures on U.S. prison construction in the late twentieth century, infrastructure has not kept pace with the punishment imperatives of mass incarceration. Dangerously overcrowded confinement conditions remain widespread in prisons and jails, raising recurring dilemmas about the judicial oversight, and legal regulation of correctional policy. Perhaps no state better exemplifies the prison overcrowding crisis than California, which operates one of the nation's and western world's largest prison systems. After several decades of rapid growth, by 2011 the state incarcerated nearly twice the number of people its prisons were designed to hold. Despite these levels, California's recidivism rate remained one of the highest in the nation; roughly 60 percent of those released from prison reoffended within three years (Pew Center on the States 2011). Such extreme prison overcrowding combined with its lack of crime control efficacy led to historic intervention by the U.S. Supreme Court in Brown v. Plata (2011). In a 5-4 decision, the Plata court found California's conditions of confinement to violate the Eighth Amendment's prohibition on cruel and unusual punishment and ordered the state to reduce its prison population to 137.5 percent of capacity (or, by roughly 40,000 people) within two years. Justice Scalia decried the order as "the most radical injunction issued by a court in our Nation's history" (Brown v. Plata 2011: 1 of Scalia dissent).Brown v. Plata has understandably been characterized as a "remarkable" case (Simon 2014). Even more remarkable is the state of California's response. While states have often sought to comply with population cap orders by expanding prison capacity (e.g., Feeley and Rubin 1998; Guetzkow and Schoon 2015; Schoenfeld 2010), California enacted legislation known as "Public Safety Realignment," or Assembly Bill 109 (AB 109 2011), which localized the onus of compliance to individual counties (Schlanger 2013). AB 109 devolves the supervision of most nonviolent offenders to the county level and, notably, delegates unprecedented discretion to local practitioners to either incarcerate those previously sent to state prison in local jails or to use alternative, community-based sanctions that do not entail incarceration (Pen. Code §1170(h); §17.5). California's unique response, thus, raises the possibility of decarceration-rather than prison expansion- as a viable mode of legal compliance with court intervention for the first time in decades.In a keynote address to the National Institute of Justice, criminologist Joan Petersilia (2012) said this about California's Realignment: "It is the biggest criminal justice experiment ever conducted in America, and most people don't even know it's happening." The Economist (May 19, 2012) has also called AB 109 "one of the great experiments in American incarceration policy," in part due to concerns about its effects on future crime levels, but also because whether it will in fact lead to decarceration, as many reformers have hoped (e.g., ACLU 2012), remains an open question. Emerging awareness of the underlying variation in California counties' historical reliance on the state prison system has raised concerns that the relatively small number of historically high prison using counties-counties that disproportionately drove the state's prison overcrowding crisis in the first place (e.g., Ball 2012)-will use the discretion afforded to them under Realignment to either subvert the law's central mandates or to simply relocate the sites of incarceration from state prison cells to local jail cells (Lynch 2013; Petersilia and Snyder 2013).Realignment's "experiment" has attracted interest from public policy scholars (e.g., Bird and Grattet 2014; Lofstrom and Raphael 2013; Males and Goldstein 2014) and legal scholars (e.g., Ball 2012; Schlanger 2013; Zimring 2014). However, this is the first study to empirically address the sociolegal questions raised by this distinctive form of regulation, which renders legal compliance possible because of-not despite-local variations in front-line implementation. …

The Law-Before: Legacies and Gaps in Penal Reform

Anjuli Verma
This article introduces the law-before as an analytic tool for enhancing explanations of legal reform. Based on an integration of neo-institutional law and organizations studies and punishment studies of local variation in penal policy, I define the law-before as the past organizational practices and power arrangements that precede law-on-the-books and shape present day implementation. I utilize the law-before as a heuristic to investigate the legacy effects of variations in local practice on the implementation of the prison downsizing law, AB 109, or "Realignment," in California. I analyze organizational documents produced by county practitioners in the aftermath of AB 109's enactment in 2011 as empirical windows into how actors shape the meaning of law in local settings. I find that practitioners in counties with divergent historical imprisonment patterns enact four processes (overwriting or underwriting law, selective magnification, and selective siting) to arrive at distinct interpretations of AB 109 as mandating system-wide decarceration or the relocation of incarceration from state prisons to county jails. Although my data do not speak to the ultimate implementation of AB 109, the processes revealed have practical implications for the reform goal of decarceration by rationalizing distinct resource allocations at an early stage in the implementation process.
Despi te vast expenditures on U.S. prison construction in the late twentieth century, infrastructure has not kept pace with the punishment imperatives of mass incarceration. Dangerously overcrowded confinement conditions remain widespread in prisons and jails, raising recurring dilemmas about the judicial oversight, and legal regulation of correctional policy. Perhaps no state better exemplifies the prison overcrowding crisis than California, which operates one of the nation's and western world's largest prison systems. After several decades of rapid growth, by 2011 the state incarcerated nearly twice the number of people its prisons were designed to hold. Despite these levels, California's recidivism rate remained one of the highest in the nation; roughly 60 percent of those released from prison reoffended within three years (Pew Center on the States 2011). Such extreme prison overcrowding combined with its lack of crime control efficacy led to historic intervention by the U.S. Supreme Court in Brown v. Plata (2011). In a 5-4 decision, the Plata court found California's conditions of confinement to violate the Eighth Amendment's prohibition on cruel and unusual punishment and ordered the state to reduce its prison population to 137.5 percent of capacity (or, by roughly 40,000 people) within two years. Justice Scalia decried the order as "the most radical injunction issued by a court in our Nation's history" (Brown v. Plata 2011: 1 of Scalia dissent).
Brown v. Plata has understandably been characterized as a "remarkable" case (Simon 2014). Even more remarkable is the state of California's response. While states have often sought to comply with population cap orders by expanding prison capacity (e.g., Feeley and Rubin 1998;Guetzkow and Schoon 2015;Schoenfeld 2010), California enacted legislation known as "Public Safety Realignment," or Assembly Bill 109 (AB 109 2011), which localized the onus of compliance to individual counties (Schlanger 2013). AB 109 devolves the supervision of most nonviolent offenders to the county level and, notably, delegates unprecedented discretion to local practitioners to either incarcerate those previously sent to state prison in local jails or to use alternative, community-based sanctions that do not entail incarceration (Pen. Code §1170(h); §17.5). California's unique response, thus, raises the possibility of decarceration-rather than prison expansionas a viable mode of legal compliance with court intervention for the first time in decades.
In a keynote address to the National Institute of Justice, criminologist Joan Petersilia (2012) said this about California's Realignment: "It is the biggest criminal justice experiment ever conducted in America, and most people don't even know it's happening." The Economist (May 19, 2012) has also called AB 109 "one of the great experiments in American incarceration policy," in part due to concerns about its effects on future crime levels, but also because whether it will in fact lead to decarceration, as many reformers have hoped (e.g., ACLU 2012), remains an open question. Emerging awareness of the underlying variation in California counties' historical reliance on the state prison system has raised concerns that the relatively small number of historically high prison using counties-counties that disproportionately drove the state's prison overcrowding crisis in the first place (e.g., Ball 2012)-will use the discretion afforded to them under Realignment to either subvert the law's central mandates or to simply relocate the sites of incarceration from state prison cells to local jail cells (Lynch 2013;Petersilia and Snyder 2013).
Realignment's "experiment" has attracted interest from public policy scholars (e.g., Bird and Grattet 2014;Lofstrom and Raphael 2013;Males and Goldstein 2014) and legal scholars (e.g. , Ball 2012;Schlanger 2013;Zimring 2014). However, this is the first study to empirically address the sociolegal questions raised by this distinctive form of regulation, which renders legal compliance possible because of-not despite-local variations in front-line implementation. I analyze organizational documents known as "Realignment plans" produced by county officials in the aftermath of the Plata order and AB 109's enactment in 2011 as empirical windows into local legal interpretation and compliance. I compare plans from groups of historically high and low prison using counties to answer emerging questions about AB 109's interpretation among counties with divergent histories of state prison reliance.
I find that historically low imprisoning counties interpret the law as mandating overall decarceration in both state prisons and county jails, while historically high imprisoning counties interpret the law as mandating a relocation of the predominant site of incarceration from state prisons to county jails. Beyond providing initial empirical support for the concern that practitioners in some locales will subvert the reform goal of decarceration under AB 109, I identify mechanisms of legal translation that begin to shape divergent understandings of the law in the early planning stages of implementation. Specifically, I trace four distinct interpretative processes in the plans: overwriting law, in which local actors render law's authorship invisible by masking the legally mandated origin of changes to local policy and practice; underwriting law, which alternatively entails openly relying on law's force to substantiate local policy changes; selective magnification, in which local actors emphasize certain statutory components to the exclusion of others; and selective siting, which locates the site of the problem law is meant to solve in ways that render certain interpretations coherent while rendering others illogical. These interpretive processes in turn reveal competing field-level logics about law and legal regulation with respect to both the fundamental legitimacy of law to regulate local penal practice (whether in the form of federal case law or state legislation) and the governmental origin of the penal policy failures that legal regulation attempts to mitigate.
I argue that through the processes and according to the distinct logics reflected in these plans, local actors produce their own meaning of Realignment in ways that facilitate the continuation of previous imprisonment practices-and the power arrangements that enabled those practices-despite attempted reform. I conceptualize these past practices and power arrangements as constituting the law-before the law-on-the-books. By bringing together previous studies that explain the political development, reproduction, and change-resistance of penal policy through historical institutional processes (e.g., Gottschalk 2006;King, Messner, and Baller 2009;Savelsberg and King 2007;Schoenfeld 2010), I develop the law-before as an analytic device, or heuristic (Abbott 2004), to enhance explanations of legal reform.
This article extends three bodies of law and society scholarship. I contribute to the gap study literature by articulating the law-before as a crucial prologue to the canonical law-on-the-books to law-in-action schematic. The study extends the neo-institutional literature on how legal meaning is shaped within organizational fields by applying insights about legal endogeneity (e.g., Edelman 2005;Edelman, Fuller, and Mara-Drita 2001;Edelman and Talesh 2011) to the criminal justice field, which remains understudied in law and organizations research. In applying these insights, this article also contributes to the punishment literature on local variation in penal policy. While much of this recent scholarship aims to explain local variation in the development of mass incarceration (e.g., Barker 2009;Campbell and Schoenfeld 2013;Goodman, Page, and Phelps 2015;Lynch 2011), this study is one of the first to examine local variation in the potentially emergent development of decarceration. I draw on and deepen Page's (2011Page's ( , 2013 account of the "penal field" by highlighting county-level practitioners as key field actors, as well as how the law-before in local penal fields shapes actors' relationships and struggles for power in responding to legal reform. Finally, my examination of the lawbefore adds to theoretical development in both the organizational compliance and punishment literatures by providing initial evidence for a typology of local legacies of penal practice and local orientations to the legitimacy of "higher order" law to regulate imprisonment.
I proceed by first more fully describing AB 109's distinctive statutory features and the underlying variation in its field of implementation. I then review theoretical perspectives on the difficulty of realizing legal reform. I describe my methodology before presenting findings, and I conclude with a discussion of practical and theoretical implications.
AB 109: California's "Public Safety Realignment" Law Through multiple amendments to the California Penal Code, AB 109 shifts from the state to its 58 counties the responsibility for supervising nonserious, nonviolent, non-sex-registerable offenders (known as "non-non-nons" in local parlance)-all of whom would have otherwise previously served felony sentences in state prison (Pen. Code §1170(h)). 1 AB 109 also makes significant changes to state parole by shifting new post-release supervision responsibilities to counties, including the requirement that nearly all violations and revocations be processed and sanctioned locally (Pen. Code §3450 Tit. 2.05 of Pt. 3). Through this devolution, the state prison system expected to shed nearly one-fourth of its inmates and three-fourths of its parolees upon full implementation (California Legislative Analyst's Office 2011).
County-level practitioners are central to AB 109's implementation and, by implication, to the state's ultimate compliance with Plata. AB 109 explicitly calls for local customization and overtly appoints practitioners to shape the law's meaning. It does so by requiring a formal county-level planning process but specifying little programmatic substance for local implementation plans. This process required a standard practitioner group within each county to produce a written implementation plan, approved by the county's Board of Supervisors, delineating how it would exercise newly acquired discretion and allocate state funding. By statute, each county's Chief Probation Officer chairs a group consisting of the Sheriff, District Attorney, Public Defender, Presiding Judge, a municipal Police Chief, and a public health agency representative (Pen. Code §1230.1). The legislation designates this group as an "Executive Committee" of each county's pre-existing "Community Corrections Partnership" (CCP). CCPs consist of more than a dozen members and are also chaired by the Chief Probation Officer of each county. They were formed two years earlier under separate legislation to implement a state incentive program for counties to reduce the number of people returned to state prison for probation violations (Senate Bill 678 2009). In selecting this particular subset of the CCPs to draft Realignment plans, AB 109 shaped a specific structure for the planning process by including certain actors while excluding others-notably, counties' chief fiscal agents, a number of broader social services department heads and community members with special interests in offender rehabilitation and victims' rights (Pen. Code 1230(b)(2)). At the same time, the statutory designation of this "Executive Committee" reflects (and was itself constrained by) institutional arrangements put in place by previous legislation, including the leadership designation of Chief Probation Officers as CCP chairs. The law imposed no state-level review, and no requirements or funding conditions on the plans' form or content, leading to wide inter-county variation in the final documents despite the standard group charged with drafting them.
Spanning nearly 1,000 pages, AB 109's purposes, intents and many of its mandates were frustratingly ambiguous to practitioners (e.g., Petersilia 2014). While largely acknowledged as the state's central mode of compliance with Plata, its statutory language explicitly states that it is not intended to reduce state prison overcrowding (Pen. Code §17.5(b)). At the same time, its Legislative Findings include bold statements decrying the state's prison overcrowding problem and calling for reduced reliance on incarceration in favor of community-based alternatives (Pen. Code §17.5(a)). AB 109 is unusual less, however, in the ambiguity of its "true" purpose (e.g., Black 1972;Feeley 1976) than in its explicit award of local discretion and overt appointment of implementers to shape its very meaning.
Like previous "grassroots" governance strategies (Selznick 1965), the discretion and devolution of authority under Realignment come at a price to locals. AB 109's distinction, however, is that it exacts a different price from each county. AB 109 ends the long-standing correctional "free lunch" (Zimring and Hawkins 1991) by requiring counties to largely assume the costs of incarcerating realigned offenders. This yields unequal local costs because counties have historically depended on the state's "free lunch" to varying degrees. Annual data reveals wide variation in the rates at which counties sent people to state prison in the decades leading up to Realignment. This variation exhibits a pattern in the presence of two distinct, relatively small outlier groups: consistently "high prison using" and consistently "low prison using" counties (Ball 2012). This variation is not explained by differences in local crime rates. In an analysis of data from 2000 to 2009, Ball (2012) shows that, net of local crime rates, the same "high use" counties repeatedly fell into the top quartile of state prison admission rates for nearly all 10 years, as did the same "low use" counties fall into the lowest quartile.
Regardless of the range of possible explanations for this variation, it leads to what Lofstrom and Raphael (2013: 8) call different "doses" of Realignment. Historically high prison using counties experience high "doses" of AB 109's reforms because they must adapt to managing the large number of people otherwise sent to state prisons, while Realignment delivers a smaller reform "dose" to counties that originally sent relatively fewer to state prison. Males and Goldstein (2014) refer to these groups as "state-dependent" versus "self-reliant" counties, which signals the different local-state power relationships that may underlie local imprisonment practices and differences in associated legitimacy claims about the proper governmental level of imprisonment regulation. The variation in past imprisonment practices can be readily observed in counties' state prison use rates over time. The power arrangements enabling these practices are more difficult to measure. However, the Realignment plans provide a textual window into the variation in power relations not only among the local practitioners selected by statute to write plans but also between these local practitioners and the state.

The Difficulty of Realizing Legal Reform
Beyond the "Gap" AB 109 can be understood as legislation with a "gap" already written into it. Unlike centralized bureaucratic models of legal regulation premised on the Fordist-era ideal of uniformity and the standardization of practices within a regulatory field (e.g., Barron, Dobbin, and Devereaux 1986;Jacoby 1985;Llewellyn 1957), California's Realignment takes for granted that standardization is not feasible (and may not be desirable). It even creates a statutory vehicle for discretionary local implementation in the mandated implementation plans. Gap studies have classically conceptualized discretion, especially in concert with legal ambiguity, as a threat to the redistributive power of law, in part, because the implementation dilemmas it presents for front-line workers in local settings often lead to unintended policy directions (e.g., Lipsky 1971). Discretion is by design, however, under AB 109, which recasts the very notion of the "gap" as a puzzle to be explained.
Related scholarship has raised normative questions about decentralized social policy in the context of American federalism, which places constitutional and political limits on federal intervention into state and local policymaking even as it facilitates a nationalized policy agenda (e.g., Feeley andRubin 1998, 2008;Miller 2008). As a somewhat paradoxical regulatory model that descends directly from federal court intervention but in which compliance is premised precisely on county-level variations in implementation, Realignment complicates and raises the stakes of these questions for the reform goal of decarceration.

Legal Regulation of Organizations
Neo-institutional analyses of how organizations reshape the meaning of legal regulations within their fields show that legal ambiguity leaves room for discretionary implementation, which facilitates the "endogenous" interpretation of law. Legal endogeneity is the process by which the meaning of law comes to be "generated within the social realm it seeks to regulate" (Edelman 2005: 337). Accounts of legal endogeneity in the regulatory environment show that, from an organizational standpoint, "to comply or not to comply-that is not the question" (Edelman and Talesh 2011); rather, the question is how the very meaning of "compliance" gets defined, and by whom.
In studies of corporate compliance with public regulation, legal endogeneity begins through the "managerialization" of law (e.g., Edelman, Fuller, and Mara-Drita 2001), where prototypical business logics such as efficiency and managerial discretion subsume the legal logics of due process and impartiality-in effect, reshaping the substantive meaning of legal rules to benefit corporate managers and elites (e.g., Krawiec 2003;Reichman 1992;Schneiberg and Bartley 2001;Talesh 2009Talesh , 2012. Organizational managers, as both the objects and designated implementers of regulation, may, thus, subvert legal mandates by articulating in their own terms what it means to comply with the law. In turn, where courts formally adopt managerial interpretations in case law (e.g., by deferring to internal corporate grievance procedures as evidence of compliance), law is ultimately rendered endogenous (e.g., Edelman, Uggen, and Erlanger 1999;. In the criminal justice context, which remains understudied in law and organizations literature (cf. Jenness and Smyth 2011), Grattet and Jenness (2005) identify legal "surplus" as another mechanism of legal endogeneity. In the case of hate crime policy, interpretations of an ambiguous California law filtered through numerous police agencies situated in diverse local contexts rather than through corporate managers articulating standard business logics. This led to a surplus in legal meaning, or "multiple legitimate expressions of the same rule"; importantly, however, Grattet and Jenness (2005: 893) found that these expressions were not perfectly idiosyncratic to locale but "clustered" as a function of distinct policy diffusion processes that generate similarity in organizational practices as organizations struggle to maintain legitimacy within their fields (e.g., DiMaggio and Powell 1983). Organizational fields are defined in the neo-institutional literature as "a community of organizations that partakes in a common meaning system and whose participants interact more frequently and fatefully with one another than actors outside the field" (Scott 1992: 56). Legitimacy is sustained, in part, by key actors who operate as "standards-bearers" (Crank 1994) with the power and influence to produce legitimate legal meaning within these fields. Jenness and Grattet (2005: 339) bridge these insights with gap studies by articulating organizational fields as constituting a law-in-between stage, where law is subject to bureaucratic quests for legitimacy as actors translate law-on-the-books into law-inaction. This suggests that, while the surplus of viable interpretations of an ambiguous law leads to variation in the law-in-between, the distinct processes by which actors sustain legitimacy within fields may ultimately render this variation patterned rather than perfect.

Local Variation in Penal Policy
Punishment and society scholarship has recently drawn attention to the puzzle of local variation in penal law and practice as a defining characteristic of modern punishment (e.g., Hannah-Moffat and Lynch 2012). While recognizing the existence of a predominant penal order, such as "mass incarceration," this literature reveals the multiple, often contradictory, ways that punishment transforms over time and across place, including how law becomes mobilized in different settings. These settings are conceptualized geographically as well as jurisdictionally, and exhibit variation at multiple units of analysis, including among nations (e.g., Savelsberg 1994Savelsberg , 1999, states (e.g., Barker 2009;Campbell and Schoenfeld 2013), counties (e.g., Arvanites and Asher 1998;McCarthy 1990;Percival 2010;Weidner and Frase 2003), and court jurisdictions (e.g., Lynch and Omori 2014;Ulmer 2005).
As a result, punishment scholars confront persistent questions about why and how similar macrolevel phenomena lead to varied punishment outcomes in specific places. Explanations have been mainly variable-centered in showing that a range of spatial sociodemographic characteristics exert causal effects on incarceration levels even when jurisdictions operate under the same criminal codes and sentencing statutes. These studies overwhelmingly find, like Ball (2012), that differences in local crime rates do not adequately explain local variation; rather, it is the interaction of political ideology, racial demography, levels of urbanization and income inequality, prison and jail capacity factors, and crime rates that produce high degrees of variation in incarceration (e.g., Arvanites and Asher 1998;Jacobs and Carmichael 2001;McCarthy 1990;Percival 2010;Stucky, Heimer, and Lang 2005;Weidner and Frase 2003).
Because these findings have falsified the hypothesis that imprisonment levels are merely the artifact of crime levels, suggesting instead that cultural differences underlie the ways that local jurisdictions translate penal law into local contexts, Lynch (2011: 674) has argued that the literature should move beyond variable-centered explanations and pay more systematic attention to "law as locale": how criminal and penal law as practiced is significantly shaped by the local (locale) such that, although law on the books might lead us to expect some homogenization of outcomes within state and federal jurisdictions, law in action indicates much more microlevel variation shaped by local norms and culture related to how the business of criminal justice happens in any given place. Page (2013: 152) makes the related argument that studies should focus analytic attention on the intervening mechanisms that translate large-scale phenomena, such as legal reform, into concrete outcomes in specific places; he offers the penal field concept as the "something missing" that explains this variation. Page (2013: 162-3) stresses the need to account for the role that people play in making decisions that translate into penal policies on the ground, and in particular, "which actors are involved (and, importantly, not involved) in the struggle over the policy matter and what is the relationship between these actors' positions in the field." Goodman, Page and Phelps (2015: 315) synthesize this into an "agonostic" field analysis framework for explaining penal change, which "posits that penal development is fueled by ongoing, low-level struggle among actors with varying amounts and types of resources." Punishment scholars have also applied historical institutionalist concepts to explain local variation, showing how legacies of lynching in Southern states (Fleury-Steiner, Kaplan, and Longazel 2015;Jacobs, Kent, and Carmichael 2005;King, Messner, and Baller 2009;Petersen and Ward 2015) and path dependence and policy feedback processes (Campbell and Schoenfeld 2013;Gottschalk 2006;Schoenfeld 2010) function to channel racialized social control and punitive penal policies over time. When Lynch and Omori (2014) tested the "law as locale" concept across federal court district responses to a line of potentially transformative U.S. Supreme Court rulings on criminal sentencing, they concluded that "local legal practices not only diverge in important ways across place but also become entrenched over time such that top-down legal reform is largely re-appropriated and absorbed into locally established practices." These findings provide empirical support for Savelsberg and King's (2007: 202) observation that "legacies of the past enable and constrain government decision making" in the present-day creation and enforcement of law. Until now, these insights have been largely oriented to explaining the punitive turn to mass incarceration. However, California's Realignment presents the puzzle of local variation in explaining a different (potential) penal development: decarceration.
The Law-Before AB 109 draws attention to a remaining limitation in these theoretical perspectives on legal reform. The origin story of the "gap" in the law and society schematic begins with law-on-thebooks, which obscures the salience of conditions that precede the codification of formal law. This schematic fails in particular to account for how past practices and power arrangements may shape gaps in implementation. Drawing on historical institutionalist accounts of social transformation (e.g., Savelsberg and King 2007;Sewell, Jr. 2005), I conceive of these past practices and power arrangements as much more than historical contextual factors fixed at a previous point in time, but as legacies that successively shape how local actors translate today's law-on-the-books into tomorrow's law-in-action. The law-before heuristic can enhance explanations of legal reform by provding an analytic lens for investigating the effects of these legacies.
California's Realignment highlights local variation as a salient feature of the law-before. The implication for understanding legal reform is that rather than a single "gap," many gaps may exist according to the degree of local variation in the law-before. At the same time, the different "doses" of reform AB 109 presents in counties with divergent historical patterns of imprisonment practice suggests that there may be patterns of variation in implementation, or multiple types of gaps, rather than perfectly idiosyncratic local responses. An "agonistic" perspective of the penal field (Goodman, Page, and Phelps 2015) further leads to the proposition that a deeper order drives this pattern of variation: the different types of gaps observed will reveal the relationships and power arrangements among local actors that underlie responses to legal reform. These underlying arrangements contribute to explaining not only the variation in legal implementation of AB 109 but also the local conditions under which different responses to reform are possible (or not).
This leads to the following research questions, refined in light of the law-before: How is AB 109 interpreted in counties with divergent historical patterns of past practice? Do counties with historically high imprisonment rates interpret AB 109 differently than those with historically low imprisonment rates? If so, what processes do local actors exhibit in the plans to arrive at these interpretations?
Methodology I analyzed 430 pages of county Realignment plans written in 2011, the first year of AB 109's enactment. I approached analysis from an institutional ethnographic perspective. Following Smith's (2006: 67) elaboration of the "Act-Text-Act" sequence, I conceptualize these texts as important occurrences in and of themselves, and as embedded within a larger sequence of actions related to the local implementation of AB 109. This leads to my understanding of local practitioners' production of plans as an initial compliance effort in and of itself. Through mandating a specific process-but not specific content-for the development of these plans, AB 109 impelled a unique moment of "legal translation" (White 1990) among the practitioners appointed to write them. In this case, the Realignment plans serve as a canvas for local actors to construct the meaning of law for the purposes of implementation, or what Schoenfeld (2010: 734) describes as "back end" legal translation. Besides each county's Board of Supervisors, audiences for the plans included the state agencies responsible for allocating Realignment funds, the press and, depending on local conditions, special interest, and social movement advocacy groups (e.g., ACLU 2012). The plans are, therefore, expressive and symbolic as well as instrumental documents and should be understood as strategic vehicles that are products of particular local dynamics at one early stage in the implementation process rather than as strictly accurate records of action or intentions to take particular actions. Therefore, I analyze them as memorializing a potentially revealing point of translation between state and local articulations of the law and for their role in organizing-not actualizing-future paths to action.
I first gathered the complete universe of plans for the 2011-2012 fiscal year (n 5 57) 2 from several Web sites: those of county governments, the Chief Probation Officers of California, and the non-profit groups, The Rosenberg Foundation and California Forward. The ACLU of California provided the plans not readily available online. The plans range from 3 to 120 pages, with an average of 34 pages.
I then developed two comparison groups by identifying the counties that fell into the upper-and lower-quartiles of state prison admissions for each of the years from 2000 to 2009. I define state prison admissions as the rate of each county's new felony admissions to California state prisons. I classified counties based on the new felony admissions rate as calculated in Ball's (2012) longitudinal data compilation, which merges annual California Department of Corrections and Rehabilitation (CDCR) and California Department of Finance population data. State prison admissions is standardized into an annual rate per 100,000 of each county's population. 3 Following previous research, I measure countyoriginated prison admissions rather than populations. Prison admissions are "flow" rates of how many county residents enter prisons rather than "stock" rates of how many county residents are incarcerated in prisons at a given time. Flow rates capture the organizational practices and decision making of county practitioners without the confounds that arise in stock measures of prison populations, which are shaped by state-level prison administrators' policies and practices (see McCarthy 1990: 330;Weidner and Frase 2003: 393). For the same reason, I did not include new parole violations leading to a new prison term in my measure, as parole determinations prior to AB 109 were made by state officials with minimal input from county officials.
Those counties with state prison admission rates in the lowerquartile (83 people or less per 100,000) for nine or more of the 10 years leading up to AB 109's enactment-what I call the Low Imprisonment Legacy group-represent an exemplary group of consistently low state prison using counties. Those counties with prison admission rates in the upper-quartile (172 people or more per 100,000) for nine of the 10 years during this period-the High Imprisonment Legacy group-by contrast, represent an exemplary group of consistently high state prison using counties (Table 1). Other measures of variation in relevant policy legacies at the county level might include prosecution rates for secondand third-strike offenses under California's "Three Strikes Law and You're Out Law," the prevalence of capital charges in deatheligible cases (e.g., Ganshcow 2008) and the proportion of drug arrests and racial disparities in arrests, prosecutions and prison admissions (e.g., King 2008). Here, however, I isolated rates of state prison use as the most directly relevant measure of the consonance or dissonance between previous county practice and the specific reforms under Realignment.
Each group exhibits distinct political, sociodemographic, jail capacity, and crime characteristics. As shown in Table 2, high imprisonment counties demonstrate robust Republican party affiliation, low voter support for progressive criminal justice reform ballot measures and low per capita income levels throughout the 2000-2010 decade compared to low imprisonment counties. The groups converge more closely, however, in urbanization, racial/ethnic heterogeneity, income inequality, crime rates and pre-AB 109 jail capacity constraints.
I did not include these measures in developing the comparison groups. This was an intentional choice. Prior studies suggest that a number of these underlying characteristics explain the variation in state prison use (e.g., Arvanites and Asher 1998). While an important research goal, the aim of this study, however, is not to explain observed variation among county imprisonment. Rather, I take this variation at face value and explore whether it is associated with different interpretations of AB 109. Findings of this association may very well be explained by the same characteristics and processes that, for example, led voters in low imprisonment counties to overwhelmingly support ballot measures to reduce penalties under the state's notoriously punitive Three Strikes law in 2004 and 2012 but led to weaker support among high imprisonment county voters in both elections. Variation could also be influenced by a number of "production process" factors (Krippendorff 2012), including but not limited to the idiosyncrasies of the local drafting procedures leading to final plans, which, in turn, may or may not be related to differences in past reliance on state prison use. Because this study is neither aimed nor designed to rule out such competing explanations, variations found within AB 109 plans cannot be exclusively or causally attributed to variations in historical state prison use.
Instead, my analysis pivots attention from the discrete characteristics of the jurisdictions in which plans were authored to the characteristics of the interpretive processes used in the plans to make sense of the law. The plans offer a rich data source for documenting the possible divergence of legal interpretations, and rather than explaining why such divergences exist, this study is designed to make mechanism-based inferences about how these interpretations are fashioned (e.g., Small 2013). The ability to engage in a deeper  (2012), all rates are calculated per 100,000 of each jurisdiction's annual population. "Statewide" measures are an average of all county measures, except for the proposition voting percentages. "Metropolitan/nonmetropolitan" is assigned to counties by the U.S. Office of Management and Budget based on decennial census data. The "racial/ethnic heterogeneity index" is calculated based on five groupings (white, African American, Latino, Asian, and other races); the index scale is 0-100, where "0" represents no heterogeneity and "100" indicates maximum heterogeneity (all groups represent equal proportions of the population). The "income inequality index" is based on the Gini coefficient; the index scale is 0-100, where "0" represents complete equality (every individual earns the same income) and "100" indicates maximum inequality (one person earns all the income while everyone else earns none).
Sources reading of plans within a distinct group of counties based on empirical indicators of divergent imprisonment legacies better served this research objective than a quantitative content analysis of the complete universe of plans (see Abarbanel et al. 2013;Bird and Grattet 2014 for such analyses). Because my aim was to compare and contrast both the content and the processes leading to local interpretations of AB 109, I used a combination of Qualitative Content Analysis (Schrier 2012) and Critical Discourse Analysis (Van Dijk 2008), which blends data-driven coding as a largely inductive method with the underlying constructivist assumption that language does not merely represent reality but also functions to help create it (see also Neunendorf 2001;Smith 2006).
I developed my coding frame through closed and open coding (Cope 2003;Schrier 2012). The closed stage began with conceptually identifying the following dimensions of how plans characterized AB 109 as a law: (1) terminology used to describe AB 109, (2) descriptions of statutory components, (3) references to legislative findings, (4) references to purposes and intents, (5) references to the magnitude of impact, (6) verbiage signifying attitudes toward AB 109, and (7) inclusion of direct quotations from the statute. I also listed each substantive statutory component and created codes to measure the extent to which the plans reference them in summaries and explanations of the law. Because some AB 109 components are required, while others are discretionary, creating particularly ripe space for multiple interpretations, I categorized coding of each component separately by whether the plans reference it as required or discretionary. I also developed subcodes to capture verbiage indicating who or which parties were responsible for AB 109 and its implementation, as well as subcodes about the people sentenced and punished under the law (the "realigned" offenders). During the open coding stage, I piloted my initial scheme on a subset of plans not included within the comparison groups and added additional subcodes that emerged. The final coding scheme was, thus, a product of both deductive and inductive processes.

Findings
Counties within the High and Low Imprisonment Legacy groups arrived at unique interpretations: low imprisonment counties interpret AB 109 as mandating overall decarceration in state prisons and county jails, while high imprisonment counties interpret AB 109 as mandating a relocation of the incarceration of realigned offenders from state prisons to county jails. These findings at an early stage in the implementation process support previous findings that organizational actors translate the meaning of legal interventions in ways that facilitate the historical continuity of penal practice (e.g., Schoenfeld 2010). More revealing, however, are the distinct processes by which these interpretations are accomplished-overwriting or underwriting law, selective magnification and selective siting-and the underlying logics and struggles for power they expose among county-level actors in the penal field (Table 3).

Overwriting or Underwriting Law
Local practitioner groups in high imprisonment counties overwrite law when they remove and replace explicit AB 109 references and quotations in their plans, citing instead local actors as the originators and authors of policy change. These plans largely render law's authorship invisible and obfuscate the legally mandated nature of changes to local policy and practice, thus, creating the appearance that local actors are the architects rather than the implementers of major organizational change. Local actors overwrite law when they redact the legal origin of policy choices even as they articulate the official substance of legal directives-in other words, removing the legal file path of regulation. I found no low imprisonment plans to overwrite law. Rather, they underwrite law by extensively quoting AB 109 and prominently citing the statute as mandating policy change and specific local implementation steps. While the overwriting process appears to insulate local authority and prerogative from being perceived as subject to higher levels of governmental regulation, the underwriting process relies on higher order law as a source of local authority. Rather than using the law to deflect local responsibility (what might be thought of as "scapegoating" the law), local actors who underwrite law in these plans appear to "co-sign" onto AB 109 as a way of enhancing legitimacy.
As illustrated in the examples below, local actors use these processes to defend and enhance what Page (2013: 159) describes as "penal capital-the legitimate authority to determine penal policies and priorities," as well as to elevate the "penal expertise" (161) of certain actors over others. The competing logics underlying the overwriting versus underwriting processes reveal that high imprisonment counties react to higher order law as a threat to local legitimacy, whereas low imprisonment counties draw on law as resource for accumulating local penal capital. This also reflects Gray and Silbey's (2014) finding in the organizational compliance literature that actors develop distinct orientations to the regulator as a "threat," "ally," or "obstacle" according to different levels of autonomy, technical expertise and proximity vis-avis regulatory entities.
The overwriting and underwriting processes are observed in the relative presence or absence of legal references in the plans. State and federal laws play a starring role in low imprisonment plans but a minimal role in high imprisonment plans. All low imprisonment plans contain conspicuous references to institutional reform litigation and/or AB 109's specific statutory language. All of these plans also specifically reference the statutory requirement that standard local practitioner groups develop written implementation plans, and a majority directly and identically quote Penal Code §1230.1 on this point (Contra Costa, p. 10; San Francisco, p. 1; Santa Cruz, p. 3; Sonoma, p. 11; El Dorado, p. 4). In contrast, only two high imprisonment plans (Sutter, p. 5; Yuba, p. 3) directly quote Penal Code §1230.1. The high imprisonment county of Shasta (p. 8) includes this language but does so without the use of quotation marks or any statutory references so as to depict a process for organizing a course of action that is, in fact, required by law, as if it emanates from local prerogatives. Similarly, all high imprisonment plans reference-often using terminology identical to that of the statute itself-specific elements contained in AB 109's Legislative Findings (Pen. Code §17.5), yet only two (Kern and Madera) cite or directly quote Penal Code §17.5.
This does not mean, however, that the substance of AB 109 is less present in high imprisonment plans. Like the Low Imprisonment Legacy group, all High Imprisonment Legacy plans clearly reference the law's major components. The difference is that they overwrite law in the process. A key accomplishment of overwriting is to elevate the authority and penal expertise of local Sheriffs by minimizing the legitimacy of state bureaucrats and federal judges as "counterfeit experts" (Page 2013: 161). A prime example of how the overwriting of law functions to shore up Sheriffs' local authority in high imprisonment counties can be observed in Kern County's plan (p. 12), which, after explaining that "the existing capacity to manage the seriously mentally ill (in a custody setting) is limited," goes on to declare that "[t]he Sheriff has consequently dedicated a portion of realignment funding to contract with the California Department of Corrections (CDCR) in anticipating this challenge." The plan obfuscates the fact that one of AB 109's major statutory components provides local correctional administrators new authority to contract back with CDCR to house inmates if local jail capacity is lacking (Pen. Code §2057); only in the final pages does the plan (p. 21) reference the law's creation of this option for counties. Similarly in this plan, even though AB 109's discretionary components are referenced by statute ("Penal Code 1203.018 allows the Sheriff to release prisoners being held in lieu of bail to an electronic monitoring program. . ." [p. 11]), the reference is immediately overwritten by the Sheriff 's authority to arbitrate these alternatives to incarceration: "The Sheriff will prescribe reasonable rules and regulations to provide a functional platform for the management of the program. . ." (p. 12). Even while using language nearly identical to AB 109's statutory language, the plan omits citation or quotes of the statute, which, in fact, authorizes the Board of Supervisors, not the Sheriff, to make such decisions (Pen. Code §1203.018 2(d)).
A majority of the low imprisonment plans includes explicit references to the role of Brown v. Plata and/or other institutional reform litigation as catalysts for the enactment of Realignment and positions such litigation as a reflection of the state's problematic overreliance on incarceration. For example, in a section introduced by the statement, "Three primary factors have driven passage of this legislation:" and the subheading "Judicial," Marin County's plan (p. 1) explains: The Coleman Plata lawsuit, filed in 2001, alleged significant deficiencies in the State's ability to provide adequate medical care to prison inmates. . .AB 109, or Public Safety Realignment is, in part, a response to these federal court orders. Nevada County's plan (p. 4) also describes institutional reform litigation in a section entitled, "California's Contribution to the Crisis" as follows: . . .the state faced a series of class action lawsuits that were initiated in 1990 and 2001 by seriously mentally ill prisoners and prisoners with serious medical conditions. Finally, in 2009, a panel of three federal judges ordered California to reduce its prison population to 110,000 from 156,000 (the official state prison capacity is 80,000) (Liptak, 2011). In May, 2011, the federal ruling was upheld by the Supreme Court decision in Brown v. Plata No. 09-1233 where the Court noted that overcrowding is the "primary cause" of "severe and unlawful mistreatment of prisoners through grossly inadequate provision of medical and mental health care. . .leading to needless suffering and death (Liptak, 2011)." AB 109 represents the state's attempt to meet the mandated population reduction through increased local control supported by flexibility and fiscal appropriations. . . By contrast, only two high imprisonment plans reference the Plata order by name (Kings and Madera), and two of the plans generically reference conditions litigation (San Bernardino and Sutter). While Low Imprisonment Legacy plans tended to position institutional reform litigation as a corrective, Kings County's plan (p. 4), for instance, references the litigation as a potential threat to be avoided: We will make sure we are compliant with Title 15 and Title 24 so we avoid what CDCR has gone through over the last 20 years that resulted in the Coleman-Plata ruling by the US Supreme Court.
These characterizations echo Page's (2011) description of how federal court intervention figured into the struggle for penal capital by California correctional officers, who gained ascendancy in the field, in part, by depicting federal judges as caring more about prisoners' rights than about the practical exigencies and safety needs of officers who daily walk "the toughest beat." Low imprisonment counties' depiction of federal courts as sources of legitimate expertise and authority on California prison conditions as contrasted to high imprisonment counties' depiction of courts as threats or burdens demonstrates the persistence and depth of this struggle, and that it diffuses beyond state-level interest groups to county-level practitioners.

Selective Magnification
Selective magnification highlights certain statutory components to the exclusion of others. Both groups utilize selective magnification to arrive at their interpretations of AB 109. This process illustrates how the legal ambiguity of AB 109 created a window of opportunity for local actors to garner legitimacy in distinct ways. High imprisonment counties responded to this window by consolidating local power and autonomy from what they characterize as the threat of state intervention; they do so by selectively magnifying AB 109's realignment of especially dangerous offenders to local custody, which portrays state bureaucracy as incompetent and lacking credibility. Low imprisonment counties, conversely, seized on AB 109's ambiguity to bolster the legitimacy of decarceration-oriented reform policies by depicting them as stringent legal requirements.
The plans' depictions of which classes of offenders will be realigned to local supervision show how the selective magnification process accomplishes divergent meanings. Among all of the plans in both groups, only two (Lake and San Bernardino, both high imprisonment counties) used terminology other than "Public Safety Realignment," "AB 109" or simply "Realignment" to refer to the law. Both counties additionally refer to the law as "parole realignment." By (re)naming the law parole Realignment, they draw attention to and emphasize the post-release, or paroled, offender. While AB 109 specifies that one class of realigned offenders must be considered nonviolent, nonserious, non-sex offender registerable and have no serious, violent or sex-registerable prior offenses (Pen. Code §1170(h)), the post-release class realigned to county supervision upon release from serving a term in state prison for a "non-non-non" offense may have committed prior offenses deemed serious, violent and/or sex-registerable (Pen. Code §3450 Tit. 2.05 of Pt. 3). By drawing attention specifically to the post-release class, San Bernardino's plan (p. 19) orchestrates a discussion of the actual dangerousness of locally realigned offenders while minimizing references to the other class: The Community Corrections Partnership is cautious about speculating the outcome of the parole realignment due to significant concerns on the types of offenders, the number of offenders, budgetary issues affecting county departments, and the potential for an increased crime rate.
Even in the plan's exceptional reference to the 1170(h) provision (p. 18), the verbiage constructs an image of the locally realigned offender as a truly dangerous criminal who has been misclassified by the state as "nonviolent": There is some solace in the concept that the offenders being directed to our local jurisdictions are "nons" -non-violent, non-serious, and non-sex offenders. However, as this plan has pointed out, CDCR classification of these offenders is based solely upon current convictions and offenses. . .
Note that the plan appears to mischaracterize the law's designation of the locally realigned offender as "based solely upon current convictions and offenses;" in fact, offenders who commit a new felony under Penal Code §1170(h) will not be realigned to local supervision if they have serious, violent, or sex-registerable prior offenses (see California Department of Corrections and Rehabilitation [CDCR] 19 Dec. 2013: 3). Nevertheless, the plan (p. 18) goes on to fashion all realigned offenders in the image of the post-release class, which it decries as having disturbing criminal histories heretofore not contemplated by the law: It is common for persons committed to state prison for a less serious offense to have significant, lengthy criminal histories that may encompass more serious or violent crimes; and to have a history of habitual non-compliant conduct and be resistive to community corrections interventions. The San Bernardino County criminal justice system should remain vigilant to potential increases in crime rates or incidents of criminal conduct that are the corollary of the re-introduction of these offenders into our communities. This process of subjecting the more problematic statutory component to the magnifying glass while omitting discussion of the less controversial component is further accomplished through subtle references that remind the reader of the state's "misclassification" of offenders as nonviolent ("The focus of AB 109 is on the California Department of Corrections (CDCR) parolees, who have been classified as 'low-level' offenders. . ." [note the use of quotation marks]; "This is accomplished by the release of those deemed to be low risk offenders by CDCR. Parolees categorized as low risk. . .after their current offense is determined to be non-serious, non-violent, and non-sex related" [emphasis added, p. 4]). This demonstrates how the construction of the high versus low risk offender is premised on a lack of expertise among state-level bureaucrats.
Selective magnification is also used to arrive at shared legal interpretations among the groups. Despite its complexity, AB 109's statutory language makes clear which components are required as opposed to merely authorized (see Byers 3 December 2011). The most significant example is that the legislation authorizes and suggests-but does not require or attach funding conditions tothe use of alternatives to incarceration, a "justice reinvestment" approach and the use of "evidence-based practices" (Pen. Code §17.5; §1203.016; §1203.018). Both low and high imprisonment county plans, however, blur this line. Among the Low Imprisonment Legacy group, in line with the tendency to underwrite law as the means for implementing favored policy changes, plans depict the use of alternatives to incarceration in local jails as a necessary requirement of the law. The High Imprisonment Legacy group, in line with the tendency to overwrite law in portraying the policy changes introduced by Realignment as emanating from local prerogative, depicts the use of alternatives to incarceration as locally derived practical necessities rather than legal mandates.
Marin County's plan illustrates how low imprisonment plans position these discretionary aspects as legal requirements by overstating AB 109's suggestion that counties implement alternatives to incarceration: By fundamentally altering sentencing laws, expanding local responsibility for custody, and requiring the use of evidence-based correctional practices the 2011 Realignment reverses more than 30 years of increasing reliance in [sic] state prison (p. 1, emphasis added). . .The legislation does not intend for prison sentences to be simply replaced by jail sentences. Rather, it requires the use of evidence-based correctional sanctions and interventions to reduce the high rate of incarceration in California. It thereby directs a significant swing from emphasis on institutional corrections towards local, communitybased strategies and interventions (p. 2, emphasis added).
Similarly, Santa Cruz County's plan (p. 7) states that "[t]he enabling legislation for realignment specifies the use of Evidence-Verma Based Practice (EBP) as a requirement for activities and services funded through AB 109" (emphasis added), and Imperial County's plan (p. 3) describes, "Key Features of AB 109" with the bullet point: "Requires Evidence-Based Practices: AB 109 requires the adoption of evidence-based practices as a condition to receiving state [funding] for realignment. . ." (emphasis added). Sonoma County's plan (p. 10) goes a step further by depicting alternatives to incarceration as central to the meaning of Realignment ("The implementation of EBP [evidence-based practices]. . . is a cornerstone of Realignment legislation. . . [emphasis added]) and characterizes the requirements of the law as directly confronting and fundamentally changing the state's historical use of incarceration: Realignment legislation anticipates that local governments will handle their new offender population in a manner different than CDCR. . .it is clear that for any County to succeed with Realignment, it must be approached in the manner the legislation envisions -by using resources wisely, basing decisions on risk, and using evidence-based practices as much as possible. For, if a County treats offenders in the same manner as the State, i.e. incarcerate for significant periods, leave criminogenic risks and needs unaddressed, and simply release, the added resources will certainly not be adequate (p. 13, emphasis added).
High and low imprisonment county plans converge in their discussion of alternatives to incarceration as necessary. They diverge, however, in the stated logic behind this necessity. Whereas low imprisonment plans utilize the legal mandate as the organizing logic and, as reflected by the underwriting process, to bolster local authority and legitimacy, high imprisonment plans rationalize reducing reliance on incarceration through the lens of the practical local necessity of managing fiscal strain and public safety risk despite the legal mandate. San Bernardino County's plan demonstrates how this is done through a section dedicated to the Sheriff 's "Issue Statement" (pp. 15-16): The realignment of state prisoners and the shifting of parole violator housing to the county jails will logically increase San Bernardino County Sheriff 's Department (SBSD) costs associated with housing, processing, feeding, and out-of-custody supervision. . .The retention of approximately eight thousand three hundred (8300) additional inmates per year within the jurisdiction of the Bureau of Detention and Corrections by virtue of AB 109 creates an enhanced need for alternative custody programs. . .The administration of these programs is vital for both inmate population management and the reduction of recidivism rates within the county.
Here, the plan specifies a large number of new inmates that will come under local supervision "by virtue of AB 109," which will "logically increase" the Sheriff 's need to utilize alternative mechanisms. The purpose behind utilizing alternatives to incarceration is not in order to comply with the law's mandates, or even its strong suggestion, but rather to address the "costs associated with housing, processing, feeding. . ." inmates and to manage risk through alternative custody programs. Kings County's plan (p. ii) similarly quantifies the practicalities at hand: As a result of the State budget, AB 109 and AB 117 were passed in which counties assumed new corrections responsibilities for people convicted of certain non-serious, non-violent felonies and new community supervision and reentry assistance for people after they are released from prison and jail. It is projected that the Kings County Jail would receive approximately 321 of both male and female inmates within four years with an average of 3 parole violators per month and 21 newly sentenced inmates per month from October 2011 through June 2012. The plan also calls for rehabilitation and diversion giving the opportunity for these individuals to become law abiding citizens.
Note also the distinction between what the law motivates ("new corrections responsibilities") and what the plan motivates ("rehabilitation and diversion") in this verbiage. Similar to another high imprisonment plan's discussion (Sutter County, p. 17), King County attributes its "new corrections responsibilities" to the state's law ("AB 109 and AB 117") and "State budget," numerically detailing the practical burden placed on local administrators as a result, while attributing the use of alternatives to incarceration to "The plan" as authored by local authorities who must manage this burden. Together, these plans illustrate how the languages of risk and cost combine to frame the use of nonincarcerative alternatives in high imprisonment counties as a matter of local-not legalnecessity, and to direct blame at higher governmental levels for burdening local jurisdictions with this budget dilemma.

Selective Siting
The final process, selective siting, refers to where plans locate the site of the problem the law is meant to solve. Here, both groups exhibit the same process in arriving at divergent interpretations. Low imprisonment plans provide robust and multidimensional discussions of the broader policy environment and legal landscape of Realignment, while high imprisonment plans tend to limit context-setting to the state's fiscal and overcrowding problems.
Some low imprisonment plans situate California's correctional system within what is depicted as the deeply problematic American criminal justice system writ large. For example, Nevada County's plan (p. 3) opens with the heading "California's Correctional Context," explaining that "the growth of U.S. prison populations and the related costs associated is well-documented. Over the past decade criminologists and legal scholars alike have repeatedly characterized the growth in prison population as 'unprecedented,' creating a dangerously overcrowded system. . ." and that "California, one of the largest correctional systems, contributes greatly to the correctional crisis facing the U.S. . ." (p. 4). Sonoma County's plan (p. 10) goes even further in contextualizing the enactment of Realignment: In the late 1970s, research indicating that 'nothing works' with offenders presented the criminal justice field with a serious challenge. This led to a period focused on increased sanctions for criminal offenders, leading to prison overcrowding, a problem targeted by Realignment.
Madera County's (p. 3) was the only high imprisonment plan to include this kind of broad contextualization. This suggests that high imprisonment counties may also, at times, locate the law's problem site in ecological, rather than local, terms.
A majority of low imprisonment plans explicitly frame the law's enactment as a direct response to the U.S. Supreme Court's ruling in Brown v. Plata, and virtually all attribute one of Realignment's main purposes to solving the broadly constructed problem of state prison overcrowding. As demonstrated above, some plans go even further in constructing the prison overcrowding problem as systemic in nature-transcending levels of government. In this way, the Low Imprisonment Legacy group appears to frame Realignment as emerging from the multi-decade-long, multi-sited problem of over-incarceration, in addition to positioning the law change as a response to a federal court order. The construction of the problem as over-incarceration appears to lay the groundwork for low imprisonment plans' construction of the meaning of the law itself: decarceration.
By contrast, few high imprisonment plans reference the role of institutional reform litigation in Realignment's enactment, while the majority attribute the law's purpose to the state-located problems of fiscal crisis and prison overcrowding. Sutter County's plan is instructive on this point in that it describes the county jail terms to be served by realigned offenders as "serving a state prison sentence in county jail" (p. 6) and points out that " [t]here are difficult challenges ahead in implementing widespread systemic change in order to avoid simply transferring the prison overcrowding problem to the local jail" (p. 11). This language sites the problem at the state level and positions the county as the recipient of a transferred state problem. Additionally, the language of "state prison sentence in county jail" explicitly resists the law's redefinition of where sentences for specified felony sentences will be served-in local jails rather than state prisons. If, as this plan's language suggests, these sentences are actually "state prison sentences," then the law's requirement that they be served in county jails appears illogical and/or as a displacement of responsibility. The use of this language throughout the plan provides subtle but repetitive prompts for the reader to spatially organize the problem and the law's solution as top-down.
While the previous processes manifest as strategies that local actors use to garner legitimacy in the penal field by consolidating penal capital and elevating certain kinds of penal expertise, selective siting points to a related but distinct phenomenon. Local actors' different characterizations of the origin of the prison overcrowding problem as residing externally, at the state level, or as a problem shared across governmental levels, suggests another important "source of gravity" (Page 2013: 153-4) in the penal field: "the taken-for-granted assumptions and categories that determine what is thinkable and unthinkable (or orthodox or heterodox) in a given field," or Bourdieu's (1977) doxa. While the relational struggle among actors for penal capital and legitimate penal expertise is more or less self-consciously and strategically waged, doxa is the very water in which actors swim; thus, they may not recognize the constraints its largely invisible logic imposes on their possible actions. In this case, the siting of the prison overcrowding problem selectively renders coherence to particular interpretations of AB 109 as, alternatively, a maneuver to displace responsibility for the state's prison overcrowding debacle onto counties, or as a promising solution to a crisis shared among governmental levels.

Conclusion
Nearly five years since the Plata order and AB 109's enactment, reformers' hopes appear largely-but not entirely-"hollow" (Rosenberg 1991). While it remains premature to definitively assess Realignment's "great experiment" on the reform goal of decarceration, there is reason to believe that the distinct legal interpretations found in this study at the initial planning phase have had practical implications. At present, the state has reduced its prison population to the court-mandated limit. Even so, California prisons remain among the nation's most dangerous and overcrowded. Overcrowding at the local level has exploded. The number of county jail systems operating above rated capacity almost doubled, from 11 in 2010 to 21 by 2014; thus, AB 109 appears to have merely relocated the sites of incarceration from state prisons to local jails in a sizeable number of counties, displacing the overcrowding problem downward (Lofstrom and Raphael 2013). This "dispersal" (Cohen 1979) may come as little surprise to social control scholars, but the puzzle of local variation remains: practitioners in some counties appear to have used their newfound discretion under AB 109 to reduce overall incarceration levels, rendering reform apparently successful in some places but not others.
I have analyzed the law-before AB 109 as one key to solving this puzzle. Because Realignment is premised on limiting the county "free lunch" (Zimring and Hawkins 1991), in this study I focused specifically on past imprisonment practices as measured by counties' state prison use rates. The local variation in this measure over time allows for a comparison of how different "doses" of reform are swallowed in places with relatively consonant or dissonant legacies of practice. In this case, implementation plans from counties with historically high or low imprisonment patterns demonstrate local practitioners' divergent orientations to the legitimacy of law itself, the opportunities created by legal ambiguity and the governmental source of the prison overcrowding problem. These competing logics underlie the interpretive processes observed-overwriting or underwriting law, selective magnification, and selective siting-which ultimately facilitate divergent interpretations of AB 109 as mandating either system-wide decarceration or the relocation of incarceration from state prisons to county jails. While this study demonstrates the salience of the law-before only on paper, the divergent interpretations contained in official planning documents rationalize-even if they do not determine-the allocation of fiscal and human resources either to build community-based capacities for alternatives to incarceration or to expand local jail capacity.
My focus on these processes contributes to the valuable research goal of explaining what drives local imprisonment variation in the first place by uncovering potential mechanisms that link predictive measures with key outcomes of interest. This may refine interpretations of results found in the variablecentered literature on inter-jurisdictional penal variation and extend the theories that underlie causal models, potentially revealing additional, alternative or more parsimonious explanations (e.g., Tavory and Timmermans 2013). By scoping my comparison to counties' state prison use rates based on AB 109's particular regulatory premise, I do not mean to imply that other local measures previously identified in this literature are irrelevant. Indeed, imprisonment is but one salient metric of the multiple practices and power arrangements that mutually constitute my conception of the law-before. The state prison use measure was pivotal in this study because it captured variation in a relational construct central to the reform in question: the relationship between county and state governments. My findings provoke future inquiry into the interaction of variation in this relational construct with the spatial sociodemographic characteristics found to explain penal reform. The mechanism-based findings of studies such as this one that examine penal practice longitudinally can also help develop existing theory by illuminating the "historical transmission" (Petersen and Ward 2015) of local variation in both carceral outcomes and the variables hypothesized to explain them.
The central thesis of Simon's (2007) Governing Through Crime is that crime control policy should be understood as a strategic vehicle for rearranging governmental power relations. If so, then attempted legal reform must be analyzed beyond its specific policy context, and in light of its broader implications for governmental arrangements, including implementers' perceptions of the legitimacy of law to realign them at all. Brown v. Plata and AB 109 were "critical events" in California's massive criminal justice system; in the institutional crisis triggered by such events, windows of opportunity emerge for actors to build legitimacy and support for reforms previously dismissed as radical or impossible (Page 2013: 163;Tonry 2004).
The notion of "legitimacy" within organizational fields is classically conceptualized as insulating organizations from the external pressures of critical events (e.g., DiMaggio and Powell 1983). However, under California's Realignment, "governing through crime" takes the shape of governing through the local, which may be breaking down the distinction between internal and external threats. While the Plata order can be classically understood as an "exogenous shock" (Edelman, Leachman, and McAdam 2010: 655) from the federal courts, AB 109 can be analyzed both as exogenous and as a shock from within. The dilemma of American federalism is most often conceptualized as a binary between states' rights and the federal power to intervene (judicially, congressionally or through administrative regulation); in this account local government is "nested" within states (Feeley and Rubin 2008). However, in identifying the "perils" of federalism for poor and minority group representation, Miller (2008) reminds us that, even if the politics of crime are most consequentially forged at the federal level, it is at the county and municipal levels that the governance of crime is often most consequential in people's daily lives. The binary account of federalism as a statefederal balancing act has therefore obscured the reality that many local-state, as well as local-federal, balancing acts simultaneously take place on criminal justice issues. In this sense, federalism may be better understood as "fractal." At a reform moment animated by governing through the local, the fractal nature of federalism is relevant for understanding the relational dynamics of penal change across governmental levels. This study has illuminated county-level actors as key players in the penal field (Page 2011(Page , 2013 and the "agonistic" (Goodman, Page, and Phelps 2015) struggle that takes place within these fields. My findings about how the law-before shapes local variation in the taken-for-granted assumptions, or "rules," of the penal field (doxa) reflect part of the structuration or institutionalization of power arrangements in the field. A proposition for future inquiry is that the law-before also functions at the individual level by shaping actors' intuitive understanding, or "feel," for how to be effective in the penal field (habitus); this would help answer questions raised but not adequately answered in this study about the strategic rationales behind different interpretative processes. Jenness and Grattet (2005: 339) define the law-in-between as "organizational structures and policies that provide the intermediary linkage between state statutes and officer discretion." This study's findings show that the law-in-between is itself variegated, and that distinct logics about the fundamental legitimacy of law to regulate policy and practice manifest in this intermediary stage, with practical implications for law-in-action. These variations can, in turn, be traced back to variations in the law-before, which leads to an understanding of the law-inbetween as partially constituted by the legacies of past practice and power arrangements among organizational actors. This insight contributes to legal endogeneity studies by showing how underlying patterns of variation within organizational fields, as well as variations in the legacies of these fields, shape responses to legal regulation and definitions of compliance. In examining the penal field, which differs from corporate fields, I discovered distinct interpretive processes and competing field-level logics that may be applicable to understanding the endogeneity of law in other policy domains, especially those where regulatory dilemmas manifest predominantly at the county or municipal level.
Contemporary sociolegal scholarship recognizes the "gap" as conceptually problematic and has moved beyond the "theatrical" (Gould and Barclay 2012: 331) revelation that a gap exists, training analysis instead on the conditions under which law is implemented and the processes and mechanisms that shape implementation gaps. Gap studies have, thus, decentered law in favor of examining the institutional factors that shape the cultural production of law. Zimring (2014: 739-41) typologizes Realignment as a criminal sentencing reform mainly of "procedure" rather than "substance" in both its means and ends. In the context of a law where the "gap" functions as the foundation rather than the unintended consequence of its statutory framework, my findings provoke law and society scholars to consider whether contemporary moves to decenter law have led gap studies to overlook the salient features of how a law is written and to consider that its statutory architecture provides important resources and constraints for actors who shape the meaning of compliance in the law-in-between. Rather than analyzing multiple interpretations of an ambiguous law as perversions, in some cases, the ideal of uniformity may itself hamper reform and complicate compliance. In this respect, California's Realignment may signal an evolving reform species that takes account of the local-state power struggles inherent in the U.S. governance of crime and punishment, implying the presence of not one, but multiple, types of gaps between law-on-the-books and law-in-action.