SCHELLING'S RULE: HOW TO HAVE SAFER
PRISONS AND LESS RECIDIVISM

"By that hidden way
My guide and I did enter, to return
To the fair world; and heedless of repose
We climb'd, he first, I following his steps,
Till on our view the beautiful lights of Heaven
Dawn'd through a circular opening in the cave
Thence issuing we again beheld the stars."


-Dante, Inferno, Canto 34

The day David was murdered just outside my cell door, I vowed to think most carefully about how to end prison violence. David seemed worried the day of his death, his eyes distant upon some troubling horizon, but concerned reflection is not unusual in maximum-security institutions. A strapping six-foot-four, thirty-five year old concluding a ten-year sentence, he soon was to return to his supportive family in Walnut Creek, California. While disputing another inmate's debt for a paltry sum of contraband tobacco, he suddenly was stabbed repeatedly by three Hispanic gang members in a drunken, malevolent, tattooed frenzy.

Only months later I saw "Fish," a young black man who lived on the tier above me, assaulted in the yard by a rival gang. He slowly crumpled down against a chain link fence beneath the razor wire as he was beaten and kicked to death, finally moving no more. His attackers fled into the staring crowds while klaxons resounded and hundreds were forced to the ground as flash-bang grenades detonated overhead.

Yet David and Fish did not have to die, if new protocols were in effect that are known to reduce community violence, gang activity, and substance abuse. Rather than providing a litany of prison horrors, it is proposed here that such needless deaths can be avoided by adopting in prisons several promising new policies in community policing and experimental drug courts.

The Problems in Prisons: An Overview

In high-security federal and state prisons, violence is the norm. It is correlated most strongly with alcohol and drug use, and often instigated by gangs. Over my past decade in penitentiaries, it has not been uncommon to encounter stabbing victims sitting in doorways, hiding in bloody urinals, running frantically across yards as attackers encircled them, or lying stunned on floors as their wounds bled. Within such brutish environments, both staff and inmates are exposed to severe and endless hazards, while institutional responses to violent events remain only partially adequate.

Triaging the Knife and Fist: "Nothing Works"

These problems are endemic to all prison settings, and they are formidable. Authorities' attempts to control violence and substance abuse are highly variable and subject to failure; in the absence of rigorous statistical research on outcomes there is little good data for penologists to consider.

Although gang memberships are scrutinized by special investigative units, responses of institutional staff tend to be only reactive. By contrast, inmates develop an intuitive sense of impending assaults, so that before an outbreak of violence one sees "grouping": men lingering and racially segregating, huddling in twos or threes and speaking little or not at all. Prisoners reflexively herald the attack like herds of gazelles alerting, sensing collectively the presence of predators. As the knifing commences, someone may announce, "There it is!" The inmate population freezes, looking on somberly at yet another small matter of life and death.

Prison staff are well-trained to respond rapidly to violence. In the federal system, one routinely observes the entirety of a compound's staff - guards, senior officials, teachers, secretaries - literally running to the aid of other staff or inmates in a show of cohesive, insurmountable force. After the participants are immobilized, both assailants and victims then are confined within Special Housing Units SHUs - prisons within prisons - subjected to their enervating hardships, and isolated for weeks or months until every social and physical aspect of the assault is understood by investigative specialists.

After the federal system experienced eighteen deaths within two years, including those of several correctional officers, the Bureau of Prisons B OP introduced a remedy for the more heinous violators. Rather than the usual procedure of transferring assailants to new institutions, thus displacing the problem but not resolving it, the BOP began to aggregate gang leaderships at a dedicated prison, the Special Management Unit SMU . Inmates at SMU are locked down twenty-four hours a day until violent behavior is renounced; individuals begin a "step down" program over years, eventually returning to a normal population, or not. While the outcome for the federal system has been beneficial in that gang leaderships are dismantled, violent gang recruits continue to replace one another in a process akin to arrested street-corner crack dealers whose shoes are filled quickly by those aspiring to the lifestyle.

The federal response to prison violence, while well-designed in some respects, is not only resource intensive, but suffers from a lack of good science and policy analysis. Unaware of sound new alternatives, institutions subject entire inmate populations to the punishment of debilitating lockdowns and searches in an effort to correct the few. With only a fraction of inmates who are frequent management problems, there are more effective procedures for reducing violence and targeting the primary offenders.

"Nothing works" still remains the conventional wisdom among penologists1. Even with the advent of SMUs, prison yards continue to pose extreme threats for inmates and staff as gang leaderships and members are replaced and alcohol and drug use remain endemic. How can these problems be solved? It is time to adapt to prisons the few novel strategies that in several communities actually do work. All are applications of game theory called "dynamic concentration."

New Community Solutions Can Be Applied to Prisons: Dynamic Concentration, H.O.P.E., South Dakota's 24/7 Sobriety Program, and High Point

Since the writings of Becarria in the 18th Century, considered the foundation of criminology, it has been known that the efficacy of deterrence is based on swiftness, certainty and severity, with severity the least important factor2. The penological literature recently has benefited from the seminal analyses by criminologist Mark Kleiman of UCLA of both Becarria's conclusions and the few successful community programs, identifying the underlying mechanism shared by these workable approaches to resolving probation recidivism, flagrant drug markets, drunk driving, and - more broadly - a spectrum of errant behavior3.

Kleiman has noted that trying to control everything and everyone - the "zero tolerance" approach most often cited in drug enforcement and prison management - leads to delayed responses, system overload, and ineffective, costly programs. Observing the inefficiencies and expense of "zero tolerance" regimes, Kleiman has shown that the alternative of swiftness and certainty in place of severity, together with "dynamic concentration" of resources through direct communication of deterrent threats to offenders, is a more effective, cheaper, short- and long-term remedy.

Schelling's Rule and Dynamic Concentration

A basis for the model of dynamic concentration is the work of economist and game theorist Thomas Schelling, who concluded at RAND during Soviet detente that "effective deterrent threats are never carried out"4. The principles of dynamic concentration and game theory demonstrates that in any group where most individuals are well-behaved, enforcement can concentrate on the few miscreants, in contrast to the typical solutions applied to entire populations of offenders wherein the amount of enforcement to control bad behavior inevitably leads to swamping of enforcement and resources, and delayed, sporadic punishment5.

Put another way, among those who always get punished for any infraction - drinking, drug use, failure to appear for court hearings - violation rates go down. Among those who never get punished, violation rates go up. Dynamic concentration means that adding extra enforcement effort to a small set of high-violators - even, it is proposed here, in prisons - results in a low-violation equilibrium among larger groups while requiring less enforcement capacity. That means fewer violations and fewer sanctions in the long-term. As Kleiman has observed in assessing successful applications in drug courts, "to the extent dynamic concentration reduces offending rates while unpunished violations increase them, dynamic concentration turns out to be the deterrent version of the Miracle of the Loaves and Fishes."

Dynamic concentration - moving from a bad, high-violation equilibrium to a good, lowviolation equilibrium - also leads to a reduction in resources needed to tip the equilibrium in the favorable direction, freeing up resources to address new problems. Indeed, varieties of dynamic concentration with maximal benefits from minimal resources now are being employed at increasing numbers of drug courts, probation settings and drunk driving reduction programs throughout the United States.

In a paper in the Proceedings of the National Academy of Sciences, Kleiman and Beau Kilmer, an economist and drug researcher at RAND, have indicated that game theory and the dynamic concentration model can be applied to many categories of errant behavior6. Thus, it is proposed that a variation of dynamic concentration may also help resolve seemingly intractable prison gang violence and alcoholism, using similar methods of dynamic concentration and Schelling's Rule that have resulted in a few startlingly successful community programs: H.O.P.E., South Dakota's 24/7 Sobriety Program, and High Point.

The H.O.P.E. Program for Probationers Can Be Adopted by Prisons

Judge Alm in Hawaii was confronted by an epidemic of methamphetamine abuse, a crowded court docket, and high rates of recidivism among probationers. Typically, in all states, probationers accumulate numerous "dirties," or failed drug tests, until they are violated and receive long prison sentences. Probationers lost their jobs and families, and the prisons became filled with recidivists. Judge Alm initiated a new program called H.O.P.E. "Hawaii Opportunity Probation with Enforcement" , streamlined the drug court process, and replaced the practice of following multiple violations by long prison terms with a new approach: swift and certain sanctions - in most instances a few days in jail - for each and every failed drug test or missed appointment7. Probationers kept their jobs, supported their families, and - in an unanticipated outcome - spontaneously stopped using methamphetamine. The experiment of assuring a brief stay in jail for every failed urinalysis or absence from a hearing significantly reduced methamphetamine use, even among probationers who were repeated recidivists and chronically failed drug tests under the old practice of lengthy prison terms after multiple violations.

This unusual outcome was predicted by dynamic concentration, as probationers moved from a high-violation equilibrium to a low-violation equilibrium through swift, certain and short-term sanctions. Schelling's Rule began to apply as probationers became aware they would be jailed for every infraction: "effective deterrent threats are never carried out."

H.O.P.E. is indisputably fair. Probationers realize their incarceration depends on their own behavior rather than the caprice of a probation officer, a realization that is essential to breaking bad habits8. H.O.P.E. sites have begun to appear in several states, and are being evaluated by public policy researcher Angela Hawken. In that the dynamic concentration analysis of Kleiman and Kilmer observed it may be applied to diverse forms of behavior, how may it be employed to make prisons safer?

Applying South Dakota's 24/7 Sobriety Program to Prisons

An unusually successful program incorporating elements of dynamic concentration is South Dakota's 24/7 Sobriety Program. In South Dakota, those with drunk driving violations are required to report twice daily to a police station for breathalyzer testing, rather than facing mandatory prison terms after accumulating multiple drunk driving citations. The results, as with H.O.P.E., are encouraging9.

H.O.P.E. and South Dakota's program should be applied to prison populations. In the community, three-quarters of those addicted to substances are addicted to one intoxicant: alcohol, which also accounts for most drug-related crime. In the prison system, the link between alcoholism, disease, and violence is even stronger.

Inmates routinely manufacture and consume alcohol in the form of "hooch," crudely fermented solutions of stolen sugar, potatoes, rice, tomato paste, and even carbonated sodas degassed and evaporated in microwaves. Rudimentary stills using plastic sheets are encountered. Yet among violent inmates and prison gangs, alcohol use is closely correlated with assaults, injuries and deaths, and multiple lockdowns of institutions for protracted periods.

In many prisons, breathalyzers are administered in an informal and random manner, or through the ad hoc solicitation of "volunteers." Seized alcohol may be discarded with no sanctions until repeated infractions lead to a lengthy term in the Special Housing Unit, a process similar to probation violators in Hawaii before Judge Alm's program. Yet, a new approach being evaluated at an Arizona federal prison indicates that dynamic concentration may be employed to reduce alcohol abuse in all state and federal prisons. After experiencing repeated incidents of alcohol-related violence and lockdowns, and upon becoming aware of the value of swift, certain sanctions and non-random testing, an Arizona warden began a "yellow card" program based in part on South Dakota's 24/7 Sobriety Program and dynamic concentration10. Inmates with histories of alcohol infractions were issued "yellow cards" and required to report three times each day for breathalyzer testing, while those failing the test or not appearing were immediately sanctioned. Yellow card inmates were summoned to testing through a public address system, reinforcing inmates' perception of the certainty of sanctions. The results, as with H.O.P.E., moved the prison from frequent violence and alcohol use the high-violation equilibrium to a relatively peaceful prison compound the low-violation equilibrium . Both institutional staff and inmates anecdotally have reported that the incidence of violence declined substantively10. This application of dynamic concentration and the H.O.P.E. and South Dakota protocols is worthy of statistical study by external researchers approved through the Bureau of Prison's Office of Research and Evaluation.

The Arizona "yellow card" program may be refined further. It has been proposed that in the community a selective prohibition of drinking of those previously convicted of alcohol-related offenses would have to be enforced by regulations on sellers rather than on buyers11. Applying dynamic concentration resources to prison alcohol manufacturers more than consumers would lead to greater efficiency of the program, while minimizing institutional burdens from assigning staff to conduct frequent alcohol testing. Nevertheless, the personnel hours required to test all prior alcohol offenders - those with yellow cards - constitute only a fraction of the staff burden from institutional searches and lockdowns following episodes of alcohol-mediated violence. Continued success in the H.O.P.E., South Dakota, or Arizona prison programs requires sustained, certain sanctions, however. Otherwise, the low-violation equilibrium will revert to high violations as subjects risk being sanctioned.

The More Difficult Problem of Opiates and Psychotropics

Oxycontin and other opiates licitly prescribed for medical conditions are diverted in prisons and abused by addicts and newly initiated inmates, as are psychotropics including Effexor, Zoloft, Welbutrin and Neurotin. Jails and prisons largely have replaced psychiatric hospitals as housing for the mentally ill12, so that inmates have a high incidence of character disorders and "Axis I" mental health problems: schizophrenia, depression, bipolar disorder, and aggression.

Because some inmates are skilled at mimicking medical or psychiatric problems, or over report real maladies to secure drugs - even feigning suicide gestures - medical staff inadvertently may over prescribe medication. Diversion is difficult to monitor: even crushed tablets whose consumption is observed by staff can be mouthed and resold for injection or nasal insufflation. Diverted opiates and psychotropics are expensive for addicts, and prison debts significantly contribute to violence.

To resolve this problem, prisons could adopt two protocols: inmates with a history of abuse could be included in a weekly yellow card program requiring urinalysis, and - more effectively - problematic medications could be distributed not in pill form, but as tamperproof skin patches13. Kilmer has proposed remote electronic alcohol and other drug testing for probationers, and such advances may be particularly useful for the recalcitrant prisoner14.

Prior heroin users can be tested through a yellow card procedure as well. Typically smuggled in through visiting rooms, heroin is endemic to many facilities. In one instance, a twenty-eight year old male died from ingestion of balloons of heroin during a visit, only one week before his release date.

Disempowering Prison Gangs through the High Point Solution

Gangs are ethnocentric, with a single BOP statistical study comparing the drug and violence propensities of white supremacist gangs such as the Aryan Brotherhood, Dirty White Boys, Nazi Low Riders, Skinheads and motorcycle gangs including the Hells Angels and Mongols. Hispanic gangs including La Eme, Tex-Mex, and the Border Brothers were reviewed, as well as black gangs including Bloods, Crips and Gangster Disciples. Prison gangs are symbiotic with those in the community15.

Prison yards essentially revert to tribal societies warring for dominance and control of drug, alcohol and gambling revenues. Directed by their "shot callers," gang "enforcers" administer beatings or "disciplinaries" to their own members for failing to carry out assaults, while also extorting the unaffiliated through "taxes."

An inmate’s age, ethnicity, background or behavior do not deter the prospect of violence. I no longer count the attacks on victims, including the illiterate, elderly or those in wheelchairs or on walkers. Blood trails from victims require dedicated inmate clean-up crews who scrub sidewalks beneath stars engraved into prison walls from the sharpening of homemade knives.

The High Point Experiment

The community of High Point, North Carolina was oppressed by burgeoning street corner crack cocaine sales bazaars, increasing drug related crime and violence, and diminishing property values. The local court and probation systems were overwhelmed by processing arrests even as new dealers replaced those who were incarcerated.

After lengthy and contentious community meetings, the High Point police department, supported by community leaders and advised by David Kennedy - then a Fellow in Criminal Justice Policy and Management at Harvard's Kennedy School of Government and now a professor of John Jay College of Criminal Justice - created a social experiment whereby street dealers were identified by the usual surveillance methods and buys were made until evidence was gathered for significant prison terms. Surprisingly, no one was arrested16.

Rather, the police went to the homes of dealers and invited them to a meeting, where they were shown three-ring binders and films of their felonies sufficient for conviction and years in prison. A few of the more heinous "bad actors" were publicly dispatched to jail and prison. The others, the "junior varsity" as described by Kleiman, were instructed to stop trafficking and given the alternatives of community support, job opportunities, tattoo removal, and dental work. Aware that the "bad actors" met severe penalties, the others simply quit. Instead of the arrest of one dealer after another only to have them replaced, the experiment yielded a different outcome: all dealers simultaneously stopped trafficking. More than five years later, wholesale crack sales have not resurfaced in High Point, North Carolina. The High Point experiment is being replicated in other locations. High Point is an example of focused and directly communicated deterrent threats. Characterized by David Kennedy as "pulling levers" in urban policing, it is another example of Schelling's Rule: because the threat was credible, it did not need to be carried out.

Applying High Point, Dynamic Concentration and Schelling's Rule to Prison Gangs

Prison gangs and their assaults, extortion and drug sales are constantly monitored by intensive surveillance, cell searches and debriefing of gang dropouts. As in High Point, not only the primary bad actors but the "junior varsity" are easily identified and potential cases developed. Employing the High Point model, prison officials and investigative staff can collect gang members together, provide audio/video surveillance and reports of criminal activity adequate for indictments, and - after visibly removing targeted gang leaders to SMU programs - the "junior varsity" or gang prospects and affiliates can be instructed to cease activity or face severe sanctions or prosecution based on already acquired evidence.

Alternatives to gangs can be provided for dropouts: anger/violence counseling, drug classes and the option to transfer to safer institutions. Former gang members can be challenged to physical competitions or boot camp disciplines requiring early rising, abstinence, group exercise, and awards for progress. Those failing to stop gang activity can be listed for SMU programs or subjected to Special Housing Unit alternatives. Schelling's Rule functions across errant behaviors, so prisons are likely to experience a significant reduction in gang activity without heavy demands on institutional resources, as former gang members elect more positive lifestyles.

Changing Inmate Attitudes: Project Ceasefire, the Good Behavior Game, Therapeutic Communities and Behavioral Segregation

Project Ceasefire, a small community program that addresses gang members' attitudes towards violence, has resulted in a fifty per cent reduction in the homicide rate in target neighborhoods17. Ceasefire procedures can be adapted to prison settings. In that inmates on average have a ninth grade educational level, the highly effective middle school "Good Behavior Game" also may be used to challenge prison populations, dividing inmates into teams that compete for rewards based on conduct18.

The federal system has several Therapeutic Communities TCs where inmates self-select for better conditions and undergo group therapy. Although studies claiming benefits from TCs are flawed due to self-selection bias, forms of behavioral segregation are essential to safer prisons. Other than aggregating the violent in SMUs, BOP has established several specialized institutions to house sex offenders, gang dropouts, and other "hard-to-place" inmates, creating a few gang-free safety zones that are far less threatening than behaviorally mixed populations.

Only New Remedies Lead to New Outcomes

There are 600,000 prison returnees annually, and two-thirds of those released are back within three years19. Anything that increases licit opportunity, self-command, and informal social controls reduces criminal behavior20. Kleiman has proposed changes to reduce recidivism, listing them in descending order of likelihood for implementation: the more improbable in the short-term are categorized under the heading "A Bridge Too Far"21. To that list are added the following:

Sentence Reductions for Ideas that Benefit the Government and the Community

Incentives can be provided in the form of sentence reductions for developing new approaches to prison, government or community problems. Although sentence reductions now are permitted only for cooperation leading to the conviction of others, the BOP Director has personal authority to reduce sentences on various grounds. This authority is underused but can be employed to encourage inmates to contribute actionable ideas and programs benefiting society. The BOP's authority could grant reductions not only for policy changes but for exceptional rehabilitation or educational accomplishments. Many prisoners now mired in hopeless pursuits would become actively engaged and goal directed toward positive outcomes.

Housing Aged or Terminally Ill Prisoners is Counterproductive

Older inmates who are incarcerated long after the age peak of criminal activity are particularly vulnerable to gang violence and extortion. Recognizing the cost of medical treatment for an aging population, BOP has created a small test program in one facility to expedite the release of elderly prisoners. This program can be expanded to include all prisoners over 65, those with serious disabilities, and the mounting number of inmates with sentences of Life Without Parole LWOP .

BOP's compassionate release system is designed to return terminally ill inmates to families, but is rarely used, so that even inmates who clearly pose no threat to the community in effect confront a death sentence. As a volunteer medical aide in one prison, I encountered inmates with only weeks to live being rejected for compassionate release. Rather than being an implacable and insurmountable hurdle, the compassionate release system should function for those non-violent inmates who are severely ill or aged to return to their families.

Education, Training and Incentives

Very small monetary or voucher rewards have been observed to reduce addictive behavior. Rewards for abstinence or educational accomplishments may deter gambling and alcoholism, and be less costly than controlling violence through lockdowns. General inmate populations experience - as punishment for the infractions of a few - the continual attrition of privileges. Good behavior, in contrast, could be rewarded, aggregating inmates who are not management problems into settings with longer recreational hours and other incentives, while segregating non-compliant inmates into disciplinary barracks and thereby making security monitoring more efficient and less costly. Prisons are encouraged to adopt free technologies for online coursework and certificates, including MIT's EdX and Stanford's Coursera.

Learning Deportment

Prisons are painfully noisy places for staff and most inmates, increasing stress, racial tensions, drug use and violence. Policies against loudness are never enforced. Sanctions should be applied for constant shouting or obscenities, so that community and workplace norms may be learned before release. Decibel meters that switch off televisions if noise limits are exceeded by inmates would be a quick, inexpensive and effective remedy.

Permit Personal Electronics

Emulating some state prisons, BOP has made email available to all inmates, enhancing investigative surveillance and even providing economic benefits for facilities. MP-3 players recently have been permitted. These programs can be expanded to include e-book readers, reducing the costs of inmate libraries, and substituting education for more deleterious pursuits. The limits on personal electronics can be suspended in part to allow laptop computers or tablets; educational games and a sanitized form of Internet access would redirect habitual prisoner activities away from gambling and chronic, passive television viewing while increasing inmates' capacities to function in the community upon release.

Open Institutions to Researchers

Although the federal system benefits from BOP's Office of Research and Evaluation ORE , which reviews research proposals concerning prisons, the ORE process can be streamlined. Lowering barriers to inquiry from graduate students, criminal justice departments and research institutions such as RAND would stimulate penological reform, resulting in better outcomes, lower costs, and more rapid innovation. The prison system is long overdue for extensive examination by behavioral scientists, economists and policy analysts.

Rethink Home Confinement for Non-Violent Offenders

The majority of non-violent offenders can be sentenced to forms of home confinement, where they may be closely monitored while employed and supporting their families, thus substantively reducing the prison population and moderating the aggregate damage to children and relatives from a parent's incarceration. The High Point program can be modified to ensure prison terms for non-compliance or flight, while a variant of H.O.P.E. can reduce drug and alcohol violations.

Expand Volunteerism

In a poignant scene at an Arizona women's prison, inmates knit dolls for the traumatized children of incarcerated illegal immigrants. Border Patrol officers now keep the dolls in trunks of investigative vehicles, to be provided to children after the arrests of immigrant groups; the dolls have comforted children found in the desert next to their dead parents. Known as RAISE Retirees Available Inmates Seeking Education , the program originated with Mrs. Jan Riding, who also hosts prison writing circles with University of Arizona professor emeritus Richard Shelton and other noted authors and poets. Expansion of RAISE-type programs in prisons can reduce costs of education, provide role models for inmates returning to the community, increase what Thomas Schelling describes as self-command and rational choice, and open dialogue between inmates, prison officials and community members on preventing recidivism. RAISE is a clearly valuable, eminently practical solution to understaffing and rehabilitation.

Yet little has changed, from the prisoner's viewpoint, since Arabist T.E. Lawrence in 1918 observed of his confinement, "There seemed a certainty in degradation"22. However, if these new policies are applied to the long moribund methods of incarceration, confrontation with the criminal justice system can become, for many, the first step to freedom.

***

The author is a graduate of the Kennedy School of Government at Harvard, and is a former research associate in neurobiology at Harvard Medical School and drug policy fellow with the Interfaculty Initiative on Drugs and Addictions under Harvard's Program on Mind, Brain and Behavior. He is incarcerated for life at a maximum security federal penitentiary. He may be reached through the website http://www.freeleonardpickard.org.

September 27, 2012

This paper was prepared as a submission for the Yale Law Journal prison writing project. The efforts of Barth Beresford in its preparation are gratefully acknowledged.

References:

1. Cook, Phillip J. "Punishment and Crime: A Critique of Current Findings Concerning the Preventative Effects of Punishment." Law and Contemporary Problems, 41:1, 164- 204 1977.

2. Becarria, Cesare. "Del delitti e dele pene" "On Crimes and Punishments" , 1764. translated with notes and introduction by David Young. Hackett, Indianapolis 1986.

3. Kleiman, Mark. When Brute Force Fails: How to Have Less Crime and Less Punishment. Princeton University Press 2009.

4. Thomas C. Schelling. Choice and Consequence: Perspectives of an Errant Economist. Harvard University Press 1984.

5. Kleiman, Mark. "Enforcement Swamping: A Positive Feedback Mechanism in Rates of Illicit Activity," Mathematical and Computer Modeling, 17, 2: 65-75 1993.

6. Kleiman, Mark and Beau Kilmer, "The Dynamics of Deterrence," Proceedings of the National Academy of Sciences USA , 106:34 2009.

7. Hawken, Angela and Mark Kleiman. "H.O.P.E. for Reform," The American Prospect, April 10, 2007, web only: http://prospect.org/article/hope-reform.

8. Rotter, Julien. "Generalized Expectancies for Internal verus External Control of Reinforcements." Psychological Monographs 80:1-28 1966.

9. Caulkins, Jonathan, Angela Hawken and Mark Kleiman. Drug Policy: What Everyone Should Know. Oxford University Press 2011.

10. Author, personnel communication.

11. Kleiman, Mark. Against Excess: Drug Policy for Results. Basic Books, New York 1992 , p. 250.

12. Gilligan, James. "The Last Mental Hospital," Psychiatric Quartelry 72 1 : 45-61 2001.

13. See also available fentanyl patches Duragesic ; Opsite occulsive dressings for tamper-proofing; and personal communication, Julie Holland, MD, Department of Psychiatry, New York University Medical School.

14. Kilmer, Beau. "The Future of DIRECT Surveillance: Drug and Alcohol Use Information from Remote and Continuous Testing," J. Drug Policy Analysis, Vol. 1, No. 1 2008.

15. Wacquant, Loic. "Deadly Symbiosis: When Ghetto and Prison Meet and Mesh," Punishment and Society 3, 18: 95-133 2009.

16. Kennedy, David. Deterrence and Crime Prevention: Reconsidering the Prospect of Sanctions. Toronto, Routledge 2009.

17. Wray, Richard J. "Public Health Communication Theory and Strategies for Interpersonal Violence Prevention," Journal of Aggression. Maltreatment and Trauma 13 3/4 , 41-61 2006.

18. Kellam, S.G., G.W. Rebok, N. Ialongo, and L.S. Mayer. "The Course and Malleability of Aggressive Behavior from Early First Grade into Middle School: Results of a Developmental Epidemiological-Based Preventative Trial," Journal of Child Psychology and Psychiatry 35: 259-81 1994.

19. Travis, Jeremy. "But They All Come Back: Facing the Challenges of Prisoner Re- Entry," Urban Institute Press, Washington, D.C. 2005.

20. Schelling, Thomas C. "Self-Command in Practice, in Policy and Theory of Rational Choice," American Economic Review 74: 1-11 1988.

21. Kleiman, Mark. Brute Force op. cit.

22. Lawrence, T.E. The Seven Pillars of Wisdom. CIII.