Emory Law Journal Emory Law Journal
Volume 70 Issue 6
2021
The Paradox of Recidivism The Paradox of Recidivism
Christopher Lewis
Follow this and additional works at: https://scholarlycommons.law.emory.edu/elj
Part of the Law Commons
Recommended Citation Recommended Citation
Christopher Lewis,
The Paradox of Recidivism
, 70 Emory L. J. 1209 (2021).
Available at: https://scholarlycommons.law.emory.edu/elj/vol70/iss6/1
This Article is brought to you for free and open access by the Journals at Emory Law Scholarly Commons. It has
been accepted for inclusion in Emory Law Journal by an authorized editor of Emory Law Scholarly Commons. For
more information, please contact law-scholarly-commons@emory.edu.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
THE PARADOX OF RECIDIVISM
Christopher Lewis
*
A
BSTRACT
The idea that we should respond more severely to repeated wrongdoing than
we do to first-time misconduct is one of our most deeply held moral principles,
and one of the most deeply entrenched principles in the criminal law and
sentencing policy. Prior convictions trigger, on average, a six-fold increase in
the length of punishment in states that use sentencing guidelines. And most of
the people we lock up in the U.S. have at least one previous conviction.
This Article shows that given the current law and policy of collateral
consequences, and the social conditions they engender, judges and sentencing
commissions should do exactly the opposite of what they currently do: impose a
recidivist sentencing discount, rather than a premium. This thesis is
counterintuitive and politically unpalatable. It goes against the grain of criminal
law and policy dating back as far as we know it, virtually the entire scholarly
literature, and millennia of social tradition. But this Article shows that it follows
logically from fairly ordinary moral premises.
*
Assistant Professor, Harvard Law School. For helpful feedback, I am grateful to audiences at the “New
Directions in Legal Philosophy” conference at All Souls College, University of Oxford; and the Criminal Law
Theory workshop at Harvard Law School. I owe special thanks to Aziza Ahmed, Rick Banks, Juliana
Bidadanure, Francois Bonnet, Guy Charles, Noah Feldman, Chad Flanders, Barbara Fried, Stephen Galoob, John
Gardner, John Goldberg, Kate Greasley, Don Herzog, Erin Kelly, Joshua Kleinfeld, Adriaan Lanni, Jed
Lewinsohn, Richard Lippke, Jonathan Masur, Tracey Meares, David Plunkett, Debra Satz, David Sklansky,
Patrick Tomlin, Adaner Usmani, Manuel Vargas, and Robert Weisberg.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1210 EMORY LAW JOURNAL [Vol. 70:1209
INTRODUCTION ........................................................................................... 1211
I. E
FFICIENCY AND PUBLIC SAFETY .................................................... 1215
A. Incapacitation ......................................................................... 1217
1. Replacement Effects .......................................................... 1218
2. The Backlash of Concentrated Incarceration ................... 1220
3. Crime Inside Prisons ........................................................ 1221
B. Deterrence .............................................................................. 1223
1. General Doubts ................................................................ 1223
2. Biases and Heuristics ....................................................... 1224
3. Probabilities of Detection .................................................. 1225
II. C
ULPABILITY ................................................................................... 1226
A. Character ................................................................................ 1227
1. Notice ............................................................................... 1227
2. “Lapse” ............................................................................ 1229
B. Additional Wrong ................................................................... 1233
1. Defiance ............................................................................ 1234
2. Omission ........................................................................... 1237
III. T
HE PARADOX ................................................................................. 1240
A. Barriers to Reentry and the Causes of Recidivism ................. 1241
1. Stigma for Job Applicants ................................................ 1241
2. Interrupting the Life Course ............................................. 1243
3. Welfare, Housing, and Other Goods ................................ 1244
B. Inequality, Incentives, and Criminal Responsibility ............... 1246
1. Moral Blame and Legal Punishment ................................ 1246
2. Kinds of Incentives ........................................................... 1248
3. Measuring Incentives ........................................................ 1250
4. Evidence of Ill Will ........................................................... 1252
5. Incentives and Deterrence Once More ............................. 1254
C. Violence and Victims .............................................................. 1256
1. Violent Crime ................................................................... 1256
2. Sexual Violence ................................................................ 1257
3. Victims’ Rights ................................................................. 1258
IV. C
OLLATERAL CONSEQUENCES: OBJECTIONS AND RESPONSE ......... 1259
A. Option Luck ............................................................................ 1259
B. Declining Severity .................................................................. 1261
C. Jurisprudence and Justification .............................................. 1264
1. Procedural Safeguards ..................................................... 1265
2. Substantive Criteria .......................................................... 1267
C
ONCLUSION ............................................................................................... 1270
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1211
INTRODUCTION
The idea that we should respond more severely to repeated wrongdoing than
we do to first-time misconduct is one of our most deeply held moral principles,
and one of the most deeply entrenched principles in the criminal law and
sentencing policy. Criminal justice systems around the world, including the U.S.
federal system and those in every U.S. state, punish repeat offenders much more
severely than first-time wrongdoers convicted of similar crimes.
1
Recidivist
sentencing enhancements are promulgated in state and federal sentencing
guidelines and statutory provisions;
2
imposed by sentencing judges as a
discretionary matter;
3
and furthered in decisions made by parole boards,
probation officers, and other corrections officials.
4
A convicted offender’s prior
criminal record is one of the two most important determinants of how long they
will spend in prison, along with the severity of the present criminal offense.
5
Prior convictions trigger, on average, a six-fold increase in the length of
punishment in states that use sentencing guidelines; some state guidelines
impose a ten-fold average increase; and for some offense categories, guidelines
recommend sentences 100 times more severe for offenders with the most serious
criminal records, compared to first-time offenders convicted of exactly the same
crime.
6
Most of the people we lock up—especially those who are Black or
Latino, and poor—already have at least one prior conviction.
7
Roughly half of
the people who are released from state prisons will return to prison within three
years.
8
The “recidivist sentencing premium,” as such, plays a large role in
1
See, e.g., NEAL B. KAUDER & BRIAN J. OSTROM, NATIONAL CENTER FOR STATE COURTS, STATE
SENTENCING GUIDELINES: PROFILES AND CONTINUUM 7, 19 (2008); JULIAN V. ROBERTS, PUNISHING
PERSISTENT OFFENDERS 93–117 (2008).
2
See, e.g., RICHARD S. FRASE, JULIAN V. ROBERTS, RHYS HESTER & KELLY LYN MITCHELL, ROBINA
INSTITUTE OF CRIMINAL LAW AND CRIMINAL JUSTICE, CRIMINAL HISTORY ENHANCEMENTS SOURCEBOOK 7
(2015) (cataloguing differences between state sentencing guidelines with respect to recidivist enhancements).
3
In United States v. Booker, 543 U.S. 220 (2005), the Supreme Court held that judges are merely
advised, not required, to follow the federal sentencing guidelines. Most judges still impose sentences within the
recommended ranges provided in the guidelines, though. See, e.g., Paul J. Hofer, Federal Sentencing After
Booker, 48 CRIME & JUST. 137, 139 (2019); Crystal S. Yang, Have Interjudge Sentencing Disparities Increased
in an Advisory Guidelines Regime? Evidence from Booker, 89 N.Y.U. L. REV. 1268, 1272 (2014). See generally
U.S. SENTENCING COMMISSION, SOURCEBOOK OF FEDERAL SENTENCING STATISTICS (2015), https://www.ussc.
gov/research/sourcebook/archive/sourcebook-2015 (exhibiting sentencing information and guideline
applications).
4
See, e.g., Joel M. Caplan, What Factors Affect Parole: A Review of Empirical Research, 71 FED. PROB.
16, 16 (2007)
5
KAUDER & OSTROM, supra note 1, at 3–6.
6
RICHARD S. FRASE & JULIAN V. ROBERTS, PAYING FOR THE PAST: THE CASE AGAINST PRIOR RECORD
SENTENCE ENHANCEMENTS 101 (2019).
7
See BRIAN A. REAVES, FELONY DEFENDANTS IN LARGE URBAN COUNTIES, 2009, at 8–9 (2013).
8
MATTHEW R. DUROSE, ALEXIA D. COOPER & HOWARD N. SNYDER, RECIDIVISM OF PRISONERS
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1212 EMORY LAW JOURNAL [Vol. 70:1209
shaping the size of the incarcerated population, and in determining its
demographic distribution.
But as this Article demonstrates, the law and policy of collateral
consequences, and the social conditions they engender, have left us in a situation
where judges and sentencing commissions ought to do precisely the opposite of
what they currently do: treat prior criminal convictions as a presumptive
mitigating factor, rather than an aggravating one—imposing a recidivist
sentencing discount, rather than a premium.
The case for this thesis unfolds in four parts.
The first two parts of the paper examine the prevailing rationales for the
recidivist sentencing premium. In Part I, I assess and cast doubt upon rationales
that appeal to the incapacitation—and deterrent—effects of the premium. I
pinpoint several important areas where existing empirical research provides
insufficient reason to be sure of these standard views and show that the recidivist
sentencing premium may be counterproductive to crime control in some
circumstances. There is an undeniable correlation between one’s past criminal
record and one’s future likelihood of reoffending. But we must weigh a number
of other determinants of public safety that the existing literature on recidivism
fails to consider—including “replacement effects,” the potential backlash of
concentrated incarceration, crime that occurs inside of our prisons and jails, the
biases and heuristics that underlie all human decision-making, and the
probability that an offender with prior criminal convictions will be detected.
Given our lack of information about the extent of these phenomena and how
they correlate with prior criminal convictions, considerations of public safety do
not weigh in favor of the recidivist sentencing premium as clearly as most people
think.
This portion of the argument should be taken as a possibility proof and a call
for further empirical research, rather than a decisive refutation of the rationales
under examination. Future empirical research could vindicate these rationales,
partially undermine their force, or potentially show that the recidivist sentencing
premium is counterproductive to crime control. But until such research is
completed, we should be skeptical of incapacitation- and deterrence-based
arguments for the recidivist sentencing premium.
RELEASED IN 30 STATES IN 2005: PATTERNS FROM 2005 TO 2010, at 1 (2014), https://www.bjs.gov/content/pub/
pdf/rprts05p0510.pdf.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1213
In Part II, I turn to backward-looking considerations of blameworthiness and
culpability. As the Federal Sentencing Guidelines put it, “a defendant with a
record of prior criminal behavior is more culpable than a first offender and thus
deserving of greater punishment.”
9
A wide range of criminal law theorists
defend various forms of this position.
10
Some argue that in cases of repeat
offending, we have more evidence of malice, ill will, or bad character than we
do when someone is convicted of an otherwise similar first offense. And others
argue that repeat offenders are guilty, by virtue of their past criminal history, of
an additional wrong on top of their current offense—for example, defiance of a
judicial mandate not to reoffend, or a failure to organize their lives in such a way
as to prevent themselves from repeating past mistakes. I argue that all of these
backward-looking rationales are either viciously circular or otherwise unsound.
This, combined with the possibility proof I give in Part I, shows that there is no
clear reason to think we are justified in punishing repeat offenders more severely
than first-time wrongdoers as a matter of course.
In Part III, I lay out the central, positive argument. The severity with which
we punish offenders, relative to one another, should track the amount of ill will
their crimes manifest. The amount of ill will an offense manifests depends in
large part on the strength of the offender’s “incentives” to commit that crime,
measured in terms of the extent to which getting away with that crime would
foreseeably add to the offender’s bundle of what Rawls called the “primary
goods”
11
—things that anybody would want, regardless of whatever else they
wanted—or in terms of Nussbaum and Sen’s “Capabilities Approach,”
12
which
tracks, roughly, one’s opportunities to live a life they have reason to value. I
canvass the wide range of barriers that people with prior criminal convictions
face to finding employment; beginning careers; getting welfare, housing, and
education; and to achieving a basic level of social status or esteem in their
communities. These barriers give people with prior convictions stronger
“incentives” than first-time offenders to commit just about any kind of crime—
with the possible exception of sexual violence. So, we cannot justifiably blame
9
U.S. SENT’G COMM’N, GUIDELINES MANUAL 392 (2016).
10
See, e.g., Youngjae Lee, Recidivism as Omission: A Relational Account, 87 TEX L. REV. 571, 621
(2009); Andrew von Hirsch, Desert and Previous Convictions in Sentencing, 65 MINN. L. REV. 591, 591–92
(1981) [hereinafter von Hirsch, Desert and Previous Convictions in Sentencing]; Andrew von Hirsch, Criminal
Record Rides Again, 10 CRIM. JUST. ETHICS 2, 55 (1991) [hereinafter von Hirsch, Criminal Record Rides Again];
PREVIOUS CONVICTIONS AT SENTENCING: THEORETICAL AND APPLIED PERSPECTIVES (Julian V. Roberts &
Andrew von Hirsch eds., 2010); RECIDIVIST PUNISHMENTS: THE PHILOSOPHER’S VIEW 1 (Claudio Tamburrini
& Jesper Ryberg eds., 2012).
11
See, e.g., JOHN RAWLS, JUSTICE AS FAIRNESS: A RESTATEMENT 58–61 (Erin Kelly ed., 2001).
12
See, e.g., MARTHA C. NUSSBAUM, CREATING CAPABILITIES: THE HUMAN DEVELOPMENT APPROACH,
at x (2011).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1214 EMORY LAW JOURNAL [Vol. 70:1209
or punish them for reoffending as severely as we could do for the same crime, if
it were a first offense. Judges and sentencing commissions, as such, have moral
reason to treat prior convictions as a presumptive mitigating factor at
sentencing—imposing a recidivist sentencing discount, instead of a premium.
In Part IV, I examine and rebut two of the most important potential
objections to the central argument: (1) that people with prior convictions are
themselves to blame for the incentives they face to reoffend, so those incentives
cannot excuse them even partially; and (2) that a recidivist sentencing premium
is necessary to “balance out” the declining marginal severity of collateral
consequences after a first conviction.
Before considering any of these claims in more depth, I want to flag two
important features of the scope and significance of the argument.
First, it is important to note that the argument I defend is comparative. That
is, the claims I make in this paper relate to the severity with which repeat
offenders should be punished, compared to first-time offenders. I do not make
any claims here about how severely people should be punished in absolute
terms. The arguments I defend in this paper are, in principle, compatible with
the idea that absolute levels of punishment should be higher, lower, or the same
as they are currently. I happen to think that prison sentences in the U.S. should
be much shorter than they are, for the most part, and that prison conditions
should be much more conducive to offenders’ future and present wellbeing (or
capabilities). And I defend those views in other work. But those views are
neither here nor there for the purposes of this paper.
Second, given that my argument for the recidivist sentencing discount
depends on the premise that collateral consequences incentivize future crime, it
is natural to wonder why we don’t simply get rid of collateral consequences,
instead of trying to compensate for them at sentencing with a discount for
recidivists. After all, passing legislation that would eliminate those collateral
consequences certainly seems more politically palatable than imposing a
recidivist sentencing discount.
But eliminating the power of the incentives that people with prior
convictions have to return to crime would require sweeping legislative,
administrative, and broader social change that neither judges nor sentencing
commissions or corrections officials can themselves unilaterally enact. Such
change is unlikely to be realized in the foreseeable future. Until it is, I argue,
judges and sentencing commissions have moral reason to treat prior convictions
as a presumptive mitigating factor, regardless of how unpalatable that may seem.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1215
Moreover, it may not be possible to eliminate all of the incentives that people
with prior criminal convictions have to return to crime without a fundamental
society-wide shift toward Scandinavian-style social democracy. A group of
economists studying Norwegian prisoners found that spending time in well-
resourced prisons with robust rehabilitative and job-training programs can
reduce offenders’ future offending and increase their employment prospects.
13
But Norway incarcerates far fewer people per capita than the U.S. (44 per
100,000, compared to almost 700 per 100,000); spends far more on each prisoner
(around $120,000/year, compared to approximately $30,000/year); and has a far
more robust social welfare state, higher minimum wages, far less poverty, and
far less inequality.
14
Even the most ambitious reentry programs in the U.S. do
not come close to achieving the results of these Norwegian programs.
15
Nor is it clear that eliminating all collateral consequences would be
desirable, on balance. Some policies designed to reduce those barriers may have
significant drawbacks. For example, a recent study found that “Ban The Box”
legislation could exacerbate racial discrimination in employment, as companies
unable to obtain information about applicants’ past criminal records may use
cues about an applicant’s racial group membership as a proxy for criminality
instead.
16
Whether or not the unfairness and potential inefficiencies associated
with that increase in racial discrimination are outweighed by the protections that
such policies ostensibly offer for formerly convicted people is unclear.
So, the unpalatability of the recidivist sentencing discount may indeed give
us—as citizens and as a society—one reason among many to reduce the barriers
formerly convicted people face to productive reentry. But unfortunately, for
judges and sentencing commissions, the paradox of recidivism is not that easy
to escape.
I. E
FFICIENCY AND PUBLIC SAFETY
Our criminal justice systems do not have unlimited resources, nor can they
prevent every possible crime from occurring. Both macro-level sentencing
policy and individual sentencing decisions affect the amount of crime that occurs
13
Manudeep Bhuller, Gordon B. Dahl, Katrine V. Løken & Magne Mogstad, Incarceration, Recidivism,
and Employment, 128 J. POL. ECON. 1269, 1272 (2020).
14
Id. at 1288–89.
15
See MARIE GOTTSCHALK, CAUGHT: THE PRISON STATE AND THE LOCKDOWN OF AMERICAN POLITICS
79–97 (2014).
16
See Amanda Agan & Sonja Starr, Ban the Box, Criminal Records, and Racial Discrimination: A Field
Experiment, 133 Q. J. ECON. 191, 193–94 (2018).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1216 EMORY LAW JOURNAL [Vol. 70:1209
in society, which in turn affects the wellbeing and functioning of the citizenry
and the society as a whole.
17
If public safety or crime control is the primary aim
of punishment, as many think it ought to be, then our sentencing policy and
sentencing decisions ought to minimize crime (or maximize public safety) as
much as possible given the fiscal, constitutional, and moral constraints within
which our systems operate.
18
A large body of empirical research shows that repeat offenders are more
likely to reoffend than those who have only been convicted of a single offense.
For example, in a meta-analysis of 131 studies, Gendreau et al. found that
“criminal history”—including both prior convictions and other contact with the
criminal justice system, such as prior arrests—was one of the most reliable
predictors of recidivism.
19
A more recent study found, similarly, that young first-
time offenders who avoid a subsequent conviction for ten years are no more
likely to commit a further crime than someone who has never been convicted at
all; older first-time offenders who avoid a second conviction reach this level of
risk even faster; but people with multiple prior convictions never reach that
level.
20
The recidivist sentencing premium is thus often seen as an efficient way to
allocate the scarce resources of our criminal justice systems, with the aim of
public safety in mind. On this view, we are justified in incarcerating repeat
wrongdoers for longer periods of time than first-time offenders convicted of
similar crimes because either (1) it takes more severe sanctions to deter them, or
17
The standard way to measure the relationship between imprisonment and crime, and between crime
and social welfare, is through cost-benefit analysis. See, e.g., John J. Donohue II & Peter Siegelman, Allocating
Resources Among Prisons and Social Programs in the Battle Against Crime, 27 J. LEGAL STUD. 1, 1–2 (1998).
This approach has a number of well-known normative limitations, however. See, e.g., DANIEL HAUSMAN,
MICHAEL MCPHERSON & DEBRA SATZ, ECONOMIC ANALYSIS, MORAL PHILOSOPHY, AND PUBLIC POLICY 158–
70 (2017); Amartya Sen, The Discipline of Cost-Benefit Analysis, 29 J. LEGAL STUD. 931, 950 (2000); MATTHEW
D. ADLER, WELL-BEING AND FAIR DISTRIBUTION: BEYOND COST-BENEFIT ANALYSIS 88–114 (2012).
18
See, e.g., Gary Becker, Crime and Punishment: An Economic Approach, 76 J. POL. ECON. 169, 170
(1968). “Side constrained” consequentialist theories hold that a sentencing scheme must also pass some
additional moral bar—for example, that it not violate anyone’s rights, that it not mandate or allow punishment
of the innocent, or that it not allow more punishment than any individual offender deserves; on these views,
passing the cost-benefit analysis is necessary, but not sufficient, for the justification of a sentencing scheme. See,
e.g., RICHARD FRASE, JUST SENTENCING: PRINCIPLES AND PROCEDURES FOR A WORKABLE SYSTEM 7 (2013);
NORVAL MORRIS, THE FUTURE OF IMPRISONMENT, at ix–x (1974); John Rawls, Two Concepts of Rules, 64 PHIL.
REV. 3–5 (1955); H.L.A. HART, PUNISHMENT AND RESPONSIBILITY: ESSAYS IN THE PHILOSOPHY OF LAW 1–28
(2008).
19
Paul Gendreau, Tracy Little & Claire Goggin, A Meta-Analysis of the Predictors of Adult Offender
Recidivism: What Works!, 34 CRIMINOLOGY 575, 582 (1996).
20
Shawn D. Bushway, Paul Nieuwbeerta & Arjan Blockland, The Predictive Value of Criminal
Background Checks: Do Age and Criminal History Affect Time to Redemption? 49 CRIMINOLOGY 27, 28 (2011).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1217
(2) they are more dangerous, so incapacitating them through imprisonment
promises a greater net benefit.
In this section, I show that neither one of these two future-oriented rationales
for the recidivist sentencing premium is definitively supported by the existing
empirical evidence. Rather, it is uncertain whether efficiency and crime control-
related considerations tell in favor of, or against, a recidivist sentencing
premium—and this section identifies areas where empirical research is needed
to fully address that question. But I show that it is at least possible that the
recidivist sentencing premium could be counterproductive to crime control.
I should note at the outset that, because of its reliance on data tracking
arrests, convictions, and other official contact points with the criminal justice
system, the empirical literature may exaggerate the connection between prior
convictions and future offending. People who are more likely to have been
caught breaking the law many times already are also more likely to be caught
doing so in the future.
21
Some of the factors that contribute to the mismatch
between crime statistics and the actual commission of crime can be controlled
for.
22
But there are a number of factors that cannot—such as criminal
skillfulness. Arrest and conviction data inevitably reflect this factor: all things
equal, more skillful offenders are caught less often than the clumsy ones. So,
while prior convictions may have some predictive value for projecting future
crime, that value is likely at least somewhat overstated. This at least slightly
mutes the power of both the incapacitation and deterrence rationale.
A. Incapacitation
Now let us turn to the incapacitation rationale specifically. Given that
(1) prior convictions are a predictor of future crime and (2) our criminal justice
21
See infra Part I.B.3 (“Probabilities of Detection”).
22
For example, some groups are much more heavily monitored and policed than others. Young Black
men in low-income areas of New York City are much more likely to be stopped, frisked, and subsequently
arrested than older white men in higher income areas of New York. See, e.g., Decio Coviello & Nicola Persico,
An Economic Analysis of Black-White Disparities in the New York Police Department’s Stop-and-Frisk
Program, 44 J. LEGAL STUD. 315, 315 (2015); Floyd v. City of New York, 959 F. Supp. 2d 540, 556 (2013). As
such, we could expect both that young Black men facing a current criminal conviction would be more likely
than older white men in the same boat to have a record of prior convictions, and we could also expect that the
same young Black men would be more likely to be arrested and convicted once again upon being released. We
can at least theoretically control for this by comparing intra-race, intra-class, and within-neighborhood
recidivism rates between first-time and repeat offenders. And we can perhaps control for much more than just
the most obvious factors that could otherwise distort data on recidivism rates, taking any statistical factors into
account that we have or could conceivable collect data on: for example, marital status, parents’ education levels,
job history, credit reports, known criminal acquaintances, and neighborhood characteristics across the lifespan.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1218 EMORY LAW JOURNAL [Vol. 70:1209
systems have limited resources, it seems logical that the most efficient way to
promote public safety may be to incarcerate repeat offenders for significantly
longer periods than first-time lawbreakers. After all, the thought goes, even if
the vast majority of the people we lock up have to return to the community
eventually, we can at least ensure that they pose no risk to the public while they
are in prison.
23
Criminologists measure the “incapacitation effects” of incarcerating any
given offender by calculating that offender’s counterfactual likelihood of
committing crime during the prospective period of incarceration were he or she
to remain free in the community. (In the literature this projection is represented
by the Greek lambda (λ)).
24
For example, imagine two people, Crabbe and Goyle, both of whom were
caught selling heroin to undercover drug enforcement agents. Crabbe had a
squeaky-clean record before his current offense, while Goyle has a rap sheet
longer than his arm. According to our best risk assessment instruments, Goyle
is predicted to commit about four times as much crime as Crabbe per year for
the foreseeable future (taking into account both the frequency and seriousness
of their predicted offending). Assume that the cost of imprisonment in their
jurisdiction, Hogsmeade, is $50,000/year per prisoner. Now consider two
hypothetical sentencing options:
(a) Crabbe and Goyle both get two years in prison—the default
sentence in Hogsmeade.
(b) Crabbe gets one year; Goyle gets three years.
Option (a) would cost the state $200,000 ($50,000/year per prisoner). Option
(b) would impose the same fiscal burden on the state as option (a) ($200,000)
but would have a 30% greater total incapacitation effect. Assuming that these
are the only two options, if sentencing officials in Hogsmeade want their
criminal justice system to promote public safety as much as possible given a
fixed budget, then option (b) seems clearly preferable to (a).
1. Replacement Effects
If incapacitating a specific offender from committing further crime is to have
any net social benefit, it cannot be the case that somebody else who would
23
For an influential defense of this thesis, see JAMES Q. WILSON, THINKING ABOUT CRIME 145–61
(1975).
24
See, e.g., Alex R. Piquero & Alfred Blumstein, Does Incapacitation Reduce Crime?, 23 J.
QUANTITATIVE CRIMINOLOGY 267, 268–69 (2007).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1219
otherwise have been law-abiding “steps up to the plate,” so to speak—
committing the same amount of crime (or more) than the incarcerated person
would have otherwise committed herself.
Unfortunately, the literature on “replacement effects” suggests that this may
often be the case—especially for crimes that are either conducted or organized
by groups, or offenses that are “market driven.”
25
Organized crime can continue
when one member of a gang or other criminal enterprise is incarcerated, but the
others are not.
26
And incarcerating one person for a market-driven offense—for
example, trafficking an addictive drug like heroin—can open up new, lucrative
criminal opportunities for someone else.
27
The stronger these replacement
effects are, the less any change in the incarceration rate is likely to influence
public safety or wellbeing at the community-level.
Replacement effects are likely to be stronger for offenders with prior
convictions than those without. Black, Latino, and poor defendants in criminal
cases are more likely to have prior convictions than their white and wealthy
peers.
28
Blacks and Latinos are overrepresented in the incarcerated population
generally,
29
but even more so among those convicted of gang-related violence
or the trafficking of addictive drugs (especially heroin, crack, and powder
cocaine).
30
And young men and boys living in poverty are much more likely to
join gangs than the better-off.
31
25
See, e.g., Thomas J. Miles & Jens Ludwig, The Silence of the Lambdas: Deterring Incapacitation
Research, 23 J. QUANTITATIVE CRIMINOLOGY 287, 288 (2007); Andrew von Hirsch & Don M. Gottfredson,
Selective Incapacitation: Some Queries About Research Design and Equity, 12 N.Y.U. REV. L. & SOC. CHANGE
11, 30 (1983); Marcia R. Chaiken & Jan M. Chaiken, Offender Types and Public Policy, 30 CRIME &
DELINQUENCY 195, 199 (1984).
26
Hirsch & Gottfredson, supra note 25.
27
See, e.g., Isaac Ehrlich, Crime, Punishment, and the Market for Offenses, 10 J. ECON. PERSP. 43, 53
(1996).
28
See BRIAN A. REAVES, FELONY DEFENDANTS IN LARGE URBAN COUNTIES, 2009 – STATISTICAL
TABLES (2013), https://jpp.whs.mil/Public/docs/03_Topic-Areas/07-CM_Trends_Analysis/20160122/04_BJS_
Report_State_Felony_Sentencing_2009.pdf.
29
See BRUCE WESTERN, PUNISHMENT AND INEQUALITY IN AMERICA 16–17 (2006).
30
See, e.g., BUREAU OF JUST. STAT., DRUG OFFENDERS IN FEDERAL PRISON: ESTIMATES OF
CHARACTERISTICS BASED ON LINKED DATA (2015), https://cdn.theatlantic.com/assets/media/files/dofp12_
sum.pdf; MALCOLM W. KLEIN & CHERYL L. MAXSON, STREET GANG PATTERNS AND POLICIES 3–15 (2006).
31
Irving A. Spergel, Youth Gangs: Continuity and Change, 12 CRIME & JUST. 171, 171 (1990) (arguing
that “[t]he gang is an important social institution for low-income male youths and young adults . . . because it
often serves social, cultural, and economic functions no longer adequately performed by the family, the school,
and the labor market”).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1220 EMORY LAW JOURNAL [Vol. 70:1209
2. The Backlash of Concentrated Incarceration
The recidivist sentencing premium also cannot be justified on incapacitation
grounds without considering how it might affect informal mechanisms of social
control. Concentrated incarceration in poor, predominantly Black, urban
neighborhoods can also cause crime by making it harder for those communities
to maintain informal mechanisms of social order and control.
32
Offenders with
prior criminal convictions are disproportionately drawn from the same
communities that are most likely to experience the backlash effect of
concentrated incarceration.
Close to 20% of adult men are imprisoned in some of our country’s least
well-off neighborhoods.
33
Almost everyone in those communities has a male
family member who either is or has been incarcerated.
34
This exerts a great deal
of strain on those families’ personal and economic resources, which in turn
keeps them in poverty.
35
In these circumstances, parents are hard-pressed to
teach their children social skills to keep them out of trouble with the law.
36
As a result, informal social control—which is more important than formal
social control for public safety—is undermined in these neighborhoods.
37
So,
increasing rates of imprisonment in communities where incarceration is already
concentrated can cause more crime than it prevents.
38
Conversely, decreasing
rates of incarceration in these communities can reduce rates of crime—or at least
not elevate them to the extent that individual assessments of released offenders’
risk of future crime would predict.
Eliminating the recidivist sentencing premium or imposing a recidivist
sentencing discount would shift the distribution of our prison populations away
from communities that have passed the “tipping point” beyond which
concentrated imprisonment has a backlash effect on crime, even if we held the
32
TODD R. CLEAR, IMPRISONING COMMUNITIES: HOW MASS INCARCERATION MAKES DISADVANTAGED
NEIGHBORHOODS WORSE 49–69 (2007).
33
See James P. Lynch & William J. Sabol, Assessing the Effects of Mass Incarceration on Informal Social
Control in Communities, 3 CRIMINOLOGY & PUB. POL’Y 267, 269 (2004).
34
See generally Todd Clear, Dina R. Rose & Judith A. Ryder, Incarceration and the Community: The
Problem of Removing and Returning Offenders, 47 CRIME & DELINQUENCY 335 (2001) (discussing data on
family members in prison).
35
DONALD BRAMAN, DOING TIME ON THE OUTSIDE: INCARCERATION AND FAMILY LIFE IN AMERICA 154–
163 (2004).
36
DON WEATHERBURN & BRONWYN LIND, DELINQUENCY-PRONE COMMUNITIES (2001).
37
CLEAR, supra note 32.
38
Dina Rose & Todd Clear, Incarceration, Social Capital, and Crime: Examining the Unintended
Consequences of Incarceration, 36 CRIMINOLOGY 441, 441 (1998).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1221
overall rate of incarceration constant. In order to have a realistic picture of the
effects of our sentencing policies and decisions on public safety, we must take
these community-level dynamics into account, yet the existing literature on
recidivism and sentencing fails to do so.
3. Crime Inside Prisons
There is a potentially more insurmountable problem with the incapacitation-
based rationale for the recidivist sentencing premium. The criminologists who
measure “incapacitation effects” almost always treat crime within prisons as
non-existent.
39
Crime that occurs in prison is underreported and under-
prosecuted.
40
And given the conditions of many of our prisons, decisions about
who we incarcerate, and for how long, may dictate who gets hurt, and whose
rights are violated—but not whether people get hurt, or how much.
41
In our popular culture and discourse, this is both known and accepted.
42
Within that popular discourse is a kind of primitive “rights forfeiture” theory:
convicted criminals, in the popular view, forfeit their rights not only to, for
example, the freedom of movement and association that incarceration inevitably
takes away, but also to bodily integrity, freedom from harm, and police
protection.
43
But even if one accepts the view that people forfeit some of their
rights when they commit an imprisonable offense, there is no good reason to
think they forfeit all of their human rights when they do so, or that their interests
can be completely discounted in a cost-benefit analysis or social welfare
functions as soon as they are locked up.
44
We accept that offenders found guilty of even relatively minor crimes might
be brutalized, beaten, and raped in prison, but we launder this out of cost-benefit
39
See, e.g., Alex R. Piquero & Alfred Blumstein, Does Incapacitation Reduce Crime?, 23 J.
QUANTITATIVE CRIMINOLOGY 267, 268 (2007) (discussing traditional models of studying incapacitation effects).
40
See, e.g., Nancy Wolff, Cynthia L. Blitz, Jing Shi, Jane Siegel & Ronet Bachman, Physical Violence
Inside Prisons: Rates of Victimization, 34 CRIM. JUST. & BEHAV. 588, 589 (2007) (citing Richard C. McCorckle,
Fear of Victimization and Symptoms of Psychopathology Among Prison Inmates, 19 J. OFFENDER REHAB. 19,
27–41 (1993)); Richard C. McCorckle, Personal Precautions to Violence in Prison, 19 CRIM. JUST. & BEHAV.
160, 160 (1992)).
41
See, e.g., Ben Gifford, Prison Crime and the Economics of Incarceration, 71 STAN. L. REV. 71, 71
(2019).
42
Id. at 113 (citing Elizabeth Stoker Bruenig, Why Americans Don’t Care About Prison Rape, NATION
(Mar. 2, 2015), https://www.thenation.com/article/archive/why-americans-dont-care-about-prison-rape/).
43
For general discussion and defense of rights forfeiture as a justification for punishment, see Christopher
Heath Wellman, The Rights Forfeiture Theory of Punishment, 122 ETHICS 371 (2012). Wellman does not
endorse the popular view under consideration here. Id.
44
Gifford, supra note 41, at 112.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1222 EMORY LAW JOURNAL [Vol. 70:1209
analyses of crime policy.
45
At the same time, we don’t include the things that
might happen to people inside of our jails and prisons in how we think about the
severity of punishment. In Farmer v. Brennan, the Supreme Court held that the
conditions of a prison count as “punishment” only if (1) a prison official knows
that there is a substantial risk of serious harm to inmates, and (2) he or she
disregards that knowledge by failing to take reasonable precautions to protect
inmates from the risk at hand.
46
If sentencing severity were measured so as to
include the kinds of brutality we expect and allow many incarcerated people to
experience, the punishments we inflict on people would be clearly
disproportionate by any reasonable standard, and would likely trigger
constitutional protections under the Eighth Amendment.
47
The phenomenon of crime within prisons shows that using “incapacitation”
as a rationale for penal policy decisions is a dubious proposition in general. Our
information about the extent of this crime is much less reliable than our data on
crime rates in the free population. Decisions about who should be imprisoned,
and for how long, cannot be made with the blind assumption that the incarcerated
will be unable to commit crime or cause harm during their imprisonment. That
assumption is both empirically and normatively implausible. Crime of all kinds
occurs within prison walls, and that crime cannot be written off or discounted in
cost-benefit analysis or social welfare functions.
To sum up, it is unclear at best whether the recidivist sentencing premium
can be justified by its incapacitation effects. It does no good for us to
“incapacitate” a large portion of the community if others will rise up to commit
the same crimes that our prisoners would have committed had we never locked
them up in the first place, or if doing so erodes informal modes of social control,
and thus causes more crime than it prevents. These negative externalities are
likely strongest for people with prior convictions. How much crime we have to
live with is what fundamentally matters for public safety; it doesn’t matter who
commits those crimes. And we should be very skeptical of the extent to which
the people we lock up are really incapacitated, rather than simply redirected, in
their criminal endeavors.
45
Cf., e.g., John Donohue, Assessing the Relative Benefits of Incarceration: Overall Changes and the
Benefits on the Margin, in DO PRISONS MAKE US SAFER?: THE BENEFITS AND COSTS OF THE PRISON BOOM 269,
272–73 (Steven Raphael & Michael A. Stoll eds., 2009) (providing overview of cost-benefit analysis that not
does factor in crimes against incarcerated offenders); William N. Trumbull, Who Has Standing in Cost-Benefit
Analysis? 9 J. POL’Y ANALYSIS & MGMT. 201, 208–15 (1990) (discussing individuals considered in cost-benefit
analysis without inclusion of incarcerated offenders).
46
Farmer v. Brennan, 511 U.S. 825, 825 (1994).
47
Sharon Dolovich, Cruelty, Prison Conditions, and the Eighth Amendment, 84 N.Y.U. L. REV. 881, 885
(2009).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1223
B. Deterrence
Now let us turn to deterrence. It is a commonplace argument that those with
longer criminal records have shown that they have a stronger taste for crime (or
a weaker aversion to punishment) by virtue of repeatedly breaking the law. They
are harder to deter, so we must punish them more harshly to get them to desist.
In this section, I bring together three sets of empirical findings that throw this
rationale into doubt. These findings, combined with the incapacitation-related
findings synthesized above, show that it is unclear whether considerations of
crime control or public safety tell in favor of the recidivist sentencing premium
at all.
1. General Doubts
Recidivist sentencing enhancements are rules of criminal law, and there are
a number of reasons to doubt that any such rule has much of a deterrent effect.
48
While the existence of a justice system that imposes criminal punishment can
deter crime, the manipulation of rules and penalties within that system can only
do so under unusual conditions.
49
In order for a rule or penalty to have a deterrent
effect, it must be well known to the public;
50
it must carry a meaningful penalty,
the perceived threat of which must exceed the perceived benefit of breaking the
law;
51
the chance of being caught must be seen as non-trivial;
52
and those
potentially subject to it must be willing and able to bring that information to bear
on their decision-making.
53
These preconditions for the criminal law to have a genuine deterrent effect
are rarely all met at the same time. For example, as Robinson and Darley note,
“people rarely know the criminal law rules, even when those rules are
48
See, e.g., Aaron Chalfin & Justin McCrary, Criminal Deterrence: A Review of the Literature, 55 J.
ECON. LITERATURE 5, 39 (2017) (suggesting “repeat offenders have already paid the informal costs associated
with being labeled a criminal” as a consideration for sentencing severity); Paul H. Robinson & John M. Darley,
Does Criminal Law Deter? A Behavioral Science Investigation, 24 OXFORD J. LEGAL STUD. 173, 191 (2004)
(arguing repeat offenders are less deterred by the risk of reincarceration after discovering prison is not “so bad
after all”).
49
Paul H. Robinson & John M. Darley, The Role of Deterrence in the Formulation of Criminal Law
Rules: At Its Worst When Doing Its Best, 91 GEO. L. J. 949, 951 (2003).
50
Paul H. Robinson, John M. Darley & Kevin M. Carlsmith, The Ex Ante Function of the Criminal Law,
35 LAW & SOC’Y. REV. 165, 166 (2001).
51
Id. (asserting criminal law can only prevent unlawful conduct if the public is aware of the penalties
attached to unlawful conduct).
52
See, e.g., Daniel S. Nagin, Deterrence: A Review of the Evidence by a Criminologist for Economists, 5
ANN. REV. ECON. 83, 85 (2013).
53
Robinson & Darley, supra note 48.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1224 EMORY LAW JOURNAL [Vol. 70:1209
formulated under the express assumption that they will influence conduct.”
54
Instead, people often use their own moral intuitions to guess at legal rules
without any actual knowledge of the legal code.
55
Over 75% of “active
criminals” and almost 90% of the “most violent criminals . . . perceive no risk
of apprehension or are incognizant of the likely punishments for their crimes.”
56
Those who commit crime tend to be impulsive and risk-seeking in general, and
often intoxicated when they do.
57
There is no prima facie reason to think the recidivist sentencing premium is
an exception to any of these general trends. Therefore, justifying that principle
on the basis of its deterrent effect is a tall task from the start.
2. Biases and Heuristics
Findings in the behavioral sciences also show that, because of the biases and
heuristics that guide human decisions, it is unlikely that the recidivist sentencing
premium could have much of a deterrence-related benefit, if any.
58
People weigh
information and experiences that are salient to them more heavily in their
decision-making than ideal models of rationality would dictate.
59
Likewise,
people tend to discount information and experiences that are less salient.
60
For
example, people weigh their own first-hand, direct experiences much more
heavily than equally reliable information about other people’s experiences.
61
They weigh recent experiences much more heavily than experiences in the
distant past.
62
And they weigh the perceived risks of vivid or dramatic
experiences and events (e.g., witnessing an explosion) much more heavily than
less dramatic ones (e.g., long-term weight gain)—even when the less dramatic
risks are actually higher.
63
54
Id. at 176.
55
John Darley, Catherine A. Sanderson & Peter S. LaMantia, Community Standards for Defining
Attempt: Inconsistencies with the Model Penal Code, 39 AM. BEHAV. SCIENTIST 405, 414 (1996).
56
David A. Anderson, The Deterrence Hypothesis and Picking Pockets at the Pickpocket’s Hanging, 4
AM. L. & ECON. REV. 295, 295 (2002).
57
See, e.g., Marianne Junger, Robert West & Reiner Timman, Crime and Risky Behavior in Traffic: An
Example of Cross-Situational Consistency, 38 J. RSCH. CRIME & DELINQUENCY 439, 439 (2001).
58
See, e.g., David A. Dana, Rethinking the Puzzle of Escalating Penalties for Repeat Offenders, 110
YALE L.J. 733, 759–72 (2001).
59
Amos Tversky & Daniel Kahneman, Availability: A Heuristic for Judging Frequency and Probability,
5 COGNITIVE PSYCH. 207, 228–29 (1973).
60
Id.
61
Dana, supra note 58, at 759.
62
Tversky & Kahneman, supra note 59, at 224.
63
Paul Slovic, Perception of Risk, 236 SCI. 280, 283 (1987).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1225
Offenders who have been imprisoned already have first-hand, direct
experience with incarceration. That experience is usually recent, given the
‘revolving door’ effect of parole and reentry processes across the country.
64
And
actually being imprisoned is much more vivid than merely imagining the
prospect of it. Incarceration, as such, is more of a salient deterrent for those have
already experienced it before than it is for those who have not. So, even if
substantive rules of criminal law could deter people from crime, people who
have been incarcerated in the past may already have a much more salient
deterrent than those who haven’t due to the actual experience of incarceration.
3. Probabilities of Detection
People assign exponentially greater weight to the likelihood of getting
caught than they do to the severity of potential penalties in deciding whether to
commit a crime.
65
Having a prior criminal record itself makes people easier to
monitor, and thus more likely to get caught. A history of prior contact with the
criminal justice system gives law enforcement agencies information about
offenders that makes any subsequent crimes they commit easier to detect. Some
of this information is propensity related—for example, police departments keep
records about previously convicted offenders’ modus operandi.
66
Some of it aids
in prosecution efforts—for example, having fingerprints, DNA samples,
photographs, and information about an offender’s acquaintances on file.
67
These
kinds of records both help law enforcement agencies link known violations to
unknown perpetrators, and also help them detect hidden or unreported
violations.
68
From the perspective of a law enforcement agency, targeting people with
prior convictions for investigation is an efficient strategy—even if those people
are no more likely than anyone else to have committed the crime in question, or
any crime at all.
69
For example, when fingerprints are recovered at a crime scene,
it is easy for police departments to check them against fingerprint databases of
64
MATTHEW R. DUROSE, ALEXIA D. COOPER & HOWARD N. SNYDER, RECIDIVISM OF PRISONERS
RELEASED IN 30 STATES IN 2005: PATTERNS FROM 2005 TO 2010, at 1 (2014), https://www.bjs.gov/content/pub/
pdf/rprts05p0510.pdf.
65
Chalfin & McCrary, supra note 48, at 7–8; Nagin, supra note 52, at 97–100.
66
See, e.g., L. ENF’T INFO. TECH. STANDARDS COUNCIL, STANDARD FUNCTIONAL SPECIFICATIONS FOR
LAW ENFORCEMENT RECORDS MANAGEMENT SYSTEMS (RMS) 4 (n.d.); Donald C. Stone, Practical Use of
Police Records System, 24 J. CRIM. L. & CRIMINOLOGY 668, 669 (1933).
67
See L. ENF’T INFO. TECH. STANDARDS COUNCIL, supra note 66.
68
Dana, supra note 58, at 744–46.
69
See WALTER L. PERRY, BRIAN MCINNIS, CARTER C. PRICE, SUSAN C. SMITH & JOHN S. HOLLYWOOD,
PREDICTIVE POLICING: THE ROLE OF CRIME FORECASTING IN LAW ENFORCEMENT OPERATIONS 2 (2013).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1226 EMORY LAW JOURNAL [Vol. 70:1209
previously convicted felons, and that is a prudent investigative strategy even if
there is no reason to think that the perpetrator had a prior conviction.
So, the nature of criminal investigation should also cast some doubt on the
idea that the recidivist sentencing premium is justified by its deterrent effects.
Repeat offenders are more likely to get caught than first-time lawbreakers; and
the probability of detection drives decisions about whether or not to commit
crime much more than the anticipated severity of punishment conditional on
being detected.
The findings canvassed here show that (1) substantive rules of criminal law
such as statutory recidivist sentencing enhancements tend to have little to no
deterrent effect in general; (2) people with prior convictions have more salient
deterrents than first-time offenders; and (3) repeat offenders are more likely than
others to get caught, which is what would deter people from breaking the law, if
anything would. So, to the very limited extent that tinkering with the rules of
criminal law or sentencing policy is likely to have any deterrent effect at all, the
recidivist sentencing premium may be counterproductive, as repeat offenders
seem to face stronger deterrents to crime than first-time offenders even without
it. These findings, combined with the incapacitation-related findings synthesized
above, give us reason to be skeptical of the idea that the recidivist sentencing
premium is justified by considerations of public safety more generally.
II. C
ULPABILITY
The recidivist sentencing premium may not be justified by considerations of
crime control or public safety. But that is not the only rationale on offer.
According to the Federal Sentencing Guidelines, “[a] defendant with a record of
prior criminal behavior is more culpable than a first offender and thus deserving
of greater punishment.”
70
A number of criminal law theorists defend variants of
this position.
71
The greater culpability of repeat offenders, according to this
outlook, could either require or permit the state to impose a recidivist sentencing
premium.
In this section, I canvas these culpability-based rationales for the recidivist
sentencing premium and show why all of them are either viciously circular or
70
U.S. SENT’G COMM’N, supra note 9.
71
See, e.g., Lee, supra note 10; von Hirsch, Desert and Previous Convictions in Sentencing, supra note
10; von Hirsch, Criminal Record Rides Again, supra note 10; PREVIOUS CONVICTIONS AT SENTENCING:
THEORETICAL AND APPLIED PERSPECTIVES (Julian V. Roberts & Andrew von Hirsch eds., 2010); RECIDIVIST
PUNISHMENTS: THE PHILOSOPHER’S VIEW (Claudio Tamburrini & Jesper Ryberg eds., 2012).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1227
otherwise unsound. We have no good reason to think that repeat offenders are—
in principle, in general, or even on average—more culpable than first-time
offenders.
A. Character
Some legal scholars argue that when a previously convicted offender is
found guilty of a subsequent crime, we have more evidence of ill will or bad
character than we do when a first-time wrongdoer commits the same crime.
These character-based views come in two main variants: (1) notice and
(2) “lapse.” The former of the two rationales is considerably weaker than the
latter, and its main proponent, Julian Roberts, has renounced it in recent years,
72
so I discuss it briefly.
1. Notice
According to the “notice” rationale, the formal censure that comes with a
criminal conviction makes, or should make, offenders more aware of the
wrongfulness of their crimes.
73
The more an offender is—or should be—aware
of the wrongness of an offense, the more severely we are justified in blaming
and punishing him for it, the idea goes.
74
So, if repeat offenders either are or
should be more aware of the wrongness of their crimes, we may be justified in
imposing enhanced criminal sentences upon them.
Victim impact statements at sentencing may have an especially strong
pedagogical function in this regard—they illustrate the consequences of crime
to the perpetrator in a way that an abstract understanding of what is wrong with
the offense cannot. As Julian Roberts puts it, “[a] first-time burglar may not be
fully aware of the harm inflicted by breaking into someone’s home. A repeat
burglar who has experienced numerous sentencing hearings and received
multiple sentences is under no illusions about the consequences of domestic
burglary.”
75
72
See FRASE & ROBERTS, supra note 6.
73
ROBERTS, supra note 1, at 38 (2008); Julian Roberts, First Offender Sentencing Discounts: Exploring
the Justifications, in PREVIOUS CONVICTIONS AT SENTENCING: THEORETICAL AND APPLIED PERSPECTIVES, supra
note 10, at 17–19.
74
See, e.g., Gideon Rosen, Culpability and Ignorance, 103 PROC. ARISTOTELIAN SOC’Y 61, 75, 83 (2003)
(arguing that those who should understand the wrongfulness of their actions, but do not, are culpable for their
normative ignorance—which in turn enhances their culpability for the action itself).
75
Roberts, supra note 73, at 19–20.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1228 EMORY LAW JOURNAL [Vol. 70:1209
The fact that an offender has a prior conviction for one thing does not make
it any more likely that he has a special knowledge of the wrongfulness of an
entirely different kind of crime, however. Some repeat offenders have an
established modus operandi, and others have addictions that push them to
commit the same offenses over and over again; but repeat offenders do not
always duplicate the crimes that led to their prior convictions.
76
And there is
little reason to think that having been convicted of burglary leaves one with a
greater understanding of the wrongfulness or the consequences of reckless
driving, arson, or assault and battery on a police officer.
Similarly, it is implausible that having a prior conviction for one kind of
crime gives an offender a stronger obligation to understand what is wrong or
harmful about an entirely different kind of crime.
77
Even if there is an implicit
message in the processes of adjudication, sentencing, or corrections, that
convicted arsonist ought to think about what he did, the accompanying
obligation cannot extend to thinking about what is wrong with sexual battery,
for example.
One might respond that a prior conviction should have prompted a repeat
offender to contemplate and understand what is wrong with breaking the law in
general. But understanding what is wrong with breaking the law in general
accounts for very little of our understanding of what is wrong with malum in se
crimes like assault or murder, the wrongness of which are independent of their
illegality.
Roberts suggests a more individualized approach to sentencing as a solution
to this problem: first-time offenders should be able to argue for leniency at
sentencing if they can show that they were not fully aware of the wrongfulness
of their crimes.
78
But this approach already seems built into the standard mens
rea inquiry. The criminal law assigns a higher degree of culpability to offenders
who intentionally commit crime in full knowledge of the wrongfulness of their
actions than those who recklessly or negligently commit similar crimes without
regard to their wrongfulness or believing their actions to be less seriously wrong
than they actually are.
79
76
See Michael M. O’Hear, Recidivism and Criminal Specialization, MARQ. UNIV. L. SCH. FAC. BLOG
(Dec. 11, 2016), https://law.marquette.edu/facultyblog/2016/12/recidivism-and-criminal-specialization/.
77
C.f., infra Part II.B.
78
Roberts, supra note 73, at 20.
79
Douglas N. Husak, The Sequential Principle of Relative Culpability, 1 LEGAL THEORY 493, 496 (1995).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1229
An individualized or heavily fact-sensitive sentencing regime is, if anything,
somewhat incompatible with the kinds of recidivist sentencing enhancements
that we have—which are for the most part mechanical or formulaic: mandatory
minimums, advisory guidelines, or judicial presumptions. And the
considerations that tell in favor of individualized sentencing procedures—
sensitivity to the particular facts of the case, including the defendant’s
motivations for the crime—apply just as well to cases where the defendant has
multiple prior convictions. So, the “notice” view cannot justify the recidivist
sentencing premium.
2. “Lapse”
Andrew von Hirsch, among others, defends a different kind of character-
based view meant to justify treating first-time offenders more leniently than
repeat offenders.
80
On his view, criminal sentencing should be sensitive to
human frailty, which may sometimes be manifested in acts that violate the
criminal law. von Hirsch argues that frailty is an inevitable part of human
agency, and one we ought to be forgiving of. A first-time offender may have
broken the law, as such, but may have done so only in a moment of weakness or
a lapse of self-control that is “out of character.”
81
An offender who repeatedly
and persistently breaks the law, however, is less plausibly acting “out of
character,” and his actions cannot so readily be interpreted as resulting from a
lapse of control or momentary weakness—instead, they likely tell us more about
who he really is, and what kinds of attitudes he has toward others and toward
the norms of the society at large. Good people can do bad things, the idea goes,
and if a first-time offender has a long track record of being a law-abiding citizen,
it is likely that his crime was an aberration, rather than a manifestation of deeply
entrenched attitudes toward other people or the society at large.
82
There is a certain amount of uncertainty in all of this, to be sure: it is possible
that a first-time offender’s crime is totally in character, and that a ten-time
offender is extraordinarily unlucky to have been in the wrong place at the wrong
time over and over again. But, on von Hirsch’s view, recidivism is at least a
good proxy for committing crime that is in character; and the possibility that an
80
For the most recent statement of this view, see Andrew von Hirsch, Proportionality and the Progressive
Loss of Mitigation: Some Further Reflections, in PREVIOUS CONVICTIONS AT SENTENCING: THEORETICAL AND
APPLIED PERSPECTIVES, supra note 10, at 1.
81
Youngjae Lee, Repeat Offenders and the Question of Desert, in PREVIOUS CONVICTIONS AT
SENTENCING: THEORETICAL AND APPLIED PERSPECTIVES, supra note 10, at 49.
82
See YUVAL FELDMAN, THE LAW OF GOOD PEOPLE: CHALLENGING STATES’ ABILITY TO REGULATE
HUMAN BEHAVIOR 1 (2018).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1230 EMORY LAW JOURNAL [Vol. 70:1209
offense resulted from a momentary lapse becomes less and less likely with
multiple prior convictions.
This is taken to justify what von Hirsch calls the “progressive loss of
mitigation”: first-time offenders are spared the full severity of punishment that
their offense might otherwise warrant, on the assumption that they may have
acted out of character.
83
As they accumulate more convictions, that mitigation is
progressively lost, and eventually they must face the fully earned punishment
that their crimes warrant.
von Hirsch emphasizes that he is in favor of leniency for first-time offenders,
rather than escalating the severity with which we punish people with prior
convictions.
84
In practice, there is little to distinguish sentencing regimes that
treat the presence of a prior conviction as an aggravating factor from ones that
treat a clean record as a mitigating factor, however. And since this Article
focuses on whether and why there might be reason to treat recidivists differently
from first-time offenders at all, that distinction has little significance here.
Something like von Hirsch’s view could potentially rationalize the kind of
guideline-based systems we see in both the U.S. and the U.K. today, where the
more prior convictions an offender has (and the more serious those convictions
are), the more severely the guidelines advise judges to punish them.
The “lapse” view is much stronger than the “notice” view, and seeing what
is wrong with it takes careful attention to both the idea of acting “out of
character,” and how that idea connects, if at all, to recidivism. I use a “proof by
cases” style argument to illustrate this here. There are three ways we could
understand the concept of “character,” and on closer examination, none of them
support von Hirsch’s justification for the recidivist sentencing premium.
First, “character” is usually understood as a dispositional property of
persons.
85
That is, one’s character traits are understood in terms of what one
83
von Hirsch, supra note 80.
84
Id.
85
For an explicit defense of this view, see GILBERT RYLE, THE CONCEPT OF MIND (1949). This
dispositional view of character is shared across competing theories of virtue and vice through the history of
philosophy. See, e.g., ARISTOTLE, NICHOMACHEAN ETHICS § II.7 (Roger Crisp trans., Cambridge Univ. Press
2014); DAVID HUME, A TREATISE OF HUMAN NATURE § 3.2 (L. A. Selby-Bigge ed., Oxford Univ. Press, 1978);
IMMANUEL KANT, CRITIQUE OF PRACTICAL REASON § 5:156-7 (Mary Gregor trans., Cambridge Univ. Press
2015). This is how the “situationist” literature in social psychology—which argues against the idea that human
behavior is determined by external, situational factors, rather than people’s character traits—understands the
concept of “character” as well. See, e.g., John M. Darley & C. Daniel Batson, “From Jerusalem to Jericho”: A
Study of Situational and Dispositional Variables in Helping Behavior, 27 J. PERSONALITY & SOC. PSYCH. 100,
107 (1973); Rachana Kamtekar, Situationism and Virtue Ethics on the Content of Our Character, 114 ETHICS
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1231
would do (or think or feel) in a range of actual or counterfactual circumstances.
For example, a coward would fail to stand up for what is right even if doing so
would objectively cost him very little; a greedy person seeks to profit even if
they already have more than enough, and even when other considerations should
outweigh that motive; a generous person tries to help others even when doing so
comes at an objectively high cost for them.
It is unclear how acting out of character is even possible given this
background view of what character is. If character is just a counterfactual
disposition to act in certain ways given certain circumstances, then we must
always act in character. We may be surprised or disappointed when people act
in ways we don’t expect, given our prior conceptions about them. But those
reactions are likely explained by the fact that character is easy to misjudge, not
by any robust sense in which people can actually act “out of character.”
Second, one could think of “character” in a purely probabilistic sense: one’s
character is just what one usually does, feels, and thinks.
86
But if “acting out of
character” is just doing something statistically unusual in one’s life course, it no
longer seems like the fact that a crime was out of character has any normative
relevance to how severely we punish them for it.
Consider the following case:
Luxury Tax: Money Bag is an ambitious and successful financier with
no criminal record, a wealthy and powerful family background, and an
immaculate professional resume. He attended the best private schools
as a child, studied at a prestigious university, and then made a name
for himself in banking for a firm on Park Place, moving up the career
ladder through his twenties. Shortly after being promoted again in his
early thirties, he is presented with an opportunity to move up even
further through an elaborate scheme to help the firm avoid the Luxury
Tax—his first opportunity of this kind. He jumps at the chance, feeling
like he is too smart to get caught and too well-connected to be punished
even if he were.
The fact that Money Bag’s offense was “out of character” in this purely
probabilistic sense, tells us little about how severely we should blame or punish
him for committing securities fraud. Money Bag jumps at his first opportunity
to commit the crime, in circumstances that give him little objective reason to do
so. Perhaps he would have committed the crime much earlier if he had the
458, 458 (2004).
86
More Definitions for Character, MERRIAM-WEBSTER, https://www.merriam-webster.com/dictionary/
character (last visited June 17, 2021) (defining character as “the way someone thinks, feels, and behaves).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1232 EMORY LAW JOURNAL [Vol. 70:1209
opportunity then. And what if we knew that Money Bag would invariably jump
at the opportunity to commit the same crime every time, if such opportunities
only came about more often? It would seem to matter little that this was his first
offense. So, if the fact that one’s action was out of character is supposed to
mitigate how much we can blame him for it, then we cannot think about “acting
out of character” in a purely probabilistic way.
Third, we might think about the connection between crime and “character”
in terms of the likelihood that one would commit any given crime, controlling
for the difficulty of one’s circumstances. I think this is the only plausible way to
interpret von Hirsch’s argument. For example, consider the following case:
Free Parking: Racecar and Scottie Dog are irregularly employed day
laborers. They often find themselves wearing out their welcome on
friends’ couches and end up sleeping in the Free Parking lot from time
to time on an empty stomach. Life is hard for both of them, but Scottie
Dog sucks it up and tries his best to find work by waiting in front of
the Electric Company with his toolbox, hoping somebody driving by
will pick him up and give him a day’s work. Racecar mostly does the
same, but he dreads the hungry and cold nights in the Free Parking lot
more than Scottie—and from time to time, he rolls down to St. James
Place looking for someone to rob. He has now been convicted of
robbery once a year for the past ten years. Scottie, on the other hand,
goes nine years without breaking the law once, but in his tenth year
living rough, he finally lapses, and decides to rob someone on St.
James Place like Racecar.
It seems that the only thing that could explain why Scottie has only broken
the law once over the last ten years, while Racecar has done so ten times, is that
Scottie has a better character; their circumstances, by stipulation, are the same.
This argument might justify a recidivist sentencing premium (or von Hirsch’s
“progressive loss of mitigation”) if first-time and repeat offenders in general
commit crime against similar circumstances, like Racecar and Scottie.
The problem is that having a criminal record totally changes one’s
circumstances. As I discuss below in Part III, the collateral consequences of a
prior criminal conviction give one much stronger incentives to return to crime
than one would otherwise have. A repeat offender who commits the same crime
over and over again is not actually in the same circumstances as a first-time
offender who commits the same crime after a long time with a clean record—
unless the first-time offender is significantly worse off, apart from his criminal
record, than the repeat offender. That would be the only way their circumstances
could be comparable, since prior convictions are such a big disadvantage.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1233
This is not at all like the patterns we see in real life. First-time offenders in
real life are not, on average, much worse off than repeat offenders, apart from
their record.
87
In fact, the opposite is the case: repeat offenders tend to be at a
relative disadvantage even without all of the burdens of their criminal record—
they are more likely to be Black or Latino;
88
from poor family backgrounds and
disadvantaged neighborhoods;
89
unemployed;
90
have lower levels of education;
and to have weaker support systems and social networks.
91
As such, while we might be able to reconstruct a logically sound
interpretation of von Hirsch’s argument, that argument is an empirically
unsound rationale for the recidivist sentencing premium in societies like our
own—where prior criminal convictions are both constitutive of disadvantage in
and of themselves, and also correlated with many other forms of deep
disadvantage, all of which incentivize crime.
B. Additional Wrong
Another family of culpability-based rationales for the recidivist sentencing
premium holds that repeat offenders are guilty of defying, or otherwise failing
to live up to, the message implicit in their first conviction: that what they did
was wrong, and that they should organize their lives so as not to break the law
again. There are two principal variants of this family. I discuss the feebler
“defiance” variant first, before turning to Youngjae Lee’s more sophisticated,
87
See Megan F. Dickson, Nesa E. Wasarhaley & J. Matthew Webster, A Comparison of First Time and
Repeat Rural DUI Offenders, 52 J. OFFENDER REHAB. 421, 426–27 (2013).
88
See, e.g., ASHLEY NELLIS, THE COLOR OF JUSTICE: RACIAL AND ETHNIC DISPARITY IN STATE PRISONS
10 (2016) (“Prosecutors are more likely to charge [B]lack defendants under state habitual offender laws than
similarly situated white defendants.”); Sarah B. Hunter, Eunice Wong, Chris M. Beighley & Andrew R. Morral,
Acculturation and Driving Under the Influence: A Study of Repeat Offenders, J. STUD. ALCOHOL 458, 458 (2006)
(“DUI recidivism rates have been found to be higher among Mexican Americans in comparison with whites.”);
Idelisse MalavÉ & Esti Giordani, Latino Populations and Crime in America, UTNE, https://www.utne.
com/community/latino-populations-ze0z1501zdeh (last visited June 17, 2021) (“Latinos are more likely than
[w]hite Americans to get arrested and account for a disproportionate share of all felony and misdemeanor
arrests.”).
89
See, e.g., Nick Tilley, Analyzing and Responding to Repeat Offending, ASU CTR. FOR PROBLEM-
ORIENTED POLICING (2016), https://popcenter.asu.edu/content/analyzing-and-responding-repeat-offending
(“There is a risk that multiple contacts with the criminal justice system unintentionally provide ‘stepping stones’
toward further criminal involvement, especially among the poor.”).
90
See, e.g., Stewart J. D’Alessio, Lisa Stolzenberg & David Eitle, “Last Hired, First Fired”: The Effect
of the Unemployment Rate on the Probability of Repeat Offending, 39 AM. J. CRIM. JUST. 77, 77–93 (2014).
91
Bill Keller, Seven Things to Know About Repeat Offenders, MARSHALL PROJECT (Mar. 9, 2016, 11:00
PM) (“Inmates who didn’t finish high school are 10 points more likely to be arrested again than those who got
a high school diploma – and 40 points more likely than those who finished college.”); Stephanie A. Spohr,
Sumihiro Suzuki, Brittany Marshall, Faye S. Taxman & Scott T. Walters, Social Support Quality and
Availability Affects Risk Behaviors in Offenders, 4 HEALTH & JUST. 1, 7 (2016).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1234 EMORY LAW JOURNAL [Vol. 70:1209
but ultimately still unsound, “omission” theory. Thinking through the problems
with the omission theory will help bring us to the moral, political, and legal
paradox at the core of this article.
1. Defiance
One frequently discussed (though not frequently defended) rationale for the
recidivist sentencing premium is that, after having been explicitly censured for
their past conduct, people with prior criminal convictions are guilty of further
defying the law, or the norms of the society in general, upon reoffending.
92
That
initial censure creates an additional (or a stronger) obligation on the offender to
be law-abiding in the future. So, the offender’s defiance, as manifested in his or
her reoffending, constitutes an additional wrong, which in turn renders the
offender more culpable for the current offense than he or she otherwise would
be.
This rationale is sometimes seen as analogous to some features of the ethics
of parenting. Parents of a rebellious teenage child who persists in misbehaving
after being punished repeatedly might think they are justified in punishing him
more severely the more he persists in acting out—not just because their initial
disciplinary measures failed to deter him, but because the persistence in and of
itself amounts to a blameworthy act of defiance against their parental authority.
Most critics are quick to dismiss this rationale out of hand for the ways in
which it seems to entail authoritarian presumptions about the role of the state
that are incompatible with liberal democratic values.
93
That the rationale is
analogous to the ethics of parenting is taken to be telling in this regard—in a free
country, the government is not supposed to be like a parent and its citizens are
not like children.
There may be some contexts in which punishing defiance seems at least
prima facie defensible, though. For example, take “obstruction of justice”
provisions, in which defiance in certain contexts is criminalized under state and
federal law.
94
One way to commit the federal crime of contempt is
“[d]isobedience or resistance to [a federal court’s] lawful writ, process, order,
rule, decree, or command.”
95
Likewise, there are a variety of state statutes that
92
See, e.g., Markus Dirk Dubber, Policing Possession: The War on Crime and the End of Criminal Law,
91 J. CRIM. L. & CRIMINOLOGY 829, 962 (2001); R. A. DUFF, PUNISHMENT, COMMUNICATION, AND COMMUNITY
168 (2001); George P. Fletcher, The Recidivist Premium, 1 CRIM. JUST. ETHICS 54, 57 (1982).
93
See DUFF, supra note 92.
94
Lee, supra note 10, at 599–601.
95
18 U.S.C.§ 401.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1235
criminalize resisting arrest or the defiance of law enforcement officers in various
other contexts, such as when police officers are trying to control large crowds or
contain riots.
96
As Lee puts it, “the fact that one did not do what one was told to
do is precisely at the heart of this type of criminality.”
97
Legal institutions, including courts and the police, require compliance to be
able to serve the public. And citizens may be morally obligated not to interfere
with institutions that are legitimate and just. So, incentivizing compliance with
law enforcement officers through these kinds of provisions could be justified by
considerations of both efficiency and justice, at least in the abstract.
98
But while there may be some plausible justification of the punishment of
defiance in some obstruction of justice contexts, it is harder to make the case
that the state’s interest in punishing defiance is a defensible underlying reason
to impose a recidivist sentencing premium.
Obstruction of justice provisions are designed to punish acts of defiance
against new legal directives that are created by legitimate legal authorities. These
new directives (whether issued by a judge during court proceedings, or by a
police officer during a routine traffic stop) create new legal (and perhaps moral)
duties. If and when an offender is justifiably charged with contempt of court,
resisting arrest, or other obstruction of justice-related offenses, they are charged
with violating specific and cognizable legal obligations that are distinct from any
other offenses they might have committed.
99
Criminal court judges often exhort first-time offenders not to do anything
that brings them back into the same courtroom again and warn them that the
consequences will be much more severe the second time around.
100
But those
warnings do not create new legal duties to avoid reoffending. Criminal conduct
is already prohibited in general. And merely warning first-time offenders of the
harsher penalties they might face if they reoffend cannot itself justify imposing
those penalties. If it could, then the state would be justified in imposing any kind
of punishment it wanted for any kind of conduct, as long as offenders received
96
Lee, supra note 10, at 599.
97
Id. at 600.
98
For the sake of argument, I want to consider the justification of obstruction of justice provisions
separately from questions about the arbitrary or disparate enforcement of those laws—for example, when people
are charged with the crime of resisting arrest, without being charged with anything else, raising the question of
what they were being arrested for in the first place.
99
See generally CHARLES DOYLE, OBSTRUCTION OF JUSTICE: AN OVERVIEW OF SOME OF THE FEDERAL
STATUTES THAT PROHIBIT INTERFERENCE WITH JUDICIAL, EXECUTIVE, OR LEGISLATIVE ACTIVITIES (2014).
100
See, e.g., Report and Recommendation at 7, United States v. Rigger, No. 3:08-cr-00027-RLJ-HBG
(E.D. Tenn. filed Dec. 30, 2008).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1236 EMORY LAW JOURNAL [Vol. 70:1209
adequate warning of the consequences of their actions—no matter how unjust or
inefficient the punishment in question.
However, there are some situations in which courts do issue distinct,
individualized legal directives to first-time offenders that create new legal
duties, which can then be violated upon a second offense. For example, courts
often sentence offenders to probation, where the terms of probation include new
legal duties not applicable to the general public, many of which look like
directives to organize one’s life to avoid recidivism—for example, prohibitions
on associating with past accomplices,
101
barring the offender from being in or
around the area of the offense or knowingly coming near the victim,
102
completing a number of hours of mandatory community service,
103
being subject
to random drug tests,
104
having proof of employment or enrollment in education
or training programs,
105
completing drug rehabilitation or anger management
therapy,
106
abiding by a curfew or house-arrest arrangement,
107
wearing a GPS
ankle-monitoring bracelet,
108
or simply having to appear for regular meetings
with a probation officer.
109
But there are already sanctions in place for people who violate the kinds of
individualized legal directives that courts issue to first-time offenders. And
recidivist sentencing enhancements apply to people facing entirely new criminal
charges unrelated to parole violations or the like. Therefore, the recidivist
sentencing premium cannot be justified by these kinds of legal directives, or by
the idea that repeat offenders are defiant.
101
See, e.g., 18 U.S.C. § 3563(b)(6) (permitting courts to instruct criminal defendant to “refrain
from . . . associating unnecessarily with specified persons”).
102
See, e.g., id. (permitting courts to instruct criminal defendant to “refrain from frequenting specified
kinds of places”).
103
See, e.g., 18 U.S.C. § 3563(b)(12) (permitting courts to instruct criminal defendant to “work in
community service”).
104
See, e.g., A Study of Drug Testing Practices in Probation, ILL. CRIM. JUST. INFO. AUTH., https://icjia.
illinois.gov/researchhub/articles/a-study-of-drug-testing-practices-in-probation (last visited June 17, 2021).
105
See, e.g., 18 U.S.C. § 3563(b)(4) (permitting courts to instruct criminal defendant to “work
conscientiously at suitable employment or pursue conscientiously a course of study or vocational training that
will equip him for suitable employment”).
106
See, e.g., 18 U.S.C. § 3563(b)(9) (permitting courts to instruct criminal defendant to “undergo available
medical, psychiatric, or psychological treatment, including treatment for drug or alcohol dependency”).
107
See, e.g., Special Conditions of Supervision, GA. DEP’T CMTY. SUPERVISION, https://dcs.georgia.gov/
special-conditions-supervision-0 (last visited June 17, 2021).
108
See, e.g., Electronic Monitoring Program (EMP), CNTY. SANTA CLARA POLICE DEP’T, https://
probation.sccgov.org/adult-services/electronic-monitoring-program (last visited June 17, 2021).
109
See, e.g., Michelle S. Phelps, Mass Probation: Toward a More Robust Theory of State Variation in
Punishment, 19 PUNISHMENT & SOC’Y 53 (2016).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1237
2. Omission
According to Youngjae Lee’s “omission” theory, criminal punishment
changes the relationship between the offender and the state, giving the offender
the duty to rearrange his or her life to stay out of trouble in the future.
110
The
recidivist sentencing premium is justified, on this view, by the repeat offender’s
failure to do so (the “omission”). The omission theory attempts to get around
some of the difficulties identified with the “defiance” view by turning to the
broader circumstances within which criminal offending occurs.
111
Lee emphasizes that crime does not occur in isolation, or in a vacuum—it
occurs in the context of a lifestyle and against the background of many other
choices an offender makes before ultimately committing the crime he or she is
convicted of. According to Lee, “[t]his in turn means that well before individuals
end up committing crimes, they can steer their lives in different directions in
order to minimize the risk of finding themselves in a position in which
committing a criminal offense becomes a compelling—or at least appealing—
option.”
112
Lee notes that one could potentially say something similar about first-time
offenders: They’ve also failed to organize their lives to avoid committing
crime.
113
But the crucial difference between first-time offenders and recidivists,
on Lee’s theory, is that “the repeat offender has gone through a process with the
state that has created a relationship with the state, and the point of that
relationship was to ensure that whatever led the offender to the status of being a
convict should be avoided in the future.”
114
Criminal punishment, according to “expressive” and “communicative”
accounts of its nature, involves a message to the offender that his act was
wrong.
115
“Implicit in that message,” Lee tells us, is the idea that “after his
punishment is complete, he shall not offend again.”
116
This, in turn “should
110
Lee, supra note 10.
111
Christopher Bennett defends an “apology” based rationale in ‘More to Apologise For’: Can We Find
a Basis for the Recidivist Premium in a Communicative Theory of Punishment?, in PREVIOUS CONVICTIONS AT
SENTENCING: THEORETICAL AND APPLIED PERSPECTIVES, supra note 10, at 73. His view is also vulnerable to the
problems identified in this section with Lee’s view.
112
Lee, supra note 10, at 609.
113
Id.
114
Id. at 614.
115
See, e.g., Joel Feinberg, The Expressive Function of Punishment 49 MONIST 397, 403 (1965); DUFF,
supra note 92, at 97; CHRISTOPHER BENNETT, THE APOLOGY RITUAL: A PHILOSOPHICAL THEORY OF
PUNISHMENT 32 (2008).
116
Lee, supra note 10, at 613.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1238 EMORY LAW JOURNAL [Vol. 70:1209
prompt a period of reflection on the part of offenders to determine how they
ended up committing the prohibited act.”
117
Offenders should identify what went
wrong in their life to lead them to crime, according to Lee, and then organize
their lives upon release so that they don’t go down the same path.
118
A repeat offense by someone who has gone through this process of
reflection, diagnosis, and prescription justifies the inference that, for
whatever reason, the prescription was not followed, and the offender
failed to prevent herself from reoffending by failing to organize her
life in a way that steers clear of criminality.
119
So, repeat offenders are guilty not only of their current offense, but also of
violating this additional obligation entailed by their changed relationship with
the state.
The “omission” theory fails to explain the source of this special relationship
between offender and state in a way that could plausibly justify the recidivist
sentencing premium, however. Where, at any stage of criminal procedure—from
arrest through sentencing and corrections—do we see evidence of the kind of
special relationship between the offender and the state that Lee’s account is
premised on?
Parolees and probationers are subject to drug testing, curfews, employment
or education requirements, and many other forms of monitoring. But many of
those convicted of crimes are never on parole or probation.
120
And the sanctions
associated with parole and probation violations are already explicitly
incorporated in criminal codes—so there would be no need for a recidivist
sentencing premium if its only purpose was to punish that kind of conduct.
The only other place in the criminal law where we can find evidence of the
kind of relationship Lee thinks previously convicted offenders have with the
state is in recidivist sentencing enhancements themselves. Recidivist sentencing
enhancements are evidence that, once convicted, lawbreakers have a changed
relationship with the state—one where any future transgressions will be
punished more harshly than if they had not already slipped up. But if the only
117
Id.
118
Id.
119
Id.
120
See Probation and Parole Systems Marked by High Stakes, Missed Opportunities, PEW TRS. (Sept. 25,
2018), https://www.pewtrusts.org/en/research-and-analysis/issue-briefs/2018/09/probation-and-parole-
systems-marked-by-high-stakes-missed-opportunities#:~:text=Nationwide%2C%204.5%20million%20people
%20are,federal%20prisons%20and%20local%20jails (noting the difference between people on probation or
parole and the incarcerated population).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1239
evidence of the offender’s special relationship with the state is our recidivist
sentencing enhancements themselves, then we have nothing upon which to
justify those enhancements. The mere fact that we have chosen one legal
arrangement over another is not sufficient reason to think that what we have
chosen is good.
However, it is possible to interpret the idea that criminal punishment changes
the relationship between offender and state in the way Lee suggests as a purely
conceptual premise, rather than an empirical or doctrinal one. On this reading,
the changed relationship between offender and state is supposed to follow from
a feature of criminal punishment that we detect analytically—namely, its
“communicative” or “expressive” function—rather than from any doctrine in the
criminal law or from observations of our criminal justice systems in action.
121
The communicative or expressive part of criminal punishment contains an
implicit message to organize one’s life so as to avoid reoffending, at least on this
interpretation of Lee’s view.
There is no evidence, conceptual or otherwise, that this implicit message is
necessarily part of what is expressed or communicated by criminal punishment,
though. Punishment is said to be distinct from other forms of state-imposed
deprivation of rights or liberty, such as preventive detention or quarantine, by
the fact (among other things) that it communicates or expresses moral
condemnation of, or blame for, the crime in question.
122
But condemning a
crime, or blaming the offender for committing that crime, does not entail passing
judgment on the prior courses of action that led the offender to it. And we should
be very hesitant to rationalize the wide range of policies and judicial decisions
that fall under the umbrella of the recidivist sentencing premium on the basis of
scant evidence (if any) that there is some implicit message contained in the very
concept of “punishment,” yet nowhere to be found in the substantive criminal
law nor observable at any stage of criminal adjudication.
121
It is possible to read expressive and communicative theories about the nature of punishment (rather
than the justification of punishment) as either conceptual or empirical. For example, Feinberg, in his classic
article, argues that “[p]unishment is a conventional device for the expression of attitudes of resentment and
indignation, and of judgments of disapproval and reprobation[.]” Feinberg, supra note 115, at 400. A
“conventional device” might refer to the actual practices that we call “punishment” in the particular kind of
contingently arranged society we happen to have. But as Feinberg appears to intend, the “conventional device”
in question is the concept “punishment” itself, which has come (by convention) to mean the kind of practice,
that, among other things, expresses moral condemnation of a criminalized action. Id. at 401–02. For a defense
of this analysis of “punishment,” see DAVID BOONIN, THE PROBLEM OF PUNISHMENT 3–28 (2011).
122
See, e.g., Feinberg, supra note 115, at 401–02; DUFF, supra note 92.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1240 EMORY LAW JOURNAL [Vol. 70:1209
We should be skeptical, as such, of the premise that people with prior
criminal convictions stand in a special relationship to the state—apart from any
probation, parole, or suspended sentence arrangement they might have with the
criminal courts, and apart from the fact that they are subject to the recidivist
sentencing premium. Since the purported special relationship is supposed to
justify the recidivist sentencing premium, the two cannot be identical. And even
if that premise were true, it would not follow that the recidivist sentencing
premium is justified. As Lee acknowledges himself, the kinds of legal and
extralegal barriers that ex-offenders face to reentering society, and to becoming
law-abiding, productive citizens, “are inconsistent with the system’s demand
that offenders set their lives straight after going through the process of
conviction and punishment.”
123
And he concedes that “the size of the recidivist
premium should reflect the ways in which each party to that relationship has
failed.”
124
As I will argue in the remainder of this paper, these barriers not only
undermine the rationale for the recidivist sentencing premium, but also give us
reason to treat prior convictions as a presumptive mitigative factor at
sentencing—what I call the recidivist sentencing discount.
III. T
HE PARADOX
Criminal convictions have severe and lasting civil, social, and economic
consequences apart from the terms of formal sentences imposed by criminal
courts.
125
People who have been convicted of a crime in the past face huge
barriers to finding legitimate employment, becoming self-sufficient, and earning
the esteem of others when they attempt to reenter the community. These
“collateral consequences” of punishment give people with prior convictions
much stronger incentives to commit crime in the future than they would
otherwise have.
126
123
Lee, supra note 10, at 620.
124
Id. at 620.
125
See generally DEVAH PAGER, MARKED: RACE, CRIME, AND FINDING WORK IN AN ERA OF MASS
INCARCERATION (2009); MARGARET COLGATE LOVE, COLLATERAL CONSEQUENCES OF CRIMINAL
CONVICTIONS: LAW, POLICY AND PRACTICE (2013); JOAN PETERSILIA, WHEN PRISONERS COME HOME: PAROLE
AND PRISONER REENTRY 105–38 (2003). More limited contact with criminal justice systems can often have
seriously damaging effects as well. The consequences of being arrested or charged with a crime—without being
convicted, and even when those charges are dismissed—can sometimes be devastating. See, e.g., Eisha Jain,
Arrests as Regulation, 67 STAN. L. REV. 809, 820 (2015). I focus on the consequences of criminal convictions
because of the way that prior convictions are used at sentencing.
126
Joshua Kaiser, among others, argues that the distinction between “direct” and “collateral”
consequences of punishment is both doctrinally and normatively suspect. See Joshua Kaiser, Revealing the
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1241
As I show below, because of these incentives, we are left with comparatively
less evidence of ill will when people with prior convictions return to crime than
when people commit otherwise similar crimes for the first time. Judges and
sentencing commissions, as such, have moral reason to treat prior convictions as
a presumptive mitigating factor—precisely the opposite of what human societies
have done for millennia. The conclusion I defend is politically unpalatable, but
it is morally unavoidable. This is the paradox of recidivism.
A. Barriers to Reentry and the Causes of Recidivism
The argument that leads us to this paradox is partly normative and partly
empirical. Before turning to the normative core of the argument, I lay out the
wide range of barriers that people with prior criminal convictions face to finding
employment, beginning careers, and getting welfare, housing, education, and
other basic goods.
1. Stigma for Job Applicants
Future employment and earning are the strongest predictors of future
crime.
127
Employers are much less likely to hire a job applicant who they know
or suspect to have a criminal record than one with a clean record because they
believe that record to be a mark of untrustworthiness.
128
Criminal convictions
for even a single low-level, nonviolent offense can have a devastating effect on
one’s employment prospects.
129
This holds true regardless of whether the
conviction is for a drug crime or a property crime, and regardless of any other
characteristics of the potential job applicants or employers.
130
Hidden Sentence: How to Add Transparency, Legitimacy, and Purpose to “Collateral” Punishment Policy, 10
HARV. L. & POL’Y REV. 123, 130–31 (2016) [hereinafter Kaiser, Revealing the Hidden Sentence]; Joshua Kaiser,
We Know It When We See It: The Tenuous Line Between Direct Punishment and Collateral Consequences, 59
HOW. L.J. 341, 343 (2016). So, I use the term “collateral consequences” with some hesitation, albeit in line with
the current scholarly literature and legal doctrine.
127
See generally Christopher Uggen, Work as a Turning Point in the Life Course of Criminals: A Duration
Model of Age, Employment, and Recidivism, 65 AM. SOC. REV. 529 (2000); Bruce Western, The Impact of
Incarceration on Earning, 67 AM. SOC. REV. 526 (2002).
128
See generally Amanda Agan & Sonja Starr, The Effect of Criminal Records on Access to Employment,
107 AM. ECON. REV. 560 (2017); JOHN SCHMITT & KRIS WARNER, EX-OFFENDERS AND THE LABOR MARKET
(2010).
129
Christopher Uggen, Mike Vuolo, Sarah Lageson & Ebony Ruhland, The Edge of Stigma: An
Experimental Audit of the Effects of Low-Level Criminal Records on Employment, 52 CRIMINOLOGY 627, 627
(2014); Agan & Star, supra note 128, at 560.
130
Agan & Starr, supra note 128, at 560.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1242 EMORY LAW JOURNAL [Vol. 70:1209
These effects are stronger for Blacks than for whites. Prior criminal
convictions reduce the likelihood of a call-back by 50% for white job applicants,
and 64% for Black applicants.
131
Interestingly, this disparity may be exacerbated
when prospective employers’ access to information about applicants’ criminal
records is limited by “Ban the Box” legislation.
132
In every U.S. state, the stigma of a criminal conviction is both legally
facilitated and legally enforced.
133
Prospective employers have easy access to
job applicants’ criminal record information, often including arrest records for
which criminal charges were either dismissed or never filed.
134
And every state
has statutory restrictions or bans on public sector employment, occupational
licenses, and entire professional fields.
135
Thirty-three states, as well as
Washington, D.C., impose legal restrictions on public sector employment for ex-
convicts.
136
Six of those states do not allow people with felony convictions to
take any kind of public employment whatsoever.
137
Every state restricts
occupational licenses for individuals with prior convictions; and these
restrictions have become more stringent over time.
138
Some of these restrictions have obvious rationales—for example, barring sex
offenders whose victims were underage from working in K–12 schools and day-
care centers. But many obstacles to employment for the formerly incarcerated
have little connection to profession-specific risk factors. For example, ex-
offenders in some states are barred from employment as septic tank cleaners,
embalmers, billiard room staff, real estate agents, plumbers, eyeglass dispensers,
and barbers.
139
And it isn’t as if the categorical bars on employment in these
professions are limited to those convicted of crimes involving scissors, toilet
plungers, or pool sticks.
131
PAGER, supra note 125, at 67, 69. But see Agan & Starr, supra note 128, at 561 (finding a stronger
effect for whites than Blacks).
132
Agan & Starr, supra note 16, at 195.
133
NAT’L INVENTORY OF COLLATERAL CONSEQUENCES OF CONVICTION, https://niccc.
nationalreentryresourcecenter.org/#about (last visited June 17, 2021) (“Collateral consequences are legal and
regulatory restrictions that limit or prohibit people convicted of crimes from accessing employment, business
and occupational licensing, housing, voting, education, and other rights, benefits, and opportunities.”).
134
Jain, supra note 125, at 824–25, 862.
135
Consequences, NAT’L INVENTORY OF COLLATERAL CONSEQUENCES OF CONVICTION, https://niccc.
nationalreentryresourcecenter.org/consequences (last visited June 17, 2021) (cataloguing employment-based
collateral consequences of conviction in every state and the District of Columbia).
136
Kathleen M. Olivares, Velmer S. Burton & Francis T. Cullen, The Collateral Consequences of a Felony
Conviction: A National Study of State Legal Codes 10 Years Later, 60 FED. PROBATION 10, 13 (1996).
137
Id.
138
PETERSILIA, supra note 125, at 113–15.
139
Elena Saxonhouse, Unequal Protection: Comparing Former Felons’ Challenges to
Disenfranchisement and Employment Discrimination, 56 STAN. L. REV. 1597, 1613 (2004).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1243
These restrictions have a disparate impact on African Americans, who are
disproportionately represented among the prison population, and who are
otherwise more likely to find work in the restricted occupations—another
example of how the stigma of incarceration has a disproportionate impact on the
Black community.
140
2. Interrupting the Life Course
Not everyone who is convicted of a crime serves time in prison. But for those
who do, periods of incarceration can undermine the development of human and
social capital that prisoners might otherwise have enjoyed during late
adolescence and early adulthood.
141
People (mostly men) entering prison for the first time are typically around
the same age as college students or those first entering the workforce.
142
Young
people outside of prison build most of their human capital during these years—
either through higher education or through apprenticeships and on-the-job
training. Most young, male, full-time workers enjoy steady income increases
throughout their 20s and 30s.
143
But for those convicted of and incarcerated for
a first offense, much of this time is spent behind bars.
Incarceration at a young age often permanently severs the transition from
adolescence to stable careers, as such. Upon being released, former convicts are
often unable to secure jobs with opportunities for advancement, if they can find
work at all, since private sector employers often prefer to hire younger job
applicants for those positions.
144
Finding employment after serving time in
prison is difficult, and those who do succeed tend to be relegated to jobs with
little chance of climbing a career ladder.
145
140
PAGER, supra note 125, at 33–34.
141
Bruce Western, Jeffrey R. Kling & David F. Weiman, The Labor Market Consequences of
Incarceration, 47 CRIME & DELINQ. 410, 413 (2001); Joel Waldfogel, The Effect of Criminal Conviction on
Income and the Trust “Reposed in the Workmen”, 29 J. HUM. RES. 62, 66–67 (1994).
142
Western, supra note 127, at 528.
143
Id.
144
Id. at 529; HARRY J. HOLZER, WHAT EMPLOYERS WANT: JOB PROSPECTS FOR LESS-EDUCATED
WORKERS (1996).
145
Daniel Nagin & Joel Waldfogel, The Effect of Conviction on Income Through the Life Cycle, 18 INT’L
REV. L. & ECON. 25 (1998).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1244 EMORY LAW JOURNAL [Vol. 70:1209
3. Welfare, Housing, and Other Goods
There are over 1,000 statutory restrictions in state and federal law on ex-
offenders’ access to public benefits.
146
These restrictions are especially stringent
for people found guilty of drug offenses.
147
People convicted of drug distribution
are ineligible for all federal benefits for five years after a first conviction, for ten
years after a second conviction, and for life after a third conviction.
148
The
Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (the
Clinton Administration’s “welfare reform”) imposed a lifetime ban on anyone
convicted of a drug-related felony from the Temporary Assistance to Needy
Families (TANF) and the Supplemental Nutrition Assistance Program (SNAP)
program (“food stamps”).
149
These bans increased recidivism.
150
And although
all but two states have opted out of the drug felony provision,
151
many have
imposed modified versions of the ban mandating people with felony convictions
to participate in drug testing, treatment, or rehabilitation to be eligible for
benefits.
152
“In most states with modified bans, failing to comply with drug
treatment or testing means the loss of benefits entirely.”
153
Close to 1,000 other sanctions restrict ex-offenders’ access to both public
and private housing.
154
Public housing agencies and owners of federally assisted
housing are required to exclude households with a member who they think is
illegally using a controlled substance or abusing alcohol.
155
Changes in tort
doctrine and public nuisance laws over the last fifty years have also incentivized
(and in some cases required) landlords to conduct criminal background checks
on prospective tenants.
156
For example, in 1970, a federal appeals court ruled in
146
Consequences, NAT’L INVENTORY OF COLLATERAL CONSEQUENCES OF CONVICTION, https://niccc.
nationalreentryresourcecenter.org/consequences (last visited June 17, 2021) (cataloguing 1,207 collateral
consequences of conviction relating to public benefits).
147
21 U.S.C. § 862.
148
Id.
149
Personal Responsibility and Work Opportunity Reconciliation Act, Pub. L. No. 104-193, 110 Stat.
2105, 2180 (1996).
150
See, e.g., Cody Tuttle, Snapping Back: Food Stamp Bans and Criminal Recidivism, 11 AM. ECON. J.:
ECON. POL’Y 301 (2019).
151
Brittany T. Martin & Sarah K.S. Shannon, State Variation in the Drug Felony Lifetime Ban on
Temporary Assistance for Needy Families: Why the Modified Ban Matters, 22 PUNISHMENT & SOC’Y 439, 458
n.6 (2020).
152
Id. at 443–45.
153
Id. at 444.
154
Consequences, NAT’L INVENTORY OF COLLATERAL CONSEQUENCES OF CONVICTION, https://niccc.
nationalreentryresourcecenter.org/consequences (last visited June 17, 2021) (cataloguing seventy-two collateral
consequences of conviction relating to public benefits).
155
42 U.S.C § 13661.
156
Barbara Glesner, Landlords as Cops: Tort, Nuisance and Forfeiture Standards Imposing Liability on
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1245
favor of a tenant who sued her landlord for damages after she was robbed in the
hallway of her apartment building.
157
This set a precedent that shifted
responsibility for public safety toward landlords, who responded in part by
screening prospective tenants to avoid future premise liability suits.
158
Similarly, people convicted of drug offenses are disqualified from a wide
range of federal financial aid programs for higher education, including grants,
loans, and work assistance.
159
For some years, the Anti-Drug Abuse Act allowed
judges to suspend eligibility for federal financial aid as part of a criminal
sentence.
160
The list does not stop here: prior convictions often bar people from obtaining
driver’s licenses, business licenses, credit union memberships, service contracts
with government agencies, and car insurance policies.
161
In short, having a
criminal conviction on one’s record makes it very hard to get things that just
about anyone would want.
Incarceration (along with other types of contact with criminal justice
systems) also exacerbates many of the preexisting mental health problems that
inmates bring to prison and can also cause new ones.
162
There are a number of other serious collateral consequences of criminal
convictions—perhaps most notably, disenfranchisement.
163
I focus on the
collateral consequences of criminal convictions canvassed above, however,
because they incentivize those who face them to commit future crimes in a
particular way, which is central to the normative argument below.
Landlords for Crime on the Premises, 42 CASE W. RES. L. REV. 679 (1992).
157
Kline v. 1500 Mass. Ave. Apartment Corp., 439 F.2d 477 (D.C. Cir. 1970).
158
David Thatcher, The Rise of Criminal Background Screening in Rental Housing, 33 LAW & SOC.
INQUIRY 5 (2008).
159
See, e.g., Higher Education Amendments of 1998, Pub. L. No. 105-244, § 483(f)(1), 112 Stat. 1581,
1736 (1998).
160
21 U.S.C. § 862; Anti-Drug Abuse Act of 1988, Pub. L. No. 100-690, § 5301, § 115, 102 Stat. 4181,
4310.
161
See, e.g., 12 U.S.C. §§ 1772d, 1786; 15 U.S.C. § 78c(39)(F); CONN. GEN. STAT. § 1-57b (2014); CAL.
INS. CODE § 11629.73 (2015).
162
Naomi Sugie & Kristin Turney, Beyond Incarceration: Criminal Justice Contact and Mental Health,
82 AM. SOC. REV. 719, 722–23 (2017).
163
Consequences, NAT’L INVENTORY OF COLLATERAL CONSEQUENCES OF CONVICTION, https://niccc.
nationalreentryresourcecenter.org/consequences (last visited June 17, 2021) (cataloguing seventy-two collateral
consequences of conviction relating to public benefits).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1246 EMORY LAW JOURNAL [Vol. 70:1209
B. Inequality, Incentives, and Criminal Responsibility
Let us now turn to the normative core of the argument for the paradox of
recidivism. Below, I argue that we cannot justifiably blame or punish repeat
offenders for breaking the law as harshly as we would if they didn’t face such
strong incentives to reoffend. Therefore, judges and sentencing commissions
have reason in principle to treat prior convictions as a presumptive mitigating
factor—giving repeat offenders a recidivist sentencing discount.
1. Moral Blame and Legal Punishment
As Judge Thurman Arnold once said, “[o]ur collective conscience does not
allow punishment where it cannot impose blame.”
164
This is true, according to
prevailing views, because criminal punishment has an “expressive” or
“communicative” function—and (one of) the things that it expresses or
communicates is blame.
165
So, if punishment automatically comes with blame,
then the blame must be justified for the punishment to be justified.
166
Consider some of the basic phenomenological features of interpersonal
moral blame.
167
Blame is often a response to the perception that someone else
164
Holloway v. United States, 148 F.2d 665, 666–67 (D.C. Cir. 1945).
165
Feinberg put it this way: “[p]unishment is a conventional device for the expression of attitudes of
resentment and indignation, and of judgments of disapproval and reprobation, either on the part of the punishing
authority himself or of those ‘in whose name’ the punishment is inflicted.” Feinberg, supra note 115, at 400.
Feinberg does not use the term “blame,” but the dominant view about the psychological nature of blame is that
it is constituted by the same “attitudes of resentment and indignation” and “judgments of disapproval and
reprobation” he says that punishment expresses. See, e.g., P.F. Strawson, Freedom and Resentment, 48 PROC.
BRIT. ACAD. 1 (1962); R. JAY WALLACE, RESPONSIBILITY AND THE MORAL SENTIMENTS (1994); cf. T.M.
SCANLON, MORAL DIMENSIONS: PERMISSIBILITY, MEANING, AND BLAME (2008); GEORGE SHER, IN PRAISE OF
BLAME (2006).
166
The underlying justification for blame or punishment need not be deontological. See, e.g., HART, supra
note 18, at 25; PAUL ROBINSON, DISTRIBUTIVE PRINCIPLES OF CRIMINAL LAW: WHO SHOULD BE PUNISHED HOW
MUCH 175–212 (2008); Charles Fried, Moral Causation, 77 HARV. L. REV. 1258, 1268 (1964).
I do not endorse this standard view of the conceptual relationship between moral blame and legal punishment.
But I do believe that examining the normative principles that govern our ordinary interpersonal practice of moral
blame can nonetheless provide some useful heuristics for understanding the justification of state-imposed
punishment under criminal law. My own view, which I defend elsewhere, is that legal punishment does not
necessarily express or communicate interpersonal moral blame. But comparative sentencing severity should
nonetheless track the relative quantum of ill will manifested by any given crime, because that is the fair way to
distribute the burdens of upholding a deterrent system of criminal justice between those who violate the rules of
that system. For the purposes of this Article, the difference between my own view and the prevailing
“expressive” and “communicative” views is of little matter—they both entail the same conclusions with respect
to the issues currently at hand.
167
The concept of “blame” is also sometimes used to express a merely causal relationship between
events—for example, I might say I “blame” the fact that my car’s engine has a rough idle on a failed spark plug.
But this Article is concerned with the nature and norms of moral blame in particular, which is what is relevant
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1247
has done something morally wrong. But we also blame ourselves for things
we’ve done. We blame people for doing the right thing for the wrong reasons.
168
And we blame people for things we think they believe or feel, not just for things
they do.
169
So blame is different from a judgment of wrongdoing: it is a response
to the perception of “ill will”—a morally objectionable attitude or motivation.
170
Because blame is a response to a perception of ill will, we must have good
evidence of ill will to be justified in blaming.
171
The severity with which we are
justified in blaming or punishing one another, as such, depends on our
assessment of one another’s “quality of will.”
172
The quality of will that a crime manifests depends upon the facts about what
the offender’s mind state and motivations were at the time of the crime
(assuming he actually committed it). But there is no mechanism through which
criminal justice officials can directly access that information.
173
The evidence
we have for attributing motives and mental states to one another is inevitably
circumstantial.
174
So the best we can do is to assess the possible motives that a
defendant may have had for committing an offense, and to assess the likelihood
that they acted on one or more of those motives.
175
Below, I show that we have less evidence of ill will when people with prior
convictions reoffend than we would if they were appearing in criminal court for
the first time. This entails the paradox of recidivism: that judges and sentencing
commissions ought to treat prior convictions as a presumptive mitigating factor,
as politically unpalatable as that might be.
to the ethics of legal punishment.
168
Some accounts of action-individuation in moral philosophy entail that “doing the right thing for the
wrong reasons” is not even possible—if, for example, the moral rightness and wrongness of our actions are
determined at least in part by our reasons for acting. See, e.g., Steven Sverdlik, Motive and Rightness, 106 ETHICS
327 (1996). But I take this phenomenon to be part of our normal understanding of our everyday moral landscape.
169
It is possible to conceptualize both belief and feelings or emotions as kinds of action. But I take it that
the distinction between actions and beliefs or feelings is, again, part of our normal phenomenological landscape.
170
There is some disagreement about what kind of response to ill will blame might be. I do not take a
position here in that debate.
171
Christopher Lewis, Incentives, Inequality, Criminality, and Blame, 22 LEGAL THEORY 153, 157–59
(2016).
172
I borrow this term from Strawson, supra note 165, at 14.
173
Lewis, supra note 171, at 168.
174
Id.
175
Id. at 165–70.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1248 EMORY LAW JOURNAL [Vol. 70:1209
2. Kinds of Incentives
Only certain kinds of incentives to commit crime—those associated with the
objective benefits or payoffs of breaking the law in question—warrant a
mitigated punishment. Consider the following cases:
Boardwalk Dreams: Thimble and Shoe are twin brothers who grew up
in the same working-class household on Baltic Avenue in Atlantic
City. They have the same job at the Water Works, the same income
($200 a week), and roughly equivalent net assets (around $1,500
worth). Both of them have dreams of moving up the career ladder and
buying a hotel on Boardwalk. For Thimble, that dream is almost an
obsession—one he yearns to realize by any means necessary. For
Shoe, the dream is much less all-consuming; he’d settle for a cozy
house on St. James Place if the only way to buy the hotel on Boardwalk
were truly unsavory. Horse, a local “wise guy,” tips the twins off to an
elaborate scheme to rob an armored truck. Shoe is worried that
someone might get hurt in the process. Thimble could not care less.
There is a sense in which Thimble has a stronger incentive to rob the local
bank than Shoe here—the payoffs of committing (and getting away with) the
crime would be greater for Thimble, given his personality and values. But the
fact that Thimble is greedy cannot be a justification or an excuse for the crime
in itself. If anything, committing a crime out of greed, like committing crime out
of hatred or malice, gives us a reason to more severely blame and punish the
offender. And we have little way of knowing the extent to which these
“subjective” incentives play into an offender’s motivation for any given crime
anyway. Consider a second case:
Park Place: Iron and Boot are best friends who grew up on
Mediterranean Avenue, around the corner from Thimble and Shoe, and
who work alongside the twins at Water Works. Like Thimble, they
both dream, almost obsessively, about owning their own hotels. Horse
approaches them with plans for another armored car heist. Iron thinks
the plan is bulletproof and is eager to make a quick buck—his uncle is
the police commissioner, and he figures that his family ties give him a
degree of impunity in their small city. Boot is more skeptical that the
plan will succeed and has no ties to law enforcement or the legal
profession.
There is a sense in which Iron has a stronger incentive to take part in the
heist than Boot here. He thinks that they have a lower chance of getting caught
than Boot does, and he thinks that if they do get caught, his uncle might be able
to protect him from being punished. But this has no bearing on how severely we
should blame or punish either of the two friends if they do go ahead with the
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1249
heist. The fact that Iron thought that they could get away with it, or figured that
he wouldn’t get punished even if he did get caught, does not excuse or justify
the crime even partially. So, incentives related to the expected costs of
committing a crime should not figure into how we calculate our moral response
to crime either.
Incentives related to the objective benefits or payoffs of committing crime
are a different story. Consider a third case that illustrates this:
Pennsylvania Avenue: On an otherwise quiet Tuesday night on leafy
Pennsylvania Avenue, two elderly couples are robbed at gunpoint
under more or less identical circumstances. Their assailants escape
with a few hundred dollars each. Both couples are well-heeled, so the
financial loss is not nearly as significant to them as the trauma of the
robbery itself. One of the stickup artists is Wheelbarrow, a day laborer
in the construction trade who has been unable to find work over the
last several months, has no savings or assets, and has been sleeping in
his broken-down van in the Free Parking lot. The other is Top Hat, a
mid-career law professor who makes a six-figure salary and lives in a
newly renovated mid-century Colonial in sunny Marvin Gardens.
Neither one of the robbers is mentally ill or incompetent; nor does
either present any special mitigating circumstances to explain why
they committed the crime.
Without knowing anything else about either of the two robbers, it may seem
intuitive that Wheelbarrow had a much stronger incentive to commit the crime
than Top Hat. After all, it is hard to imagine a well-off, mid-career law professor
committing an armed robbery. When we do, the first association that is likely to
come to mind is mental illness. But we have already ruled out mental illness as
a motivation for, or cause of, the crime here by stipulation. What other
explanation can we come up with—what kind of motive can we impute to the
rogue Professor Top Hat?
No obvious story is even mildly exculpatory. Perhaps Top Hat did it for the
thrill; maybe he even did it for the money, despite being financially secure.
Either way, absent some special explanation, it seems he must have weighted an
extremely trivial interest of his own over the much more fundamental interests
of his victims in their property rights, their right to bodily integrity, and their
right to be free of harm, trauma, and the risk of those things. In order to place
such disproportionately strong weight on his own interests relative to others’,
Top Hat must have some extremely loathsome moral attitudes—callousness,
cruelty, or lack of empathy.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1250 EMORY LAW JOURNAL [Vol. 70:1209
It is much easier to imagine an irregularly employed day laborer committing
an armed robbery. Some of this may have to do with the implicit psychological
association of crime with disfavored groups.
176
But much of it likely has to do
with it being easier to give reasons why someone in those straits would do such
a thing. He has an immediate and strong incentive to commit the robbery—he
has no money and is afraid of being homeless. Now it may be that this doesn’t
give him an all-things-considered moral justification for the offense; perhaps
there are other things he could have done within the bounds of the law to keep a
roof over his head; or perhaps his interest in having a roof over his head just
doesn’t outweigh his victims’ interests in being left alone. But we need not
impute any extreme immoral attitude to Wheelbarrow in order to understand
why he likely committed the crime. He might not have been sufficiently caring
or respectful of his victim’s rights, but there is little reason to think he has the
kinds of morally abhorrent attitudes we are likely to impute to Top Hat.
3. Measuring Incentives
The difference between Top Hat’s incentives to rob the elderly couple, and
Wheelbarrow’s, has to do with the objective benefits that each of the two
associates with the crime in question. It does not stem from their respective
personalities, values, or assessments of the risks or expected costs of committing
that crime.
But the objective benefits of that crime also seem to be identical—at least in
some sense. Both Top Hat and Wheelbarrow rob elderly couples under similar
circumstances in a similar area. And both get away with a few hundred dollars.
Yet it also makes sense to think that the objective benefit or payoff to the
crime is exponentially greater for Wheelbarrow than it is for Top Hat. After all,
a few hundred extra dollars can make a huge difference in the life of someone
who is destitute and without realistic hopes or plans to escape poverty; while it
would make little objective difference in the life of a well-off, mid-career law
professor on a six-figure salary.
One way to explain the difference in strength between Top Hat and
Wheelbarrow’s incentives to commit the crime in question would be to appeal
to the diminishing marginal utility of income and wealth. In general, the more
income and wealth people have, the less any additional financial gains contribute
176
See, e.g., Ted Chiricos, Kelly Welch & Marc Gertz, Racial Typification of Crime and Support for
Punitive Measures, 42 CRIMINOLOGY 359, 363–64 (2004).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1251
to their subjective sense of wellbeing.
177
We could stipulate that Top Hat and
Wheelbarrow have identical utility curves with respect to income and wealth,
and so Wheelbarrow is likely to realize greater subjective utility from robbing
the bank than Top Hat.
But this explanation alone gives us no reason to think that we should blame
or punish Wheelbarrow any less than Top Hat. For if Top Hat has more of a
greedy personality than Wheelbarrow—and thus a different utility curve—then
he stands to gain more subjective utility from the robbery, even if we would
expect most people in his circumstances not to.
Another way to quantify the difference between Wheelbarrow and Top Hat’s
incentives to rob the elderly couples would be to take a view about what is
objectively good for people, and then show how the crime would bring about a
greater quantum of good for Wheelbarrow than it would for Top Hat. But there
is no clear consensus on the true nature of wellbeing.
178
And there may be good
reasons for governments and government officials to remain neutral between
competing conceptions of the good.
179
This kind of neutrality between
conceptions of the good is likely to be especially important for the judiciary, and
perhaps doubly so in the context of criminal courts. Judges could render wildly
different sentencing decisions depending on their own views about the good life
or the nature of wellbeing.
180
So political liberalism—state neutrality between
competing conceptions of the good—is likely to go hand in hand with the rule
of law in the context of criminal adjudication.
181
Luckily, there are at least two plausible ways to explain the difference in
strength between Top Hat and Wheelbarrow’s incentives to commit the crime
here, without appealing to subjective utility or relying on a theory of objective
wellbeing. First, we could measure the strength of their incentives in terms of
what Rawls called the “primary goods”—things that anybody would want,
regardless of whatever else they wanted—including freedom of thought, liberty
177
Ed Diener & Robert Biswas-Diener, Will Money Increase Subjective Well-Being? A Literature Review
and Guide to Needed Research, 57 SOC. INDICATORS RSCH. 119, 145 (2002).
178
See, e.g., RICHARD KRAUT, AGAINST ABSOLUTE GOODNESS (2011); DANIEL HAUSMAN, VALUING
HEALTH: WELLBEING, FREEDOM, AND SUFFERING (2015); cf. ANNA ALEXANDROVA, A PHILOSOPHY FOR THE
SCIENCE OF WELL-BEING (2017).
179
See, e.g., JOHN RAWLS, POLITICAL LIBERALISM (1993); CHARLES LARMORE, PATTERNS OF MORAL
COMPLEXITY (1987); RONALD DWORKIN, TAKING RIGHTS SERIOUSLY (1977); cf. JOSEPH RAZ, THE MORALITY
OF FREEDOM (1986).
180
Richard A. Posner, The Role of the Judge in the Twenty-First Century, 86 B.U. L. REV. 1049, 1066
(2006).
181
Richard C. Sinopoli, Liberalism and Contested Conceptions of the Good: The Limits of Neutrality, 55
J. POL. 644, 646 (1993).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1252 EMORY LAW JOURNAL [Vol. 70:1209
of conscience, freedom of movement, free choice of occupation, income and
wealth as means for achieving a wide range of other ends, and, importantly, “the
social bases of self-respect.”
182
Second, we could also measure the strength of
those incentives in terms of what Amartya Sen and Martha Nussbaum called the
“capabilities approach” in development economics.
183
On this metric, we could
measure the relative strength of Top Hat and Wheelbarrow’s incentives
according to the likelihood that committing the crime would enhance their
respective opportunities to live a life they have reason to value.
I am agnostic between these two approaches. Both give judges the normative
basis for rendering consistent sentencing decisions, regardless of their own
views about the nature of the good life or the basis of human wellbeing. And
when we measure the strength of one’s incentives to commit any given crime in
terms of the capabilities-related (or primary good-related) payoffs associated
with that crime, we should be able to see why having a stronger incentive to
break the law can mitigate the extent to which one is liable to moral blame and
criminal punishment.
4. Evidence of Ill Will
Leaving aside “victimless” crimes and crimes against the self, paradigmatic
cases of malum in se crime generally pit the interests of a wrongdoer against the
interests of one or more victims. An offender who acts knowingly or
intentionally, rather than recklessly or negligently, takes their own interests in
committing the crime—whatever they might be—to outweigh the victim’s
interests in being free from harm, or having their rights violated.
184
If the offender has weaker incentives to commit the crime—measured in
terms of primary goods or capabilities—they need to weigh their own interests
more heavily relative to the interests of the victim in order to go through with
the crime. Someone with strong incentives to commit the crime need not place
as heavy a weight on their own interests relative to the interests of the victim,
compared to an offender whose incentives are weaker.
The way that an offender would have to weigh his own interests relative to
the interests of his victim gives us some information about how we ought to
182
JOHN RAWLS, JUSTICE AS FAIRNESS: A RESTATEMENT 58–61 (Erin Kelly ed., 2001).
183
See, e.g., MARTHA NUSSBAUM, CREATING CAPABILITIES: THE HUMAN DEVELOPMENT APPROACH
(2011); AMARTYA SEN, THE IDEA OF JUSTICE 225–91 (2009).
184
He may also take his interests to outweigh some collective interests of the society at large, but we can
set that to the side here for the sake of simplicity.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1253
evaluate his motives for the crime. In Pennsylvania Avenue, we have less
evidence of ill will in Wheelbarrow’s case than we do in Top Hat’s case. Because
Wheelbarrow had much stronger incentives to commit the robbery than Top Hat,
we need not impute the same degree of ill will to his crime that we do to Top
Hat’s.
It is possible, however unlikely, that Wheelbarrow could have committed
the robbery out of the same kind of extremely unsavory motives that we initially
imputed to Top Hat.
185
For example, Wheelbarrow may have placed no weight
on the fact that he was homeless in deciding to rob the elderly couple. He might
instead have been motivated by thrill-seeking, or the prospect of seizing
momentary power and control over someone else, or a sadistic urge to scare the
wits out of a pair of old people.
This kind of far-fetched possibility does not undermine the main import of
the account given above. When we seek to evaluate the motives of wrongdoers
in a criminal context, the kind of evidence we have available is inevitably
circumstantial and imperfect.
186
We have no direct mechanism through which
we can, with complete certainty, find out why people commit crime. There is no
way, for example, to use fMRI technology or other neuroscience techniques to
retroactively find out whether someone committed an armed robbery for the
thrill of it, or out of desperation.
Perhaps the most natural way to proceed, given the circumstantial nature of
this evidence, would be to simply assess the range of possible motives that an
offender might have had, and evaluate the likelihood that he or she might have
acted on one or more of those motives. But we might add some normative
constraints to this inquiry as well, as Gideon Yaffe and Alex Sarch argue we
ought to.
187
Yaffe and Sarch separately defend variants of what they call the “principle
of lenity.” In Yaffe’s formulation, the principle of lenity “requires us to
determine what the defendant’s conduct says about his [motives] under the
assumption that he is as little different from the law-abiding citizen as possible,
given his behaviour.”
188
185
But see, e.g., Peter Chau, Temptation, Social Deprivation and Punishment, 30 OXFORD J. LEGAL STUD.
775 (2010) (arguing against this logic).
186
See, e.g., James Morsch, The Problem of Motive in Hate Crimes: The Argument Against Presumptions
of Racial Motivation, 82 J. CRIM. L. & CRIMINOLOGY 659, 661 (1992).
187
GIDEON YAFFE, THE AGE OF CULPABILITY: CHILDREN AND THE NATURE OF CRIMINAL RESPONSIBILITY
(2018); ALEX SARCH, CRIMINALLY IGNORANT: WHY THE LAW PRETENDS WE KNOW WHAT WE DON’T (2019).
188
YAFFE, supra note 187, at 89.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1254 EMORY LAW JOURNAL [Vol. 70:1209
In Sarch’s formulation, under the principle of lenity, “D’s action, A, only
manifests the least amount of insufficient regard for legally protected interests
or values (i.e., the least amount of error in weighing the legal recognized reasons
that is needed to explain why a rational and otherwise well-motivated person
would do A (i.e., what D did under the relevant description) in the circumstances
as D believed them to be.”
189
The argument here does not rely on a normative constraint like Yaffe or
Sarch’s lenity principle. In the Pennsylvania Avenue hypothetical sketched
above, for example, it is not even possible that Top Hat could have committed
the robbery out of desperation or need; whereas for Wheelbarrow that seems like
the easiest conclusion to draw from the evidence available. But such a principle
would likely make the argument here even stronger. For if we take the most
charitable possible interpretation of why any given offender might have
committed a crime, we can rule out the more unsavory possible motivations that
people with prior convictions might have for returning to crime, and simply
assume that they do so because their incentives are so strong.
We can presume that in most cases, we will have less evidence of ill will
when people with prior convictions reoffend than when others are convicted for
the first time, ceteris paribus. It is possible, but unlikely, that any given
individual places no weight on the incentives to break the law that the collateral
consequences of his or her prior convictions give rise to. In some criminal cases,
we will have additional evidence (beyond the presence or absence of prior
convictions) that the defendant acted out of extreme malice, hatred, or disregard
for the interests of others. So prior convictions should not serve as an automatic
or indefeasible mitigating factor.
190
But criminal courts have limited resources;
we have imperfect and inevitably circumstantial evidence of other peoples’
quality of will; and the collateral consequences of past criminal convictions
strongly incentivize future crime. So, judges and sentencing commissions should
treat prior convictions as a presumptive mitigating factor.
5. Incentives and Deterrence Once More
One might immediately wonder why the incentives that people with prior
convictions have to return to crime don’t bolster the deterrence-based rationale
for the recidivist sentencing premium, even if they also underlie a backward-
189
SARCH, supra note 187, at 51.
190
Cf., e.g., Miriam Gohara, Grace Notes: A Case for Making Mitigation the Heart of Noncapital
Sentencing, 41 AM. J. CRIM. L. 41 (2013) (arguing that noncapital sentencing should include the kind of
individualized consideration of mitigating factors that is routine in capital cases).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1255
looking rationale for the Recidivist Sentencing Discount. After all, if the
collateral consequences of a first conviction do so much to strengthen one’s
incentives to reoffend, then wouldn’t it be wise to balance those incentives out
to prevent future crime?
I have two responses to this worry.
The first is primarily empirical. The recidivist sentencing premium is likely
to be an extremely inefficient way to compensate for the incentives that people
with prior criminal convictions have to return to crime. Given the fact that the
perceived certainty of apprehension, not the perceived severity of potential
penalties, drives deterrence from crime,
191
we should probably look to electronic
monitoring and policing if we wanted to find a way to offset those incentives
efficiently.
192
The second, and more fundamental, response to this worry is primarily
normative. The fact that the collateral consequences of conviction give people
stronger incentives to commit crime than they would otherwise have should
make us reconsider the moral limitations of how severely we can permissibly
punish them—not just the economic efficiency of our sentencing severity.
Consider an analogous case. We know that young teens (say, between
fourteen and sixteen) are much harder to deter from crime than older adults (say,
between fifty and sixty-five).
193
But the courts and criminal law theorists agree
that the former are less culpable, all else equal.
194
And one would be hard-
pressed to find a justification for punishing a young teen more harshly than a
fully formed adult were the two convicted of identical crimes.
Similarly, if we cannot blame people with prior convictions for reoffending
as severely as we could blame first-time wrongdoers convicted of the same kinds
of crime, then we cannot punish the former more severely than the latter, even
if doing so would serve the aim of crime control—and it is unclear whether it
would.
191
See supra Part I.B.2 and accompanying footnotes.
192
See supra Part I.B.3 and accompanying footnotes.
193
W.G. Jennings & J.M. Reingle, On the Number and Shape of Developmental/Life-Course Violence,
Aggression, and Delinquency Trajectories: A State-of-the-Art Review, 40 J. CRIM. JUST. 472, 473 (2012).
194
This observation usually attributed to the stages of neurological development. See, e.g., Miller v.
Alabama, 567 U.S. 460, 465 (2012). Gideon Yaffe argues that it is due to minors being denied the vote. YAFFE,
supra note 187.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1256 EMORY LAW JOURNAL [Vol. 70:1209
C. Violence and Victims
The argument for treating prior convictions as a presumptive mitigating
factor is most intuitive in cases where the crime in question (1) has a clear
financial payoff, and is (2) financially motivated, (3) intentional, and
(4) committed with full awareness of the consequences for the victim. But the
theory can be extended to a broader range of offenses, including violent crimes
committed recklessly or negligently and for no financial gain.
195
1. Violent Crime
Consider the following scenario as an illustration of this broader application:
Warning Shots: Across town from one another in the wee hours
of a Saturday evening or Sunday morning, Battleship, a thirty-year-old
graduate student in economics, and Cannon, a thirty-year-old line cook
at Chipotle, are each walking home after a long day of work.
Battleship’s Ventnor Avenue neighborhood is a tony community of
academics and young professionals. Cannon’s neighborhood around
Oriental Avenue, on the other hand, is stricken with concentrated
poverty and violence. On this night, both Battleship and Cannon
encounter groups of mildly inebriated young men coming out of their
respective local bars, who proceed to hassle and taunt them—assaults
that turn physical when each the two groups proceed to surround
Battleship and Cannon, pushing them around and roughing them up.
Drunken bystanders at both locations watch attentively and laugh,
some beginning to record the altercations on their cell phones.
Battleship and Cannon both respond quickly, pulling pistols from their
waistbands and firing warning shots into the air, which puts an
immediate halt to each of their respective altercations. Neither one of
them anticipates the kickback when he fires, however, and in both
cases an innocent bystander is struck and injured (but not killed).
In Warning Shots, we can assume that neither Battleship nor Cannon
intentionally or knowingly hurt the bystanders who they accidentally end up
shooting. Their actions certainly seem reckless, or at least negligent, however.
We can also assume that there was no perceived financial payoff for either crime.
Battleship, as such, would seem not to have any substantial reason for acting as
recklessly as he did, other than avoiding temporary embarrassment and
discomfort. Cannon, on the other hand, may have had more serious concerns in
mind (either consciously or unconsciously).
195
Lewis, supra note 171, at 165–70.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1257
In neighborhoods where poverty is concentrated and violence is common,
mutual respect often becomes a zero-sum game, leaving residents with
incentives to adopt a threatening demeanor and to behave in ways that are often
unfriendly, uncivil, and disrespectful—sometimes breaking the law in doing so.
Violent crime, and the reputation it often comes with, can sometimes be the best
(or the only) way to secure one’s social standing, especially in response to other
acts or threats of violence.
196
So Cannon could have been worried that being
perceived as weak could make him a target for further (and perhaps more
serious) victimization, humiliation, or ostracism long into the future—especially
if video recordings of him being pushed around without standing up for himself
were to go viral or spread in his community.
Insofar as the collateral consequences of punishment cause (or themselves
constitute) social and economic disadvantage, as such, they will in turn
incentivize crime—including, in many cases, violent crime without any clear
financial payoff, like we see in Warning Shots.
2. Sexual Violence
It is doubtful that the collateral consequences of conviction incentivize
sexual violence in the way that matters here, however. Only certain kinds of
incentives to break the law—namely, those related to the objective benefits of
crime—mitigate the extent to which we can justifiably blame and punish people
for doing so. The strength of those incentives should be measured in terms of
either the Rawlsian Primary Goods
197
or the Sen Capabilities Approach
198
—not
in terms of the offender’s own preferences or tastes.
Across the social and behavioral sciences, empirical research shows that the
vast majority of sexual violence is motivated by a desire to dominate, demean,
humiliate, hurt, or exploit the victim (most often a woman or girl), rather than a
desire for intimacy or sexual gratification.
199
And while intimacy and sexual
196
This is especially, but not exclusively, true for men (in particular, young men), because neighborhood
violence can threaten their sense of masculinity. Elijah Anderson documents this phenomenon in detail in his
book THE CODE OF THE STREET: DECENCY, VIOLENCE, AND THE MORAL LIFE OF THE INNER CITY. ELIJAH
ANDERSON, THE CODE OF THE STREET: DECENCY, VIOLENCE, AND THE MORAL LIFE OF THE INNER CITY 66–107
(1999).
197
JOHN RAWLS, A THEORY OF JUSTICE (1971).
198
Amartya Sen, Equality of What?, in TANNER LECTURES ON HUMAN VALUES 197 (Sterling McMurrin
ed., 1979).
199
See, e.g., David Lisak & Susan Roth, Motivational Factors in Nonincarcerated Sexually Aggressive
Men, 55 J. PERSONALITY & SOC. PSYCH. 795, 796 (1988); David Lisak & Susan Roth, Motives and
Psychodynamics of Self-Reported, Unincarcerated Rapists, 60 AM. J. ORTHOPSYCHIATRY 268, 269 (1990);
Sarah K. Murnen, Carrie Wright & Gretchen Kaluzny, If “Boys Will Be Boys,” Then Girls Will Be Victims? A
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1258 EMORY LAW JOURNAL [Vol. 70:1209
gratification could perhaps be thought of as Primary Goods (things that anyone
would want, regardless of whatever else they wanted) or Capabilities (things that
we have good reason to value, or which make a life worth living), the
domination, subordination, demeaning, humiliation, or exploitation of others
cannot. Those are not the kinds of things that anyone would want, regardless of
whatever else they wanted; and they are not the kinds of things that make a life
worth living, or which the Capabilities Approach is meant to capture. So, the
perceived “payoff” to sexual violence is not likely to increase, in terms of either
the Capabilities or the Primary Goods, as a result of a first criminal conviction.
Sex crime that has a clear financial payoff, or a payoff in social standing,
could be a potential exception to this. Pimping and sex trafficking are more
likely to be motivated by financial considerations than rape or sexual assault, for
example. And in settings where extreme forms of “rape culture” or “toxic
masculinity” are the norm, boys and men may face pressure to commit acts of
sexual violence to secure their own social standing—pressures that could be
exacerbated by the difficulty of finding alternative paths to social standing when
one has a criminal record.
Given the abject moral disgust many of us—including myself—associate
with sexual violence, the implication that people with prior criminal convictions
could be less blameworthy than first-time offenders for a crime of sexual
violence is likely to seem repugnant. But the repugnance of the implication
should be softened by noticing that the argument at hand entails only that a
repeat sex offender ought to be punished less severely (and perhaps not much
less severely) than a first-time sex offender convicted of an otherwise similar
offense. This is compatible with the thought that our penal systems ought to
respond more harshly to crimes of sexual violence overall—though that is a
question that falls outside of the scope of this paper.
3. Victims’ Rights
One might also worry that a recidivist sentencing discount would express a
lack of sufficient concern for the rights or interests of the victims of repeat
offending, or otherwise unfairly devalue those rights and interests.
200
But here
Meta-Analytic Review of the Research That Relates Masculine Ideology and Sexual Aggression, 46 SEX ROLES
359, 364 (2002); Neil Malamuth, Predictors of Naturalistic Sexual Aggression, 50 J. PERSONALITY & SOC.
PSYCH. 953, 954 (1986).
200
See, e.g., Press-Enter. Co. v. Superior Court of Cal., Riverside Cnty., 464 U.S. 501, 509 (1984)
(describing the “concerns of the victims . . . in knowing that offenders are being brought to account for their
criminal conduct”).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1259
again it is important to emphasize the comparative nature of the argument at
hand. People with prior criminal convictions have stronger incentives to commit
just about any kind of crime than those without any prior convictions. So,
compared to a person with prior convictions, a first-time offender would need to
weigh their own interests more heavily relative to the rights and interests of the
victim in order to go through with almost any otherwise similar crime. As such,
thinking clearly about the weight of victims’ rights and interests tells in favor of
the argument I have defended above, not against it.
In sum, the core argument I have advanced likely does not apply to crimes
of sexual violence, except when those crimes are driven by financial gain or
social status. But it should certainly extend to just about any other kind of
criminal offense, including (1) violent offenses that lack any financial motive or
payoff and (2) crimes that are committed recklessly or negligently rather than
intentionally, and without knowing what the likely consequences are for the
victims. This does not disrespect or unfairly devalue the rights or interests of
victims. Rather, it puts victims’ rights and interests in proper balance with the
interests of offenders and the quality of will that their offenses reveal to us.
IV. C
OLLATERAL CONSEQUENCES: OBJECTIONS AND RESPONSE
Here, I respond to two of the most important potential objections to this
account—both of which relate to the nature of, and justification for, collateral
consequences. These objections both fail for the same reason. So, my strategy
here will be to lay these two objections out, show how they depend on a crucial
shared premise—namely, that collateral consequences are a justified part of how
we punish people, rather than mere civil sanctions that are triggered by criminal
convictions—and then show that this premise is false.
A. Option Luck
Perhaps the most obvious and powerful objection to the argument thus far is
that repeat offenders are themselves responsible for the incentives they have to
return to crime, given that the collateral consequences that those incentives arise
from are a result of their own previous choices. As such, we might think that
people with prior convictions have obligations to resist the incentives they have
to reoffend, and that these incentives cannot exculpate them from punishment in
the present.
201
Call this the “Option Luck” objection.
202
201
I am grateful to Patrick Tomlin for forcefully articulating this objection to me.
202
I borrow the term “option luck” from Ronald Dworkin, who viewed the distinction between option
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1260 EMORY LAW JOURNAL [Vol. 70:1209
The mere fact that someone chose to commit a crime is not enough to show
that they alone are responsible for bearing the burdens of the collateral
consequences associated with that offense, however. For that to be true, the
collateral consequences attached to the offense need to be justified themselves
first. Individual offenders do not unilaterally create the barriers they face to
reentering society. Rather, those barriers are a joint product of the choices that
offenders make, the legislative and regulatory landscape our governments
create, and the general conditions of our society and economy. Consider the
following (admittedly strange) hypothetical illustration of this:
Bubblegum: In the year 2090, federal law criminalizes the use or
possession of bubblegum. Judges have wide discretion in sentencing
bubblegum-law offenders, and are generally lenient, preferring
probation or community service to imprisonment. But the collateral
consequences of violating the bubblegum law are extremely harsh:
anyone caught with bubblegum on their person receives a lifetime ban
from subsidized housing, government assistance, and employment of
any kind.
It is nearly impossible to imagine a legitimate state interest in criminalizing
the use or possession of bubblegum or in imposing such extreme collateral
consequences on bubblegum-law offenders. So, it makes little sense to say that
offenders who break the law are responsible for bearing the burdens of that
choice simply by virtue of the fact that it was a choice. The burdens that the law
attaches to different choices are in need of justification themselves.
203
Furthermore, collateral consequences need to be justified as part of the
punishment for the crime in question for the “Option Luck” objection to work.
For if collateral consequences were justified as a purely preventive measure—
without reference to the crime committed—then there would be no reason to
think that the offender is responsible for bearing the burdens associated with
those consequences himself.
204
In that case, collateral consequences would have
a similar underlying normative justification to the law of eminent domain. Under
the law of eminent domain, the state can unilaterally and coercively “condemn”
(take ownership of) private property for public use without the previous owner’s
consent.
205
But the burdens of eminent domain fall on the state, rather than the
luck (“how deliberate and calculated gambles turn out”) and “brute luck” (“how risks fall out that are not in that
sense deliberate gambles”) as central to the nature of distributive justice. See RONALD DWORKIN, SOVEREIGN
VIRTUE 73 (2000).
203
See, e.g., DOUGLAS HUSAK, OVERCRIMINALIZATION 6–7 (2008).
204
See Sandra G. Mayson, Collateral Consequences and the Preventive State, 91 NOTRE DAME L. REV.
301, 303–04 (2015) (defending—with caveats—collateral consequences as a regulatory mechanism of this kind).
205
Chicago, Burlington & Quincy R.R. v. City of Chicago, 166 U.S. 226, 258 (1897).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1261
previous owner of the condemned property—whom the state is required under
the Fifth Amendment to pay “just compensation” for the property that it takes.
206
So, the “Option Luck” objection only has force against the argument for the
recidivist sentencing discount if the collateral consequences we impose on
criminal offenders are justified as part of the punishment for their crimes. Below,
I show that the other central objection to my argument also depends on this same
premise. Then, I show that this premise is almost certainly false.
B. Declining Severity
David Dana argues that recidivist sentencing enhancements can be at least
partially justified by the fact that collateral consequences have “declining
marginal severity.”
207
Collateral consequences, that is, often have an extremely
severe effect after a first conviction, and little to no further effect after a second,
third, or fourth conviction. The collateral consequences one faces after a first
conviction can often ruin careers, relationships, opportunities, and dreams—
putting those things beyond the point of repair in one’s lifetime. Progressively
increasing the marginal severity of formal sentences could be justified as a way
of “balancing out” the declining marginal severity of their collateral
consequences, as such.
208
According to Dana’s model, this balancing effect
should promote both optimal deterrence and proportional punishment.
209
Acknowledging the declining marginal severity of collateral consequences
tells neither in favor of, nor against, the recidivist sentencing premium per se. In
principle, that is, we could punish first-time offenders primarily with the
sanctions we now call “collateral consequences,” and punish repeat offenders
primarily with imprisonment, while adjusting the severity with which we punish
them—compared to one another—in any way we like. The means by which we
punish people does not by itself determine how harshly we punish them.
But one way to put this as an objection to this paper’s argument would be to
say that, even if we think that judges and sentencing commissions should treat
prior convictions as a presumptive mitigating factor, as I have argued, we would
still be justified in imposing harsher criminal sentences on repeat offenders than
we do on first-time convicts because the effects of the collateral consequences
of a first conviction are so much more severe than those of a second, third, or
206
Id. at 228–29.
207
Dana, supra note 58, at 773.
208
Id.
209
Id.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1262 EMORY LAW JOURNAL [Vol. 70:1209
fourth conviction. And it is at least plausible that the collateral effects of a first
conviction are more severe than the additional prison or jail time that might be
triggered by existing recidivist sentencing enhancements for a second, third, or
fourth conviction.
So, if we take the declining marginal severity of collateral consequences into
account, then this paper might give us a reason to think about the justification of
our current regimes differently, without giving us a reason to actually change
existing sentencing practices.
210
One problem with this inference is that recidivist sentencing enhancements
in most jurisdictions—especially those with guidelines-based sentencing
regimes—often take the form of incremental, ladder-style increases in the
recommended range of severity for any given offense, the more prior convictions
the offender in question has on his or her record (and the more severe the prior
offenses were).
211
The severity of collateral consequences seems to decline
precipitously after a first conviction, by contrast. To put it somewhat
hyperbolically, one’s life could be ruined in some respects after a first
conviction, and subsequent convictions can do little to make things worse in
those respects. As such, even “three strikes” legislation likely could not “balance
out” the declining marginal severity of collateral consequences in the way that
Dana suggests.
212
Recidivist sentencing enhancements could do so only if they
took the form of an extremely generous one-time discount for first-time
offenders—or, less appealingly, an ultra-punitive “two strikes”-style regime. So,
Dana’s reasoning may not be able to vindicate statutory or guidelines-based
recidivist sentencing enhancements as we know them today but could potentially
vindicate some form of the recidivist sentencing premium.
Some courts have held, consonant with Dana’s way of thinking, that under
federal law, judges are permitted to give first-time offenders much more lenient
sentences than would otherwise be advised under relevant guidelines, on the
grounds that the collateral consequences of the conviction must be considered
part of the punishment.
213
In United States v. Nesbeth, for example, the
defendant was convicted of both possession with intent to distribute, and
210
Id.
211
See, e.g., FRASE ET AL., supra note 2, at 22–25.
212
Dana, supra note 58, at 773.
213
Courts are divided as to the permissibility of this approach, however. See, e.g., United States v.
Musgrave, 647 Fed. App’x 529, 532 (6th Cir. 2016); United States v. Stefonek, 179 F.3d 1030, 1038 (7th Cir.
1999); United States v. Morgan, 635 Fed. App’x 423, 450 (10th Cir. 2015); United States v. Kuhlman, 711 F.3d
1321, 1329 (11th Cir. 2013).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1263
importation of, cocaine.
214
But the court justified a non-custodial sentence on
the grounds that “the collateral consequences [she] will suffer, and is likely to
suffer—principally her likely inability to pursue a teaching career and her goal
of becoming a principal[]—has compelled me to conclude that she has been
sufficiently punished, and that jail is not necessary. . . .”
215
Similarly, in United
States v. Pauley, the defendant (a West Virginia high school art teacher) was
convicted of possessing child pornography.
216
But the Fourth Circuit found that
the lower court was permitted in imposing a sentence thirty-six months shorter
than the very bottom of the range recommended by the sentencing guidelines on
the grounds that he would lose his teaching certificate and state pension as a
result of the conviction.
217
And in United States v. Stewart, another comparable
case, the Second Circuit has remarked that “[i]t is difficult to see how a court
can properly calibrate a ‘just punishment’ if it does not consider the collateral
effects of a particular sentence.”
218
There are reasons to be skeptical of the idea that courts should include the
likely effects of collateral consequences in how they calculate sentencing
severity.
219
But even if we assume that this approach is justified, it does not
follow that sentencing judges ought to make the severity with which we punish
first-time offenders the baseline from which they derive the severity of
punishment for repeat offenders. Consider the Bubblegum thought experiment
detailed above once again to see why. In Bubblegum, the collateral consequences
of a harmless and extremely trivial offense are exceptionally severe. Judges
faced with sentencing first-time “bubblegum offenders” might thus include the
likely effects of those collateral consequences in how they calculate sentencing
severity, like the Nesbeth, Pauley, and Stewart courts.
220
But how should they
sentence repeat bubblegum offenders? Consider two possible options:
214
United States v. Nesbeth, 188 F.Supp.3d 179, 180 (E.D.N.Y. 2016).
215
Id. at 194.
216
United States v. Pauley, 511 F.3d 468, 469 (4th Cir. 2007).
217
Id.
218
United States v. Stewart, 590 F.3d 93, 141 (2d Cir. 2009).
219
Perhaps the most obvious reason for skepticism in this context is that the effects of collateral
consequences are likely to be felt most acutely by those who are the most socially and economically advantaged
beforehand. As Judge Posner commented in United States v. Stefonek—a case where the Seventh Circuit rejected
the approach in question: “It is natural for judges, drawn as they (as we) are from the middle or upper-middle
class, to sympathize with criminals drawn from the same class. But in this instance, we must fight our nature.
Criminals who have the education and training that enables people to make a decent living without resorting to
crime are more rather than less culpable than their desperately poor and deprived brethren in crime.” Stefonek,
179 F.3d at 1038.
220
See Nesbeth, 188 F.Supp.3d at 180; Pauley, 511 F.3d at 469; Stewart, 590 F.3d at 14.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1264 EMORY LAW JOURNAL [Vol. 70:1209
Bubblegum 2: Repeat offenders under the bubblegum law, like
first-time offenders, are generally sentenced to probation, a few hours
of community service, or whatever the sentencing judge thinks of as
the mildest available sanction, on the grounds that the offense itself is
trivial.
Bubblegum 3: Judges give repeat bubblegum offenders sentences
that they consider roughly equivalent in severity to how they would
punish a first-time offender for the same crime—perhaps adjusting
slightly upward in order to impose a recidivist sentencing premium, or
slightly downward, in order to impose a recidivist sentencing discount.
But because the collateral consequences of a first conviction for
chewing bubblegum are so severe, judges tend to give repeat
bubblegum offenders between five and ten years in state prison.
Bubblegum 3 promotes one value better than Bubblegum 2—namely,
equality between repeat- and first-time bubblegum-law offenders. But just about
any sane reader will find Bubblegum 2 preferable to Bubblegum 3. Judges in
Bubblegum 3 compound the injustice of the way first-time offenders are treated
in treating repeat offenders equally unfairly.
The mere fact that the marginal effects of collateral consequences decline
precipitously after an offender’s first conviction cannot justify imposing harsher
sentences on repeat lawbreakers to make up for the fact that they no longer feel
these effects, given that their lives were already ruined in many respects. Like
the “Option Luck” objection, the “Declining Severity” objection would only
work if the collateral consequences of criminal conviction are a justified part of
how we punish people.
C. Jurisprudence and Justification
The idea that the collateral consequences we impose on people with criminal
convictions are a justified part of how we punish them is wildly implausible in
the current doctrinal landscape.
Indeed, it is questionable that collateral consequences should be thought of
as a form of punishment at all; courts have long treated them as mere civil
sanctions that attach to a criminal conviction, but which are not part of the
criminal sentence per se.
221
If collateral consequences are not—or should not be
regarded as—forms of punishment, then the “Option Luck” and “Declining
Severity” objections have no leg to stand on from the start. For collateral
221
For an historical analysis of the development of this doctrine, see Kaiser, Revealing the Hidden
Sentence, supra note 126.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1265
consequences could not be justified forms of punishment if they are not
punishment at all. There is a large and growing literature examining this
question;
222
however, and I do not attempt to adjudicate between competing
strains of that literature here. For even if collateral consequences are (or should
be regarded as) forms of punishment, they could not plausibly be thought of as
justified forms of punishment without systematically restructuring current
doctrine and legislation.
1. Procedural Safeguards
The primary reason for this is that collateral consequences lack the kinds of
procedural safeguards that are necessary for punishment to be justified. The
Supreme Court has ruled that collateral consequences are not “punishment” for
the purposes of constitutional protections against cruel and unusual punishment,
double jeopardy, bills of attainder, and excessive fines.
223
And circuit courts
have created the “collateral consequences rule,” under which judges and defense
attorneys have no duty to inform defendants of the collateral consequences of
plea deals or trial convictions.
224
In Padilla v. Kentucky, the Supreme Court
allowed for an exception to this rule, ruling that defense attorneys must tell
noncitizen clients if pleading guilty to a criminal charge could result in
deportation. The Court explicitly refused to evaluate the collateral consequences
rule in Padilla, however, arguing that “[t]he question whether that distinction
[between direct and collateral consequences] is appropriate need not be
considered in this case because of the unique nature of deportation.”
225
Together, these two doctrines make it impossible to justify collateral
consequences, at least as we know them, as a form of punishment. Less than 3%
of criminal cases in both our federal and state courts ever go to trial.
226
Over
222
See, e.g., Mayson, supra note 204 and accompanying text; Gabriel J. Chin, The New Civil Death:
Rethinking Punishment in the Era of Mass Conviction, 160 U. PA. L. REV. 1789, 1790 (2012); Kaiser, Revealing
the Hidden Sentence, supra note 126 and accompanying text; Carol S. Steiker, Foreword: Punishment and
Procedure: Punishment Theory and the Criminal-Civil Procedural Divide, 85 GEO. L. J. 775, 777 (1997);
Andrew von Hirsch & Martin Wasik, Civil Disqualifications Attending Conviction: A Suggested Conceptual
Framework, 56 CAMBRIDGE L.J. 599, 599 (1997).
223
See, e.g., Smith v. Doe, 538 U.S. 84 (2003) (holding that the collateral consequence humiliation
resulting from convicted sex offenders filing in a public registry does not constitute ex post facto punishment).
224
United States v. Parrino, 212 F.2d 919, 920–21 (2d Cir. 1954); Meaton v. United States, 328 F.2d 379,
380–81 (5th Cir. 1964); United States ex rel. Durante v. Holton, 228 F.2d 827, 830 (7th Cir. 1956); Munich v.
United States, 337 F.2d 356, 361 (9th Cir. 1964); Hutchison v. United States, 450 F.2d 930, 931 (10th Cir. 1971).
225
Padilla v. Kentucky, 559 U.S. 356 (2010).
226
U.S. SENT’G COMM’N, OVERVIEW OF FEDERAL CRIMINAL CASES—FISCAL YEAR 2018, at 8 (2019); Jed
Rakoff, Why Prosecutors Rule the Criminal Justice System—and What Can Be Done About It, 111 NW. L. REV.
1429, 1432 (2017).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1266 EMORY LAW JOURNAL [Vol. 70:1209
97% of them, that is, are resolved through a plea-bargaining process. At the same
time, on average, each offender is subject to over 2,000 different kinds of
legislative and administrative penalties, and collateral consequences are encoded
in over 43,000 legislative and administrative provisions.
227
The vast majority of
defendants in our criminal courts plead guilty, bargaining away their right to trial
in exchange for a reduced sentence, and they do this without having to be
informed about any of the collateral consequences that come along with the
criminal conviction they have ostensibly agreed to (except, under Padilla, if they
are at risk of deportation).
228
The way we adjudicate the vast majority of criminal cases—combined with
the fact that we attach so many hidden consequences to criminal convictions—
violates the fundamental principle that, in order for punishment to be a justified
response to what someone has done, he or she must have had a fair opportunity
to avoid that punishment by choosing differently.
229
This principle can be
understood as a more general constraint on the justification of any policy
decision; as T.M. Scanlon puts it, “If a policy imposes burdens on some people
in order to provide some general social benefit, then, wherever possible,
individuals must be given adequate opportunity to avoid bearing these burdens
by choosing appropriately.”
230
Some scholars argue that the Padilla decision will—or at least should—
force courts to abandon the collateral consequences rule altogether, giving
judges and defense attorneys the duty to inform defendants of a much wider
range of consequences that might result from entering a guilty plea or being
convicted at trial.
231
The Padilla Court’s characterization of deportation as
“unique” and “particularly severe” rings hollow, they argue, given how severe
227
Kaiser, Revealing the Hidden Sentence, supra note 126, at 157.
228
559 U.S. 356 (2010).
229
See, e.g., HART, supra note 18, at 22–23 (suggesting that this principle does not require the existence
of free will; its main expositors have been free will skeptics who are compatibilists about the relationship
between free will and moral responsibility); T.M. Scanlon, The Significance of Choice, 8 TANNER LECTURES ON
HUMAN VALUES 149 (1986) (discussing the necessity of free will for just punishment).
230
T.M. SCANLON, WHY DOES INEQUALITY MATTER? 123 (2018); cf. T.M. SCANLON, WHAT WE OWE TO
EACH OTHER 256–94 (1998) (presenting an extended analogy for understanding the social costs and benefits of
punishment).
231
Gabriel Chin, Making Padilla Practical: Defense Counsel and Collateral Consequences at Guilty Plea,
54 HOW. L.J. 675 (2011); Margaret C. Love, Collateral Consequences After Padilla v. Kentucky: From
Punishment to Regulation, 31 ST. LOUIS PUB. L. REV. 87 (2011) (arguing that “[Padilla’s] logic extends beyond
deportation to many other severe and certain consequences of conviction that are imposed by operation of law
rather than by the sentencing court”).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1267
and lasting other collateral consequences can be, and how closely those
consequences are (and have been) tied to “the criminal process.”
232
But, as courts have recognized, we are far from a world in which we could
easily implement even the minimal procedural safeguards necessary—but not
sufficient—for the sanctions we currently classify as “collateral consequences”
to be justified forms of criminal punishment.
233
Michael Pinard argues that
abandoning the collateral consequences rule would be “made even more
complicated by the fact that collateral consequences are not centralized, but
rather are scattered throughout federal and state statutes, state and local
regulatory codes, local rules, and local policies.”
234
And even if courts were to
universally abandon the collateral consequences rule, criminal defendants would
still lack constitutional protections against cruel and unusual punishment, double
jeopardy, bills of attainder, and excessive fines. Without minimal procedural
safeguards, collateral consequences could not possibly be justified as a form of
punishment.
2. Substantive Criteria
Even if we could implement those procedural safeguards, as some argue we
can, collateral consequences as we know them—though they could no longer be
thought of as “collateral”—would still be extremely difficult to justify as
punishment, on substantive grounds.
235
First, given the strong way in which they incentivize future crime, it is
implausible that many of the collateral consequences we impose on people who
have been convicted of criminal offenses could serve the aim of crime control.
As discussed in Part I.B, increasing the certainty that people will be caught—
not the severity with which they will likely be punished on the chance that they
are caught—is what deters people.
236
Changing the rules of criminal law or
sentencing policy is unlikely to have much, if any, deterrent effect.
237
So
collateral consequences are not likely to be good deterrents, regardless of
whether we treat them as part of how we formally punish people or not.
232
Love, supra note 231, at 103 (quoting Padilla, 559 U.S. at 365–66).
233
Padilla, 559 U.S. at 376 (Alito, J., concurring); Fruchtman v. Kenton, 531 F.2d 946, 949 (9th Cir.
1976); United States v. Cariola, 323 F.2d 180, 186 (3d Cir. 1963).
234
Michael Pinard, An Integrated Perspective on the Collateral Consequences of Criminal Convictions
and Reentry Issues Faced by Formerly Incarcerated Individuals, 86 B.U. L. REV. 623, 646 (2006).
235
Cf. ZACHARY HOSKINS, BEYOND PUNISHMENT? A NORMATIVE ACCOUNT OF THE COLLATERAL LEGAL
CONSEQUENCES OF CONVICTION (2019).
236
Chalfin & McCrary, supra note 48; Nagin, supra note 52, at 85.
237
Nagin, supra note 52, at 85.
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1268 EMORY LAW JOURNAL [Vol. 70:1209
The fact that collateral consequences have declining marginal severity also
suggests that they may encourage more crime than they deter, on balance. The
majority of those sentenced to jail or prison terms already have a prior
conviction.
238
Since collateral consequences have little to no effect on the
second, third, or fourth time one is convicted of a crime, they would not be a
deterrent to repeat offenders even if people were deterred by the harshness of
potential sanction, rather than the certainty of those sanctions (which they are
not). So, the collateral consequences of punishment as we know them almost
certainly encourage more crime than they deter, on balance.
Some collateral consequences serve an easily identifiable incapacitation-
related purpose, though the vast majority do not. Restrictions on sex offenders
from working in daycares, K–12 schools, and other settings with vulnerable
populations; removal from public office for officials convicted of bribery;
restrictions on access to firearms for people convicted of violent crimes; and the
loss of licensure for people convicted of crimes related to the licensed activity
are all easy to justify on incapacitation grounds.
239
But these restrictions alone would not give people with prior convictions
much more of an incentive to reoffend than they would have if there were no
legal barriers at all to reentry, however—so they are not really what is at stake
in the “Option Luck” and “Declining Severity” objections to the argument of
this Article. And in any case, there is a whole smorgasbord of restrictions that
together make it very difficult for people with prior convictions to live a normal
life, and which serve no plausible incapacitation-related purpose.
240
There is no
reason to think, for example, that preventing people with a criminal conviction
from becoming barbers (one of the trades that is easiest to pick up in prison)
stops them from committing crime; the vast majority of criminal offenders are
not Edward Scissorhands or Sweeney Todd. Permanent barriers to employment
across multiple sectors and industries cannot be justified on an incapacitation
rationale.
241
Preventing the formerly convicted from accessing public housing
and financial aid for higher education; suspending their driver’s licenses; and
making their criminal record easily accessible to both public and private-sector
238
BRIAN A. REAVES, FELONY DEFENDANTS IN LARGE URBAN COUNTIES, 2009, at 17 (2013)
https://jpp.whs.mil/Public/docs/03_Topic-Areas/07-CM_Trends_Analysis/20160122/04_BJS_Report_State_
Felony_Sentencing_2009.pdf.
239
Cf. AM. BAR ASS’N, ABA STANDARDS FOR CRIMINAL JUSTICE: COLLATERAL SANCTIONS AND
DISCRETIONARY DISQUALIFICATION OF CONVICTED PERSONS 24 (3d ed. 2004).
240
See discussion supra Part III.B.
241
See, e.g., Miriam Aukerman, The Somewhat Suspect Class: Towards a Constitutional Framework for
Evaluating Occupational Restrictions Affecting People with Criminal Records, 7 J.L. & SOC’Y 18 (2005).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1269
employers, where it is often an automatic bar to being hired (either de jure or de
facto) does not to incapacitate them from returning to crime—rather, it gives
them much stronger incentives to do so.
There is little to be said in favor of collateral consequences from a
retributivist perspective either. Retributivism is not in principle committed to
any particular form of punishment, so long as offenders are punished to the
extent they deserve. The severity of any given criminal could be thought of as a
function of how burdensome or unpleasant it is, and the amount of time for
which we impose it. A prison sentence of any given amount of time would be a
more severe sanction than imposing the kinds of collateral consequences we
currently attach to criminal convictions for that same amount of time. But the
average prison sentence in our state systems is only 2.6 years, while the median
length of time is 1.3 years.
242
Collateral consequences, by contrast, often persist
for decades or even indefinitely.
243
There are reasons internal to the retributivist outlook, however, to favor
punishments that are short and sharp over those that are long and gentle.
244
For
retributivists, not punishing those who deserve punishment is bad.
245
In other
words, it is bad when someone who deserves punishment is left unpunished, and
the longer they are left unpunished, the worse things get. If that is true, then we
ought to punish them as quickly as possible. Another way to put the point is to
say that “[i]t is bad that the deserving go unpunished, but as soon as they get
their just deserts, all is well with the world (from the perspective of retributive
justice).”
246
If that is true, then, ceteris paribus, we should prefer forms of
punishment that are more likely to deliver the full amount of hard treatment the
offender deserves. Forms of punishment that take longer to impose leave open a
greater risk that the offender might become ineligible for punishment before the
sentence is served in full—for example, by dying or going insane.
247
242
DANIELLE KAEBLE, TIME SERVED IN STATE PRISON, 2016, U.S. DEP’T OF JUST. (Nov. 2018),
https://www.bjs.gov/content/pub/pdf/tssp16.pdf.
243
See Love, supra note 231.
244
Patrick Tomlin, Time and Retribution, 33 LAW & PHIL. 655 (2014).
245
Id. at 664. As Kant famously put it, “Even if a civil society were to be dissolved…the last murderer
remaining in prison would first have to be executed, so that each has done to him what his deeds deserve and
blood guilt does not cling to the people for not having insisted upon this punishment.” IMMANUEL KANT, THE
METAPHYSICS OF MORALS 106 (Mary Gregor ed., Cambridge Univ. Press, 1996).
246
Tomlin, supra note 244, at 669.
247
See, e.g., Ford v. Wainwright, 477 U.S. 399 (1986) (holding that the Eight Amendment protection
against cruel and unusual punishment prohibits capital punishment for prisoners who have lost their sanity);
Panetti v. Quarterman, 551 U.S. 930 (2007) (holding that capital punishment is prohibited when a prisoner’s
mental illness prevents them from having a “rational understanding” of the state’s reason for executing them).
Similarly, both Tomlin and Parfit argue that an offender might also become morally ineligible for punishment
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
1270 EMORY LAW JOURNAL [Vol. 70:1209
Collateral consequences, as such, cannot be justified as a form of
punishment—for both substantive and procedural reasons. So, the “Option
Luck” and “Declining Severity” objections cannot be sustained. And it appears
that we are stuck with the paradox of recidivism.
C
ONCLUSION
The argument that led us to this paradox has taken a number of twists and
turns, so it is worth looking back at the road we have traveled.
We started from the ubiquitous and abiding principle that we should respond
more severely to repeated wrongdoing than we do to first-time misconduct—
one of our most deeply held moral and legal principles. We then considered the
tremendously intuitive reasons for why we seemingly ought to extend that
principle to the criminal law in the form of a recidivist sentencing premium:
compared to first-time lawbreakers, repeat offenders are harder to deter, need to
be incapacitated for longer, and are more morally culpable.
On closer examination, however, we discovered that these reasons fall apart.
We saw that, insofar as the rules of criminal law and sentencing policy have a
deterrent effect at all, repeat offenders are in many ways easier to deter than first-
time lawbreakers—given what we know about the biases and heuristics that
underlie all human decision-making, and the probability that an offender with
prior criminal convictions will be detected. We saw that it is unclear at best
whether we should “incapacitate” repeat offenders for longer periods of time
than first-time offenders, given the extent of crime inside of our prisons and jails,
the phenomena of “replacement effects,” and the backlash of concentrated
incarceration—the latter two of which are likely to be more powerful in the
communities that offenders with prior convictions disproportionately come
from. And we saw that when someone with prior convictions breaks the law
again, we have no more evidence of ill will or bad character than when someone
with a clean record commits the same offense for the first time.
In fact, as we found, the opposite is the case. We saw that the collateral
consequences of criminal convictions incentivize people to return to crime in a
morally significant way. As a result, when people with prior convictions return
to crime, we have less evidence of ill will or bad character than when others are
by through a process of personal change—e.g., repentance—so that he no longer deserves to be punished as
much punishment as he did before. Tomlin, supra note 244, at 680; DEREK PARFIT, REASONS AND PERSONS 326
(1984) (“When some convict is now less closely connected to himself at the time of his crime, he deserves less
punishment. If the connections are very weak, he may deserve none.”).
LEWISFINAL_7.20.21 7/20/2021 2:29 PM
2021] THE PARADOX OF RECIDIVISM 1271
caught committing similar offenses for the first time. We did not need to invoke
any particular theory of wellbeing to reach this conclusion, and we found that it
applies to almost any kind of crime, including violent offenses committed
recklessly or negligently—though probably not sexual violence. Finally, we
considered the two most important potential objections to this position. We saw
that both of them, however, depend on the premise that the collateral
consequences we attach to criminal convictions are justified parts of how we
punish people. And we saw that this premise is implausible.
Thus, given the way we have structured our social conditions and legal
doctrine, judges and sentencing commissions ought to turn one of our most
ubiquitous and abiding moral and legal principles upside down. Rather than
responding more severely to repeated wrongdoing than we do to first-time
misconduct, they should do the exactly the opposite. This may be an incredibly
unpalatable conclusion, but there is no easy way to avoid it. Judges and
sentencing commissions cannot unilaterally eliminate the incentives that people
with prior criminal convictions have to return to crime. That would require
sweeping legislative, administrative, and broader social change. And it is not
clear that such sweeping changes are even possible, let alone desirable, all things
considered. This is the paradox we are left with.
