¹ Marc Mauer, ‘The Changing Racial Dynamics of the War on Drugs’, The Sentencing Project, <http://sentencingproject.org/doc/dp_raceanddrugs.pdf> (2009).

² Nazgol Ghandnoosh, ‘Race and Punishment: Racial Perceptions of Crime and Support for Punitive Policies’ (Washington, DC: The Sentencing Project, 2014), p. 21. The situation is much the same for Latinos, though I will keep to a narrow path.

³ This article will be concerned with positive retributivism, which is the view that offenders deserve to be punished, and that their desert furnishes a sufficient reason to punish. Positive retributivists also believe that an offender's desert constitutes a strong (but defeasible) reason to punish him. Negative retributivism is the more modest view that punishment may be imposed only on those who have committed a crime; it does not insist on punishing them. Negative retributivists sometimes use consequentialist considerations to guide determinations about whether and how much to punish, though they maintain that punishment must be at least roughly proportionate to an offender's desert.

Not all philosophers of punishment believe desert to be a conceptual component of retributive justice; see e.g. Matravers, Matt, ‘Is Twenty-First Century Punishment Post-Desert?’, Retributivism Has a Past, Has It a Future? ed. Tonry, Michael (Oxford, 2011), pp. 30–45 CrossRefGoogle Scholar. My argument is compatible with some of these alternative brands of retributivism, most notably the communicative theory of R. A. Duff (Punishment, Communication, and Community, Studies in Crime and Public Policy (New York, 2001)), but I must refrain from discussing these matters. I will also set aside consequentialist theories of punishment, which raise an entirely different set of issues.

⁴ Although Ernst van den Haag is the best-known advocate for this view, which I later dub non-comparative rigourism, there are a few others. See Haag, van den, ‘The Ultimate Punishment: A Defense’, Harvard Law Review 99.7 (1986), pp. 1662–9CrossRefGoogle Scholar; Meyers, Christopher, ‘Racial Bias, the Death Penalty, and Desert’, The Philosophical Forum 22.2 (1990), pp. 139–48Google Scholar. Kant is often thought to fit this mould, though I harbour serious doubts about this interpretation.

⁵ Respectarians include Avraham, Ronen and Statman, Daniel, ‘More on the Comparative Nature of Desert: Can a Deserved Punishment Be Unjust?’, Utilitas 25.3 (2013), pp. 316–33CrossRefGoogle Scholar; Berman, Mitchell N., ‘Punishment and Justification’, Ethics 118.2 (2008), pp. 258–90CrossRefGoogle Scholar; Frankfurt, Harry, Necessity, Volition, and Love (Cambridge, 1999)Google Scholar; Pojman, Louis, ‘Does Equality Trump Desert?’, What Do We Deserve? ed. Pojman, Louis and McLeod, Owen (Oxford, 1999), pp. 283–97Google Scholar.

⁶ Comparativists include Cholbi, Michael, ‘Race, Capital Punishment, and the Cost of Murder’, Philosophical Studies 127 (2006), pp. 255–82CrossRefGoogle Scholar; Gargarella, Roberto, ‘Penal Coercion in Contexts of Social Injustice’, Criminal Law and Philosophy 5 (2011), pp. 21–38 CrossRefGoogle Scholar; Hurka, Thomas, ‘Desert: Individualistic and Holistic’, Desert and Justice, ed. Olsaretti, Serena (Oxford, 2003), pp. 45–68 Google Scholar; Kelly, Erin, ‘Desert and Fairness in Criminal Justice’, Philosophical Topics 40. 1 (2012), pp. 63–77 CrossRefGoogle Scholar; Lippert-Rasmussen, Karl, ‘Punishment and Discrimination’, Punishment and Ethics: New Perspectives, ed. Ryberg, Jesper and Cortlett, J. Angelo (New York, 2010), pp. 169–88CrossRefGoogle Scholar; McLeod, Owen, ‘On the Comparative Element of Justice’, Desert and Justice, ed. Olsaretti, , pp. 123–44Google Scholar; Miller, David, ‘Comparative and Noncomparative Desert’, Desert and Justice (Oxford, 2003), pp. 25–44 Google Scholar; Pittman, John, ‘Punishment and Race’, Utilitas 9.1 (1997), pp. 115–30CrossRefGoogle Scholar; Shelby, Tommie, ‘Justice, Deviance and the Dark Ghetto’, Philosophy & Public Affairs 35.2 (2007), pp. 126–60CrossRefGoogle Scholar.

⁷ See, for example, Lippert-Rasmussen, ‘Punishment and Discrimination’.

⁸ For our purposes, total punishment is the product of two factors, the number of people punished and the severity of their punishment.

⁹ Of course, it must be a salient group; see Lippert-Rasmussen, ‘Punishment and Discrimination’, p. 170. A salient group is one that structures social interactions across a variety of contexts, e.g. women, LGBTQ, deaf – rather than Camry owners, people born on 22 April and so on.

¹⁰ NCVS 2006, Table 40, <http://www.bjs.gov/content/pub/pdf/cvus0602.pdf> (2006); UCR 2006 Table 43 <http://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2006> (2006). It should be noted that the UCR counts Hispanic offenders as white, and the NCVS counts some Hispanics as white, some as ‘other’.

¹¹ Mauer, ‘The Changing Racial Dynamics of the War on Drugs’.

¹² Ghandnoosh, ‘Race and Punishment’, p. 21.

¹³ Alexander, Michelle, The New Jim Crow: Mass Incarceration in an Age of Colorblindness (New York, 2010), p. 97 Google Scholar. In Ferguson, Missouri police officers stop, search and arrest black drivers more frequently than white drivers, even though contraband hit rates are higher for white drivers; see Nazgol Ghandnoosh, ‘Black Lives Matter: Eliminating Racial Inequality in the Criminal Justice System’ (Washington, DC: The Sentencing Project, 2015), pp. 6–7.

¹⁴ Owen McLeod correctly observes that it is hard to get a handle on exactly what it means for a fact to be ‘about’ a person (‘On the Comparative Element of Justice’). But this difficulty need not detain us any more than it has detained anyone else in the last forty-odd years.

¹⁵ Some facts about other people are irrelevant to P's desert. For example, the fact that an assistant professor's friends desperately want her to be tenured has no bearing on whether she deserves tenure.

¹⁶ One can be a comparativist or non-comparativist about either DB or T, where the disjunction is inclusive. So we have the following set of views: (4) is self-explanatory; it represents orthodox non-comparativism. There is a slight but noticeable difference between (2) and (3). Imagine four friends help you move on a sweltering August day, and you want to reward them with beer. While each friend has worked hard enough to deserve four beers, you have only two six-packs in your refrigerator, and the stores are closed. A proponent of (2) will say that this creates an unavoidable injustice, given the non-comparative nature of desert bases. No matter how you divide the beers, someone will get less than their DB requires, even though an equal distribution is just qua distribution. By contrast, advocates of (3) can say that so long as everyone receives three beers, everyone gets the treatment they deserve, and justice prevails. It is important to note that the comparative egalitarianism discussed earlier is a version of (2), in so far as it asserts a non-comparative conception of desert bases.

The difference between (1) and (3) can be seen in the grading example. Those favouring (1) and (3) would both grade on a curve. But those favouring (3) would not raise Q's grade to match P's, while friends of (1) would.

¹⁷ Initially, the contrast between comparative and non-comparative justice was invoked in debates about the role of desert in distributive justice. In this literature, retributive justice is often held up as a paradigm case of non-comparative justice, and retributive and distributive justice are said to be asymmetrical. See Scheffler, Samuel, ‘Justice and Desert in Liberal Theory’, California Law Review 88.3 (2000), pp. 965–90CrossRefGoogle Scholar. My suspicion is that these debates utilized a cramped notion of retributive justice because they were preoccupied with arguing for the comparative elements of distributive justice. The non-comparative characterization of retributive justice responded to the need for an intuitive contrast, and expressed no deep philosophical commitment; on this point, see Moriarty, Jeffrey, ‘Smilansky, Arneson, and the Asymmetry of Desert’, Philosophical Studies 162 (2013), pp. 537–45CrossRefGoogle Scholar.

¹⁸ A comparativist of type (3) could be a rigourist. A pluralist comparativist who also endorsed, mutatis mutandis, respectarianism would count as a comparative non-rigourist, as would a comparative egalitarian of type (2), in so far as she believes that the treatment of P can be just even when it is not completely congruent with P's desert basis. I have elected not to represent these nuances in my chart above.

¹⁹ See also Meyers, ‘Racial Bias, the Death Penalty, and Desert’.

²⁰ Avraham and Statman, ‘More on the Comparative Nature of Desert’; Frankfurt, Necessity, Volition, and Love, p. 150; Pojman, ‘Does Equality Trump Desert?’, p. 291; White, Mark D., ‘ Pro Tanto Retributivism: Judgment and the Balance of Principles in Criminal Justice’, Retributivism: Essays on Theory and Policy, ed. White, Mark D. (New York, 2011), pp. 129–45CrossRefGoogle Scholar.

²¹ For example, Russ Schafer-Landau argues that while retributivism requires an accurate assessment of an offender's non-comparative and comparative desert, there is no right answer to the question ‘what does P deserve?’. Retributivism, he concludes, is an incoherent justification of punishment (‘Retributivism and Desert’, Pacific Philosophical Quarterly 81.2 (2000), pp. 189–214).

²² Frankfurt, Necessity, Volition, and Love, p. 150; Pojman, ‘Does Equality Trump Desert?’, p. 291.

²³ Avraham and Statman, ‘More on the Comparative Nature of Desert’, p. 326.

²⁴ Egalitarian considerations, broadly speaking, also figure in criticisms of differential punishment that are grounded on political or distributive values. But positions like Elizabeth Anderson's do not portray egalitarianism as a matter of retributive justice (‘What Is the Point of Equality?’, Ethics 109.2 (19990, pp. 287–337, at 288, 312). Since I am interested in a retributivist approach, I will set these brands of egalitarianism aside.

²⁵ Hurka, ‘Desert: Individualistic and Holistic’, pp. 52–7.

²⁶ See also McLeod, ‘On the Comparative Element of Justice’, pp. 129–30; Miller, ‘Comparative and Noncomparative Desert’, pp. 34–7.

²⁷ Kelly, ‘Desert and Fairness in Criminal Justice’, p. 71; Shelby, ‘Justice, Deviance and the Dark Ghetto’, p. 152.

²⁸ Gargarella, ‘Penal Coercion in Contexts of Social Injustice’, pp. 24–6; see also Anderson, Elizabeth, ‘Outlaws’, The Good Society 23.1 (2014), pp. 103–13CrossRefGoogle Scholar; Duff, Punishment, Communication, and Community; Kelly, ‘Desert and Fairness in Criminal Justice’, p. 68; McDermott, Daniel, ‘A Retributivist Argument against Capital Punishment’, Journal of Social Philosophy 32 (2001), pp. 317–33CrossRefGoogle Scholar; Pittman, ‘Punishment and Race’.

²⁹ Since my goal is to convince retributivists that differential punishment is a problem, and a problem of a different sort than is sometimes imagined, I make use of some relatively conservative presuppositions. I assume that punishment is legitimate, and that retributivism offers a reasonably satisfactory general justification of punishment. These are contestable assumptions, of course, and adopting them risks obscuring the extent to which the administration of the American criminal justice system is a system of racialized social control. However, they are required by my dialectical aims. Furthermore, I find it interesting to see how differential punishment violates retributive norms, even if retributive institutions might be unjust in their application.

³⁰ Avraham and Statman, ‘More on the Comparative Nature of Desert’.

³¹ Avraham and Statman, ‘More on the Comparative Nature of Desert’, p. 328.

³² Avraham and Statman, ‘More on the Comparative Nature of Desert’, p. 323.

³³ Avraham and Statman, ‘More on the Comparative Nature of Desert’, p. 326.

³⁴ Avraham and Statman, ‘More on the Comparative Nature of Desert’, p. 323.

³⁵ Avraham and Statman, ‘More on the Comparative Nature of Desert’, p. 323.

³⁶ Avraham and Statman, ‘More on the Comparative Nature of Desert’, p. 319. Avraham and Statman label themselves comparativists to register their belief that a comparative review of sentencing data can help reveal when sentences are a product of disrespect. But if this epistemic conviction is all there is to comparativism, the concept is so loose as to be meaningless. Even non-comparative rigourists like van den Haag count as comparativists in this sense.

³⁷ <http://www.washingtonpost.com/blogs/wonkblog/wp/2013/08/13/heres-what-you-need-to-know-about-stop-and-frisk-and-why-the-courts-shut-it-down/>. The title of this article is inaccurate, as the decision was overturned on appeal.

³⁸ See the expert report of Jeffrey Fagan, available at <http://ccrjustice.org/files/Expert_Report_JeffreyFagan.pdf>.

³⁹ Some argue that implicit cognitions are beliefs, others that they are aliefs (Michael Brownstein, ‘Implicit Bias', Stanford Encyclopedia of Philosophy, <http://plato.stanford.edu/entries/implicit-bias> (2015)). Neither this nor related controversies have any bearing on my argument.

⁴⁰ <https://implicit.harvard.edu/implicit/selectatest.html>. For a detailed discussion of IATs, see Kang, Jerry et al., ‘Implicit Bias in the Courtroom’, UCLA Law Review 59 (2012), pp. 1124–86Google Scholar; Kelly, Daniel and Roedder, Erica, ‘Racial Cognition and the Ethics of Implicit Bias’, Philosophy Compass 3.3 (2008), pp. 522–40CrossRefGoogle Scholar, at 524–5.

⁴¹ Kang et al., ‘Implicit Bias in the Courtroom’, pp. 1129–31; Brownstein, ‘Implicit Bias’, p. 7.

⁴² <http://www.nyclu.org/content/stop-and-frisk-data>.

⁴³ Ghandnoosh, ‘Black Lives Matter’, p. 9.

⁴⁴ Goff, Phillip et al., ‘The End of Innocence: Consequences of Dehumanizing Black Children’, Journal of Personality and Social Psychology 106.4 (2014), pp. 526–45CrossRefGoogle Scholar.

⁴⁵ Kang et al., ‘Implicit Bias in the Courtroom’, pp. 1136–9; Kelly and Roedder, ‘Racial Cognition and the Ethics of Implicit Bias’, p. 526.

⁴⁶ Ghandnoosh, ‘Race and Punishment’, p. 16.

⁴⁷ Ghandnoosh, ‘Race and Punishment’, pp. 14–16.

⁴⁸ Muhammad, Khalil, The Condemnation of Blackness: Race, Crime, and the Making of Urban America (Cambridge, 2010), p. 7 Google Scholar.

⁴⁹ Muhammad details how the 1890 Census inaugurated the use of statistics to link blackness and criminality. Not surprisingly, the data were dramatically flawed (The Condemnation of Blackness, pp. 2–5, et passim). For example, statistical reports failed to mention that blacks were liable for far more criminal offences than whites. Black intellectuals like W. E. B. Du Bois were some of the first to identify these problems.

⁵⁰ Muhammad, The Condemnation of Blackness, pp. 41, 76, 103, 273.

⁵¹ Correll, Joshua et al., ‘The Influence of Stereotypes on Decisions to Shoot’, European Journal of Social Psychology 37 (2007), pp. 1102–17CrossRefGoogle Scholar, at 1103. Blacks commit some crimes more frequently than whites, but whites also commit some crimes more frequently than blacks. Drunk driving, which is linked to the deaths of an estimated 10,000 people a year, is committed far more frequently by whites, who make up roughly 85 per cent of drunk driving arrests. It is quite telling that whiteness is not linked to the notion of ‘vehicular criminality’.

⁵² Ghandnoosh, ‘Race and Punishment’, p. 13.

⁵³ Goff et al., ‘The End of Innocence’.

⁵⁴ See, for example, Hawthorne, John and Stanley, Jason, ‘Knowledge and Action’, Journal of Philosophy 105 (2008), pp. 571–90CrossRefGoogle Scholar. I want to thank Geoff Pynn for helpful discussion of these issues.

⁵⁵ Haslanger, Sally, ‘Oppressions: Racial and Other’, Racism in Mind, ed. Levine, Michael P. and Pataki, Thomas (Ithaca, 2004), pp. 97–123 Google Scholar, at 100–7.

⁵⁶ See, for example Anderson, ‘What Is the Point of Equality?’ and Scheffler, Samuel, ‘What Is Egalitarianism?’, Philosophy & Public Affairs 31.1 (2003), pp. 5–39 CrossRefGoogle Scholar.

⁵⁷ Racial hierarchy is complex. A race can be lower on the hierarchy in some domains of social life, but not others. And social position is strongly affected by the intersection of race, class and gender; see Crenshaw, Kimberlé, ‘Mapping the Margins: Intersectionality, Identity Politics, and Violence against Women of Color’, Stanford Law Review 43.6 (1991), pp. 1241–99CrossRefGoogle Scholar.

⁵⁸ Mallon, Ron, ‘Passing, Traveling and Reality: Social Constructionism and the Metaphysics of Race’, Noûs 38.4 (2004), pp. 644–73, at 644CrossRefGoogle Scholar.

⁵⁹ Mills, Charles, Blackness Visible: Essays on Philosophy and Race (Ithaca, NY, 1998), p. 47 Google Scholar.

⁶⁰ Haslanger, ‘Oppressions: Racial and Other’, p. 114.

⁶¹ Matthew Fogg, a former US marshal and DEA agent, reports that DEA agents intentionally avoid wealthier neighbourhoods. See <http://thefreethoughtproject.com/dea-agent-drug-laws-intentionally-rich-communities/>. Former New York Police Chief Lee Brown casts the choice of enforcement targets in similarly repugnant terms: ‘it's easier for police to make an arrest when you have people selling drugs on the street corner than those who are [selling or buying drugs] in the suburbs or in office buildings. The end result is that more blacks are arrested than whites’; quoted in Fellner, Jamie, ‘Race, Drugs, and Law Enforcement in the United States’, Stanford Law and Policy Review 20.2 (2009), pp. 257–92Google Scholar, at 271.

⁶² Fellner, ‘Race, Drugs, and Law Enforcement in the United States’, p. 261.

⁶³ Of course, class also plays a role; see Reiman, Jeffrey and Leighton, Paul, The Rich Get Richer and the Poor Get Prison: Ideology, Class, and Criminal Justice (Boston, 2010)Google Scholar. Overall, the types of crimes poor people commit are more likely to be punished with incarceration than those committed by the rich. And blacks tend to be less well-off than whites. But class alone cannot explain drug sentencing disparities; there are almost twice as many whites as blacks living under the poverty line.

⁶⁴ Kang et al., ‘Implicit Bias in the Courtroom’, p. 1151.

⁶⁵ Kang et al., ‘Implicit Bias in the Courtroom’, p. 1151.

⁶⁶ The combined federal and state criminal caseload hovered around 21 million annually from 2001–2010 <http://www.courtstatistics.org/~/media/Microsites/Files/CSP/DATApercent20PDF/CSP_DEC.ashx>.

⁶⁷ I want to thank a reviewer for urging me to clarify this point.

⁶⁸ ‘A Living Death: Life without Parole for Nonviolent Offences’, American Civil Liberties Union, <https://www.aclu.org/criminal-law-reform/living-death-life-without-parole-nonviolent-offenses-0> (2013).

⁶⁹ ‘A Living Death’.

⁷⁰ Although this is a bit of armchair philosophizing, I would venture that non-comparative injustice flows from the fact that implicit stereotypes and attitudes are not practically inert and therefore influence how black Americans are arrested, charged and sentenced: the imagined gun generated by weapon bias turns a misdemeanour assault into felony aggravated assault; and the improperly high ascription of culpability renders black children more liable to be tried as adults. In both cases, black offenders are punished disproportionately to their non-comparative desert.

⁷¹ ‘A Living Death’.

⁷² Alexander, The New Jim Crow.

⁷³ The ‘Ban the Box’ campaign has been making strides in this area. See <http://www.nelp.org/campaign/ensuring-fair-chance-to-work/>. For a philosophical treatment of collateral consequences, consult Lafollette, Hugh, ‘Collateral Consequences of Punishment: Civil Penalties Accompanying Formal Punishment’, Journal of Applied Ethics 22.3 (2005), pp. 241–61Google Scholar.

⁷⁴ Hart, H. L. A., Punishment and Responsibility: Essays in Legal Philosophy (Oxford, 1968), pp. 5–6 Google Scholar.

⁷⁵ <http://www.npr.org/2014/05/19/312158516/increasing-court-fees-punish-the-poor>.

⁷⁶ In principle, whites suffer the same consequences. But the background fact that whites generally enjoy more wealth and economic opportunity ensures that the consequences are usually not as severe.

⁷⁷ Pittman, ‘Punishment and Race’, p. 118, see also 29.

⁷⁸ Susanna Siegel describes a feedback loop that operates at the individual level, whereby the ‘cognitive penetration’ of perceptual experience by pre-existing beliefs illicitly reinforces those beliefs (‘Cognitive Penetrability and Perceptual Justification’, Noûs 46.2 (2012), pp. 201–22, at 202). In the situation I am discussing, cognitive penetration leads to arrest data that illicitly reinforce the original beliefs about blackness and criminality.

⁷⁹ <http://www.chicagotribune.com/news/opinion/chapman/ct-garner-brown-ferguson-police-brutality-crime-blacks-perspec-1207-jm-20141205-column.html>. For a philosophical version of this response, see Boonin, David, Should Race Matter? Unusual Answers to Unusual Questions (Cambridge, 2011), p. 346 CrossRefGoogle Scholar.

⁸⁰ This admission highlights, rather than diminishes, the significance of the argument on offer, which establishes the responsiveness of retributivism to racial oppression, and steers clear of controversies regarding the priority of distributive or retributive justice.

⁸¹ Michael Cholbi develops an interesting argument along these lines regarding disparities in capital sentencing (‘Race, Capital Punishment, and the Cost of Murder’).

⁸² By my lights, when a white offender is sentenced too leniently, there is a comparative and non-comparative injustice, but the offender is not wronged. But those who subscribe to the view that offenders have a right to be punished will maintain that such offenders are retributively wronged, because they do not receive the punishment they deserve. A well-known proponent of this view is Morris, Herbert, ‘Persons and Punishment’, The Monist 52 (1968), pp. 475–501 CrossRefGoogle Scholar.

⁸³ To repeat, egalitarianism is a very large family of views, not all of which are committed to the spartan apparatus of comparative egalitarianism. My view could be labelled egalitarian in the broad sense; see, e.g., Anderson, ‘What Is the Point of Equality?’. John Pittman helpfully pushed me to precisify my criticism of egalitarianism.

⁸⁴ Hurka asserts that comparative injustice should be ameliorated by levelling down (‘Desert: Individualistic and Holistic’, pp. 54–6), but his argument for this view is not terribly persuasive.

There is a more general problem here. The comparative egalitarian principle can demand treating P too harshly just because Q has been treated too harshly, heaping wrong on top of wrong. Some egalitarians try to get around this unwelcome conclusion by grounding the like cases principle on the claim that individuals deserve equal treatment by virtue of their equal moral standing; see, e.g., Knight, Carl, ‘Describing Equality’, Law and Philosophy 29 (2009), pp. 327–65CrossRefGoogle Scholar, at 338 ff.; O'Brian, William E. Jr, ‘Equality in Law and Philosophy’, Inquiry 53.3 (2010), pp. 257–84CrossRefGoogle Scholar, at 261. This more robust view enables egalitarians to deny that legal institutions must replicate disproportionately harsh treatment, but it also squeezes a healthy dose of respectarianism into egalitarianism. It requires egalitarians to endorse the claim that the problem with differential punishment lies in treating an offender as less worthy than he is. The result is a hybrid egalitarianism that inherits the deficiencies of respectarianism.

⁸⁵ See, for example, the curriculum at <http://www.fairimpartialpolicing.com>, which has been adopted by several police departments. The National Center for State Courts has promoted a variety of strategies for mitigating implicit bias and its effects (Pamela M. Casey et al., ‘Helping Courts Address Implicit Bias: Resources for Education’, (National Association for State Courts, 2012), Appendix G).

⁸⁶ Destiny Peery, ‘Implicit Bias Training for Police May Help, but It's Not Enough’, <http://www.huffingtonpost.com/destiny-peery/implicit-bias-training-fo_b_9464564.html> (2016).

⁸⁷ Peruche, B. Michell and Plant, E. Ashby, ‘The Correlates of Law Enforcement Officers’ Automatic and Controlled Race-Based Responses to Criminal Suspects’, Basic and Applied Social Psychology 28.2 (2006), pp. 193–99CrossRefGoogle Scholar.

⁸⁸ Casey et al., ‘Helping Courts Address Implicit Bias’.

⁸⁹ Ghandnoosh, ‘Black Lives Matter’, pp. 19–20.

⁹⁰ I am attracted to a slightly stronger position, namely, that the argument developed above furnishes what Joseph Raz calls a ‘protected reason’ for equalizing punishment (Practical Reason and Norms, 2nd edn. (Oxford, 1999)). A protected reason is a first-order reason for action that is accompanied by a second-order exclusionary reason for not acting on (a specified range of) competing first-order reasons. More specifically, the claim that tempts me is this: when we are uncertain about crime rates, differential punishment provides a first-order reason for equalizing punishment rates, and a second-order reason that prohibits the development of policies based on crime data.

That said, I am sympathetic to comparativism regarding desert bases, and so my own view is that the retributive principle in question may be overly restrictive.

⁹¹ As I noted above, whites are much more likely to be arrested for driving under the influence. According to the National Highway Traffic Safety Administration's self-report survey, whites are responsible for 84 per cent of drink-driving trips; offence rates thus seem to closely track arrest rates. See <http://www.nhtsa.gov/About+NHTSA/Traffic+Techs/current/Racial+And+Ethnic+Differences+In+Drinking+And+Driving+Attitudes+And+Behaviors>. Whether self-report surveys regarding criminal activity constitute good evidence is not a matter that can be discussed here.

⁹² I want to thank Anthony Reeves and John Pittman for their insightful comments and suggestions. The article also benefited from discussion with Geoff Pynn and others at the Bowling Green State University Workshop on the Ethics of Policing and Prisons, as well as from Utilitas reviewers’ constructive criticisms.