The harm we can’t agree on

In June 1994, Congress passed and President Clinton signed the Violent Crime Control and Law Enforcement Act — the largest crime bill in American history. $30 billion, 100,000 new police officers, expanded federal death penalties, mandatory minimum sentencing provisions, the “three strikes” clause for federal felonies. The bill commanded bipartisan support and was backed by a majority of the Congressional Black Caucus, whose members represented the communities most devastated by the crack cocaine epidemic and escalating violent crime of the preceding decade.

By 2016, Bill Clinton was publicly apologizing for it. By 2020, “the crime bill” had become shorthand for mass incarceration — its mandatory minimums, its sentencing enhancements, its contribution to a prison population that peaked at 2.3 million in 2008. The same legislation designed to protect Black communities was being cited as evidence of systematic racial harm to those communities. Both claims were grounded in something real.

By 2023, the United States held 1.9 million people behind bars — more than any country on earth in absolute numbers, and among the highest rates per capita at around 530 per 100,000 residents. The Bureau of Justice Statistics finds that 44 percent of state prisoners are rearrested within one year of release and 64 percent within three years. The costs to governments exceed $80 billion annually. Crime fell sharply from the early 1990s through the mid-2010s, then rose in some categories through 2021 before declining again — a trajectory that confounds simple narratives in every direction.

What is striking about these numbers is not what they say about any particular policy — it is what every position in the criminal justice debate does with them. Advocates for tougher enforcement cite incarceration’s contribution to the crime drop. Advocates for reform cite recidivism rates as evidence that incarceration produces more crime than it prevents. Abolitionists cite mass incarceration itself as the harm that policy must address. Every position looks at the same numbers and finds confirmation.

This is not because the evidence is ambiguous, though it often is. It is because the evidence cannot answer the prior question: what is the criminal legal system for? Until that question is answered — and it has not been, collectively, by any democratic process — no evidence about outcomes can settle the dispute. The maps Ripple has been building across this cluster, taken together, show why.

The hidden common structure

Mapped individually, these debates look like separate policy questions. Criminal sentencing is a debate about mandatory minimums and how long sentences should be. Police reform is a debate about use-of-force accountability and whether policing can be fixed from within or must be fundamentally reimagined. Drug policy is a debate about prohibition versus harm reduction versus legalization. Prosecutorial discretion is a debate about DA accountability and the scope of charging authority. Predictive policing is a debate about algorithmic tools and civil liberties. Juvenile justice is a debate about how the system should treat young people who break the law.

But every one of these maps, looked at structurally, is an argument about an unresolved question that American democracy has never settled: what is the coercive apparatus of the state for, when someone has done harm?

That question has five canonical answers in the philosophical and criminological literature, each with its own logic and its own policy implications.

Retribution holds that people who commit serious wrongs deserve to suffer proportionate consequences — not to deter future crime, not to rehabilitate the offender, not to protect society, but because justice requires it. The Kantian formulation: the offender who escapes punishment is the one who has been wronged by the universe. The criminal legal system reform map surfaces this in the opening vignette about a murder victim’s sister who does not want revenge but does want the sentence served — because a reduced sentence would mean her brother’s death was not taken seriously.

Deterrence holds that punishment prevents future crime by raising its expected cost — rational actors weigh costs and benefits, so making crime more costly reduces crime. Three-strikes laws, mandatory minimums, truth-in-sentencing provisions are all justified primarily on deterrence grounds. The criminological evidence is nuanced: certainty of punishment deters more than severity, which means most mandatory minimum laws are structured precisely backwards relative to what actually deters crime.

Incapacitation holds that separating dangerous people from society protects potential victims regardless of whether incarceration deters, rehabilitates, or delivers just deserts. People in prison cannot victimize free people — at least not free people outside the prison. The argument drives selective incapacitation strategies, career-criminal statutes, and life sentences for persistent recidivists.

Rehabilitation holds that the goal should be transforming offenders so they do not reoffend — addressing the addiction, trauma, mental illness, skill deficits, and social conditions that produced the behavior in the first place. Drug courts, education programs, therapeutic communities, reentry supports. The argument that prison as currently designed is a machine for manufacturing more crime rather than preventing it.

Restoration holds that the goal should be repairing the harm to victims and communities through accountability, acknowledgment, and reparative action — not primarily through punishment. Howard Zehr’s restorative justice model. Community conferencing. Victim-offender mediation. The argument that the current system largely excludes actual victims from meaningful participation in their case while delivering neither healing nor genuine accountability.

The American criminal legal system, built through two centuries of legislation, judicial decisions, and administrative practice, attempts to serve all five purposes simultaneously. It does not. A sentence optimized for retribution is often terrible for rehabilitation — long sentences in conditions that are frequently degrading and sometimes violent tend to increase recidivism, not reduce it. Mandatory minimums, designed for deterrence, eliminate the judicial discretion required for proportionality, producing sentences that many participants in the system describe as unjust by any theory. Plea bargaining, which resolves 97 percent of federal criminal cases, is neither retributive (sentences are negotiated, not measured to the offense), nor restorative (victims rarely participate), nor rehabilitative (the process offers no programmatic intervention).

This is not primarily a design flaw — it is an accumulation of responses to perceived failures. When rehabilitation programs produced disappointing recidivism results in the 1970s, Robert Martinson’s “nothing works” pessimism shifted policy toward retribution and incapacitation. When decades of mass incarceration produced the collateral consequences documented in Michelle Alexander’s The New Jim Crow — the civic exclusion, the neighborhood destabilization, the multigenerational harm — the pendulum shifted back toward rehabilitation and, more recently, restoration. Each shift layered new tools onto old ones without resolving the underlying disagreement about purpose. The result is a system that tells itself it serves all five goals while the specific tools it uses often work against each other.

The behavioral attribution problem

The second structural tension cuts beneath the purpose question to something more fundamental: how much of criminal behavior is chosen, and how much is caused?

The choice framing holds that crime is primarily a product of decision-making. People weigh costs and benefits, make judgments about risk and reward, and act on the basis of their preferences and values. The policy implication follows: raise the expected cost of crime, and crime will decline. The offender is morally responsible because they chose; the sentence is justified because they chose. This is the animating logic of deterrence theory, mandatory minimums, and the tough-on-crime tradition.

The circumstance framing holds that behavior is substantially shaped by conditions the actor did not choose: poverty, neighborhood environment, exposure to violence in childhood, educational opportunity, lead exposure, parental incarceration, untreated mental illness, addiction. The policy implication also follows: address the root causes, and crime will decline more sustainably than punishment alone can achieve. Punishing people for behavior significantly determined by conditions they did not choose is, on this view, at best ineffective and at worst itself an injustice.

Both framings are true simultaneously. This is what makes the tension structurally intractable. A person who commits robbery under economic duress still, in some sense, chose to do so — others in similar circumstances do not commit robbery. A person who commits robbery with what looks like full agency still had a developmental history, a neighborhood, an economy they did not choose, that shaped who they became. Neither the choice framing nor the circumstance framing is wrong. They are selecting different aspects of a reality in which behavior is both chosen and caused.

The philosophical literature on free will and moral responsibility maps almost exactly onto the criminal justice policy debate — and arrives at the same non-resolution. Compatibilists hold that genuine choice and causal determination are compatible. Hard determinists hold that if behavior is fully caused, it is not morally responsible in a meaningful sense. Libertarians (in the philosophical sense) hold that some acts are genuinely uncaused. The criminal justice argument between “personal responsibility” advocates and “root causes” advocates follows this same structure, with the same inability to resolve.

What this means practically: no amount of evidence about deterrence effects, recidivism rates, neighborhood conditions, or childhood trauma resolves the behavioral attribution question. Evidence can shift how much weight someone gives to choice versus circumstance. It cannot determine how much weight they should give, because that determination is a moral judgment, not an empirical finding. This is why criminal justice debates often feel like the participants are speaking different languages — because they are not reasoning from different facts toward the same conclusion; they are starting from different premises about human agency and moral desert, and interpreting the facts through those premises.

Where both tensions operate at once: drug policy

The most clarifying example of both structural tensions operating simultaneously is drug policy. The drug policy map and the drug sentencing disparity map surface four distinct positions: prohibition (crime and punishment), harm reduction (public health response), decriminalization (remove criminal sanctions while maintaining social response), and legalization (treat like alcohol or tobacco).

The prohibition position is coherent on its own terms: drug use is a choice; that choice has serious consequences for the user and for others; criminal sanctions deter use and protect communities. This position is consistent with a deterrence or retributive account of what criminal justice is for.

The harm reduction position is also coherent: addiction is a disease process, not a moral failing; punishing illness is wrong and counterproductive; the goal should be reducing death and suffering through treatment, needle exchanges, naloxone distribution, and supervised consumption sites. This position is consistent with a rehabilitation or restoration account.

What makes drug policy uniquely clarifying is that it surfaces the behavioral attribution question with unusual sharpness. The neurobiological literature on addiction suggests it involves both: addictive substances alter the neural reward circuitry in ways that progressively compromise the experience of choice, even as choice-making processes are never entirely absent. The political debate cannot hold this complexity because it requires taking a prior position on desert — whether the addicted person is someone doing something wrong to themselves and others, or someone to whom something wrong is being done.

The racial disparity data makes this sharper still. Black and white Americans use drugs at roughly similar rates. Black Americans are arrested for drug offenses at significantly higher rates, prosecuted more aggressively, and sentenced more severely — documented at every stage in the sentencing disparity map. If drug use is primarily a choice, these disparities require an explanation that the usage-rate data does not provide. If drug use is substantially shaped by circumstance, the circumstances differ significantly by race in ways that historical policy — redlining, disinvestment, the war on drugs itself — substantially produced. Neither framing settles the policy question. But making the framing visible changes what the disagreement is actually about.

The prosecutorial discretion problem

A third structural feature of the cluster, less visible than the first two, is the concentration of consequential judgment in actors who are nearly unaccountable for its exercise.

The prosecutorial discretion map documents that American prosecutors — particularly district attorneys at the county level — have near-total authority over whom to charge, what charges to bring, what pleas to offer, and when to decline. That authority is nominally constrained by law and professional ethics rules. It is practically constrained by political culture, public attention, and the rare case that produces appellate scrutiny of prosecutorial conduct.

What the map surfaces: a prosecutor who declines to bring charges against a politically connected defendant faces almost no institutional consequence. A prosecutor who brings controversial charges against a popular figure faces intense political scrutiny. The accountability mechanisms are weakest precisely where the power is greatest. And because most cases end in plea bargains — negotiated before any public proceeding — prosecutorial judgment is exercised mostly invisibly, with no record and no review.

The reform-versus-status-quo DA elections of the 2020s (Larry Krasner in Philadelphia, George Gascón in Los Angeles, Kim Foxx in Chicago, Chesa Boudin in San Francisco — and Boudin’s subsequent recall) map exactly onto the purpose plurality problem. Progressive DAs who prioritize rehabilitation and restoration face critics who prioritize incapacitation and retribution. Their recall campaigns are conducted as if the question is about competence, when the actual question is about which theory of what a prosecutor is for should govern charging decisions.

This structure appears across the cluster. The predictive policing map shows algorithmic risk tools concentrating surveillance on communities whose historical over-policing generates the data that justifies further policing — a feedback loop in which the past choices of the system become the predictors of future individual behavior. The juvenile justice map shows young people, whose brains are still developing in ways that reduce impulse control and risk assessment, being processed through systems designed for adults and emerging with records that follow them into adulthood. The disability and criminal justice map documents that roughly one-third to one-half of people in jails and prisons have a disability — a concentration that reflects both the criminalization of symptoms (psychosis, untreated mental illness, homelessness) and the system’s failure to provide appropriate accommodation.

In each case, the question underneath the policy debate is the same: what does the system owe the people it processes, and what do those people owe the system? The answers differ because the purposes differ, and the purposes differ because the underlying account of moral responsibility differs.

What this means for the debates

None of this settles the genuine disagreements within the cluster. The retributive case is real: there are victims for whom the experience of being taken seriously by the system matters — for whom a sentence commuted or a charge declined constitutes a second injury, a statement that what happened to them did not merit the full weight of social response. A system that cannot produce that experience is failing something important.

The rehabilitation case is also real: a system that produces 64 percent rearrest rates within three years is not protecting communities. It is cycling people through conditions that increase the probability of future harm to the same communities that experience the most crime. If the goal is public safety, the current system is a poor tool for achieving it.

The structural critique is also real: the communities that bear the highest burden of both violent crime and mass incarceration are often the same communities — predominantly low-income and disproportionately Black and Brown — and the system designed to protect them frequently does not distinguish between protection and predation in how it deploys its coercive authority. Patrick Sharkey’s research on the crime decline of the 1990s and 2000s documents that falling homicide rates saved tens of thousands of lives, most of them Black lives — which means that the political failure to take violent crime seriously also has racial justice consequences. This is the genuine complexity that neither pure reform nor pure status-quo positions can hold.

What the cluster reveals is the structural question all of these positions are answering differently: what do we do with people who have harmed others, when the harm, the response, and the prevention are all entangled with conditions of inequality and deprivation that the harmed and the harming often share?

The framing matters. Debates within the cluster are often conducted as if the question is a narrow policy question — what sentences should be, whether police should be armed, how many prosecutors a county needs. They are also that. But beneath the specific question in each case is the same philosophical one: what is the criminal legal system for, and does the answer change when the behavior being punished was substantially caused by conditions the punishing society helped create?

The purpose plurality problem — five incompatible goals layered on a single system — means that every reform that solves one problem risks creating another. Eliminating mandatory minimums restores judicial discretion and proportionality, but reintroduces the racial disparities in sentencing that mandatory minimums were partly designed to address. Expanding plea bargaining reduces case backlogs but concentrates consequential judgment in prosecutors with no public record. Community-based alternatives to incarceration serve rehabilitation and restoration but leave open the retributive and incapacitation questions for those who prioritize them. No policy choice can satisfy all five purposes simultaneously.

The behavioral attribution problem means that every policy choice embeds an implicit claim about human agency that the policy cannot prove. Mandatory minimums assume that the expected cost of a long sentence affects decision-making in the moment of a crime — an assumption that criminological research on impulsive, intoxicated, and emotionally dysregulated decision-making substantially challenges. Root-cause investment assumes that better circumstances reliably produce better choices — an assumption that is broadly supported but that cannot account for the choices of individuals within improved circumstances who still cause serious harm.

What making both problems visible does: it allows participants in the debate to be more honest about what they are actually arguing. The argument for longer sentences is primarily a retributive or incapacitative argument — not a deterrence argument, which the evidence does not support. The argument for decarceration is primarily a rehabilitation or restoration argument — and needs to answer the retributive and incapacitative cases directly, not simply cite recidivism rates. The argument for abolition is primarily a structural argument about an institution that has metastasized beyond any defensible purpose — and needs to be specific about what would replace the incapacitative and retributive functions that communities affected by serious violence actually need.

The harm we can’t agree on is both real. The harm done by crime to the people it touches is real and serious and deserves a serious societal response. The harm done by the system’s response to crime — to the incarcerated, to their families, to their neighborhoods, to the civic fabric — is also real and serious and deserves a serious societal response. These are not equivalences. They are both things that have to be in the picture. What the maps in this cluster are trying to build is a picture large enough to hold both.