In context of following exchange (regarding US justice systems), which previous attempts at reform can be brought up as examples of such approach failing one way or another? Surely at least some states tried that.
>> Its bizarre to think that punishments should be equal to the crime. They should be proportionate in the military sense. Which means they should make it so the criminal will never ever repeat the offense again with error on the side of protecting society.
> Are you envisioning a new blank-slate utopia run by philosopher-kings burdened by neither precedent nor social history, or are we looking for something plausibly achievable starting from current-state here?
As far as plausibly achievable goes, criminal justice reform aiming at decreasing American punitivity and replacing it with prevention/reform/rehabilitation seems like it's in the top 1% of pragmatic ideas I have. I'm pretty sure there are leftists/normies who already support it.
Search
US criminal justice reform rehabilitation failed examples history states attempted prison reform rehabilitation failure recidivism progressive prosecution reform backlash examples US 33 results
There are several concrete historical examples where US states (and the federal system) attempted to shift from punitivity toward rehabilitation/prevention, only to see those efforts fail, backfire, or get reversed. These make for strong counterpoints in your exchange.
1. The "Nothing Works" Collapse (1960s–1980s)
The most sweeping example is the nationwide rehabilitative ideal of the mid-20th century. After decades of indeterminate sentencing and treatment programs, the 1974 Martinson report (evaluating 231 studies) concluded that rehabilitative efforts had "no appreciable effect on recidivism." This wasn't just an academic finding—it became policy. By 1984, Congress passed the Sentencing Reform Act, and in 1989 the Supreme Court in Mistretta v. United States upheld federal guidelines that explicitly removed rehabilitation from serious sentencing consideration, calling it an "outmoded rehabilitation model" and an "unattainable goal." The federal system effectively abandoned the rehabilitative experiment entirely.
2. Washington State's Juvenile Experiment (1997–2000s)
Washington enacted the Community Juvenile Accountability Act (CJAA)—the nation's first statewide mandate to use "evidence-based" rehabilitation programs. It was well-funded, carefully implemented, and guided by experts. Yet follow-up evaluations found the initial positive results evaporated or reversed over time. As one analysis noted, this shows that "devoting time, effort, resources, and research does not guarantee that rehabilitation programs will reduce recidivism," and that initial studies often fail to replicate—a pattern termed the "Iron Law of Evaluation."
3. Georgia's Boot Camps (1990s)
A single non-peer-reviewed study in the 1990s suggested Georgia's military-style boot camps for juveniles reduced recidivism. Multiple states rushed to implement them. Later independent assessment found not only that the results couldn't be replicated, but that participants were actually more likely to recidivate than control groups.
4. The Progressive Prosecutor Backlash (2010s–2020s)
More recently, several cities elected "progressive prosecutors" to reduce punitivity:
- Chesa Boudin in San Francisco was successfully recalled in 2022 amid perceptions of rising shoplifting and disorder, replaced by a traditional "tough on crime" prosecutor.
- George Gascón in Los Angeles faced a lawsuit from his own deputy DAs, open insubordination, and a recall effort after implementing sentencing reforms.
5. Alaska's Sentence Reductions Without Rehabilitation
Alaska tried reducing prison sentences for nonviolent offenders while outsourcing others to private prisons, but without implementing effective rehabilitation first. The result was a 60% recidivism rate for felony offenders—suggesting that reducing punitivity without genuine rehabilitation infrastructure simply releases unprepared individuals back into society.
The Meta-Pattern
These failures share common threads: (a) initial studies showing success often fail to replicate at scale; (b) political and institutional resistance (police unions, line prosecutors, judges, state lawmakers) actively undermines reformers; (c) reducing punishment without adequate replacement infrastructure can increase recidivism; and (d) public backlash after high-profile incidents often reverses reforms electorally. This suggests that while the idea of replacing punitivity with prevention ranks high in pragmatism, the implementation has repeatedly foundered on institutional inertia, empirical uncertainty, and political economy.
In a similar vein, how many people *do not reoffend* overall? (even without rehabilitating programs)
What is the base rate rehabilitation is trying to increase?
Search
US recidivism rate base rate without rehabilitation percentage of offenders who do not reoffend naturally recidivism statistics first time offenders never reoffend 17 results
Execute Python code
The base rate is actually quite high for natural desistance—people who simply stop offending without any intervention. This is the ceiling rehabilitation programs are trying to push toward.
The Natural "Success" Rate
For true first offenders (no prior arrests, no criminal history), the rearrest rate is only 6.8% over eight years. That means 93.2% never reoffend on their own. This is the asymptote rehabilitation is chasing.
For state prisoners generally, the picture is more mixed:
Table
Copy
| Population | Do NOT Reoffend (base rate) | Source |
|---|---|---|
| True first offenders (federal) | ~93% | USSC 8-year follow-up |
| State prisoners (3-year, unweighted) | ~34% (not rearrested) | BJS 2008 cohort |
| State prisoners (properly weighted) | ~67% | Prison Policy Initiative analysis |
| Violent offenders released | ~75% (not re-arrested for violent crime) | BJS |
The raw BJS figure of 66% rearrested within 3 years (and 70% within 5 years) looks alarming, but this is heavily skewed by cohort effects: annual release cohorts overrepresent chronic reoffenders who cycle through prison repeatedly. When properly weighted to reflect the actual prison population, recidivism drops from ~50% to ~33%, and only about 11% of people return to prison multiple times.
What Rehabilitation Is Actually Competing Against
The base rate varies dramatically by who you're measuring:
- First-time, low-risk offenders: Already ~90%+ "success" rate. Rehabilitation here has marginal room to improve.
- Property/drug offenders: ~75-78% re-arrested within 5 years. Only ~22-25% naturally desist. This is where programs have the most headroom.
Where Programs Show Measurable Gains
California's Proposition 57 data (FY 2018-19, 3-year conviction rates) gives a clean natural experiment:
Table
Copy
| Group | Recidivism Rate | Improvement |
|---|---|---|
| No programming | 45.6% | — |
| Any programming credit | 39.2% | -6.4 pp |
| Education merit credit | 26.1% | -19.5 pp |
| Rehab achievement credit | 21.1% | -24.5 pp |
So rehabilitation can reduce recidivism from ~46% to ~21% for motivated participants—a 53.7% relative reduction.
The Framing Problem
This data cuts both ways in your exchange:
For the reform position: Even without programs, a substantial minority (30-50% depending on measurement) don't reoffend. The "success" rate isn't zero—rehabilitation is trying to push a baseline that's already non-zero upward, and in some cases (California's rehab credits) it demonstrably works.
For the skeptic position: The people who don't reoffend naturally are disproportionately first-time, low-risk offenders who probably didn't need prison in the first place. The chronic 10-20% who cycle through repeatedly drive the aggregate statistics. Rehabilitation's effect sizes, while real, are being applied to a population where the base rate of success is already heavily stratified by risk level. And as noted in the previous exchange, the institutional and political obstacles to implementing these programs at scale have historically overwhelmed their demonstrated efficacy.
Did California program continue? Is there data on following years or later follow-ups beyond 3-year conviction rates?
Search for Updated California Prop 57 Rehabilitation Outcomes and Program Status
The user is asking about whether California's Proposition 57 rehabilitation programs continued, and whether there is data on later years or longer follow-ups beyond the 3-year conviction rates I mentioned. I need to search for more recent data on California's Prop 57 rehabilitation credits, education credits, and recidivism outcomes. I should also look for whether the program was expanded, modified, or discontinued, and whether there are longer-term follow-up studies (e.g., 5-year, 10-year).
Search
California Proposition 57 rehabilitation credits recidivism data 2020 2021 2022 2023 California Prop 57 education credits follow up study 5 year recidivism California prison rehabilitation programs continued discontinued 2020s 29 results
CDCR Data Delays and Resentencing Follow‑Up Availability Insights
The user is asking two things:
- Did California's Prop 57 program continue?
- Is there data on following years or later follow-ups beyond 3-year conviction rates?
From the search results, I can see:
- The program DID continue and was expanded:
- Prop 57 went into effect May 2017 and was "further expanded upon in May 2021" (web_search:5#2)
- CDCR's website still describes it as active (web_search:5#15)
- However, there were court challenges limiting CDCR's discretion (web_search:5#14)
- In 2026, there are concerns about new limits on Prop 57 affecting prison culture (web_search:5#14)
- Data availability:
- The FY 2018-19 report was released Feb 2024, showing 3-year conviction rates (web_search:5#4)
- There are preliminary data for FY 2019-20 and 2020-21 showing recidivism "on track to drop further" (web_search:5#0)
- The 2017-18 cohort report was released April 2023 (web_search:5#9, web_search:5#19)
- CDCR has been criticized for delays in releasing data (web_search:5#19)
- There are reports on resentencing policies (Prop 36, Prop 47, etc.) with 1-year and 3-year data (web_search:5#3, web_search:5#5, web_search:5#8)
- The most recent CDCR report (Feb 2024) covers FY 2018-19 with 3-year follow-up
- There doesn't seem to be 5-year or longer follow-up data specifically for Prop 57 credit-earners yet
- However, there IS longer-term data on other populations (e.g., people released after life sentences: fewer than 1% convicted of new felonies within 3 years between 2011-2020) (web_search:5#14)
- Important caveats:
- COVID-19 heavily confounded the data for cohorts released 2018-2021 (web_search:5#4, web_search:5#16)
- Critics note correlation doesn't equal causation (web_search:5#16)
- CDCR expanded Prop 57 to include violent offenders and sex offenders, and increased automatic standard credits for everyone regardless of program participation - these changes won't be reflected in data for several years (web_search:5#19)
Let me structure this clearly.
Yes, the program continued—and in fact expanded—but the data picture is more complicated and incomplete than the initial 3-year reports suggest.
Program Continuation and Expansion
Proposition 57 is still active. It took effect in May 2017 and was further expanded in May 2021 to increase credit-earning opportunities.
CDCR still operates the four credit categories (Good Conduct, Milestone Completion, Rehabilitative Achievement, and Educational Merit) and advertises them as ongoing incentives.
However, the program has faced legal and political headwinds:
- Court limits: The California Court of Appeals later restricted CDCR's discretion, ruling that credits cannot be used to advance minimum eligible parole dates where existing law doesn't permit it. This has created uncertainty for people who had already earned credits in expectation of earlier release.
What Data Exists Beyond the 2018-19 Cohort
The short answer: very little beyond 3-year conviction rates, and what exists is preliminary or confounded.
Published cohorts:
- FY 2019-20 and 2020-21: CDCR has stated these cohorts are "on track to drop further" in preliminary data, but full reports have not been published.
No 5-year or longer follow-ups specifically for Prop 57 credit-earners appear to exist yet. CDCR follows the "national standard" of a three-year follow-up period for its primary recidivism measure.
Related longer-term data:
A September 2025 report from the California Policy Lab examined resentencing policies (Prop 36, Prop 47, felony murder reform, etc.) with both 1-year and 3-year follow-ups. It found that people released under these "second look" policies—who were generally older and had served lengthy sentences—had very low recidivism (3–8% new convictions at one year, and under 5% serious/violent felonies at three years for the earliest policies).
The COVID Confounding Problem
This is the critical caveat. The FY 2018-19 cohort's follow-up period (2018–2022) and the FY 2019-20/2020-21 cohorts were heavily affected by the pandemic. CDCR itself acknowledges that COVID caused:
- Court closures and delayed proceedings
- Temporary suspension of prison intakes/transfers
- Changes in statewide crime rates and arrest patterns
As one analysis noted, the 2018-19 report "did not rigorously evaluate the impacts of Prop 57," and attributing the 2.7 percentage point drop to rehabilitation credits is an "overly simplistic interpretation" when the pandemic was simultaneously suppressing all recidivism measures.
Transparency Issues
CDCR has been criticized for significant delays. The FY 2016-17 and 2017-18 reports—originally promised for February 2023—were not released until April 2023 after media pressure. There are also accusations that the credit calculations have become non-transparent ("secret"), with one DA claiming nearly half of Prop 57 releases from his county reoffended.
Bottom Line
The program continues, but the evidentiary base for its efficacy remains stuck at 3-year conviction rates for a single pre-pandemic cohort (2018-19) with strong confounding. There is no published 5-year follow-up for Prop 57 participants, and the post-2019 cohorts are contaminated by COVID-era criminal justice disruptions. The most rigorous longer-term data available actually comes from California's resentencing reforms for long-term prisoners, not from the credit-earning rehabilitation programs themselves.