#Resource The criminal justice system is the state's machinery of punishment — the police who arrest, the Crown Prosecution Service that charges, the courts that try, the prisons that hold, the probation service that watches on release. What it is *for* is a live argument: public safety, deterrence, rehabilitation, retribution, or the plainer management of the poor and the unruly. This note gathers the shape of it in England and Wales, and the case that it runs less as a system of *justice* than as an instrument of state power and social sorting. ## By the numbers The prison population of England and Wales reached roughly **87,300 in 2024** — the highest since records began in 1900, after a near-quadrupling from about 17,400 at the turn of the twentieth century. That is around **170 people imprisoned per 100,000**, among the highest rates in Western Europe. The system now runs permanently at the edge of capacity: operational capacity of about 89,000, a record population of **88,521 on 6 September 2024**, and official projections of **98,000–103,600 by March 2030** if nothing changes. The pressure valve is now release rather than reform. In **October 2026** the government began freeing prisoners early in stages — around **5,000 in the first tranche**, cut from an initial 6,000 once rapists and the most serious child-sex offenders were excluded, which is its own comment on the first version. The population dipped through the summer and then began to climb again. ## The courts The backlog is the bottleneck. At the end of **June 2026** there were **80,829 open Crown Court cases** — a record — up from about 80,400 in March and 78,200 a year earlier. A record **23,706 had been waiting a year or more**, up 22% in twelve months. The median wait reached **203 days**, against 197 in March and 168 the year before. The Ministry of Justice's own July 2025 modelling projected the backlog could pass **104,000 by 2029**, and the Justice Secretary has conceded it will take **a decade** to return to pre-Covid levels. The proposed fix — judge-only trials, shrinking the number of jury trials — was assessed by the **Institute for Government** as saving under **2%** of court time: a marginal gain bought by "damaging public confidence". And the crisis is uneven: roughly four in ten Crown Courts actually reduced their own backlog in the second quarter of 2026, while a small number grew steeply. Backlog is not a force of nature; it is a distribution. ## Reoffending Prison mostly fails on its own terms. The **proven reoffending rate** in England and Wales sits around **28–29%** overall. For people released from sentences of **under twelve months** — the majority of those sent to prison each year — it is about **56%**, roughly double. **Community sentences** are consistently associated with lower reoffending than equivalent custodial ones. Reoffending is estimated to cost around **£18 billion a year** once criminal-justice, victim and wider social costs are counted. Read together: the most common prison sentence is the shortest, the least rehabilitative, and the most reoffended-after. The system keeps handing down the thing that doesn't work, because the thing that does — early intervention, housing, treatment, community supervision — is the thing that doesn't fit the political grammar of *tough on crime*. ## Disproportionality The **2017 Lammy Review**, commissioned by David Cameron and led by David Lammy MP, found the system sorted by race at every stage. Black people were about **3% of the population but 12% of adult prisoners**; BAME people were **14% of the population but 25% of prisoners** and over **40% of young people in custody**. Had the prison population simply reflected the make-up of England and Wales, there would have been **more than 9,000 fewer people in prison** — twelve average-sized prisons' worth. Lammy found *greater* disproportionality in Black imprisonment here than in the United States, and concluded there was overt racial prejudice in the system — declining, with covert and unconscious bias taking its place. The disparities run the whole length of the pipe: - **Stop and search** — 52.6 searches per 1,000 Black people against 7.5 per 1,000 white (later 24.5 against 5.9), plus the strip-searches known as MTIPS, where Black Londoners were 46% of those searched in the months after Chris Kaba's death, from a 13.5% share of the population. - **Sentence length** — average custodial sentence **27.1 months for BAME offenders against 19.5 for white**, on the same offence categories. - **Children** — a third of children in custody were Black in 2020, against 13% of the adult prison population. ## Miscarriages of justice A system that convicts at this scale will convict the innocent; the question is what it costs to notice. **Peter Sullivan** served **38 years** — the longest-serving known miscarriage in British history — on bitemark "evidence" that DNA finally quashed in 2025. **Andrew Malkinson** served **17 years** for a rape he did not commit, after the Criminal Cases Review Commission missed repeated chances to correct it; a KC-led review later called those failures serious. The compensation scheme caps and contests the payouts it does make. Each case is a reminder that the state's power to punish is also a power to be *wrong* at scale — and that the errors cluster on the poor, the racialised, and the unrepresented. ## The critical frame Put the pieces side by side and the "justice system" starts to look like something else. It is **vast, expensive, and not very good at its stated purpose** — yet very effective at sorting along class and race, and at absorbing the fallout of poverty, addiction, and mental illness that other services have been starved of. Its failures are **structural, not individual**: a backlog is a funding choice, disproportionality is a policy outcome, a 38-year miscarriage is an institution protecting itself. And it is one face of the same thing that runs through this vault's politics — **the state's monopoly on legitimate violence**. The courtroom is the polite end of it and the cell is the blunt end, and the logic is continuous with [[R. J. Rummel|Rummel]]'s democide and with [[Structural Violence|structural violence]]: power that harms, then writes itself a warrant. It is also the machine the *legibility* critique predicts — see [[Seeing Like a State]] (James C. Scott the author — a different Scott from the barrister above): a system built to process categories rather than people will process people as categories, and call the wreckage "the backlog". For the specific argument that the state should not be trusted with the power to *kill*, not merely to cage, see [[The State Was Not a Party]] and [[Capital Punishment.canvas]]. And the enforcement problem is [[Making Human Rights a Reality|Hafner-Burton]]'s: international human rights law is built to catch direct abuses — the beating, the cell, the killing — while the quieter, structural harm a system like this administers is, by design, not the kind of thing a treaty or a tribunal is built to reach. ## Sources & reading - [BarristerBlogger](https://barristerblogger.com/) — **Matthew Scott**, a criminal barrister (Pump Court Chambers) who writes the most useful running commentary on the English system: miscarriages of justice, the futility of short prison sentences, and the slow strangulation of legal aid. Won best independent blog at the 2015 Comment Awards. This note began as that link; the blog is the place to watch the machinery in real time. - [[The Critic's Blind Spot]] · [[Cognitive Dissonance]] — for how people who are otherwise sharp about state power can go quiet when it is *their* state doing the harm. `Concepts:` [[Prison]] · [[Courts - Law]] · [[Violence]] · [[Structural Violence]] · [[Bureaucracy]] · [[Hierarchical]] `Knowledge Base:` [[Sociology]]