My own view on the prison system has to start with one of my favourite articles on
drug peace so far. In particular, this passage;
Fourth, all drug control policies, including enforcement, should be subjected to cost-benefit tests: We should act only when we can do more good than harm, not merely to express our righteousness. Since lawbreakers and their families are human beings, their suffering counts, too: Arrests and prison terms are costs, not benefits, of policy. Policymakers should learn from their mistakes and abandon unsuccessful efforts, which means that organizational learning must be built into organizational design. In drug policy as in most other policy arenas, feedback is the breakfast of champions.
I think you can expand this even more broadly, incorporating the concept into at least all non-violent crime.
Imprisoning someone always has a negative cost both to the individual and to society at large. The current justice system does not take that into account at all, treating it as a net positive for society (criminals off the street, regardless of if it's necessary or not) and ignoring any cost to the prisoner themselves (after all, they deserve it...). A custodial sentence should be strongly justified by the realities of the situation, not just the crime committed.
The problem is that for this to work you would need judges and prosecutors who took the same view. Those are the positions where discretion lies and any such system would rest strongly on liberal discretion being applied. In particular sentencing guidelines would need to be opened up to allow for a wider range of situations to be covered by a wider range of punishments. Narrow guidelines (and especially mandatory minimums) result in absurd results and more time behind bars than could ever make sense. But abolishing them means putting a lot of power into the judges' hands. Simply reviewing sentencing guidelines and not changing anything else would have a completely unpredictable effect as each judge applies their own views to the new situation. However, new guidelines intending to tie their hands and make them all progressive liberals would likely be just as problematic.
To crime prevention I have my own thoughts
TM, especially with the UK/US comparison. But broadly I don't think those alone solve the problems in the USA. The prison/crime feedback loop is far too entrenched. Focusing more on prevention alongside prison reform makes sense and is broadly the only way you could do such reform, but without the prison reform I would expect it to fail, hard. Practically and politically.