I've written several times that the most undercovered story in America is what's been called the conservative war on prisons. And while it's still criminally undercovered (no pun intended), things are beginning to bubble up.

Sen. Rand Paul (R-Ky.) is well-known for his efforts to end mandatory minimum sentencing — but he's not the only conservative honing in on the issue. Just last week, Sen. Mike Lee (R-Utah), who is shaping up to be the biggest intellectual force in the conservative movement and the Tea Party, went to Heritage Foundation to pitch his "conservative case for criminal justice reform."

Yes, the conservative war on prisons is real. This is because every branch of the conservative movement is starting to see eye to eye.

Christian conservatives are appalled at the inhumane treatment — and worse — of American prisoners. Fiscal conservatives realize that prison is an enormously expensive way to deal with crime in an era of straitened state budgets, an issue exacerbated by those darlings of the left, public sector unions. And even law and order conservatives are coming around to the idea that prisons too often end up being like a professional school for crime, breeding recidivism. Conservatives are increasingly realizing that a successful pitch to minorities can't be done without addressing the disproportionate racial impact of America's sentencing policies. And everybody's conscience is, or ought to be, shocked by horror stories about mandatory minimum sentences that send someone to prison for decades for shoplifting or buying marijuana.

All of this is great and good, and I applaud it. But there are a couple things that should be kept in mind.

Policy too often see-saws between extremes rather than taking the prudent middle path. We never regulate finance smartly; we just swing between underregulating it and overregulating it. Conservatives are supposed to be the reality-based community (to coin a phrase), those who are unmoved by the utopian schemes and whiggish narratives of the progressive mind, but instead are attuned to stubborn facts. And one stubborn fact is that there are psychopaths, there are sociopaths, and there are people who are and will remain violent. While it may be good policy overall to shorten prison sentences, it may lead at the margins to more crimes being committed, as some people are already fretting about California's Proposition 47.

So what's the solution? It's for conservatives to focus not just on prison sentences, drug sentences, mandatory minimums, and prison overcrowding — those are all important issues — but also on rehabilitation (and, in a dream world, policing, but we're not there yet). Pulling people out of jail is great if they lead productive lives, but not so great if they commit more crime.

How do we rehabilitate prisoners?

Anybody who tells you they know how to do that is either deluded or selling you the Brooklyn Bridge. Some approaches like Hawaii's Project HOPE are very promising, but that is no guarantee that something similar rolled out nationally would work.

Luckily, conservatives happen to have a very good approach to solving problems nobody knows the solution to: decentralized experimentation. In fact, this is how conservatives midwived one of the greatest reductions in crime in history.

The famous "Broken Windows" approach to policing, which has greatly reduced crime rates nationally over several decades (even though it is arguably reaching the point of diminishing returns now) was born out of a series of social science experiments on policing. As the entrepreneur and scholar Jim Manzi pointed out in the single most important article anyone can read about social science, those experiments were so vital because they were randomized field trials. Randomized field trials are the gold standard for evidence in social science because, like controlled trials in medicine, they make it much less likely that hidden variables other than what the experimenter is trying to measure are affecting the result. This "hidden variable bias" problem is why most social science can't be trusted.

Randomized field trials in policing in the 1960s led to James Q. Wilson's pioneering work on broken windows, which led to the revival of cities like New York and Los Angeles and a rare triumph for public policy, one midwived by the conservative movement.

It's time to do it again. Conservatives shouldn't just focus on making prisons acceptably humane, and on shortening prison sentences. They should also promote a federal effort to allow states to experiment with different rehabilitation policies — so long as those policies are run as randomized field trials and the efforts published and peer-reviewed.

That way, they won't just ensure the political viability of their efforts for criminal justice reform. They'll also help countless former prisoners actually flourish.