Should prison terms end once criminals seem "too old" to recidivate?

The question in the title of this post is prompted by this intriguing recent New York Times piece headlined "Too Old to Commit Crime?".  Here are excerpts:

Dzhokar TsarnaevV is facing the death penalty or life in prison for the Boston Marathon bombing.  But what if, instead, the maximum prison sentence were just 21 years? That was the sentence that Anders Behring Breivik received in 2012 after killing 77 people, most of them teenagers attending a summer program, in Norway in 2011.  It was the harshest sentence available.  That doesn’t mean Mr. Breivik will ever walk free. Judges will be able to sentence him to an unlimited number of five­year extensions if he is still deemed a risk to the public in 2033, when he is 53.

The idea of a 21-­year sentence for mass murder and terrorism may seem radically lenient in the United States, where life without parole is often presented as a humane alternative to the death penalty.  Yet in testimony last week to a congressional task force on reforming the federal prison system, Marc Mauer, the director of the Sentencing Project, an advocacy group, suggested exactly that approach.  He made the case for a 20­-year cap on federal prison terms with an option for parole boards or judges to add more time if necessary to protect the public.  Such a policy would “control costs” in a system that is now 40 percent over capacity, Mr. Mauer told the task force, and would “bring the United States more in line with other industrialized nations.”

This proposal has little chance of becoming law. But a compelling case can be made for it nonetheless. Research by American social scientists shows that all but the most exceptional criminals, even violent ones, mature out of lawbreaking before middle age, meaning that long sentences do little to prevent crime....

Some crimes are simply too physically taxing for an older person to commit. Regardless of why offenders age out of trouble, American sentencing practices are out of whack with the research on criminal careers. Between 1981 and 2010, the average time served for homicide and nonnegligent manslaughter increased threefold, to almost 17 years from five years. Over 10 percent of federal and state inmates, nearly 160,000 people, are serving a life sentence, 10,000 of them convicted of nonviolent offenses. Since 1990, the prison population over the age of 55 has increased by 550 percent, to 144,500 inmates. In part because of this aging population, the state and federal prison systems now spend some $4 billion annually on health care.... [A] sentence that outlasts an offender’s desire or ability to break the law is a drain on taxpayers, with little upside in protecting public safety or improving an inmate’s chances for success after release. Mr. Mauer’s proposal for a 20­-year sentence cap, applied retroactively, would free 15 percent of federal prisoners — some 30,000, except for those few whom judges or parole boards might deem unfit to re-­enter society.

This is much more aggressive than the Smarter Sentencing Act, a bipartisan proposal in Congress which would lower mandatory minimum sentences only for nonviolent drug crimes. Both the House and Senate versions of the bill keep mandatory minimum sentences of 20 or 25 years for third­-time drug offenders, and most of the bill’s provisions would not benefit current inmates. Of course, for many Americans the prison system is not only about preventing crime by getting criminals off the street, but also about punishment. Long sentences send a clear message that certain acts are unacceptable. Some conservatives who support sentencing reform say that Mr. Mauer’s proposal goes too far, offering a one­-size-­fits-­all leniency to even violent offenders.

Mr. Mauer responds that given the immense scale and cost of incarceration, “modest reforms” would be insufficient. “How much punishment is enough?” he asked. “What are we trying to accomplish, and where does redemption come into the picture?”