Anders Behring Breivik, the right-wing Norwegian extremist who massacred 77 people in shooting and bomb attacks last July, has been sentenced to 21 years in jail, the maximum term allowed under Norwegian law. He might serve longer than that. He was sentenced to "containment," a peculiar feature of Norwegian sentencing in which, after the end of 21 years in prison, a convict is eligible for unlimited five-year extensions of his prison term if he is determined to still be a threat to society. But the American practice of almost automatically handing out life terms to murderers doesn't apply. For instance, Varg Vikernes, a Norwegian black metal musician who performs as Burzum, stabbed his friend to death and committed a series of church arsons and was released after only 16 years of his 21 year term.

However long Breivik serves, he'll be in a cell that looks like this:

Ila Prison/AP

And that's only one of Breivik's three rooms (and yes, his will have a laptop too). As Breivik will be in solitary confinement, he also gets his own exercise room and a bedroom, to compensate him for his lack of access to regular parts of prison life, such as university courses, work in internal shops, the prison gym and the library. You can see a short documentary touring a Norwegian prison here.

Many have an intuitive repulsion to the idea of giving a mass murderer that kind of square footage, with Internet access to boot. But is that gut feeling backed up by the evidence?

It turns out there's a pretty extensive literature on the effects of harsh prison conditions. One finding that is growing more and more accepted is that harsh sentences, if anything, increase recidivism (or, the propensity of prisoners to reoffend once released). The economists Keith Chen and Jesse Shapiro exploited (pdf) the fact that the federal prison system assigns prisoners to different security levels based on a numeric score indicating  how much supervision that inmate needs. There are then cutoffs for assignment to each security level. Those scoring above a certain cutoff get medium security, those below it get low security, and so forth. Chen and Shapiro compared prisoners with scores just above and just below cutoffs to see how being assigned to a higher security level affected recidivism.

Their finding? Those at the border who end up placed in a higher security prison reoffend at a significantly higher rate than those at the border place in lower-security prisons. For those right at the border, the increase is about 10 to 15 percent, but if you take a broader view, it could reach as high as 42 percent. That's a serious increase in crime that, if Chen and Shapiro are right, is easily preventable by putting prisoners in less-harsh conditions.

A bevy of studies have used the security level approach and found similar results. Gerald Gaes and Scott Camp found that higher security levels increase recidivism by about 31 percent. Lawrence Bench and Terry Allen randomly assigned prisoners to medium and maximum security sectors of a prison and found that prisoners in maximum security were no less likely to commit in-prison offenses.

A more recent paper from Italian researchers Francesco Drago, Roberto Galbiati and Pietro Vertova measured the effects of prison overcrowding, prison death rates and distance of prisons from major cities on recidivism. They found no significant changes in recidivism in prisons that were more or less crowded or more or less deadly, and found that geographic isolation increases recidivism. A study from Rafael Di Tella and  Ernesto Schargrodsky found that people who are sentenced to house arrest with ankle monitors reoffend at a much lower rate than those sentence to traditional prison. And a wide array of studies have found that in-prison education programs reduce recidivism while improving quality of life. The findings on the effects of prison conditions on recidivism, in short, are a matter of scholarly consensus.

What's tricky is how to square these findings on recidivism with contrary findings on the deterrence effects of prison conditions. One prominent study, by economists Lawrence Katz, Steven Levitt (he of "Freakonomics" fame) and Ellen Shustorovich, found that higher death rates in prisons were associated with lower recidivism. There are some serious methodological problems with this approach; it is not based on an experiment or natural experiments, unlike most of the recidivism studies, for instance, and the authors do not identify a mechanism for why prison deaths should have this deterrent effects.

In any case, the policy implications of the Katz et al study are limited, given that a policy of increasing prison deaths is (thankfully) not really in the cards. But the implications of the consensus on recidivism are big and important. Maybe Breivik's cushy suite isn't such a bad thing after all.