There's almost certainly going to be more widespread news coverage of one large batch arrest. Research shows the perception of getting caught influences decisi

There's almost certainly going to be more widespread news coverage of one large batch arrest. Research shows the perception of getting caught influences decisions to commit crimes much more than the penalties. Potential criminals don't tend to maximize the ratio of expected reward to expected penalty. I think the best strategy would probably be to keep the network of informants intact and look at the social graph. Once they find a minor dealer within N (probably 2) degrees of separation of M people who weren't previously within N degrees of separation of an arrestee, nab them. Intentionally spreading arrests across the social graph would probably also help reduce the disproportionate cost borne by poor and minority communities. I'd be willing to bet that hearing from a friend about their friend's arrest has a higher psychological impact than hearing about 50 random strangers on the news. On the other hand, it seems law enforcement is more incentivized to grab headlines and maximize published arrest counts rather than actually maximize psychological deterrence. There are certainly plenty of idealists who would maximize deterrence anyway, but I fear they would be unlikely to be promoted as rapidly as those playing the system. The real question is how to properly incentivize law enforcement so even the cynics are targeting the best outcomes for society rather than the biggest media impact. Unfortunately, humans are just generally bad at optimizing over the decades-long horizons involved in incentivizing the maximization of society's long-term best interest. Making an exotic 30-year crime rate derivative (even if we found non-gamable metrics) part of the police pension plan doesn't seem like the right sort of solution. The impact decades out isn't likely to have much effect on day-to-day decision making, and even if it did, it exposes the officers to a lot of extra variability beyond their control. On the other hand, immediately incentivizing the implementation of good-looking ideas before there's time to accurately measure their impact opens one up to Enron-like gaming. So, it looks very difficult to properly incentivize long-term law enforcement goals via either long-term incentives or near-term incentives.