Santa Cruz Tests Predictive Policing Program

The police department of Santa Cruz, California is testing a new method for apprehending criminals: beating them to the crime scene. No, they haven't harnessed a group of pre-cogs; they're relying on a computer program that analyzes past crime statistics. "Based on models for predicting aftershocks from earthquakes, it generates projections about which areas and windows of time are at highest risk for future crimes by analyzing and detecting patterns in years of past crime data. The projections are recalibrated daily, as new crimes occur and updated data is fed into the program. ... For the Santa Cruz trial, eight years of crime data were fed into the computer program, which breaks Santa Cruz into squares of approximately 500 feet by 500 feet. ... Officers are given a list of the 10 highest-probability 'hot spots' of the day at roll call. They check those areas during times that they are not out on service calls. Before the program started, they made such 'pass through' checks based on hunches or experience of where crimes were likely to occur."

Comments

My prediction

[spazdor]

Anything which replaces officer "hunches" with something more probabilistically sound* is fine by me.

*given the very low predictive value of their hunches and the high potential for 'hunches' to obfuscate prejudice or patterns of harassment in their investigations("my gut told me hassling this poor neighbourhood for the eighth time this month might turn up some crimes"), a dice roll would be sound enough for my purposes. Can you come up with an even more accurate model than pure randomness? bonus!

Re:Unintended consequences.

[artor3]

People are really bad at being random. I'm sure many criminals already think they're picking random targets.