Chasing down that list of potential Predpol customers reveals dozens of cities that have secretly experimented with "predictive policing"

[doctorow]

Originally published at: https://boingboing.net/2019/02/06/secret-predpol-experiments.html

…

[Papasan]

I’m experimenting with “predictive drinking”, and it’s really going very well too.

[Pensketch]

@doctorow

Now, Motherboard’s Caroline Haskins has used that list as a jumping-off point for a wide array of public records requests that have revealed a pattern of extensive, secret experimentation with precrime in dozens of cities that allowed Predpol to access policing data and predict where more crimes would occur.

I wonder if Predpol predicted that?

[PsiPhiGrrrl]

Can’t wait for PredSuit, a tool for defense lawyers to use for clients who were targeted by mistake.

[anon47741163]

This sounds an awful lot like the computer methods used in the Viet Nam war, to help decide where to bomb. Similar mindset, similar disastrous outcome.

[wait_really]

@PsiPhiGrrrl

Can’t wait for PredSuit, a tool for defense lawyers to use for clients who were targeted by mistake.

Something tells me there’s more money to be made the other way around…

[knappa]

Looks like their model is more or less “crime diffuses the same way heat does” plus some terms which look like point sources (if the δ is a Dirac delta) and another term so that crime doesn’t go to infinity due to the point sources. It sounds like a reasonable thing to try if you are interested in things at the population level, but isn’t particularly nuanced.

Either that or their model is “people are impressed by Greek letters and a picture of a harp”. (∇: nabla, a harp)

[Chad_Boudreau]

@knappa

Either that or their model is “people are impressed by Greek letters and a picture of a harp”. (∇: nabla, a harp

That one.

The idea that criminality is contagious is ooooold, and doesn’t work.

Or rather, it DOES kind of work, but only where the underlying systems are deeply broken.

I might be interested in seeing what there models come up with in a place that has insufficient police (as defined by whatever the ratio is in say…Tokyo) and a verified lack of racial profiling at any time during the training data set, and see what happens.

No really, what would this program do if it began with ideal conditions?

Sadly though, the only demand for this program would be for police departments trying to protect themselves from lawsuits for racial profiling, which they wouldn’t be interested in unless they knew they were guilty, so literally every police department using this type of software is suspect…

Keep up the solid sluething work, guys

[HMSGoose]

they know that the prefix “Pred” has…other connotations, right? Yeah, they know.

[PsiPhiGrrrl]

Probably so, especially with the civil asset forfeiture programs in many states.

[abutilon]

Good work, Cory & Caroline!