Stop Me Before I Vote Again

I gotta object to this one, Al. You can predict human behavior at the individual level? Really? Let's hear a few predictions.

Human behavior is way too complicated to ever be that predictable. Economics and social science have been trying this game for 50 years, with zero results. Absolutely zero.

This is not to say that there aren't patterns within groups that are extremely predictive. But using those to police anybody is more simply known as profiling.

This PP effort also ignores the ecological fallacy, which is all too real.

I sure can. I accurately predicted that Barack Obama would continue, consolidate and expand the Bush Administration policies (themselves inherited from Clinton, Papa Doc Bush, Reagan...), and do so with a special emphasis on the ones that are criminal in both the real and procedural senses of the word. I predict, with confidence, that his successor will do the same. The only thing I'm lacking is the ability to effectively police that.

But, Al, that's rather different than predicting which individuals will commit street crimes. No?

Absolutely. The "predictive policing" for street crimes is invariably going to be a mass targeting, replete with post hoc justifications for the rubes when the harassment and entrapment action takes place. Any claims for efficacy on the individual level are bogus. A true street crime reduction program would never waste money on it. The whole thing looks like a crackpot scheme, if taken at face value.

so PP will help reduce electricity consumption for evermore pervasive CCTV. only run the cams when the crime is likely to be committed.

the real battle is to define crime appropriately and teach it! in my m + n years of formal education (an embarrassingly large number) I never was actually told what were and were not crimes. it was all informal, like sexual education or lore before Woody Allen.

but seriously, if the prediction is limited to crime and not extended to 'candidate criminals', what is the problem?

I doubt it will include a reduction in the use of CCTV. Predictive policing is based on the mass aggregation of statistics and targeted at what amounts to captive populations. The statistical hot spots are mapped and, presto, you have your nodes of criminal conduct. The people living there, working there, traveling through, visiting etc. are suspects before the fact of criminal conduct. Some will actually commit crimes. Most will not. Those that are arrested prove the system works. Those that arrested under false positives are proof the system needs more money. Those that are framed shouldn't have been there in the first place.

It is, in short, the addition of statistical modeling to the already existing presuppositions of nightstick fetishists.

Paging Hari Seldon.

Al, friends of mine who got arrested and had their bus impounded at the G20 in Pittsburgh would agree completely with your statement about mass targeting and post-hoc rationalization/justification. Meaning it's already happening. Dig on Fusion Centers, eh?

Of course people who travelled to see the Dead when Jerry was still alive and kicking, they know about mass targeting -- everyone in NJ I knew who were Deadheads told me about avoiding US 95, the NJ Turnpike and GS Parkway during a Dead tour. Troopers, patrollers and coppers were combing the highways looking for vehicles with peace signs, Dead stickers, etc... for VW microbuses generally, too. Common advice: "Don't be holding if you're driving one of those vehicles during a Dead tour of the mid-Atlantic."

Black kids and adults will tell you, if they spent any time living in predominantly Black neighborhoods in most US cities or outskirts, that they get pegged for more crap than the White kids.

power grows out of a barrel of
night...sticktistics

Ox, that's it in a nutshell. The tech aspect is so much flummery for people who can be persuaded to believe that the desire for more elaborate tools is, itself, indicative of a desire for better outcomes. But the rub is of course, better in what way? And the irony of the statistical models is they work really, really well for white collar and political criminality. The nodes could be seined and never fail to capture entire gangs of thugs.

'And the irony of the statistical models is they work really, really well for white collar and political criminality.' And for deciding whose taxes to audit so as not waste auditors' efforts on low-profit-returning audits.

But all THIS has been policy-ed and deregulated out of court so to speak.

Trivia prompted by Oxy's Head reminiscences: Saw a pre-Code film yesterday, "Murder at the Vanities" 1934 (not good), and was amused to hear one the second leads (female) sing a paean to the herb entitled "Marijuana, Marijuana" I imagine this doesn't get a lot of replay on Turner Classic Movies.

"In a sense, crime is just a physical process,"

Great, another fuckin' math nerd. There seems to be a certain mindset that believes everything can be reduced to numbers and functions. Take a bunch of subprime morgages that people can't afford at prevailing wages, spend 18 hours running a bunch of useless math and they think they've accomplished something.

All they really need to know is economic/housing/food insecurity causes stress. People under stress make a lot of stupid decisions.