Please create an account to participate in the Slashdot moderation system


Forgot your password?
Crime Stats Technology

Santa Cruz Tests Predictive Policing Program 228

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."
This discussion has been archived. No new comments can be posted.

Santa Cruz Tests Predictive Policing Program

Comments Filter:
  • Kind of Interesting (Score:3, Interesting)

    by Anonymous Coward on Tuesday August 16, 2011 @07:39PM (#37113604)

    It is kind of interesting on one level because it doesn't violate anyone's civil rights nor do anything odious. That much said, I am no fan of proactive policing. Proactive policing usually means law abiding citizens get harassed for walking through a "known" crime area even though they have no criminal intent. And please spare me the tired old line that only criminals go through bad areas and if you are in a bad area you must be up to something. Having been in law enforcement myself, cops are really rarely out to help which is why they call it "law enforcement" versus "peace officer." If you want proactive policing, hire private security.

    • It doesn't intrinsically violate someone's civil rights but what about feedback loops?

      If someone commits a crime in your neighborhood and it gets more policing then the policing will catch more criminals and by extension increase policing. Rinse and repeat until it reaches equilibrium.

      For instance it would suck if your street through ticket based feedback became a speed trap and you had no choice but to go through it every day.

      • by shadowrat ( 1069614 ) on Tuesday August 16, 2011 @08:56PM (#37114206)
        If i live in a neighborhood that has a lot of crime, I'd like the police to come and catch lots of criminals. I like the sight of cops walking the beat around my block.

        I'd also welcome a speed trap right outside my front door. Speeding on the highway is ok within reason. 10 - 15 over is probably fine. but a residential neighborhood is another matter entirely. We've got kids playing and people backing out of driveways.
        • If there *actually* is crime there above average that's one thing. But simply *finding* crime in one area because of greater patrolling isn't necessarily confirming the prediction's effectiveness.

          It's the same argument as profiling. If you search an African American and find drugs on them and then decide that it means African Americans have more drugs on them you'll find an excuse to search more African Americans and find more drugs. At some point you're only searching dark skinned people and only findi

      • The cops in Santa Cruz are not dumb, or at least not any dumber than any other cops, and they know which few streets encourage speeding already.

      • by ajs ( 35943 )

        Actually, I'd say the police should review the output of the model and patrol the areas least likely to produce crime. On the assumption that smart criminals will use the same modeling to predict where police coverage will be decreased, this allows you to determine where to find the smart criminals. An excellent tool!

  • My prediction (Score:5, Interesting)

    by spazdor ( 902907 ) on Tuesday August 16, 2011 @07:40PM (#37113608)

    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!

    • by dougmc ( 70836 )

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

      I have a "hunch" that this black man over here is about to commit a crime ...

      "Hunches" very often lead to profiling. Unfortunately, profiling is fairly effective -- it's just really, really unfair to those who are profiled and yet innocent.

    • by Sycraft-fu ( 314770 ) on Tuesday August 16, 2011 @09:20PM (#37114370)

      Police departments already tried to do this, as noted with the hunch thing. The cops would go around and show presence in areas to try and deter crime (they also do things like park their cars in mall parking lots when they do paperwork). Well and good but of course it is all based on what humans feel is correct. while there's some validity to that since we do notice patterns, better to have a computer work it out, if possible.

      Supposing the algorithm is tuned well, it could really do good. The patrols will go in areas where they are most needed. Also presumably a good model that is given new data daily (as this one is) will notice when things change and thus change patrols. Humans may be much slower to react.

      Have to see what the actual stats are on it, but I think it could be a real win long term for law enforcement.

    • I think we call that, "intent". ;)
      • by mrmeval ( 662166 )

        *squeeekily* erases that from the dictionary of jurisprudence and schedules anyone who disagrees with retroactive abortions!

    • by TWX ( 665546 )

      Somehow I don't think you'd get any successful prosecutions if you did that. Additionally, you'd discredit any possible hot-spot monitoring program for the same reasons.

      This only works if you let the perpetrator actually become a perpetrator. If you stop them before they've provided evidence of an intent to commit or evidence in the commission, you'll get thrown out of court, and if you do it way too much and catch too much public attention, you'll have other law enforcement entities investigating you, pr

      • by mrmeval ( 662166 )

        Blasphemer! Infidel! YOU are the criminal! DIE!

      • by HiThere ( 15173 )

        If the intent is to prevent crimes, you don't need to arrest someone, only to, *IF* they were planning to commit a crime, discourage them. Discourage them for awhile, and they'll form a different habit patten, and no crimes occurred.

        FWIW, this is a one sentence summary of an analysis of New York's, apparently working, crime reduction strategy. My summary of an article from a recent Scientific American. Is it working? Apparently statistics from several sources say that it is.

    • by sjames ( 1099 )

      So do you think we should investigate based on exit polls or will the morning after the election be soon enough?

      • by mrmeval ( 662166 )

        Election? Election! What fucking election? I say we kill the accused because we are GOVERNMENT GODS!

  • by kwiqsilver ( 585008 ) on Tuesday August 16, 2011 @07:52PM (#37113730)
    So how many small crimes would you have to commit in other areas to reduce the police coverage in your targeted area before you commit the big crime at the real target?

    Ooh! Did I just write the plot for Oceans N+1?

    • So how many small crimes would you have to commit in other areas to reduce the police coverage in your targeted area before you commit the big crime at the real target?

      wouldn't it just be easier to hack into the system and see where it is directing officers, and go break the law away from them? Or better yet, feed the computer false info, so it predicts crimes in areas you want the police to be when you are doing your big heist.
      • by adamofgreyskull ( 640712 ) on Tuesday August 16, 2011 @08:26PM (#37113934)
        Yeah man! We just need to hack the Gibson and reverse the polarity on the mainframe firewall in order to drop a logic bomb through the backdoor. Alternatively, paying street-kids to commit petty thefts in areas away from your target area is much much simpler than "hack into the system" and/or "feed the computer false info".
        • All you would really have to do is "report" crimes in the other areas... "My car stereo was stolen on tuesday at 10:00 on the corner of X and Y" if these were in fact false, that would be feeding the computer false info, you just let the agency feed your info for you, no hacking needed and no paying street kids either.
      • by dbc ( 135354 )

        It's much simpler in my town. We have "public safety officers" -- they are trained as both police and firemen, and get better pay than the surrounding communities because our town has fewer people on the payroll over all. So the paramedics drive the fire trucks to the fire, and the police driver their cruisers and grab a fire suit off the truck. So.... the secret is if you want to rob a bank, start a fire across town first :)

    • by artor3 ( 1344997 ) on Tuesday August 16, 2011 @08:17PM (#37113860)

      Enough that you'd leave a trail of evidence a mile wide before you even got to the big one. If you really, really want to commit a crime, do the big one first, and then be a model citizen forever after. Repeat offenders eventually get caught.

      Alternatively, pursue a career in finance and/or politics. You know what they say... the best way to rob a bank is to own it.

  • I wonder if they've looked at predicting how this will play out with the new program in place -- they have the basic problem that they're affecting what they're observing, and thus will change what will happen.

    If the algorithms predict crimes in certain areas, you'll end up with officers in the area, likely preventing a crime before it even happens. That is, the potential criminal will notice the police presence and decide its not a good time. Thus there would be some feedback from the prediction method b

    • SELECT grid_id FROM streets WHERE streetname LIKE 'Martin L%';

      • by maeglin ( 23145 )

        SELECT grid_id FROM streets WHERE streetname LIKE 'Martin L%';

        Damn! How did you know I was going to knck over that 7-Eleven on Martin Landau Boulevard!?

    • That is, the potential criminal will notice the police presence and decide its not a good time. Thus there would be some feedback from the prediction method back onto itself.

      Or the criminals don't notice the police presence and get caught in the act, reinforcing the prediction?

    • I was about to post the same logical problem.. it would seem the system is self-defeating by nature. I think the only thing you could accurately predict is there would be less crime where you place the cops. As Carlos Mencia would say.. deeetdeedeee...
  • I don't mean to go quantum physics on police work, but this is slashdot. As soon as the police insert themselves into the equation, the social dynamics will change and eventually invalidate their predictions. It will take a while, especially when compared to the orbit of an electron, but it will happen. If they are good, their model will adapt, but it may not work as well in such a dynamic feedback loop.
    • by artor3 ( 1344997 )

      Just FYI, that's not the Heisenberg uncertainty principle. You're thinking of wave function collapse, in which the act of measuring a particle determines its state. Heisenberg uncertainty is a mathematical proof that shows that the uncertainty in a particle's location and the uncertainty of its momentum have a non-zero product. It also applies to other pairs of properties, such as energy and time.

      People get the two confused all the time, probably because the one that's more useful to talk about doesn't h

      • Hmmm. Seems like Wikipedia [] got that wrong as well:

        The uncertainty principle says, for instance, that it is impossible to measure a particle's velocity in any moment and then have any hope of measuring its location for that moment (since the act of measurement of velocity immediately changed that particle's location).

        I read that to mean by interactively observing a particle's velocity (i.e., "pinging," it with a measurement device) one necessarily changes that particle's behavior in such a way that doe

        • by artor3 ( 1344997 )

          The two are related in that thought experiments related to changing something by measuring it led to the development of the Uncertainty Principle. However, it can be reached from other angles, including deriving it mathematically. It exists independent of any actual measurement. Even if you imagine an omniscient god thinking about the particle, it's impossible to know both the momentum and position simultaneously.

          Hyperphysics has an excellent summary [] showing where the uncertainty arises without any measu

          • Thanks for reminding me of the talk page on Wikipedia. I have to get back into the habit of checking that out on a regular basis. Thanks, also, for the link. I'm always interested in broadening my understanding in all things scientific (although I'm an engineer by training and profession as well). Again, thanks for the post! :-)
    • Urban campers will just take their thieving further from their camp sights once they realize they are bringing heat down on themselves.

      Until then the cops have a statistical homeless camp locator.

    • by Jeremi ( 14640 )

      As soon as the police insert themselves into the equation, the social dynamics will change and eventually invalidate their predictions.

      Hopefully the form of the changes will be fewer people committing crimes, because it's harder to get away with crimes after the program is in place.

      Remember, the police system doesn't have to be perfect, it just has to be effective enough to deter your average potential criminal. It's not like people have an an infinitely large incentive to commit crimes at all costs; they choose to commit a crime, or not, based largely on risk and cost-benefit analysis.

    • As soon as the police insert themselves into the equation, the social dynamics will change and eventually invalidate their predictions.

      That's the goal. The whole purported reason for putting traffic cameras at intersections with lots of crashes is to make people more cautious to reduce the crashes.

      Or is the argument just, "why fight crime when you can never eradicate it?"

      • That's the goal. The whole purported reason for putting traffic cameras at intersections with lots of crashes is to make people more cautious to reduce the crashes..

        Accidents increased. When they put up cameras to stop red light runners, people began to suddenly stop as the light turned yellow, only to get rear-ended in the process. That wasn't the goal.

        • by artor3 ( 1344997 )

          I know you're thinking that the goal was to get ticket revenues. Sometimes I suspect that myself. But the government would argue, perhaps truthfully, that getting rear-ended will at worst give you a bit of whiplash, whereas getting t-boned by some asshole running a red light can kill. Has there ever been a study on the rate of injuries before and after red light camera installation?

  • Thanks Jean and Fred (cue fake smiles and laughs all around)...

    Well today we sure did have some isolated crimestorms dotted around the metro with scattered crimebursts in the outlying areas. Your forecast for tomorrow is a 40% chance of crime in the downtown area with a peak of 80% occurring around 4th and Vine. Out in the suburbs, we're looking at a 10% chance of domestic disputes, 40% chance of mom scoring some weed from the high school pimp, and about an 80% chance of teenage drinking as we head toward

  • and totally get it now that I've read this. I got pulled over by a Sheriff last month who was three cars ahead of me and pulled off the road to get in front of my truck. I hadn't broken any laws and the Cop told me he pulled me over because I have a full beard and fitted a description of someone he was looking for.

    The Officer ran my info and came back telling me I was clean. He then asked if he could take my picture in case he found the perp he was looking for. Being a stand-up citizen I agreed and let him

    • The computer might have told him to be in the area, but it didn't tell him to pull you over. That was his decision.

      I grew up in Santa Cruz (I was just there this last weekend, actually... got some awesome flea market scores including finally picking up an electric bass... and not billy bob) and it has always been a pretty racially divided place. Very tolerant of minorities who know their place. You know, Mexicans belong in the flats or in front of Kmart (well, there's a Home Depot there now, but they were a

    • No, to your question, it was exactly what he said, you fit the description of someone they were looking for. Probably someone that just committed some type of violent crime in the area. Either that or your civil rights were violated by the officer. But if he were violating your rights he would have known it at would almost definitely not have stopped at taking your picture for the case file. The picture was most likely for if you decided to sue or claim that your rights had been violated, the officer could
  • by grouchomarxist ( 127479 ) on Tuesday August 16, 2011 @10:00PM (#37114590)

    The article doesn't go into it, but is the earthquake aftershock prediction actually any good? I haven't heard about it and the article doesn't mention anything about the accuracy.

    • ... and more importantly, how can a simulation intended to a physical phenomenon be applied to a social phenomenon governed by an entirely different mechanisms?
  • As long as they're targeting times and places, not people (individuals or groups), this is totally reasonable. In small towns with townie cops on the beat for years, the cops know where and when the crime "hotspots" are. But they're subjective, and are easily turned into just harassing people (and the neighborhoods they live or hang out in). Indeed, bad cops say that's what they're doing, when they're really just racists or settling some old grudge (often against totally different people), or just on a powe

  • So is this the Santa Cruz Operation?
    • Encinal street is in an area which is already heavily patrolled, as it includes a park and a Costco, to say nothing of the Goodwill bargain barn...

  • Ooh! Ooh!

    Bettor Cop: There were a rash of crimes around Broadway and Seabright! I'm placing my bet on Broadway and Seabright!
    Crime Dealer: Oh sorry! The next crime happened over on Center and Church! YOU LOSE!
    Bettor Cop: Damn! That sucks!
    Victim: You're telling ME asshole!

    • by artor3 ( 1344997 )

      It's only the gamblers' fallacy if you're sure the odds really are even. If I flip a coin twenty times, and it comes up heads every time, you'd be a fool to think that the odds on the next flip are 50-50. It's far more likely that I'm using a trick coin.

      See also: Bayesian inference

  • Two statistics majors went on a police ride-along to see how the new crime prediction model was working. They went to the first predicted location on 200th st., but there was no crime. Then they went to the second second predicted location on 100th st, but again no crime. Just when they were about to admit defeat, a call came in about a crime on 150th st and they both yelled "we did it!"

  • This only allows prediction based on past crimes that are known to Law Enforcement. That means the successful criminals who commit crimes without them being noticed (like crimes that no one reports, drug dealing, prostitution, etc) will have even less of a chance of having police run across them.
  • Why is this news? Proactive policing is hardly new. Doing the predictive work on a computer isn't new either. (NYC was doing it back in the 70's.)

  • "California is testing a new method for apprehending criminals: beating them" - that's what I read. Have to be careful with such headlines...

Time to take stock. Go home with some office supplies.