NYPD

NYPD is Developing an Algorithm to Measure Police-Community Relations

'Sentiment meter' would help law enforcement understand where and how to improve cooperation with the public.

|

How's my sentiment?
RJ/Flickr

The NYPD has relied on CompStat—a data-driven tool for addressing hot spots for crime—for more than two decades. Now it is developing a "sentiment meter," intended to gauge the areas in the vast metropolis where police-community relations could stand to be improved.

NYPD consultant John Linder tells The Marshall Project that his still-in-development algorithm will be a system to deliver to "real-time measures of public attitudes — whether trust is going up or down, whether the sense of safety is going up or down, and whether the job approval of the NYPD is going up or down—by neighborhood."

Linder says the people working on the project's development accurately predicted the Brexit vote and both Michigan and Ohio's 2016 presidential election results. According to Linder, his developers "may have found a way around selection bias in polling, which was a major reason most pollsters missed the Trump phenomenon."

The project comes with the blessing of NYPD Commissioner James O'Neill who also promised increased NYPD transparency when he took the job last October.

"The public will soon have the names, email addresses, and increasingly, believe it or not, the cell numbers of the individual police officers who patrol their streets every single day," CBS2 quotes O'Neill saying at his swearing-in ceremony. O'Neill hopes "that personal connection" will encourage increased cooperation from the public, who might now be able to send a text or an email to report pertinent information to officers.

O'Neill asked Linder and his development team for "real time data on what people feel," Linder says. By setting up an "algorithmically governed sentiment meter that is gathering tens of thousands of data points 24/7, 365 days a year," the hope is that the NYPD leadership can direct its rank-and-file to adjust its tactics accordingly to improve both public relations and more effectively tamp down on crime.

Advertisement

NEXT: For a Study in Failed Foreign Policy, Look at the Seven Countries in Trump's Refugee Ban

Editor's Note: We invite comments and request that they be civil and on-topic. We do not moderate or assume any responsibility for comments, which are owned by the readers who post them. Comments do not represent the views of Reason.com or Reason Foundation. We reserve the right to delete any comment for any reason at any time. Report abuses.

  1. According to Linder, his developers “may have found a way around selection bias in polling, which was a major reason most pollsters missed the Trump phenomenon.”

    Any information on the technique? Or are they keeping it a secret so it cna’t be assessed?

    1. Waterboarding

      1. Do don’t want to do that on the Hudson or East Rivers… oh

    2. Yeah. Way to bury the really interesting part. The NYPD won’t change unless they feel like it, better info is about as useful as tits on a boar.

      How everyone else can use that algorithm, now… that sound intriguing as fuck.

    3. It’s very simple to use. X times 0.

    4. Any information on the technique? Or are they keeping it a secret so it cna’t be assessed?

      I think it involves one bag full of shit and at least one briefcase full of cash. At least if pulling a texas sharpshooter on the 2016 election results and claiming Brexit as some manner of predictive analytic for crime (prevention) is any indication.

    5. I wonder how it relates to being successful in just those two states. Does that mean they failed in the other 48? I realize those two states surprised most pollsters, but it would be more informative to see what their exact predictions were in all 50 states, not just yea/nay in two states.

      Of course, this is for the NYPD, who aren’t exactly famous for releasing any information. I suspect it’s a fine match for obscurantists.

      1. I wonder how it relates to being successful in just those two states. Does that mean they failed in the other 48?

        This is actually why I’m nearly certain that it’s complete horseshit.

        If you were right in 45 of 50 states or went 6-for-6 in all the tossup states, you might have something. More importantly, you’d be saying that you predicted a Trump win and only missed two tossup states (or whatever). The fact that you laud your ability to predict two states that were already acknowledged as being contested, and nothing else, means you’re somewhere near 50/50 at calling coin tosses.

        What I really want to ask them is can we use it to keep Muslims out of NY?

    6. You’re friendly neighborhood SWAT team kicks down each and every door and hands residents a survey. It’s pre-filled for your convenience.

      It’s probably measuring sentiment on social media. That’s the new hot method – use big data and aggregate sentiment based on key words and phrases. Correlated with location this would work, but it certainly doesn’t eliminate selection bias. Plus I have always wondered how well this method can understand context. I’m sure the NYPD hasn’t thought that far. Follow the shiny ball!

  2. Just download the NYPD app to your phone and allow it access to your personal info, contacts, social media feeds, location, and camera roll so we can know just how you feel about our performance and just where to find you to address any problems!

    1. #deleteblueber

    2. #deleteblueber

  3. The NYPD has relied on CompStat?a data-driven tool for addressing hot spots for crime?for more than two decades.

    And then Bunny Colvin ruined it.

    1. New Hamsterdam FTW

  4. Finally a tool for police to determine which neighborhoods are most in need of a morale-boosting no-knock swat raid.

    1. Whose spirits are not bouyed by a sea of flashing blue and red lights?

    2. I once lived in a small, very small little town. The cops would call everyone in town and ask for donations. If you gave them one, they’d send you a ‘I support local law enforcment’ sticker for your car. If you didn’t donate, you’d get pulled over for rolling a stop or a tail light out.

  5. A Kinder, Gentler Stop & Frisk. They are now required give you a “customer feedback” card after they arrest you for sitting on your own stoop.

  6. “Not killing them just for the fuck of it seems to make them like us more, but I assume you gentleman see the glaring flaw in that strategy.”

  7. And then they’ll game it just like COMPSTAT?

  8. the hope is that the NYPD leadership can direct its rank-and-file to adjust its tactics accordingly to improve both public relations and more effectively tamp down on crime.

    More effectively tamp down on crime committed by cops.

    There. Problem half-solved.

  9. Ah, so they have another thing for the union leader to ignore. Ignoring stuff is hard work.

  10. The public will soon have the names, email addresses, and increasingly, believe it or not, the cell numbers of the individual police officers who patrol their streets every single day…

    By public he means underage prostitutes.

    Putting anonymous out of business?

    Cell numbers are really only useful to the corrections officers guarding those cells.

    And so on…

  11. I’m not seeing how this will be useful. If it returns accurate info that the cops don’t like, they’ll just change the parameters and send out a press release.

    “Look! Our polling algorithm shows that we are universally beloved and need make no further efforts for improvement!”

  12. I’m going to take a wild guess here that their sentiment meter is higher in wealthy white neighborhoods than it is in the ghetto?

    1. So you’re suggesting that this algorithm that predicts crime would effectively be a Minority Report?

  13. “Mayor, the PD has spiked to 3 woodchippers after shooting a dog and a toddler last weekend.”
    “We’ll work up a Number 6 on ’em.”

  14. I hope it’s better than the algorithm Hillary used that told her to ignore the Rust Belt and spend millions in California and New York.

  15. The polling bit – whatever. I’ll believe it when I see it.

    The “personal connection with an officer” bit could be good or bad, depending on if it leads the officer to act more like Andy Taylor or Norman Stansfield.

  16. RE: NYPD is Developing an Algorithm to Measure Police-Community Relations
    ‘Sentiment meter’ would help law enforcement understand where and how to improve cooperation with the public.

    Look, this isn’t exactly rocket science.
    If you want good relations with the community, stop beating the shit of the citizens, taking bribes, shooting them without just cause, etc. Have the police act in a professional manner instead of acting like the KGB thugs.
    Secondly, which may be a good idea, is to have minority cops in minority neighborhoods.
    Lastly, the cops should have a “report card.” That is, they will be judged not only by their superiors but also by the community as well.
    Lastly, and perhaps most importantly, eliminate gun control. Some of the highest crime areas have draconian gun laws. Most criminals, at least criminals with a half a brain cell in their head, will not rob, murder, rape or committ heinous crimes if they know the population carry.
    A little common sense goes a long way.
    Maybe this is why no one will put any of these suggestions into practice.
    Common sense and politcs rarely go together.

  17. Or they could just encourage people to rate and review their department on PoPoSearch.com, which is basically a Yelp-like website where people can rate their local police departments. Problem solved

Please to post comments

Comments are closed.