Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Santa Cruz, California bans predictive policing in U.S. first

reuters.com

Post reply on HN
Print view View original

381–390 of 643 posts

Re: Santa Cruz, California bans predictive policing in U.S. first

#381

Post by 11thEarlOfMar »

ryankemper wrote: post #356
bhntr3 wrote: post #184

The problem with predictive policing is in the name. Inference (ML) predicts the future from the past. If the past is racist, then inference will create a racist future. Since racism is systemic[1], especially when it comes to policing, predictive policing is actively working against an anti-racist future. There may be statistical ways to factor out systemic racism. There are two reasons I don't think that works: 1.…

> Since racism is systemic[1] > [1] If you don't believe this, you're in the minority now: https://www.vox.com/2020/6/11/21286642/george-floyd-protests... . This part was a little odd. It sounds like you're saying "because this belief is widely held by people, we should treat it as true"? I definitely am in the minority, but I find a much more logically consistent view than the modern definition of systemic racism to…

This struck me as well. Systemic racism means (to me) that if you took the people out of the system, you'd still get biased outcomes based on race. In my view, it's a different matter to say that a system produces biased results because it is designed to do so, rather than it is systemic because the people operating it make biased choices where their discretion is called on.

Are there actual, current examples of systemic racism where an entirely non-racist staff would still produce racially-based biased outcomes?

Re: Santa Cruz, California bans predictive policing in U.S. first

#382

Post by yurlungur »

Narhem wrote: post #192
bhntr3 wrote: post #184

The problem with predictive policing is in the name. Inference (ML) predicts the future from the past. If the past is racist, then inference will create a racist future. Since racism is systemic[1], especially when it comes to policing, predictive policing is actively working against an anti-racist future. There may be statistical ways to factor out systemic racism. There are two reasons I don't think that works: 1.…

These systems are mostly pointless. If you really want to stop crimes, you have to get involved with the community and understand why they are happening.

I think preventative measures to reduce community crime levels are not a bad idea. It just sounds very inappropriate if it was the police department doing it, especially if the community has fear and antagonism towards the police. If these were social workers knocking on doors and showing people how they can improve their lives by new education opportunities etc this would be much less problematic to me.

Re: Santa Cruz, California bans predictive policing in U.S. first

#383

Post by Misdicorl »

banads wrote: post #377
Misdicorl wrote: post #374

Earlier quoted context omitted.

My job isn't to respond in detail to FUD. Thats how trolls win; by engaging people who are willing to put in that effort while they can respond with anything resembling coherence. My job is to point out the FUD so people might not notice (for whatever reason) can at least be aware of its presence (or if in doubt, check some other reputable source themselves). You don't need data on this one though. What the fuck is c…

>What the fuck is calling the police going to do to prevent violence after a gang has done a drive by shooting? To find the criminals that did that and put them in prison to prevent them from doing that again?

But the original argument is talking about worrying response times, not conviction rates!

Re: Santa Cruz, California bans predictive policing in U.S. first

#384

Post by tpmx »

strbean wrote: post #298
franksvalli wrote: post #149

Just for the record, in 2011 Santa Cruz was proud to be the first in the US to try predictive policing: https://www.santacruzsentinel.com/2011/01/14/crime-predictio... https://www.santacruzsentinel.com/2020/06/20/santa-cruz-cons...

And somehow in this small sleepy town, the same 2 blocks have consistently been the "you can always get heroine/meth/crack here" blocks for about 30 years, and they can't predict the need for a squad car parked there.

So,

a) an incompetent or politically stymied police force

b) predictive policing from prior crime data works

Re: Santa Cruz, California bans predictive policing in U.S. first

#385

Post by raarts »

andrewprock wrote: post #376
umvi wrote: post #368

Earlier quoted context omitted.

> Why does a standard highway patrolman or cop doing routine speed traps need a gun? Because civilians have shot and killed police during traffic stops? And not just with guns. I seem to remember a road rage case in Colorado where the raging driver shot and killed another driver with a crossbow.

It's not clear how "there is a non-zero chance a motorist has a gun" leads to the conclusion that "all traffic enforcement must have a gun".

As someone from the Netherlands I can say there's a non-zero chance that a motorist has a gun. It's smaller than the US, but it's not zero. Still our cops carry guns. In the US the chances are way, way higher. I can understand that cops are armed there.

Re: Santa Cruz, California bans predictive policing in U.S. first

#386

Post by alfalfasprout »

mnm1 wrote: post #366
ryankemper wrote: post #357

Earlier quoted context omitted.

> These things are evenly distributed across society This is an unproven assertion. You should provide evidence of it, because on its face it is a completely ludicrous claim. Rates of illicit drug activity, violent crime, etc are _not_ evenly distributed in society. Not even close. > Exhibit B) Silicon Valley openly brags about microdosing acid to perform at work at their half million $ jobs. Overpolicing would ensna…

Actually, black people use drugs at slightly lower rates than whites. This is common knowledge or should be. Here's 3 of the top 6 google links for "illicit drug use by race": https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2377408/ https://www.hamiltonproject.org/charts/rates_of_drug_use_and... https://www.sciencedirect.com/science/article/pii/S004723521...

We're talking about violent crime... drug prosecutions and other nonviolent crime are notorious for selective enforcement.

A bullet fired, a body, etc. are much harder to hide.

Re: Santa Cruz, California bans predictive policing in U.S. first

#387

Post by Misdicorl »

Misdicorl wrote: post #352
banads wrote: post #336

Earlier quoted context omitted.

What is horseshit? Nothing in your response is contrary to anything the person said that you're responding to

>There's a worrying trend in the conversation these days that has shifted from "there's a serious problem with racism in how police go about their job" to "we need less police". Folks in neighborhoods living with the constant threat of gang violence don't have the luxury to sit in their aeron chairs and argue about defunding the police. They face the very real threat of themselves or their loved ones being shot on th…

I'm done responding to the FUD hydra here. If someone wants to come up with a single consistent argument for why above isn't bs, I'll engage with them

Re: Santa Cruz, California bans predictive policing in U.S. first

#388

Post by solzhenitsyn »

andrewprock wrote: post #376
umvi wrote: post #368

Earlier quoted context omitted.

> Why does a standard highway patrolman or cop doing routine speed traps need a gun? Because civilians have shot and killed police during traffic stops? And not just with guns. I seem to remember a road rage case in Colorado where the raging driver shot and killed another driver with a crossbow.

It's not clear how "there is a non-zero chance a motorist has a gun" leads to the conclusion that "all traffic enforcement must have a gun".

It's to mitigate that risk factor. You wear your seatbelt every time you drive as "there is a non-zero chance you will crash" even though the vast majority of the time wearing the seat belt was unnecessary.

Re: Santa Cruz, California bans predictive policing in U.S. first

#389

Post by iguy »

baddox wrote: post #62
iguy wrote: post #43

Earlier quoted context omitted.

More policing may indeed catch more minor crimes that would otherwise go unobserved. But the morgue has a pretty close to 100% set of data on murder victims. It's hard to argue that those are caused by catching more criminals. (And in fact, murders in high-crime neighborhoods are less likely to be solved. If anything, murders are caused by this failure to catch first-time murderers, before their second go. Their vict…

But how does the level of policing in an area affect murder rates? I don't know the answer, but from statistics I've seen, it seems pretty clear that police are not exactly great at solving murders.

Are you seriously claiming that the presence of police increases the murder rate? Or are you actually asking to know by what mechanisms policing can prevent murders?

Once you are talking about solving a murder, clearly that one was not prevented. It may draw detectives to the neighborhood looking for clues. It's pretty hard to argue that the causality runs from these detectives to the murder though.

Re: Santa Cruz, California bans predictive policing in U.S. first

#390

Post by aeturnum »

alfalfasprout wrote: post #232
bhntr3 wrote: post #184

The problem with predictive policing is in the name. Inference (ML) predicts the future from the past. If the past is racist, then inference will create a racist future. Since racism is systemic[1], especially when it comes to policing, predictive policing is actively working against an anti-racist future. There may be statistical ways to factor out systemic racism. There are two reasons I don't think that works: 1.…

I agree that, in general, you're going to have biased training data (with a bias that's difficult to measure) and so inherently policing recommendations will be biased. This is particularly problematic when it comes to arrests for crimes that are inherently 'selective' in their enforcement. Eg; drug-related crimes, public intoxication, loitering, trespassing. But the fact is... violent (fatal and nonfatal) crimes do…

I think you make a good point about the need for law enforcement in high-crime areas, though I disagree strongly that the only solution to that problem is a police force as currently constituted in the US. I think it's worth pointing out that police often has abysmal clearance rates and response times and often do not provide effective relief from the real problems you outline.

However, what I wanted to raise was this:

> the fact is... violent (fatal and nonfatal) crimes do happen at a much higher rate in poor neighborhoods

I think this statement is worth examining.

We certainly know that poor neighborhoods contact the police more often and that the police arrest more people (and record more crimes) in poor neighborhoods. As I said above, I have ever reason to believe residents of poor neighborhoods suffer crimes more often and that each crime does proportionally greater damage to their lives.

I also think that, as people get access to more resources, they are reasons to believe they are less likely to contact the police in all circumstances. That, of course, is the other way to read the stats you linked about poor neighborhoods contacting police. Police clearance rates are often low and contacting the police rarely directly addresses the harm caused by a crime. Also, because we are most likely to be harmed by people in our social circles, wealthier people are more likely to be harmed by more powerful people. These are all reasons that wealthier victims of crimes might choose not to contact the police and generate statistics about them. It's difficult to point to data about what isn't recorded (of course), but I think the feminist movement has done some great work trying to document how under-reported rapes (and other sexual crimes) are.

I just want to encourage skepticism about the systems through which data is collected. Structural elements strongly bias the data that's collected and, if we're serious about changing the system, I think it's advantageous to be skeptical of every element of the system.

Post reply on HN