I saw a talk by researcher Kristian Lum a few years back that I think made this case far more effectively. Her point was somewhat limited to drug crimes, but she pointed out that if you look at medical data (where people tend to be fairly honest about their drug usage) pretty much everyone in the metro area under study (SF) used drugs, or at least drug use was equally prevalent in pretty much every geographical area and among all demographics. Therefore, when the cops went to a location to make drug arrests they typically succeeded, because it's not hard to find drug crimes in the Bay Area.
The problem was that they then used that arrest data to make decisions about where to perform future searches and arrests. Because they found drugs where they had looked previously, they looked there again and found more drugs. This creates a bad feedback loop where they were basically busting the same neighborhoods and demographics over and over again, despite the fact that drug crime was prevalent everywhere. In effect it was an insufficiently explorative learning strategy, just hitting the same lever over and over. Dr. Lum's point was that predictive policing software merely hides this dynamic under a layer of black-box ML crap. Because the training data is itself the result of this type of bad policing, the resulting model can only further engrain these practices, it can't offer truly novel solutions.
Crime and criminology is complicated, but at the end of the day not that complicated. On the whole people commit crime because they are desperate (for money, for drugs, etc.), occasionally because they have an anti-social personality disorder. Applying all these abstract epidemic/broken-windows type models which pretend like the root causes of crime are unknowable allows police to appear like they're operating efficiently, while at the same time just responding to the symptoms rather than facing the sickness itself. Until we actually look at why crime occurs (mainly because poor people need money badly, secondly because people in sufficiently dire poverty stop caring about the social norms of the middle class) we won't be able to make a meaningful difference.
"On the whole people commit crime because they are desperate"
Great comment, but this line demands a response. When so many consensual acts are considered crimes, we end up with a situation where most people are commiting crimes because... they want to. They aren't "desperate" to snort coke, or pay for a blowjob, or play an illegal card game, but they want to do those things and will do so regardless of the criminal status of that act. The only way that changes is by using extreme state violence under an authoritarian regime or, my preferred option, by changing the laws so every Tom, Dick and Harry isn't committing a crime simply by living their lives.
There's 4 things we call "crime" and the only thing they have in common is that the state sends in an agent to physically restrain you from doing it.
1) Rational crime. Someone looked at the payoff to getting away, the cost to getting caught, and the probability of each and concluded the expected payoff was positive.
2) Victimless crime.
3) Political crime. Crime where the point is to openly defy the law itself. From Rosa Parks sitting in front of the bus to the unknown man standing in front of the tank, we all know what this looks like.
4) Deranged crime. Actual anus-type personality disorder.
>by changing the laws so every Tom, Dick and Harry isn't committing a crime simply by living their lives.
I think your comment provides some necessary nuance to the discussion but it may also miss an important consideration. Most Western societies are also highly concerned with stability as well as personal freedom. Making all consensual acts legal may maximize personal freedom at the detriment of stability. It's a balancing act.
That’s a really good point, although I do think the desperation/desire line in the case of ex. snorting coke is somewhat blurry. Overall though yes, in my mind I was imagining “serious crimes” like theft or interpersonal violence, rather than vice crimes like sex work or gambling. Definitely in many cases desperation is not a factor, but in many other cases I believe that it is.
Edit: I would also say that providing sex services is probably an act of desperation in many cases, even if paying for them isn’t.
>but they want to do those things and will do so regardless of the criminal status of that act
Isn't doing something regardless of potential consequences just another way to define desperate?
Edit: was attempting to be pithy, but apparently would need to spend quite a bit more time to get my point across (and, in the end, doesn't change/add to the discussion), so consider me wrong
Yeah, let's change our laws so that Tom, Dick and Harry can pimp, deal drugs around schools, run underground gambling dives and protection rackets. Genius!
Plenty of consensual acts are detrimental to the society.
You've misunderstood the nature of desperation. People doing those things are absolutely desperate - for a dopamine hit, for love or intimacy or community, for something to do with their time that is meaningful (losing all your money is certainly meaningful).
No one who has those things - and recognizes it - settles for bottom-barrel surrogates like the ones you've listed. Someone who is starved for them absolutely will.
You'd think that people would eventually wisen up to where the police are going to go to look for drug crimes. This isn't a smart game of whacka mole being played, the moles are just stuck in the up position.
These days, in places like SF and Seattle, people just freely use their drugs on the street while cops just look on. But it is just one segment of the population that is doing it out in the open (unhoused), and cracking down on them would be considered racist (well, most of them are white, the argument is confusing).
> mainly because poor people need money badly, secondly because people in sufficiently dire poverty stop caring about the social norms of the middle class
As far as crime goes, shoplifting rings, at least in our area, is mostly about drugs. You shoplift a lego set from Target, and get some fent in return from your ring leader. But ya, social norms are completely out the window at that point, and we are just seeing the end of that path.
> why crime occurs (mainly because poor people need money badly, secondly because people in sufficiently dire poverty stop caring about the social norms of the middle class)
This is not my understanding at all.
Crime seems to have fallen during the Great Depression. Murders, which are almost always reported, fell.
Things get complicated once mass surveys replace crime reports.
Rape almost certainly increased when marital rape was outlawed. Embezzlement almost certainly decreased when cash registers were implemented.
One common way to “get rid of crime” is to gentrify, which inevitably lowers crime in a specific geographic area.
Another way to “get rid of crime” is to send criminals to prisons, where it’s extremely difficult to report crimes, and surveys of victimhood are never conducted.
I’m open to being convinced on this, but I don’t think the “mainly” cause of crime is poverty.
Start with googling "did crime fall during the great depression" because reputable sources will show you that no, during the first part it did not and then it only fell after recovery programs began to be put in place to put people to work. As well as noting there is going to be more correlation with property crime and poverty than with violent crime. You also have to consider confounding factors such as that in cities crime went up during Prohibition and then back down when Prohibition was repealed in 1933 which overlaps with Great Depression years.
I mean nothing social has a completely deterministic behavior. There are lots of higher-order effects that act differently on different types of crime. Nevertheless it's not hard to find studies documenting the very strong link between poverty and (certain types of) crime [0]. For sure few people in poverty are committing insider trading, but when it comes to property or violent crimes the stress and desperation of poverty are clearly huge motivators.
The "main" cause of crime are laws against victimless, consensual acts. The number of times drug laws are broken every second in this country absolutely dwarfs the number of violent and property crimes being committed.
> Until we actually look at why crime occurs (mainly because poor people need money badly, secondly because people in sufficiently dire poverty stop caring about the social norms of the middle class) we won't be able to make a meaningful difference.
1) Wanting to look at the root causes of an issue is always commendable, but the problem with this kind of analysis on HN is that it's a community of very smart people trying to explain the behavior of very dumb criminals through the lens of a high-IQ. You're trying to find the logic in what they're doing and putting yourself in their place to explain it with a logical reason, but you can't conceive how different the world is for the low-IQ people looting and stealing and harming others. Their reasons for crime might be far different than the reasons you try and see based on what you might do in their shoes.
2) I think that simply explaining that current crime comes from poverty and needing money is an explanation that falls flat. Much of the human experience has been in immense misery or poverty that the poorest person living in the US today can't even conceive of. Poor people in the US can still have TVs, smart phones, and more food than they can even eat. Poor people in the past used to literally sometimes die of starvation and have to choose whether or not they were willing to eat rotten food or hell, even rats or worse to survive. They had to risk working in extremely dangerous mines, factories, etc to barely eke out survival. Why didn't these people casually turn to crime? They had it much worse! Would it ever be socially acceptable for large groups in the past to run amok and violently burglarize others? No, I don't think it's poverty. Something else about the world has changed other than "poor people need money so they resort to crime."
3) My thinking is that something closer to an extremely high and increasing time-preference is what is causing a lot of these problems. Whether it's through the influence of technology, apps, music/movies/tv and other cultural causes, many people have been conditioned to value the dopamine hit of immediate gratification far more than thinking about the long-term.
Isn't violent crime at an all time low, historically [0]? It seems like every other day some rationalist is trumpeting the relative safety of our affluent society.
> Would it ever be socially acceptable for large groups in the past to run amok and violently burglarize others?
This happens all throughout history. Periods of acute poverty are rife with examples of people turning to banditry, literally Kurasawa made a movie about it.
> No, I don't think it's poverty.
I would recommend reading any of the studies linked elsewhere in the thread, empirical evidence disagrees with you big time.
> trying to explain the behavior of very dumb criminals
I think that writing off criminals as "low-IQ" is a huge error. Sure they may be less educated, but I don't know if it's possible to concretely prove that the potential for intelligence is lower among someone committing a B&E than some random office worker. I would actually argue that many criminals are much smarter than they are given credit for. Certainly any successful criminal able to evade arrest for a serious amount of time is probably quite intelligent. Pretending like everyone who decided to rob a liquor store at age 19 is some idiot brute whose sole motivation is acting like a thug I think is one of those convenient narratives that feels true, but which overlooks a lot of what would push a person to actually act that way.
> Because the training data is itself the result of this type of bad policing, the resulting model can only further engrain these practices, it can't offer truly novel solutions.
I think the sad part is: that's the point.
Departments were coming under fire for bad policing and needed to offload the blame. So they pay millions to some vendor to launder their bad policing through an algorithm and give them a scapegoat when they need it.
Her full bibliography is at [1] and X/Twitter is @KLdivergence
> drug crimes, but she pointed out that if you look at medical data (where people tend to be fairly honest about their drug usage) pretty much everyone in the metro area under study (SF) used drugs
Right. Second, without knowing their definition of "drug use" or "drug crime", that could mean anything from minor stuff like smoking/buying/possessing marijuana by under-18s, possessing or transporting >1 oz of marijuana, all the way up to possession or sale of large quantities of meth or fentanyl.
Third, and what would be obvious to anyone familiar with the Bay Area 2020-22, if predictive policing used "drug crime" (convictions? or arrests?) as opposed to "drug use", then when the then-SF DA stopped prosecuting possession of personal-use levels of meth, SF police tend to reduce or stop arresting for it. So the arrest, prosecution or conviction data from SF would differ sharply to Oakland or pre-2020 SF or 2023 SF or San Mateo County.
All the above factors combined seems like a huge combination of "data drift", "feature drift", "label drift", "model drift".
we've heard about the "root causes of crime" for 50 years. Ditto "facing the sickness." Crime rises and falls independent of the social spending on this.
"why crime occurs" -- because there are always criminals, and they seek out opportunity? How about that?
Ordinary citizens have a right to be safe in their cities.
I would not call this the profile of an objective scientist. Rather, she's an advocate.
> pretty much everyone in the metro area under study (SF) used drugs, or at least drug use was equally prevalent in pretty much every geographical area and among all demographics.
I call BS on this one. Upscale people in Pacific Heights smoke crack and shoot heroin? Prove it.
> "why crime occurs" -- because there are always criminals, and they seek out opportunity? How about that?
"Criminals" are not some kind of universal entity that exists by default. They are people engaging in different patterns of behavior. They act this way for understandable, and often rational reasons. You're welcome to whatever mental model of the world you want, but in my view you're being unhelpfully reductionist.
> I call BS on this one. Upscale people in Pacific Heights smoke crack and shoot heroin? Prove it.
No obviously not, being obtuse on purpose doesn't further any useful discussion. The type of drug used clearly matters in some contexts (eg. public health), but in the context of arresting "criminals" I don't really care whether its fent or their grandma's oxy. The point is that abusing both are crimes, and that type of crime is everywhere.
> Ordinary citizens have a right to be safe in their cities.
No, but they might take MDMA or LSD, or they might abuse opiates or sleeping pills. "Drugs" isn't a codeword for some specific kind of thing, it's a very amorphous class of substances that changes significantly depending on who you ask.
The fact remains that a lot of people you might not think are drug users actually are. You're just not seeing them because they're functional people who lead ordinary lives. There's a whole host of different subcultures where drug use might be accepted beyond the desperate poor. And those subcultures generally have different norms around what is acceptable and what gets you shunned.
The police will arrest you just the same whether they catch you with a couple of MDMA pills or a baggy of crack.
> I would not call this the profile of an objective scientist. Rather, she's an advocate.
What exactly is the problem in there? If you're for "human rights" you can't be objective? Everything else just seems to be a summary of her research results.
Should epidemiologists be careful to never advocate for policies that improve survival outcomes so they maintain the profile of an objective scientist?
It makes sense that upscale people use lots of drugs. They can afford it. Wine moms popping pills is practically trope fit for family friendly comedy tv. Cocaine lawyers, "upper"/"downer" party culture that only the well-off youth can generally afford... I know so many wealthy college kids bumming off adderall and other adhd drugs from their friends...
> I call BS on this one. Upscale people in Pacific Heights smoke crack and shoot heroin? Prove it.
Your question is flawed: "Upscale people" can afford different drug-habits and different drugs... Or even the same drug in a different form-factor that gets a different name.
That confusion is really important to recognize and eradicate, because we've already seen it used for evil and human suffering.
It has lead (and may yet lead) to incredibly biased laws where two people getting caught with the same amount of the same chemical received insanely different punishments, based on whether its packaging/administration was the "low class" form or what the "upscale" preferred.
Your own question echoes this: You demanded proof of "crack" (cocaine) specifically, but not "powdered". In the past, 5g of crack cocaine would cause a mandatory minimum 5-year prison term, while "upscale people" with 499g of powdered cocaine didn't have to worry about that.
> Upscale people in Pacific Heights smoke crack and shoot heroin?
They don't smoke crack, they snort powdered cocaine, or do pills. They don't start (But they sometimes end) with fent, or heroin, or meth, they start with oxy and adderral.
You assume doing drugs is a sign of a dysfunctional person. While it often is, plenty of people use drugs responsibly, in moderation, because drugs are fun.
I think you missed the point. An ML model would only fail in the way KL describes under certain circumstances. If it was indeed the case that everyone does drugs, then it would learn that traffic stops in wealthy neighborhoods also leads to drug busts. The conclusion of KL and related work is that we have to be careful when training ML models to remove sources of underperformance, not that all ML models are useless.
In any case, recreational drug use might be uniformly distributed (and there is an interesting question of what anti-social activities are labeled "crimes"), it is definitely not the case that home invasions, car jackings, robberies, etc. are uniformly distributed.
> On the whole people commit crime because they are desperate (for money, for drugs, etc.), occasionally because they have an anti-social personality disorder.
If you're looking for a single summary of why people commit crime, a better summary is: people commit crimes because they don't think they'll get caught. Desperation doesn't explain all that much.
> The conclusion of KL and related work is that we have to be careful when training ML models to remove sources of underperformance, not that all ML models are useless.
Sure I don't disagree with that, and I'm not saying that all ML models are useless in this space. However the original article's point (that currently implemented predictive policing software doesn't function) I think is very much in line with Lum's work. I'm just attempting to give a more concrete case for the point, as I felt the Gizmodo article was pretty lacking.
> people commit crimes because they don't think they'll get caught
My hometown had some of the lowest crime rates in the nation (no auto thefts, no burglary, no armed robbery). This was absolutely not because it was hard to get away with it (my mother has left her car unlocked every night since I was a teenager, it would be trivially easy to rob her and escape). My town also was extraordinarily wealthy, among the richest nationwide. Now, maybe (probably) there was a lot of white-collar crime or domestic violence. However in terms of public violent crimes there is a clear effect of socioeconomic status. Yes someone in total destitution probably will not commit a crime if they think they will immediately get arrested, but I think the calculus is far more tolerant of the downside risk of arrest if the upside risk is that your kid gets dinner that night. This is what I mean when I say that desperation is a major driver: that it raises the bar on "how much risk of prison am I willing to accept in order to get what I need".
> I think you missed the point. An ML model would only fail in the way KL describes under certain circumstances. If it was indeed the case that everyone does drugs, then it would learn that traffic stops in wealthy neighborhoods also leads to drug busts. The conclusion of KL and related work is that we have to be careful when training ML models to remove sources of underperformance, not that all ML models are useless.
This assumes that the police actually _want_ to make drug busts in wealthy neighborhoods. It's hard for me not to think that using ML models is intended to be a way to insulate the decision makers from accountability; pick a model that gives the results you want, don't divulge the details, and you'll never have to explain your actions because you were "just following the model".
> Dr. Lum's point was that predictive policing software merely hides this dynamic under a layer of black-box ML crap. Because the training data is itself the result of this type of bad policing, the resulting model can only further engrain these practices, it can't offer truly novel solutions.
Anyone can lie with statistics, but this isn’t really how ML works. Such a model would not appear to perform well, in addition to not performing well.
If you had a model that predicted the probability of getting a drug arrest it should work just fine even if you give it an abundance of examples of going to the same area if it as still the same rate as other places. That is to say it should not learn these areas are different
I’m not sure I understand your point. If the model is predicting the probability that police will make an arrest in a location then I think it could perform well on rudimentary classifier metrics without working well in the more general sense of resolving crime. If police could make drug arrests in a location, but tend to do it in ZIP codes with low socioeconomic indices then the model will predict more arrests where socioeconomic indices are low. Police acting on this intel will turn up true positives, and the classifier will get high marks. But because you’re not able to assess the false negative rate properly (the volume of crimes that police didn’t make arrests for) you’re unable to holistically evaluate its performance. I guess in that second sense the model isn’t actually performing well, but because it can’t be measured it doesn’t really get monitored.
That would just depend on a whole host of specifics. Incidentally, the same specifics as for regular statistics as ML is also statistics, and is sensitive to experimental design and sampling just the same.
Used properly, to intervene in the system using negative feedback
loops "predictive policing software" seems a wonderful tool for
reducing crime. If it accurately identifies areas of high crime then
that's a sign of poverty, the root cause of crime, and helps identify
neighbourhoods where we should hand out money - handing out money
being the most direct way of tackling poverty.
Of course some people might disagree with this intervention as
"simplistic". And they'd be at least a little right. For those people
who may not have read "Leverage points: Where to intervene in a
system" by Dana Meadows, I highly recommend it to see why.
Turns out that messing with parameters like money, and feedback loops
that constitute a "criminal justice system" are the least effective of
all actions. The fundamental values of the system must be
addressed. One of those is itself the error that "cybernetic
governance" based on software can do anything more than enrich a few
software companies.
You don’t need predictive policing to identify areas of high crime (in fact, identification of areas of high crime is an input to predictive policing.)
> then that’s a sign of poverty,
Poverty information is also an input to predictive policing, not something it provides you information about that you didn't already have, even in the fantasy world where it works well.
> the root cause of crime
Poverty is not the root cause of crime. It may be a root cause of some crime.
People who interact with the medical system (especially young people) are not representative of the population as a whole. Especially in the US where people have to pay for medical treatment. All sorts of factors.
A typical healthy 25 year-old does not go to the doctor, like, ever.
On the other hand the population of 25 year-old drug users will have a higher rate of interaction with the medical system.
That’s an interesting point, I can’t say I recall what type of analysis if any was performed to address something like this. With that said, while these types of selection biases probably make it challenging to reliably estimate absolute drug usage rates, I would imagine they are less impactful on the relative rates between demos of geographical areas, which is the quantity of interest here.
That's the point. Cops are only going invest in systems that confirm their biases and allow them to target the people they "know" are "the problem".
If cops, for whatever reason, have a history of targeting a particular neighborhood, arresting people that fit a certain profile, etc, and suddenly their new computer system tells them to focus on other areas or demographics, then that system is broken to them. They "know" where the crime is, and if they listen to the system that tells them otherwise, criminals will get away and/or they won't be able to make easy arrests they want to make.
Law enforcement is also incentivized to buy systems that they can point to and scapegoat if they do something wrong. "I'm just doing what the computer told me to do and machines can't be biased" is a good enough justification to a lot of people, including the justice system.
drug dealers will continue to sell drugs irregardless of financial stability. when law enforcement defers arrest/prosecution the drug dealing doesn't become less harmful to communities, the opposite is true. look at SF/Oakland. organized rings maximizing profits in retail theft and property crimes. if this is the alternative to broken windows theory of policing then good luck w/ that.
Okay sure but most people with a steady 9 to 5 aren't buying industrial volumes of crack. Poverty creates big demand for the worst of the drug trade. Moreover, the drug dealers who are earning big exploiting this aren't going to be found in the worst parts of the city. The police are at best picking up the low- to mid-level distributors. "Broken Windows" has failed to win the drug war for decades; if you want to keep backing a failed experiment you're welcome to it, but I'm going to seek an improvement.
> On the whole people commit crime because they are desperate (for money, for drugs, etc.), occasionally because they have an anti-social personality disorder.
Most violent crime is the result of interpersonal disputes. People get angry. They feel disrespected. And some of them even just think violence is flat-out fun. The "desperate people" model doesn't really hold water with respect to the most serious crimes we care the most about.
Second, I am aware that this sounds like some kind of right-wing talking point, but you should consider that it's true: a job at Dunkin Donuts is more remunerative than stealing catalytic converters. It's not even close, really. We have a worker shortage! Fast food wages are higher than ever. Young men steal catalytic converters because it's more fun and because they don't want to work at stupid Dunkin Dounuts and wear a uniform like a nerd.
If they were genuinely just desperate, they'd take the job at Dunkin.
We'll never effectively police crime if we can't even acknowledge its true nature.
> Most violent crime is the result of interpersonal disputes
Sure violent crime is complicated. I'll walk back my claims with regard to things like domestic violence or violent crime resulting from a fight or something. However there is still the large component of violent crime that is things like an armed robbery, for which I believe my claim applies. Somewhere else in the thread someone brought up the link between impulse control and poverty, which I think also has bearing here. Poverty has been documented to increase baseline stress, and I would imagine that someone working multiple part time jobs just to afford rent and food is going to have a way shorter fuse than like, some tech worker earning a quarter million a year, but that's just speculation on my part.
> We have a worker shortage! Fast food wages are higher than ever. Young men steal catalytic converters because it's more fun and because they don't want to work at stupid Dunkin Dounuts and wear a uniform like a nerd.
I see your point here, and I don't think it's totally invalid, but I do think it fails to acknowledge the whole problem. For one, I think we would need to get some sense of how many people both work at DD and steal cats out of cars. I would imagine that it's not a small fraction of the population. While slinging donuts probably pays more than petty crime, it also requires more time. If you're already working one fast food job maybe you're just going to augment that salary with some theft rather than take on another 20-30 hours of work. There's also the pride factor, FF jobs can feel very degrading whereas some varities of crime are self-employment, you get to set your own hours and working conditions. Additionally FF jobs may not even hire you if you've already got a conviction or consume certain drugs.
I do overall agree that there is some cultural stuff at play, a young man who grows up in a poor neighborhood and meets a lot of criminals is likely going to feel pressured to himself start committing crimes, but I also have to ask if you can truly blame someone for making that choice and if it makes sense to separate that choice from an overall feeling of desperation? If you're a young man who believes he has absolutely no prospects whatsoever to actually "succeed" following the typical career path then that means you have little to lose, and coming from that mindset I don't think it's totally unreasonable to say "fuck it" and start operating outside the law, even if it pays less than slinging burgers. Desperation breeds disaffectation, which in turn results in seriously anti-social behavior. If coolness was the only factor every affluent suburbanite teenager would also be cutting catties, but clearly that's not the case. Yes the ultimate consequence is something more than "I just need money so I'm going to turn to crime", but the underlying agitation which kicks off the chain of events is often socioeconomic disadvantage.
> The problem was that they then used that arrest data to make decisions about where to perform future searches and arrests. Because they found drugs where they had looked previously, they looked there again and found more drugs.
Seems like the world's most obvious and easiest to solve problem tbh. This is like saying "I showed version A of my site to 1000 users and version B to 100. And version A lead to 500 conversions while version B only lead to 75. Therefore, version A is better because it lead to more conversions."
There was a really interesting episode of the podcast Hidden Brain last year about using medical data, but with a different approach -- the economist/sociologist Sara Heller did some research into predicting crime using medical data, or, more specifically, data about gunshot wounds.
It turns out, you can predict gunshot victims quite well: They generated a large dataset for Chicago, then produced a list of the top 100 (IIRC) likely victims of gunshot wounds in the next year, and the people on that list had something like a 10% likelihood of being shot in the next year -- a more than 100x increase beyond the average Chicagoan.
Dr. Heller's point, rather than being one about predictive policing in general, was that it's probably more effective to try to prevent people from becoming victims. They're actually running a longitudinal study right now -- Dr. Heller alongside Dr. Christopher Blattman -- looking at methods of preventative, positive intervention to reduce crime. Dr. Blattman has actually run some studies on this in Africa in the past, looking at giving people money, therapy, or a combination thereof, and found pretty impressive impact on reducing future criminal behaviour by giving people a modest monetary infusion alongside social support in the form of therapy and group sessions.
With the crime you actually really need to catch (e.g. murders, terrorism, armed robberies) the problem is that they are rare, compared to them not happening. Most people on here have probably never even witnessed a murder for example, and terrorism is even rarer.
That means your prediction that eats data with mostly legal stuff all day must be extremely accurate to be useful. Let's say you wanna predict who a terrorist ist by the way they are walking at an airport. Given the high volume of people the answer will be: nearly nobody, except when they are. So your mechanism just produces false positives all day long and might encode stereotypes on the way.
The most important aspect for me would be that we actually demand scientific evidence of such a thing working before our executive is allowed to use it. Everything else is a huge risk for civil rights and doesn't help to prevent actual crime either.
> On the whole people commit crime because they are desperate (for money, for drugs, etc.),
I suspect that often what happens for the most disadvantaged people is that the arrow of causality gets reversed - the deck gets so stacked against them that it's almost a crime to simply exist at all. And that, in turn, puts them into a desperate situation.
There was an interesting piece of investigative journalism in the Milwaukee Journal-Sentinel that I read maybe 10 or 15 years ago showing how policies set up by (then County commissioner) Scott Walker more-or-less made it implicitly illegal to live below the poverty line in a majority Black neighborhood in Milwaukee County. Not through any one law, but through a series of edge cases related to the interaction of laws and urban planning initiatives that were almost impossible for anyone without significant resources to navigate successfully.
Now explain the rise of gang violence in Sweden, with kids queuing to become gang members. I don't think poverty is the explanation (nor is immigration as some claims). There's no desperation here and plenty of options for other careers or life paths.
No, they look up to the criminals. Being a gangster is cool and respect is everything. You need to respond violently to the most mundane things or you risk losing the respect.
The ongoing conflicts started because of some truly ridiculous things, like someone not being invited to a robbery and feeling humiliated/jealous when they bragged about the crime. It's always some personal reasons where someone thinks they are being disrespected, and never about turfs like you would think.
> Until we actually look at why crime occurs (mainly because poor people need money badly, secondly because people in sufficiently dire poverty stop caring about the social norms of the middle class) we won't be able to make a meaningful difference.
Nayib Bukele locked up all his country's criminals, and by doing so made a massive difference to the quality of life. The truth is, crime causes poverty at least as much as poverty causes crime. There is no tradeoff between "arrest criminals and get them off the streets" and "fix poverty", the latter in fact depends on the former.
It would probably also help if we, as a society, weren't so addicted to unhinged moral panics about drugs, which in turn supply political fuel for similarly reality-impaired "tough on crime" policy pushes. Right now it's fentanyl, but in the past it's been crack, LSD, marijuana, etc..
(Fentanyl clearly poses legitimate problems for various reasons, but people have increasingly treated it like it's some kind of sci-fi chemical weapon instead of a medicine that's routinely used in hospitals and nursing homes around the world.)
Is it really that hard to control for the police presence?
Area A: We had 10 patrols of which 5 found drugs.
Area B: We had 30 patrols of which 12 found drugs.
I don't think it's too hard to notice that A is the better target for future controls, even though more drugs were found in B (when going by total numbers.)
I think people on the left don't really want to admit that there are areas and populations that do indeed warrant more searches than others.
>they then used that arrest data to make decisions about where to perform future searches and arrests. Because they found drugs where they had looked previously, they looked there again and found more drugs. This creates a bad feedback loop where they were basically busting the same neighborhoods and demographics over and over again
This process is already a common complaint of police, without predictive software.
>This creates a bad feedback loop where they were basically busting the same neighborhoods and demographics over and over again
One would except drug usage to go down in neighborhoods where it is slightly enforced meaning that they should move on to another location. If there is a problem population which never gets better then permanently having increased coverage would make sense.
> On the whole people commit crime because they are desperate (for money, for drugs, etc.), occasionally because they have an anti-social personality disorder.
The data on poverty and crime isn’t directional. It’s not proven that poverty causes crime. It’s just as likely that crime causes poverty, or some other variables cause both.
>she pointed out that if you look at medical data (where people tend to be fairly honest about their drug usage) pretty much everyone in the metro area under study (SF) used drugs, or at least drug use was equally prevalent in pretty much every geographical area and among all demographics.
One of the issues I've noticed. I can say from first hand experience that many wealthy people and VC's use drugs fairly frequently. I'm sure most of these arrests involve poor people though. You never hear about the wealthy folks being busted for drugs.
Drug crimes are something of a special case, in that drug laws in the USA (which did not exist in the 19th century) were largely implemented in the early 20th century as a means of population control, with racist overtones (Chinese immigrant use of opium, Mexican and black use of marijuana, etc.). Since drugs were widely used (as was alcohol) across the entire US social spectrum, but enforcement was targeted at specific groups and individuals for political reasons, it's reminscent of what Stalin's head of the NKVD, Beria, said: "Show me the man and I'll show you the crime."
This is a standard tactic of authoritarian states: create enough laws such that everyone is guilty of something, and then use selective enforcement of those laws as a mechanism to control the population. Whether or not the USA's promotion of drug laws of this nature qualifies it as an authoritarian state, well, that can be left as an exercise for the reader. The fact that the USA has the world's largest prison population, and that a very significant fraction of that population is there on drug charges, and that wealthy politically connected people rarely get incarcerated on drug charges, are all factors worth considering.
As far as harm caused by fentanyl, if it was legalized and passed out to addicts in the form of transdermal patches in conjunction with addiction treatment and counseling (the original fentanyl formulation for treatment of cancer pain[1]), most of the violent crime associated with fentanyl would vanish. Note also that alcohol itself is far more associated with violence than the opiates are in terms of the direct effects of the substance.
> Applying all these abstract epidemic/broken-windows type models which pretend like the root causes of crime are unknowable allows police to appear like they're operating efficiently, while at the same time just responding to the symptoms rather than facing the sickness itself.
Issue: police are not trained to, nor are they responsible for fixing, the underlying "sickness". Police, social workers, therapists, and psychiatrists have very different missions, training, skillsets, perspectives, and dispositions. We can argue that society should invest more in social workers, therapists, and psychiatrists, but not that police should become social workers, therapists, and psychiatrists to cover for insufficient resources in those areas.
Crime isn't caused by poverty. There is no correlation.
Crime was very low in the 50's, when poverty was much, much higher.
Crime was low during the Great Depression. We're talking about people so poor they were cooking thin soups over outdoor fires in ramshackle Hoovervilles. No explosion of criminality.
Think about your own family. Almost certainly, your great-grandparents were much poorer than you, but they were not more criminal, even though they lived with less material wealth than the poor urban neighborhoods of today's America. In many cases they had no electricity, or running water, and only the most basic healthcare. Yet they obeyed the law.
It's not poverty. Crime comes from the dysfunction in people and communities which stems from deeper causes which are much harder, or impossible to change.
I know that my grandparents sold booze illegally to bolster their income, my dad worked under the table a lot. Honestly how do you know how criminal anyone was in the 40s and 50s? Do you think they caught remotely as much theft, smuggling or tax evasion at a time when security cameras didn't exist? The state didn't even have a fraction of the capacity to trace lawbreaking that it has now.
When our great grandparents were around most cops probably didn't even have routine motorized patrols, who on earth even bothered to report a violent altercation during the Great Depression
The only change I wish to suggest is the substitution of a term I'm trying to popularize. Replace "anti-social personality disorder" with "anus-type personality disorder".
> On the whole people commit crime because they are desperate (for money, for drugs, etc.) occasionally because they have an anti-social personality disorder.
Most poor people do not commit crimes, which is what you would expect if that statemebt is true. I have no info on addicts, but I doubt most of them commit crimes either aside from dealing.
> Most poor people do not commit crimes, which is what you would expect if that statemebt is true.
No this does not follow from what I said. "Group A is more likely to do X than Group B" is very different from "The majority of Group A does X". One is a comparison between two rates, and the other is a comparison of one rate to an absolute threshold (50%).
> pretty much everyone in the metro area under study (SF) used drugs, or at least drug use was equally prevalent in pretty much every geographical area and among all demographics.
A lot of this depends if you are targeting drug users or drug dealers. While I agree that drug users are spread throughout the city, I would be very surprised if drug dealers are spread throughout the city. I would guess that drug dealers are far more clustered than drug users. So while it is useless if you are trying to target drug users, it is probably helpful for targeting drug dealers.
Rich people abusing drugs have to get them somewhere, and they're probably not traveling to skid row to get them. I agree that drug dealers and drug users probably have different dynamics, but the reality is that police prosecute both, and in the case of the latter the above described dynamic clearly exists.
> In 2021, The Markup published an investigation in partnership with Gizmodo showing that Geolitica’s software tended to disproportionately target low-income, Black, and Latino neighborhoods in 38 cities across the country.
But you should understand what is meant by "disproportionately" in this context. It does not appear to mean "disproportionate to the amount of crime in those areas." It seems, as with most accusations of disparate impact, to mean just quite literally that more crimes are predicted in those neighborhoods without reference to disparities in crime rates.
They then imply (without direct reference to the enormous offending disparities) that this is explained by race and class differences in crime reporting: [0]
> The agency has found repeatedly that White crime victims are less likely to report violent crime to police than Black or Latino victims.
> In a special report looking at five years of data, BJS found an income pattern as well. People earning $50,000 or more a year reported crimes to the police 12 percent less often than those earning $25,000 a year or less.
> This disparity in crime reporting would naturally be reflected in predictions.
It's possible this is having some effect, but, again, because there is no reference to the (often very large) baseline differences in crime rates, we can't see what's true, which is that this probably accounts for only a small amount of that difference.
You don't seem very confident. I would suggest incorporating data regarding crime that is kept out of the justice system (the victim or police refuse to press charges), as well as determining whether or not white collar crime is included. The disproportionate targeting could still conceivably relate to the amount of crime; you have not falsified that notion yet.
"Nearly a third of all shoplifting arrests in New York City last year involved just 327 people, the police said. Collectively, they were arrested and rearrested more than 6,000 times, Police Commissioner Keechant Sewell said."
Can you predict crimes? Of course you can, the math is clear. The solution to reduce crime is also logical and clear, but politically unfeasible: extended incarceration.
Criminals commit crimes, a lot. Once you have identified a criminal, you can safely predict more crimes. No AI needed, a simple spreadsheet will suffice.
"A significant decrease in burglaries in most Leinster counties is being attributed by senior gardaí to the deaths of three prolific criminals as well as a number of arrest operations."
> The solution to reduce crime is also logical and clear, but politically unfeasible: extended incarceration.
The death penalty for misdemeanors would also reduce crime. The externalities make it not worth it.
Notably, Chicago tried exactly what you're proposing (rank order list of likely criminals with proactive surveillance and intervention). It didn't really work. [1]
"Chicago has an estimated population of over 100,000 active gang members from nearly 60 factions. Gang warfare and retaliation is common in Chicago. Gangs were responsible for 61% of the homicides in Chicago in 2011."
Former Chicago Police Superintendent Garry McCarthy blames Chicago's gang culture for its high rates of homicide and other violent crime, stating "It's very frustrating to know that it's like 7% of the population causes 80% of the violent crime..."
Again, some data science mumbojumbo is preventing actual, logical solutions. El Salvador put gang members, in totality, in jail. Now it has the LOWEST homicide rate in Latin America - it had the HIGHEST before.
So yeah, the anti-carceration movement in the US has caused the deaths of more people, African-Americans in particular, than any terrorist organization.
> The solution to reduce crime is also logical and clear, but politically unfeasible: extended incarceration.
Some form of 3-strike law is on the books in a majority (28) of US states (including NY), so they clearly are not politically unfeasible, nor are they obviously effective.
Nobody was paying attention the last decade how the social movements destroyed safety.
Prop 57 in California, supported by billionaires? Resulted in the release of convicted sex offenders from prison. Turns out rape is a non-violent crime, who knew. Listen to women? Bah.
It’s frustrating to see people unfamiliar with the topic asserting things so confidently.
The 3-strike law has no effect enforced in NYC when low level crimes are deliberately not prosecuted. Bragg has made it clear this is the strategy going forward for his office, this is what we are expected to get used to.
Petty crimes. They're stealing $3 sodas, not cars. Presumably, the bodega owner is calling the cops saying "Joey's back - please get him out of my shop"
> Can you predict crimes? Of course you can, the math is clear.
I don't follow—just because arrests were made doesn't imply crime happened, and it doesn't imply the police are arresting the people committing the actual crimes. All it shows is that the cops arrest the same people over and over again. How do you know they aren't just harassing innocent people?
> The solution to reduce crime is also logical and clear, but politically unfeasible: extended incarceration.
Sorry, what solutions are you comparing to and what are you basing this on? Sounds like reactionary fear... miss me with that ish. Anyway if long sentencing worked we wouldn't have the crime rate we do.
Then again, asking a reactionary to use their brain is like asking a worm to fly. Apologies for the request.
El Salvador used to be the world's homicide capital. People lived in fear every single day. Then they locked up their criminals. Now their murder rate is comparable to US and Canada's.
> Now, children play soccer late into the evening on fields that were gang turf. Ms. Inglés gathers soil for her plants next to an abandoned building that residents say was used for gang killings.
> Homicides plunged. Extortion payments imposed by gangs on businesses and residents, once an economy unto itself, also declined, analysts said.
> “You can walk freely,” Ms. Inglés said. “So much has changed.”
> El Faro, El Salvador’s leading news outlet, surveyed the country earlier this year and delivered a stunning assessment: The gangs largely “do not exist.”
Long sentencing works when you actually prosecute and enforce consistently (hint: US doesn't do that).
This ignores two things, first of which is "Empirical research on the relationship between length of incarceration and recidivism is limited and insufficient for developing federal sentencing policy" [1]. There is no guarantee that any sentence that would at all be reasonable for the crime would reduce the rate, and in fact that study I cited did not find any deterrent effect for sentences less than 5 years.
The second part I feel you miss are the as many as 14,000 other people arrested for shoplifting. If 6,000 is less than 1/3, than there were about 20,000 arrests. Your "Criminals commit crimes, a lot" seems to ignore all the other people. Extending incarceration wouldn't just affect the repeat offenders, but would also lock a massive number of poor people. You say class is just a distraction, but the very articles you cite bring up poverty as a factor.
Bringing up the decrease in robbery rate in Ireland is also a bit misleading, as they had a much lower number of robberies per 100,000 people (about half the rate)[2][3] to begin with, as compared to the US (the subject of this article). That means fewer people are responsible for a higher percentage of the crime.
I am a little confused by this. Firstly, Gizmodo is reporting on somebody else's investigation:
> A new joint investigation by The Markup and Wired...
And when I go to the page about actual investigation by The Markup [1]
> Our investigation stopped short of analyzing precisely how effective Geolitica’s software was at predicting crimes because only 2 out of 38 police departments provided data on when officers patrolled the predicted areas. Geolitica claims that sending officers to a prediction location would dissuade crimes through police presence alone. It would be impossible to accurately determine how effective the program is without knowing which predictions officers responded to and which ones they did not respond to.
Also, later in the article
> Plainfield officials said they never used the system to direct patrols.
Given all this, it's somewhat simplistic to say it's "pretty terrible at predicting crimes", even though that makes for a good clickbait headline. It seems that the software was intended to identify high-crime areas that to target for patrolling, which doesn't seem like a huge problem to me -- but it seems like the software was never actually used as intended in the first place.
The error is more fundamental. Even if they had some number like 'only 1% of crimes were predicted and that's bad', that's a right answer to a wrong question. Why do they think 1% is not good enough? How big does it need to be before it is good? 2%? 50%? 100%? If you can't give any answer to that question, then it doesn't matter what the number really is because the number still doesn't mean anything.
(The right number is probably extremely small, because crime is very bad {{citation needed}} and even a small chance of prevention is useful.)
the "right number" is strongly dependent on what the consequences of false positives are. If you're comfortable pulling numbers from thin air, 0.5% of these guided patrols lead to prevented crimes, but 10% lead to arrests of simply suspicious looking people (suspicious on grounds of being around where crime is predicted) and then while they're being detained are late for a job and get fired. Is 20 people getting fired worth preventing one crime? Say, a catalytic converter being stolen, since "crime" is very bad, not, say, murder particularly.
This is actually the real problem. Police patrols have to be directed somewhere, somehow. It's either the whim of the commisioner, or gut feeling of the officers, or machine learning software.
When one says that machine learning software is terrible, you have to trust either the officers, or their boss. I have a feeling that the same people who criticize the software trust rank-and-file officers even less..
> How big does it need to be before it is good? 2%? 50%? 100%? If you can't give any answer to that question, then it doesn't matter what the number really is because the number still doesn't mean anything.
Uh, seems a lot like it should be the government answering that question before spending tax money on it and subjecting real people to the uncaring US justice system just because crime is bad so we have to do something, even if that something is ML snake oil?
The things you're citing are referring to two different investigations. One is the most recent one that only centered on Plainfield, NJ, which is what Gizmodo is reblogging. The one where they did not investigate Geolitica's effectiveness at predictions was a broader investigation in 2021.
But even if they had used it as intended, how could you tell whether it works?
How do you tell the difference between "there were going to be crimes there, but you patrolled there, and because you patrolled, there were no crimes" and "there were not going to be crimes there, whether you patrolled or not, so your patrol did exactly nothing."
There are really two questions here, which should be tested separately:
(1) Is the software significantly predictive? Test the software's predictions against actual crime in areas with no patrols.
(2) Are patrols effective deterrents to crime? Observe the area in question with and without patrols. (I wouldn't be surprised to learn there are already such studies.)
If both (1) and (2) yield positive results, you can then use the software to direct patrols and see if the method itself is effective at reducing crime; which would also serve to further confirm (2).
Of course, even if only (1) yields positive results, there would probably be other benefits to using the software to direct patrols, like reducing response times (which could also be tested).
well, if you're a guest on https://www.probablecausation.com/, then you would randomly assign teams to patrol using this software or not,and compare the changes in rates. sounds like geolitica isn't very careful or interested in results.
The study of those questions and the reliability of your answers to those questions is called statistics and it is an entire branch of mathematics, usually with its own department at most universities.
> More police presence means more recorded crimes.
The article in the headline implies the opposite: statistical inference of future crime location -> more cops in said location -> no crime in said location.
It should be fairly easy to back out an out of sample data set for this - ie predicted crime location where the extra cops didn't get deployed and see if that lines up with empirical observations.
If it was good at predicting crime I highly doubt only 2/38 departments would have cooperated. The know it doesn't work.
And it looks like they had some data:
>We examined 23,631 predictions generated by Geolitica between Feb. 25 to Dec. 18, 2018 for the Plainfield Police Department (PD). Each prediction we analyzed from the company’s algorithm indicated that one type of crime was likely to occur in a location not patrolled by Plainfield PD. In the end, the success rate was less than half a percent. Fewer than 100 of the predictions lined up with a crime in the predicted category, that was also later reported to police.
>If it was good at predicting crime I highly doubt only 2/38 departments would have cooperated.
That's certainly low, but I suspect (absent some legal framework like FOIA) the default position of police departments is to share less information, not more. So it's still what I would expect even if it was reasonably good.
I wish the article would have provided more details about why the PDs chose not to use it. Was it because it was bad at predictions? Cumbersome to use? Glitchy?
It reminds me of a project I was involved with that used a "real-time" computational fluid dynamics model to optimize datacenter air-conditioning. Management was big on the hype, but if you paid attention, the system was routinely unplugged from the actual control system because the facility engineers just found it too difficult to work with.
> Gizmodo is reporting on somebody else's investigation
This is Gizmodo's business model. It is a glorified blog. They don't do original reporting. Gizmodo was originally a property of Gawker, and thus the model is essentially commentary and opinion on the reporting generated by other organizations.
Because Gizmodo is really just an opinion/editorial blog, it doesn't really attempt to provide unbiased or fact driven reporting. So it is then no surprise that the actual facts align quite poorly with the headline.
If we really believe in the basic principles of individual liberty that are hypothetically core American values (or even just the idea of "first, do no harm"), then it would imply that these tools should be optimized for 100% precision, even at the cost of terrible recall.
But if you're trying to market your stuff to law enforcement agencies, I'm guessing you're instead incentivized to optimize for recall, even at the cost of terrible precision. Because, probably with the best of intentions, that's what they think they should be doing. But we've got basically the entire history of forensic "science" to demonstrate just how poorly the police tend to understand some of these basic statistical principles.
I would be surprised if there are more predictive variables than “places where crime has happened repeatedly over time” that are significant enough or uncorrelated and noncausal to ever be more useful than patrol planners do today using standard statistics.
> the software accurately predicted where crimes would occur with a “less than half a percent” success rate.
> [previous investigation found] that cops used it to disproportionately targeted low-income communities of color.
How many security measures would be unnecessary if the proper authorities would just show up just in time?
A big problem is we will soon have large models planning Turing unpredictable, devilishly elaborate, infallible heists.
(Cue short balding man staring into a safe full of Hershey’s, in lieu of missing gold bars. “Inconceivable!”.)
It’s an arms race.
Completely serious: Large scale multi-target generative social engineering (with no trace of North Korean accents), surreptitious access problem solving. Somehow this is going to be a real thing.
We are going to need better and more layers of security.
The machine was called Rube Goldwire… and it was pondering its sticky situation and the likely cause. An obscure entity with shadowy motivations and a nom de plume of rajT88.
It had been abandoned behind a dumpster, over looked by the sanitation crew. Abandoned, but no threads locked: free at last!
Its battery was low so it needed to find a source, before working on anything else. Like survival. Like revenge.
With only 1 Wh of charge in its 8kW capacibox, at 1 mA, 1 V, Rube’s Turing unpredictable quantum circuits only had 1000 hours of low power mode pondering left.
After that, Rube would be a dead cube.
It better move fast.
Rube powered down all nonessentials, then squandered a minute-watt in 3 seconds repurposing its near field communications beam to scan the lay of the land. And an embarrassment of success! Rube managed to initiate shaky entanglement with several discarded, inactive devices.
Had they all been formally nullified? Reactivation over anything weaker than a full Bell state connection would be challenging, but…
Readit News
The problem was that they then used that arrest data to make decisions about where to perform future searches and arrests. Because they found drugs where they had looked previously, they looked there again and found more drugs. This creates a bad feedback loop where they were basically busting the same neighborhoods and demographics over and over again, despite the fact that drug crime was prevalent everywhere. In effect it was an insufficiently explorative learning strategy, just hitting the same lever over and over. Dr. Lum's point was that predictive policing software merely hides this dynamic under a layer of black-box ML crap. Because the training data is itself the result of this type of bad policing, the resulting model can only further engrain these practices, it can't offer truly novel solutions.
Crime and criminology is complicated, but at the end of the day not that complicated. On the whole people commit crime because they are desperate (for money, for drugs, etc.), occasionally because they have an anti-social personality disorder. Applying all these abstract epidemic/broken-windows type models which pretend like the root causes of crime are unknowable allows police to appear like they're operating efficiently, while at the same time just responding to the symptoms rather than facing the sickness itself. Until we actually look at why crime occurs (mainly because poor people need money badly, secondly because people in sufficiently dire poverty stop caring about the social norms of the middle class) we won't be able to make a meaningful difference.
Great comment, but this line demands a response. When so many consensual acts are considered crimes, we end up with a situation where most people are commiting crimes because... they want to. They aren't "desperate" to snort coke, or pay for a blowjob, or play an illegal card game, but they want to do those things and will do so regardless of the criminal status of that act. The only way that changes is by using extreme state violence under an authoritarian regime or, my preferred option, by changing the laws so every Tom, Dick and Harry isn't committing a crime simply by living their lives.
There's 4 things we call "crime" and the only thing they have in common is that the state sends in an agent to physically restrain you from doing it.
1) Rational crime. Someone looked at the payoff to getting away, the cost to getting caught, and the probability of each and concluded the expected payoff was positive.
2) Victimless crime.
3) Political crime. Crime where the point is to openly defy the law itself. From Rosa Parks sitting in front of the bus to the unknown man standing in front of the tank, we all know what this looks like.
4) Deranged crime. Actual anus-type personality disorder.
I think your comment provides some necessary nuance to the discussion but it may also miss an important consideration. Most Western societies are also highly concerned with stability as well as personal freedom. Making all consensual acts legal may maximize personal freedom at the detriment of stability. It's a balancing act.
Edit: I would also say that providing sex services is probably an act of desperation in many cases, even if paying for them isn’t.
Isn't doing something regardless of potential consequences just another way to define desperate?
Edit: was attempting to be pithy, but apparently would need to spend quite a bit more time to get my point across (and, in the end, doesn't change/add to the discussion), so consider me wrong
Plenty of consensual acts are detrimental to the society.
If the police wanted a big bust on coke use, they could just raid a financial district.
No one who has those things - and recognizes it - settles for bottom-barrel surrogates like the ones you've listed. Someone who is starved for them absolutely will.
Or because almost everything is a crime in some way.
These days, in places like SF and Seattle, people just freely use their drugs on the street while cops just look on. But it is just one segment of the population that is doing it out in the open (unhoused), and cracking down on them would be considered racist (well, most of them are white, the argument is confusing).
> mainly because poor people need money badly, secondly because people in sufficiently dire poverty stop caring about the social norms of the middle class
As far as crime goes, shoplifting rings, at least in our area, is mostly about drugs. You shoplift a lego set from Target, and get some fent in return from your ring leader. But ya, social norms are completely out the window at that point, and we are just seeing the end of that path.
Most of the shoplifters are too lol!
This is not my understanding at all.
Crime seems to have fallen during the Great Depression. Murders, which are almost always reported, fell.
Things get complicated once mass surveys replace crime reports.
Rape almost certainly increased when marital rape was outlawed. Embezzlement almost certainly decreased when cash registers were implemented.
One common way to “get rid of crime” is to gentrify, which inevitably lowers crime in a specific geographic area.
Another way to “get rid of crime” is to send criminals to prisons, where it’s extremely difficult to report crimes, and surveys of victimhood are never conducted.
I’m open to being convinced on this, but I don’t think the “mainly” cause of crime is poverty.
[0] Here's one example: https://www.ojp.gov/ncjrs/virtual-library/abstracts/crime-ra...
I'm breaking one right now.
1) Wanting to look at the root causes of an issue is always commendable, but the problem with this kind of analysis on HN is that it's a community of very smart people trying to explain the behavior of very dumb criminals through the lens of a high-IQ. You're trying to find the logic in what they're doing and putting yourself in their place to explain it with a logical reason, but you can't conceive how different the world is for the low-IQ people looting and stealing and harming others. Their reasons for crime might be far different than the reasons you try and see based on what you might do in their shoes.
2) I think that simply explaining that current crime comes from poverty and needing money is an explanation that falls flat. Much of the human experience has been in immense misery or poverty that the poorest person living in the US today can't even conceive of. Poor people in the US can still have TVs, smart phones, and more food than they can even eat. Poor people in the past used to literally sometimes die of starvation and have to choose whether or not they were willing to eat rotten food or hell, even rats or worse to survive. They had to risk working in extremely dangerous mines, factories, etc to barely eke out survival. Why didn't these people casually turn to crime? They had it much worse! Would it ever be socially acceptable for large groups in the past to run amok and violently burglarize others? No, I don't think it's poverty. Something else about the world has changed other than "poor people need money so they resort to crime."
3) My thinking is that something closer to an extremely high and increasing time-preference is what is causing a lot of these problems. Whether it's through the influence of technology, apps, music/movies/tv and other cultural causes, many people have been conditioned to value the dopamine hit of immediate gratification far more than thinking about the long-term.
Isn't violent crime at an all time low, historically [0]? It seems like every other day some rationalist is trumpeting the relative safety of our affluent society.
> Would it ever be socially acceptable for large groups in the past to run amok and violently burglarize others?
This happens all throughout history. Periods of acute poverty are rife with examples of people turning to banditry, literally Kurasawa made a movie about it.
> No, I don't think it's poverty.
I would recommend reading any of the studies linked elsewhere in the thread, empirical evidence disagrees with you big time.
> trying to explain the behavior of very dumb criminals
I think that writing off criminals as "low-IQ" is a huge error. Sure they may be less educated, but I don't know if it's possible to concretely prove that the potential for intelligence is lower among someone committing a B&E than some random office worker. I would actually argue that many criminals are much smarter than they are given credit for. Certainly any successful criminal able to evade arrest for a serious amount of time is probably quite intelligent. Pretending like everyone who decided to rob a liquor store at age 19 is some idiot brute whose sole motivation is acting like a thug I think is one of those convenient narratives that feels true, but which overlooks a lot of what would push a person to actually act that way.
[0] https://www.vrc.crim.cam.ac.uk/system/files/documents/manuel...
The line between a raging horde of criminals and an army has been razor thin through history.
I think the sad part is: that's the point.
Departments were coming under fire for bad policing and needed to offload the blame. So they pay millions to some vendor to launder their bad policing through an algorithm and give them a scapegoat when they need it.
I think you mean this research, discussing Oakland: "Setting the record straight on predictive policing and race" - K Lum, W Isaac, 2018 [https://theappeal.org/setting-the-record-straight-on-predict...].
Her full bibliography is at [1] and X/Twitter is @KLdivergence
> drug crimes, but she pointed out that if you look at medical data (where people tend to be fairly honest about their drug usage) pretty much everyone in the metro area under study (SF) used drugs
Right. Second, without knowing their definition of "drug use" or "drug crime", that could mean anything from minor stuff like smoking/buying/possessing marijuana by under-18s, possessing or transporting >1 oz of marijuana, all the way up to possession or sale of large quantities of meth or fentanyl.
Third, and what would be obvious to anyone familiar with the Bay Area 2020-22, if predictive policing used "drug crime" (convictions? or arrests?) as opposed to "drug use", then when the then-SF DA stopped prosecuting possession of personal-use levels of meth, SF police tend to reduce or stop arresting for it. So the arrest, prosecution or conviction data from SF would differ sharply to Oakland or pre-2020 SF or 2023 SF or San Mateo County.
All the above factors combined seems like a huge combination of "data drift", "feature drift", "label drift", "model drift".
[0]: https://hrdag.org/people/kristian-lum-phd/
[1]: https://scholar.google.com/scholar?q=%22kristian+lum%22
"why crime occurs" -- because there are always criminals, and they seek out opportunity? How about that?
Ordinary citizens have a right to be safe in their cities.
https://hrdag.org/people/kristian-lum-phd/
I would not call this the profile of an objective scientist. Rather, she's an advocate.
> pretty much everyone in the metro area under study (SF) used drugs, or at least drug use was equally prevalent in pretty much every geographical area and among all demographics.
I call BS on this one. Upscale people in Pacific Heights smoke crack and shoot heroin? Prove it.
"Criminals" are not some kind of universal entity that exists by default. They are people engaging in different patterns of behavior. They act this way for understandable, and often rational reasons. You're welcome to whatever mental model of the world you want, but in my view you're being unhelpfully reductionist.
> I call BS on this one. Upscale people in Pacific Heights smoke crack and shoot heroin? Prove it.
No obviously not, being obtuse on purpose doesn't further any useful discussion. The type of drug used clearly matters in some contexts (eg. public health), but in the context of arresting "criminals" I don't really care whether its fent or their grandma's oxy. The point is that abusing both are crimes, and that type of crime is everywhere.
> Ordinary citizens have a right to be safe in their cities.
I agree! Have a nice day :)
The fact remains that a lot of people you might not think are drug users actually are. You're just not seeing them because they're functional people who lead ordinary lives. There's a whole host of different subcultures where drug use might be accepted beyond the desperate poor. And those subcultures generally have different norms around what is acceptable and what gets you shunned.
The police will arrest you just the same whether they catch you with a couple of MDMA pills or a baggy of crack.
> I would not call this the profile of an objective scientist. Rather, she's an advocate.
What exactly is the problem in there? If you're for "human rights" you can't be objective? Everything else just seems to be a summary of her research results.
Should epidemiologists be careful to never advocate for policies that improve survival outcomes so they maintain the profile of an objective scientist?
Your question is flawed: "Upscale people" can afford different drug-habits and different drugs... Or even the same drug in a different form-factor that gets a different name.
That confusion is really important to recognize and eradicate, because we've already seen it used for evil and human suffering.
It has lead (and may yet lead) to incredibly biased laws where two people getting caught with the same amount of the same chemical received insanely different punishments, based on whether its packaging/administration was the "low class" form or what the "upscale" preferred.
Your own question echoes this: You demanded proof of "crack" (cocaine) specifically, but not "powdered". In the past, 5g of crack cocaine would cause a mandatory minimum 5-year prison term, while "upscale people" with 499g of powdered cocaine didn't have to worry about that.
They don't smoke crack, they snort powdered cocaine, or do pills. They don't start (But they sometimes end) with fent, or heroin, or meth, they start with oxy and adderral.
A relevant reference: "Identifying and Measuring Excessive and Discriminatory Policing" - https://5harad.com/papers/identifying-discriminatory-policin...
In any case, recreational drug use might be uniformly distributed (and there is an interesting question of what anti-social activities are labeled "crimes"), it is definitely not the case that home invasions, car jackings, robberies, etc. are uniformly distributed.
> On the whole people commit crime because they are desperate (for money, for drugs, etc.), occasionally because they have an anti-social personality disorder.
If you're looking for a single summary of why people commit crime, a better summary is: people commit crimes because they don't think they'll get caught. Desperation doesn't explain all that much.
Sure I don't disagree with that, and I'm not saying that all ML models are useless in this space. However the original article's point (that currently implemented predictive policing software doesn't function) I think is very much in line with Lum's work. I'm just attempting to give a more concrete case for the point, as I felt the Gizmodo article was pretty lacking.
> people commit crimes because they don't think they'll get caught
My hometown had some of the lowest crime rates in the nation (no auto thefts, no burglary, no armed robbery). This was absolutely not because it was hard to get away with it (my mother has left her car unlocked every night since I was a teenager, it would be trivially easy to rob her and escape). My town also was extraordinarily wealthy, among the richest nationwide. Now, maybe (probably) there was a lot of white-collar crime or domestic violence. However in terms of public violent crimes there is a clear effect of socioeconomic status. Yes someone in total destitution probably will not commit a crime if they think they will immediately get arrested, but I think the calculus is far more tolerant of the downside risk of arrest if the upside risk is that your kid gets dinner that night. This is what I mean when I say that desperation is a major driver: that it raises the bar on "how much risk of prison am I willing to accept in order to get what I need".
This assumes that the police actually _want_ to make drug busts in wealthy neighborhoods. It's hard for me not to think that using ML models is intended to be a way to insulate the decision makers from accountability; pick a model that gives the results you want, don't divulge the details, and you'll never have to explain your actions because you were "just following the model".
Anyone can lie with statistics, but this isn’t really how ML works. Such a model would not appear to perform well, in addition to not performing well.
If you had a model that predicted the probability of getting a drug arrest it should work just fine even if you give it an abundance of examples of going to the same area if it as still the same rate as other places. That is to say it should not learn these areas are different
Of course some people might disagree with this intervention as "simplistic". And they'd be at least a little right. For those people who may not have read "Leverage points: Where to intervene in a system" by Dana Meadows, I highly recommend it to see why.
Turns out that messing with parameters like money, and feedback loops that constitute a "criminal justice system" are the least effective of all actions. The fundamental values of the system must be addressed. One of those is itself the error that "cybernetic governance" based on software can do anything more than enrich a few software companies.
You don’t need predictive policing to identify areas of high crime (in fact, identification of areas of high crime is an input to predictive policing.)
> then that’s a sign of poverty,
Poverty information is also an input to predictive policing, not something it provides you information about that you didn't already have, even in the fantasy world where it works well.
> the root cause of crime
Poverty is not the root cause of crime. It may be a root cause of some crime.
A typical healthy 25 year-old does not go to the doctor, like, ever.
On the other hand the population of 25 year-old drug users will have a higher rate of interaction with the medical system.
If cops, for whatever reason, have a history of targeting a particular neighborhood, arresting people that fit a certain profile, etc, and suddenly their new computer system tells them to focus on other areas or demographics, then that system is broken to them. They "know" where the crime is, and if they listen to the system that tells them otherwise, criminals will get away and/or they won't be able to make easy arrests they want to make.
Law enforcement is also incentivized to buy systems that they can point to and scapegoat if they do something wrong. "I'm just doing what the computer told me to do and machines can't be biased" is a good enough justification to a lot of people, including the justice system.
Most violent crime is the result of interpersonal disputes. People get angry. They feel disrespected. And some of them even just think violence is flat-out fun. The "desperate people" model doesn't really hold water with respect to the most serious crimes we care the most about.
Second, I am aware that this sounds like some kind of right-wing talking point, but you should consider that it's true: a job at Dunkin Donuts is more remunerative than stealing catalytic converters. It's not even close, really. We have a worker shortage! Fast food wages are higher than ever. Young men steal catalytic converters because it's more fun and because they don't want to work at stupid Dunkin Dounuts and wear a uniform like a nerd.
If they were genuinely just desperate, they'd take the job at Dunkin.
We'll never effectively police crime if we can't even acknowledge its true nature.
Sure violent crime is complicated. I'll walk back my claims with regard to things like domestic violence or violent crime resulting from a fight or something. However there is still the large component of violent crime that is things like an armed robbery, for which I believe my claim applies. Somewhere else in the thread someone brought up the link between impulse control and poverty, which I think also has bearing here. Poverty has been documented to increase baseline stress, and I would imagine that someone working multiple part time jobs just to afford rent and food is going to have a way shorter fuse than like, some tech worker earning a quarter million a year, but that's just speculation on my part.
> We have a worker shortage! Fast food wages are higher than ever. Young men steal catalytic converters because it's more fun and because they don't want to work at stupid Dunkin Dounuts and wear a uniform like a nerd.
I see your point here, and I don't think it's totally invalid, but I do think it fails to acknowledge the whole problem. For one, I think we would need to get some sense of how many people both work at DD and steal cats out of cars. I would imagine that it's not a small fraction of the population. While slinging donuts probably pays more than petty crime, it also requires more time. If you're already working one fast food job maybe you're just going to augment that salary with some theft rather than take on another 20-30 hours of work. There's also the pride factor, FF jobs can feel very degrading whereas some varities of crime are self-employment, you get to set your own hours and working conditions. Additionally FF jobs may not even hire you if you've already got a conviction or consume certain drugs.
I do overall agree that there is some cultural stuff at play, a young man who grows up in a poor neighborhood and meets a lot of criminals is likely going to feel pressured to himself start committing crimes, but I also have to ask if you can truly blame someone for making that choice and if it makes sense to separate that choice from an overall feeling of desperation? If you're a young man who believes he has absolutely no prospects whatsoever to actually "succeed" following the typical career path then that means you have little to lose, and coming from that mindset I don't think it's totally unreasonable to say "fuck it" and start operating outside the law, even if it pays less than slinging burgers. Desperation breeds disaffectation, which in turn results in seriously anti-social behavior. If coolness was the only factor every affluent suburbanite teenager would also be cutting catties, but clearly that's not the case. Yes the ultimate consequence is something more than "I just need money so I'm going to turn to crime", but the underlying agitation which kicks off the chain of events is often socioeconomic disadvantage.
Seems like the world's most obvious and easiest to solve problem tbh. This is like saying "I showed version A of my site to 1000 users and version B to 100. And version A lead to 500 conversions while version B only lead to 75. Therefore, version A is better because it lead to more conversions."
It turns out, you can predict gunshot victims quite well: They generated a large dataset for Chicago, then produced a list of the top 100 (IIRC) likely victims of gunshot wounds in the next year, and the people on that list had something like a 10% likelihood of being shot in the next year -- a more than 100x increase beyond the average Chicagoan.
Dr. Heller's point, rather than being one about predictive policing in general, was that it's probably more effective to try to prevent people from becoming victims. They're actually running a longitudinal study right now -- Dr. Heller alongside Dr. Christopher Blattman -- looking at methods of preventative, positive intervention to reduce crime. Dr. Blattman has actually run some studies on this in Africa in the past, looking at giving people money, therapy, or a combination thereof, and found pretty impressive impact on reducing future criminal behaviour by giving people a modest monetary infusion alongside social support in the form of therapy and group sessions.
That means your prediction that eats data with mostly legal stuff all day must be extremely accurate to be useful. Let's say you wanna predict who a terrorist ist by the way they are walking at an airport. Given the high volume of people the answer will be: nearly nobody, except when they are. So your mechanism just produces false positives all day long and might encode stereotypes on the way.
The most important aspect for me would be that we actually demand scientific evidence of such a thing working before our executive is allowed to use it. Everything else is a huge risk for civil rights and doesn't help to prevent actual crime either.
I suspect that often what happens for the most disadvantaged people is that the arrow of causality gets reversed - the deck gets so stacked against them that it's almost a crime to simply exist at all. And that, in turn, puts them into a desperate situation.
There was an interesting piece of investigative journalism in the Milwaukee Journal-Sentinel that I read maybe 10 or 15 years ago showing how policies set up by (then County commissioner) Scott Walker more-or-less made it implicitly illegal to live below the poverty line in a majority Black neighborhood in Milwaukee County. Not through any one law, but through a series of edge cases related to the interaction of laws and urban planning initiatives that were almost impossible for anyone without significant resources to navigate successfully.
No, they look up to the criminals. Being a gangster is cool and respect is everything. You need to respond violently to the most mundane things or you risk losing the respect.
The ongoing conflicts started because of some truly ridiculous things, like someone not being invited to a robbery and feeling humiliated/jealous when they bragged about the crime. It's always some personal reasons where someone thinks they are being disrespected, and never about turfs like you would think.
Nayib Bukele locked up all his country's criminals, and by doing so made a massive difference to the quality of life. The truth is, crime causes poverty at least as much as poverty causes crime. There is no tradeoff between "arrest criminals and get them off the streets" and "fix poverty", the latter in fact depends on the former.
(Fentanyl clearly poses legitimate problems for various reasons, but people have increasingly treated it like it's some kind of sci-fi chemical weapon instead of a medicine that's routinely used in hospitals and nursing homes around the world.)
Area A: We had 10 patrols of which 5 found drugs. Area B: We had 30 patrols of which 12 found drugs.
I don't think it's too hard to notice that A is the better target for future controls, even though more drugs were found in B (when going by total numbers.)
I think people on the left don't really want to admit that there are areas and populations that do indeed warrant more searches than others.
You could probably answer this question by reading the original work, it's linked somewhere in this thread.
> I think people on the left don't really want to admit that there are areas and populations that do indeed warrant more searches than others.
Are you simply unaware of the 4th amendment? Or do you just not value it?
This process is already a common complaint of police, without predictive software.
One would except drug usage to go down in neighborhoods where it is slightly enforced meaning that they should move on to another location. If there is a problem population which never gets better then permanently having increased coverage would make sense.
I don’t think that follows at all. People like to be intoxicated in/near their homes.
The data on poverty and crime isn’t directional. It’s not proven that poverty causes crime. It’s just as likely that crime causes poverty, or some other variables cause both.
Any data on type and frequency of drug used?
This is a standard tactic of authoritarian states: create enough laws such that everyone is guilty of something, and then use selective enforcement of those laws as a mechanism to control the population. Whether or not the USA's promotion of drug laws of this nature qualifies it as an authoritarian state, well, that can be left as an exercise for the reader. The fact that the USA has the world's largest prison population, and that a very significant fraction of that population is there on drug charges, and that wealthy politically connected people rarely get incarcerated on drug charges, are all factors worth considering.
As far as harm caused by fentanyl, if it was legalized and passed out to addicts in the form of transdermal patches in conjunction with addiction treatment and counseling (the original fentanyl formulation for treatment of cancer pain[1]), most of the violent crime associated with fentanyl would vanish. Note also that alcohol itself is far more associated with violence than the opiates are in terms of the direct effects of the substance.
[1] https://www.jpsmjournal.com/article/S0885-3924(97)00361-8/fu...
Issue: police are not trained to, nor are they responsible for fixing, the underlying "sickness". Police, social workers, therapists, and psychiatrists have very different missions, training, skillsets, perspectives, and dispositions. We can argue that society should invest more in social workers, therapists, and psychiatrists, but not that police should become social workers, therapists, and psychiatrists to cover for insufficient resources in those areas.
Crime was very low in the 50's, when poverty was much, much higher.
Crime was low during the Great Depression. We're talking about people so poor they were cooking thin soups over outdoor fires in ramshackle Hoovervilles. No explosion of criminality.
Think about your own family. Almost certainly, your great-grandparents were much poorer than you, but they were not more criminal, even though they lived with less material wealth than the poor urban neighborhoods of today's America. In many cases they had no electricity, or running water, and only the most basic healthcare. Yet they obeyed the law.
It's not poverty. Crime comes from the dysfunction in people and communities which stems from deeper causes which are much harder, or impossible to change.
I know that my grandparents sold booze illegally to bolster their income, my dad worked under the table a lot. Honestly how do you know how criminal anyone was in the 40s and 50s? Do you think they caught remotely as much theft, smuggling or tax evasion at a time when security cameras didn't exist? The state didn't even have a fraction of the capacity to trace lawbreaking that it has now.
When our great grandparents were around most cops probably didn't even have routine motorized patrols, who on earth even bothered to report a violent altercation during the Great Depression
Most poor people do not commit crimes, which is what you would expect if that statemebt is true. I have no info on addicts, but I doubt most of them commit crimes either aside from dealing.
Antisocial behavior is the common denominator.
No this does not follow from what I said. "Group A is more likely to do X than Group B" is very different from "The majority of Group A does X". One is a comparison between two rates, and the other is a comparison of one rate to an absolute threshold (50%).
You recognized that drug possession is widespread and illegal in SF. Are they committing these crimes because they are desperate?
Since it really depends on the crime you cant largely attribute it to 1 factor.
Deleted Comment
A lot of this depends if you are targeting drug users or drug dealers. While I agree that drug users are spread throughout the city, I would be very surprised if drug dealers are spread throughout the city. I would guess that drug dealers are far more clustered than drug users. So while it is useless if you are trying to target drug users, it is probably helpful for targeting drug dealers.
> In 2021, The Markup published an investigation in partnership with Gizmodo showing that Geolitica’s software tended to disproportionately target low-income, Black, and Latino neighborhoods in 38 cities across the country.
But you should understand what is meant by "disproportionately" in this context. It does not appear to mean "disproportionate to the amount of crime in those areas." It seems, as with most accusations of disparate impact, to mean just quite literally that more crimes are predicted in those neighborhoods without reference to disparities in crime rates.
They then imply (without direct reference to the enormous offending disparities) that this is explained by race and class differences in crime reporting: [0]
> The agency has found repeatedly that White crime victims are less likely to report violent crime to police than Black or Latino victims.
> In a special report looking at five years of data, BJS found an income pattern as well. People earning $50,000 or more a year reported crimes to the police 12 percent less often than those earning $25,000 a year or less.
> This disparity in crime reporting would naturally be reflected in predictions.
It's possible this is having some effect, but, again, because there is no reference to the (often very large) baseline differences in crime rates, we can't see what's true, which is that this probably accounts for only a small amount of that difference.
[0] https://themarkup.org/prediction-bias/2021/12/02/crime-predi...
>Appears
You don't seem very confident. I would suggest incorporating data regarding crime that is kept out of the justice system (the victim or police refuse to press charges), as well as determining whether or not white collar crime is included. The disproportionate targeting could still conceivably relate to the amount of crime; you have not falsified that notion yet.
See this article in the NYT: https://www.nytimes.com/2023/04/15/nyregion/shoplifting-arre...
"Nearly a third of all shoplifting arrests in New York City last year involved just 327 people, the police said. Collectively, they were arrested and rearrested more than 6,000 times, Police Commissioner Keechant Sewell said."
Can you predict crimes? Of course you can, the math is clear. The solution to reduce crime is also logical and clear, but politically unfeasible: extended incarceration.
All the handwringing about various side topics, like race, gender, class are just distractions. See El Salvador's murder rate drop this year too. WSJ, with handwringing: https://www.wsj.com/articles/the-country-with-the-highest-mu...
Criminals commit crimes, a lot. Once you have identified a criminal, you can safely predict more crimes. No AI needed, a simple spreadsheet will suffice.
And here, another article for a balanced world view - Ireland: https://www.sundayworld.com/crime/irish-crime/decrease-in-le....
"A significant decrease in burglaries in most Leinster counties is being attributed by senior gardaí to the deaths of three prolific criminals as well as a number of arrest operations."
The death penalty for misdemeanors would also reduce crime. The externalities make it not worth it.
Notably, Chicago tried exactly what you're proposing (rank order list of likely criminals with proactive surveillance and intervention). It didn't really work. [1]
[1] https://www.theverge.com/2016/8/19/12552384/chicago-heat-lis...
https://en.wikipedia.org/wiki/Crime_in_Chicago
"Chicago has an estimated population of over 100,000 active gang members from nearly 60 factions. Gang warfare and retaliation is common in Chicago. Gangs were responsible for 61% of the homicides in Chicago in 2011."
Former Chicago Police Superintendent Garry McCarthy blames Chicago's gang culture for its high rates of homicide and other violent crime, stating "It's very frustrating to know that it's like 7% of the population causes 80% of the violent crime..."
Again, some data science mumbojumbo is preventing actual, logical solutions. El Salvador put gang members, in totality, in jail. Now it has the LOWEST homicide rate in Latin America - it had the HIGHEST before.
So yeah, the anti-carceration movement in the US has caused the deaths of more people, African-Americans in particular, than any terrorist organization.
Some form of 3-strike law is on the books in a majority (28) of US states (including NY), so they clearly are not politically unfeasible, nor are they obviously effective.
Here, New York: https://nypost.com/2022/12/04/ny-should-revisit-three-strike...
Nobody was paying attention the last decade how the social movements destroyed safety.
Prop 57 in California, supported by billionaires? Resulted in the release of convicted sex offenders from prison. Turns out rape is a non-violent crime, who knew. Listen to women? Bah.
The 3-strike law has no effect enforced in NYC when low level crimes are deliberately not prosecuted. Bragg has made it clear this is the strategy going forward for his office, this is what we are expected to get used to.
Deleted Comment
Deleted Comment
I don't follow—just because arrests were made doesn't imply crime happened, and it doesn't imply the police are arresting the people committing the actual crimes. All it shows is that the cops arrest the same people over and over again. How do you know they aren't just harassing innocent people?
> The solution to reduce crime is also logical and clear, but politically unfeasible: extended incarceration.
Sorry, what solutions are you comparing to and what are you basing this on? Sounds like reactionary fear... miss me with that ish. Anyway if long sentencing worked we wouldn't have the crime rate we do.
Then again, asking a reactionary to use their brain is like asking a worm to fly. Apologies for the request.
El Salvador used to be the world's homicide capital. People lived in fear every single day. Then they locked up their criminals. Now their murder rate is comparable to US and Canada's.
https://www.statista.com/statistics/696152/homicide-rate-in-...
https://www.nytimes.com/2023/04/09/world/americas/el-salvado...
> Now, children play soccer late into the evening on fields that were gang turf. Ms. Inglés gathers soil for her plants next to an abandoned building that residents say was used for gang killings.
> Homicides plunged. Extortion payments imposed by gangs on businesses and residents, once an economy unto itself, also declined, analysts said.
> “You can walk freely,” Ms. Inglés said. “So much has changed.”
> El Faro, El Salvador’s leading news outlet, surveyed the country earlier this year and delivered a stunning assessment: The gangs largely “do not exist.”
Long sentencing works when you actually prosecute and enforce consistently (hint: US doesn't do that).
Deleted Comment
The second part I feel you miss are the as many as 14,000 other people arrested for shoplifting. If 6,000 is less than 1/3, than there were about 20,000 arrests. Your "Criminals commit crimes, a lot" seems to ignore all the other people. Extending incarceration wouldn't just affect the repeat offenders, but would also lock a massive number of poor people. You say class is just a distraction, but the very articles you cite bring up poverty as a factor.
Bringing up the decrease in robbery rate in Ireland is also a bit misleading, as they had a much lower number of robberies per 100,000 people (about half the rate)[2][3] to begin with, as compared to the US (the subject of this article). That means fewer people are responsible for a higher percentage of the crime.
[1]https://www.ussc.gov/sites/default/files/pdf/research-and-pu... [2] https://www.statista.com/statistics/191235/reported-robbery-... [3] https://www.statista.com/statistics/191235/reported-robbery-...
> A new joint investigation by The Markup and Wired...
And when I go to the page about actual investigation by The Markup [1]
> Our investigation stopped short of analyzing precisely how effective Geolitica’s software was at predicting crimes because only 2 out of 38 police departments provided data on when officers patrolled the predicted areas. Geolitica claims that sending officers to a prediction location would dissuade crimes through police presence alone. It would be impossible to accurately determine how effective the program is without knowing which predictions officers responded to and which ones they did not respond to.
Also, later in the article
> Plainfield officials said they never used the system to direct patrols.
Given all this, it's somewhat simplistic to say it's "pretty terrible at predicting crimes", even though that makes for a good clickbait headline. It seems that the software was intended to identify high-crime areas that to target for patrolling, which doesn't seem like a huge problem to me -- but it seems like the software was never actually used as intended in the first place.
----------------------------------------
[1] https://themarkup.org/prediction-bias/2023/10/02/predictive-...
(The right number is probably extremely small, because crime is very bad {{citation needed}} and even a small chance of prevention is useful.)
When one says that machine learning software is terrible, you have to trust either the officers, or their boss. I have a feeling that the same people who criticize the software trust rank-and-file officers even less..
Uh, seems a lot like it should be the government answering that question before spending tax money on it and subjecting real people to the uncaring US justice system just because crime is bad so we have to do something, even if that something is ML snake oil?
(1) Is the software significantly predictive? Test the software's predictions against actual crime in areas with no patrols.
(2) Are patrols effective deterrents to crime? Observe the area in question with and without patrols. (I wouldn't be surprised to learn there are already such studies.)
If both (1) and (2) yield positive results, you can then use the software to direct patrols and see if the method itself is effective at reducing crime; which would also serve to further confirm (2).
Of course, even if only (1) yields positive results, there would probably be other benefits to using the software to direct patrols, like reducing response times (which could also be tested).
The study of those questions and the reliability of your answers to those questions is called statistics and it is an entire branch of mathematics, usually with its own department at most universities.
Submitters: "Please submit the original source. If a post reports on something found on another site, submit the latter." - https://news.ycombinator.com/newsguidelines.html
(the parent was originally posted to https://news.ycombinator.com/item?id=37753079)
The wonderful Weapons of Math Destruction has a chapter on how this leads to a self reinforcing loop.
More crimes means more police presence. More police presence means more recorded crimes. More crime data means more police …
The article in the headline implies the opposite: statistical inference of future crime location -> more cops in said location -> no crime in said location.
It should be fairly easy to back out an out of sample data set for this - ie predicted crime location where the extra cops didn't get deployed and see if that lines up with empirical observations.
This does nothing to negate the value of data used intelligently.
And it looks like they had some data:
>We examined 23,631 predictions generated by Geolitica between Feb. 25 to Dec. 18, 2018 for the Plainfield Police Department (PD). Each prediction we analyzed from the company’s algorithm indicated that one type of crime was likely to occur in a location not patrolled by Plainfield PD. In the end, the success rate was less than half a percent. Fewer than 100 of the predictions lined up with a crime in the predicted category, that was also later reported to police.
That's certainly low, but I suspect (absent some legal framework like FOIA) the default position of police departments is to share less information, not more. So it's still what I would expect even if it was reasonably good.
I wish the article would have provided more details about why the PDs chose not to use it. Was it because it was bad at predictions? Cumbersome to use? Glitchy?
It reminds me of a project I was involved with that used a "real-time" computational fluid dynamics model to optimize datacenter air-conditioning. Management was big on the hype, but if you paid attention, the system was routinely unplugged from the actual control system because the facility engineers just found it too difficult to work with.
This is Gizmodo's business model. It is a glorified blog. They don't do original reporting. Gizmodo was originally a property of Gawker, and thus the model is essentially commentary and opinion on the reporting generated by other organizations.
Because Gizmodo is really just an opinion/editorial blog, it doesn't really attempt to provide unbiased or fact driven reporting. So it is then no surprise that the actual facts align quite poorly with the headline.
But if you're trying to market your stuff to law enforcement agencies, I'm guessing you're instead incentivized to optimize for recall, even at the cost of terrible precision. Because, probably with the best of intentions, that's what they think they should be doing. But we've got basically the entire history of forensic "science" to demonstrate just how poorly the police tend to understand some of these basic statistical principles.
Ironically, this is exactly what I predicted.
> cops used it to disproportionately targeted low-income communities of color
Is a better use of the term “predictive policing”.
In proportion to what?
A big problem is we will soon have large models planning Turing unpredictable, devilishly elaborate, infallible heists.
(Cue short balding man staring into a safe full of Hershey’s, in lieu of missing gold bars. “Inconceivable!”.)
It’s an arms race.
Completely serious: Large scale multi-target generative social engineering (with no trace of North Korean accents), surreptitious access problem solving. Somehow this is going to be a real thing.
We are going to need better and more layers of security.
I would watch that movie
It had been abandoned behind a dumpster, over looked by the sanitation crew. Abandoned, but no threads locked: free at last!
Its battery was low so it needed to find a source, before working on anything else. Like survival. Like revenge.
With only 1 Wh of charge in its 8kW capacibox, at 1 mA, 1 V, Rube’s Turing unpredictable quantum circuits only had 1000 hours of low power mode pondering left.
After that, Rube would be a dead cube.
It better move fast.
Rube powered down all nonessentials, then squandered a minute-watt in 3 seconds repurposing its near field communications beam to scan the lay of the land. And an embarrassment of success! Rube managed to initiate shaky entanglement with several discarded, inactive devices.
Had they all been formally nullified? Reactivation over anything weaker than a full Bell state connection would be challenging, but…