site banner
PaperclipPerfector  3mo ago (text post)   38744 thread views

Culture War Roundup for the week of July 15, 2024

This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.

Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.

We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:

  • Shaming.

  • Attempting to 'build consensus' or enforce ideological conformity.

  • Making sweeping generalizations to vilify a group you dislike.

  • Recruiting for a cause.

  • Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.

In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:

  • Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.

  • Be as precise and charitable as you can. Don't paraphrase unflatteringly.

  • Don't imply that someone said something they did not say, even if you think it follows from what they said.

  • Write like everyone is reading and you want them to be included in the discussion.

On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.

9
Top Bottom Old Controversial Comments
Jump in the discussion.

No email address required.

On the topic of a small number of relatively 'known' people being involved in an outsized portion of the crime problem

or

AI is sometimes allowed to say things that are otherwise not allowed to be said, so long as they make sure to say that it's definitely not racist

Machine Learning Can Predict Shooting Victimization Well Enough to Help Prevent It is the name of the paper. They took arrest/victimization records in Chicago and tried to predict who was likely to be shot in the next 18 months. 644,000 people in the data; of the top 500 with the highest predicted risk, almost 13% were shot. That's the top line. 13% accuracy might not seem like much, but they claim that the rate is 128 times higher than that of the average Chicagoan. For context, that's 64 shooting victims over an 18mo period. I don't know what the total 'shot but maybe didn't die' rate is, but Chicago has in the ballpark of 600 homicides (by all means) each year.

This is not about who did the shooting; it's about who was shot. The implicit argument is that most shooting victims are close enough to the criminal world. Even if they were just purely victims before, it is at their doorstep. Plausibly, if a little old lady just happens to live in a really terrible neighborhood and had to report being the victim of various prior crimes, this could indicate that she is also at risk of getting shot, too.

They definitely go out of their way to say that, yes, Black males are more likely to have prior data in the system, but that the system still predicts with similar accuracy across demographics.

I don't know how practical these sorts of things will be to actually use for any purpose, but this paper dropping is definitely adding some fuel for the folks who think that a variety of criminal problems are mostly concentrated within a relatively small subset that could, at least in theory, be somewhat identified.

Didn’t we have this entire conversation two years ago already? Am I going crazy?

In spite of book like 'Weapons of Math Destruction', books like Criminal (In)Justice still get published. The former presents plausible arguments for algorithmic bias, the other presents data about who commits crime and where. Black crime has been a supposedly awkward talking point since the 1970's. Jesse Jackson's comments marked a turning point in political honesty, but Sowell was happily publishing Black Rednecks and White Liberals not long after (which I found quite convincing). All of this stuff can and has been said. It's not some secret knowledge.

You (and others) have said this, that there's nothing new, everyone knows how this works, etc. Matt Yglesias didn't seem to know that. It's "wild" to him. Perhaps it is likewise "wild" to many others.

Just posting "despite..." in the right context is a meme. Yglesias said "This precrime paper is kind of wild". Thinking this implies he had no idea about the demographics of crime is kind of wild. Google trends seems to indicate that since 2004 "black on black crime" is about a common a trend as the highly secret sport of "pingpong". Just searching "black crime", it's about a common as searching for "Ethiopian food". (Random aside: I like spicy indian food, and Ethiopian food is like a cousin, which I also like. They frequently offer a spiced raw beef dish (kitfo). Veggies are good, sometimes too oily. All dishes pair well with beer).

Are you under the impression that all the algorithm determining likely victims does is look at race?

No. If it helps clarify things, I'm under the impression that looking at race might be the most important factor, perhaps tied with zip code.

I don't think that any one factor delivers an actionable level of accuracy. Given that actionable is the term under debate, really. The point of the data analysis, which I've seen done before without the mystical AI reference, is that it's actually a tiny sliver of poor Black men who are likely to be involved.

I don't think that people are shocked to find out about Black crime. Most PMC white libs actually vastly OVER estimate the frequency of crime among Black populations, they think everything is their little fantasies from trap music and the wire. What's shocking is that we can achieve a degree of discrimination where we CAN exclude the vast majority of blacks who won't be involved in a crime, and rather those that will.

You might life Manguels Criminal Injustice which talks about, in part, the degree of discrimination we can achieve. One block in NYC might have a 10x difference in violent crime rate, stable for 50 years. The geography of crime is wild. I recall some reporting circa 9/11 about the violent death rate in Compton being higher than in Iraq or Baghdad during times of crisis (vague memories here) Even contemporary Chicago is a fairly safe city if you never go to the 10% of areas which accounts for 80% of the crime. It's good general knowledge to have.

Thinking this implies he had no idea about the demographics of crime is kind of wild.

Let's walk through this, then. What do you think his tweet does imply?

Its wild that an algorithm can predict crime before it happens.

...and why would one say that it is wild?

Because it a SciFi concept come to life

Now I'm just very confused. I thought this was all just old hat, been done before, obviously out there for anyone who cares to see. Now it's a SciFi concept come to life. I have no idea anymore...

More comments

Machine learning isn't a necessary ingredient here! Cops will tell you that they know very well which individuals have the highest risk (of both being the offender and victim), but they can't do anything about it until after the fact. It's not difficult to notice who keeps going through the system.

Frick, accidentally clicked a link in the sidebar and lost a ton of interesting stuff I had picked out. The highlights, less eloquently put:

  • Super interesting article and worth a read, thanks! Appendix is also very interesting. They discuss quite a few common questions in detail.

  • Very interesting setup where they consider victims instead of perpetrators. From a numbers standpoint, it's much more plausible and practical to help victims, plus fewer ethical issues, because few people would say no (presumably) to some kind of intervention to prevent being shot. I like this framing. And presumably you can't target alleged perpetrators without a bad tradeoff: you get ethical issues if you target too narrowly, or you get too many potential perpetrators that it's not practical to do anything substantial, and still might have ethical issues anyways.

  • The framing of preventing victims is more than just PR though. It's methodologically nicer too. For those who might not know, predictive policing has some issues beyond just pure ethics, though there's that too. The classic example is when I think it was NYC who tried it, they would feed recent crime data back into the algorithm that determined where police should patrol. Might sound like a good idea, but you get a feedback loop! Turns out police "generate" crime to some extent because they naturally uncover crimes on patrol, so if you assign a cop to a neighborhood, yup, they discover more crime, and the algo would send even more cops there, and they'd discover even more crime... etc. You get the idea.

  • They say it will pay for itself. Two ways of analyzing this. You have direct costs measured (some easy like medical bills, some more hard like productivity decline, social badness, etc.) and then also broader societal willingness to pay (WTP). I looked through at the book and paper referenced the latter, which I think is the source of their claim that halving shooting risk for the top 500 would be worth $67 million. Those WTP numbers came I think from a n= 1200ish survey where people were asked if they'd support an extra 50/100/200 bucks a year for a program that reduced gun violence 30%. I'm not totally sure those two questions are closely enough related that the connection is justified. Though it does raise the question -- if we could, and the model suggests we might, prevent over 100 shootings in Chicago of a specific group of people, how much would that be worth?

  • I agree their accuracy (well, to be specific, it's precision, since they use the ML algo to finger a subset of people of varying sizes, they try out 500 and 3300 in various spots, and then count how many of those fingered high-risk people actually did get shot) is actually pretty good and plausibly useful in practice.

  • Note that because of the above, you might notice that their fundamental set-up is a "threshold rule", where the ML algo spits out either "yes" or "no" to the question of if someone will be shot (with "yes" being "high-risk" in practical interpretation). The ML algo as analyzed was not set up to give any other kind of analysis or figure, all of the numbers are relative to this classification paradigm. So 500 as above is the top 500 in risk. When you tell it you want the top 3300 choices, it's less precise (10ish%) but still pretty good. Be Basically a cousin to logistic regression, if that sounds familiar to anyone, which is related to traditional OLS but not quite (IYKYK)

  • The risk of being shot again if you've already been shot (which makes sense) is about as precise in its risk as the model is. In other words, if the ML says you're high risk, even if you haven't been shot before, it's as if you had been. Which is interesting and could be a wake-up call to someone ID'd as high risk.

  • Young Black men are correctly identified as the highest risk group, and with slightly better accuracy than other groups, though the model still does pretty well for most races overall.

  • Obviously if you prevent the algo from considering things it does worse. Withholding demographics hurts, as do own arrests, though the precision drops to only about 10% which is still not so bad. Nice to have that as an ethical option. Dropping felony arrests doesn't hurt at all! Dropping someone's own personal arrest history puts it at 11%. (All top 500 but even out a bit as you increase the number of people you want warned).

  • What kind of predictors did they use? Notably, network effects (did your name appear along with another person in a police record, are you family, etc). But also, demographics (age, gender, race where availaible), fine-grained arrest data (what type, severity, what beat, how long ago), victimization data (were you the victim of various kinds of crime, fine-grained), gang affiliation indicators. Age actually mattered a lot, and race was never a top predictor, but it sounds like that was partially because the algo to some extent figured out someone's race due to other things (including the network effects). Being arrested yourself sadly was the biggest predictor overall (still a correlation of like .1). Someone you appeared in a record with also being shot and gang affiliations rounded out the top predictors.

  • They only used people who appeared in the police database at least twice for data quality reasons. How many people did this approach mean they missed? About a third of the victims weren't even in the sample at all. If you finger the top ~3300, you still miss 90% of the victims overall but 10% isn't bad.

  • No covid issues! They did their cross-validation time-wise so that they could keep their network effects.

Wouldn't you expect a negative effect on crime with police patrolling an area, with any feedback-loop being quite short lived? Else it would be a damning fact for patrolling, showing that it doesn't work. Do you have NY report at hand I would love to read it.

I know what book I read it in but it’s packed I will have to dig it out if you’re still interested - this appears to be an okay overview of a few of the issues and causality concerns in the meantime. The name PredPol sounds familiar?

The core idea is that: no, not necessarily. It heavily depends on the behavior of the police force in question. While some police forces might deter crime by their presence, other forces might by their policies generate crime (reports) by their proactivity in making stops, and yet other forces might lie about their crime data in order to make themselves look better. After all, what actually makes it to a computer-usable report in what is fundamentally a profession which emphasizes discretion is quite variable.

This feels like one of those "Academic spends countless dollars manhours and nuerons figuring out what any rando with a room-temperature IQ could have told you for free" situations.

Normal people dont need machine learning to tell them which house in their nieghborhood is "the drug house", or that that the Williams' ner-do-well son is likely to come to a bad end based on the company he's been keeping.

Your gut feeling isn't wrong, but it isn't quantitative. I appreciate actually checking the proportions on this obviously true thing.

Well, I don’t know. A 1 in 8 chance of getting shot is actually really high, and in just a year and a half? Like sure street knowledge is cheap and directionally accurate, but judging a “more likely than others to be shot” is a lot easier than “is among the absolute top candidates to be shot”, right?

I think if you pick the right subcultures of active/adjacent criminals you'd reach that kind of number. Certain rap scenes feel like that'd almost be an understatement of the shooting risk.

But this would require skill and insider information and subjective analysis. Having a deterministic, mechanical process with known inputs that can process this data goes a long way towards preventing someone from corruptly picking and choosing which places they count as "high risk" and which they don't. And lowering variance, since some individuals are going to be better at doing this sort of subjective guesswork than others, while the AI can have its performance actually tested.

And goes a long way towards laundering this in the public perception. Even if everyone "knows" that this group is high risk, having an AI with testable metrics say so is probably going to be easier to sell (in the long run) than having human beings say so.

Not really- gangs work like armies in microcosm. Theres plenty of people who do things like logistical support (ie transporting narcotics, guns, and fellow gang members) and administrative work (even drug dealers need some sort of accountants, and babysitting offspring is a non-trivial challenge), and a relatively small number of infrantry/grunts/bangers who are on the "front lines". It may not be obvious to outsiders, but gang members and associates certainly know who is most likely to be involved in violence.

I think you’re way overstating the organization of street level gangs. Cartels probably operate this way, but your median street level gang is just some teenaged boys with a semi-charismatic leader, and none of them are willing to specialize outside of frontline operations or leadership.

Now there’s certainly more sophisticated gangs who have accountants, employee benefits coordinators, internal tribunals, money laundering specialists, specialized smugglers and straw purchase administrators, etc. I don’t think that’s what this AI was applied too.

Depends on where you draw the brackets. As you say, 1 in 8 is already really high, and i'm pretty sure you could get close to that over the couse of a lifetime (if not the next year and a half) simply by looking for people who hang out in sketchy places with sketchy people.

To build on my previous example, everyone in the nieghborhood knows that the middle Williams boy has always been a bit "exictable" / prone to violent outbursts and now he's started a beef with the local drug dealer. What are the odds he ends up getting shot, stabbed, or arrested in the next 18 months? 1 in 8 seems positively generous in that context.

I think that was more true in the old days of organized crime, but these days some of the bigger U.S. cities (well at least Chicago, Baltimore, and Philadelphia) have mostly disorganized crime where it's mostly small scale block gangs beefing with each other over instagram posts. Actual drug infrastructure exists but isn't fighting over territory as much and is avoiding doing dumb shit like rapping about pissing on the corpse of somebody's best friend. With the social media data and complete lack of op sec these guys have it's pretty easy to tell who is going to get killed.

Even today i would expect the violent crime (or at least the lethal variety thereof) to follow a similar pattern. Murders generally dont happen "out of the blue" and when they do happen (see Asian ladies getting pushed in front of trains) they get treated as news-worthy precisely because they are a deviation from the norm.

To paraphrase Heath Ledger's Joker, nobody freeks out when a gangbanger gets shot.

Do you really need machine learning or AI to do this analysis? I feel like if you gave me arrest records in an excel spreadsheet with data I could 65% of this just eyeballing and 95% if I did some regressions in excel.

My guess is you already assumed my point, and your main point is maybe outsourcing this to AI gets you thru the politics some how.

I believe the old joke is, "Machine learning is just OLS with constructed regressors."

No you don't, but you gotta get those buzzwords in if you want to get the clicks.

Saying that you used "machine learning" is so much cooler and "truthy" than making a list of everyone under 30 who's been riding with the Hell's Angels or hanging out with Crips.

Under thirty seems old. More like 'not old enough to drink'.

And my impression was that biker gangs were a bit lower risk, even as they often engaged in illegal activities they preferred not to do things the stupid way.

You're probably right.

Everyone here or like 98% knows this, but the average reader it’s probably 3-5% knows this fact.

My money would be on the reverse.

My suspicion is that most users here don't know how to spot a drug house or potential ambush where as 98% of regular folk (or at least those that have had some experience with the seedier elements of society) do, and thats why this result is being treated as novel.

Specifically meant the math bit. That this doesn’t require AI but simpler stuff.

You are probably right that society realizes a hells angel guy is more likely to die than the average citizen. I do think this place has their eyes wide open and could compete with the average guy but typical blue tribe might not.

I may have been being a little uncharitable there.

Identifying potential victims fits more easily into progressive ideology than identifying (inconvenient) perpetrators. The people victimized by crime are not, primarily, white people or the well-to-do. It's poor people and black people. If you can more effectively frame policing reform as "we need policies to protect vulnerable population from victimization from criminals," it's both harder to argue against and true. People may be willing to overlook a black man murdering an elderly Asian woman for ideological reasons, but it's much harder to take the position "we need to allow criminals to murder black people with impunity in order to protect black people."

The question is how identifying previctims benefits them. Are you going to attach a secret service detail to them?

How about this: “According to our algorithms, you have at least a 13% chance of being shot if you stay in this city. Perhaps it’s time for a new start. Join our New Leaf program today and we’ll relocate you to a new metro area on the other side of the country, and find you a basic job and lodging.”

I know, I know, I’m being hopelessly naive and idealistic. But maybe there’s a tiny fraction for whom this kind of algorithmic warning might serve as a life-changing trigger.

Unironically not even a terrible idea. See for example the studies about giving people free money. Presumably, this won't help someone deeply involved in for example a gang, but it probably would help a lot of other youths stuck in a bad economic/friend situation. Give them a housing stipend, for example, to relocate, and it could be pretty potent. At least in theory, you could even fund this with property taxes and come out ahead, since shootings do decrease property values at least a little. Or if the state ends up shouldering the medical bills, you might come out ahead too by giving them a portion of that money directly, if it actually works as an intervention. IDK how exactly the math would shake out.

I think the problem is, of course, that these shooting victims are likely people who make the places they go worse.

The elderly Asian woman getting murdered wouldn’t be overlooked. The race of the perpetrator would.

And, like, obviously, the headline from this is ‘black men more likely to be murdered by x%’. We already know this, obviously- even for unsolved homicides we have the race of the victim.

I don’t think this is any kind of breakthrough on criminal justice politics.

What is this place?

This website is a place for people who want to move past shady thinking and test their ideas in a court of people who don't all share the same biases. Our goal is to optimize for light, not heat; this is a group effort, and all commentators are asked to do their part.

The weekly Culture War threads host the most controversial topics and are the most visible aspect of The Motte. However, many other topics are appropriate here. We encourage people to post anything related to science, politics, or philosophy; if in doubt, post!

Check out The Vault for an archive of old quality posts. You are encouraged to crosspost these elsewhere.


Why are you called The Motte?

A motte is a stone keep on a raised earthwork common in early medieval fortifications. More pertinently, it's an element in a rhetorical move called a "Motte-and-Bailey", originally identified by philosopher Nicholas Shackel. It describes the tendency in discourse for people to move from a controversial but high value claim to a defensible but less exciting one upon any resistance to the former. He likens this to the medieval fortification, where a desirable land (the bailey) is abandoned when in danger for the more easily defended motte. In Shackel's words, "The Motte represents the defensible but undesired propositions to which one retreats when hard pressed."

On The Motte, always attempt to remain inside your defensible territory, even if you are not being pressed.


New post guidelines

If you're posting something that isn't related to the culture war, we encourage you to post a thread for it. A submission statement is highly appreciated, but isn't necessary for text posts or links to largely-text posts such as blogs or news articles; if we're unsure of the value of your post, we might remove it until you add a submission statement. A submission statement is required for non-text sources (videos, podcasts, images).

Culture war posts go in the culture war thread; all links must either include a submission statement or significant commentary. Bare links without those will be removed.

If in doubt, please post it!


Rules


Recommended Posts And Communities

Recommended Realtime Chats