Another post starting with "I liked when Scott punctured liberal orthodoxies, but now that I notice he's also puncturing conservative orthodoxies, I'm having to re-evaluate him".
But they're coming from the same place! The liberal with the "lived experience" that police are commiting genocide against black people (even though statistics say such killings are rare and around expected base rates), and the conservative with the "lived experience" that their friend's nephew got autism at around the same time as getting a vaccine (even though studies say there's no link) are coming from the same place. And I accuse this article of coming from that place too.
I know you think you're smarter than that because you use Greek words and appeal to the James Scott tradition of the fallibility of metrics. You would like to think that *you* are wisely balancing the value of data with a well-honed understanding of its potential failure modes, and *I* am blindly subscribing to a Church Of Graphs.
But sorry, I think I'm the one doing the wise balancing. Yes, I posted a graph, but then I immediately talked about the many ways the graph could be wrong or biased, and discussed each of them and whether they're true. Then I talked about ways why people might have vibes that disagree with the data, and came up with theories about those, and test them. In the end, I end up concluding that some types of crime and disorder are up over certain time periods, and other types are down over the same time period, in ways that I hope don't do much violence either to the consensus interpretation of the data or to most people's lived experiences. Then I wrote another post about the conflict between graphs and lived experience and why I think it's important to look at the graphs even when the lived experience also has value (https://www.astralcodexten.com/p/malicious-streetlight-effects-vs) . I wrote five thousand words over three posts entirely devoted to trying to rebut the accusations you're levying here. All of the counterarguments you're bringing up, I examined as honestly as I could and weighed against other things.
I don't think you read the majority of this, because you still make the same counterarguments that my posts were meant to analyze, as if they're totally new things I never considered. For example, you bring up the "shoplifting reports from one Target" thing as proof that I might be missing something, but I discussed it in the post in a section starting "Could stores be failing to report to police" (CTRL-F "there was an embarrassing incident . . . where two stores briefly changed their reporting policy and nearly doubled the total report number). Then I talk about the alternate collateral source I'm using to try to avoid that problem.
Or: you cite a four year trend in motor vehicle theft that looks pretty bad. But when you zoom out to see the years before and after your selection, you find the overall trend looks like https://www.statista.com/statistics/191216/reported-motor-vehicle-theft-rate-in-the-us-since-1990/ , which is exactly the same as the overall property crime trend that my post analyzed. You just took a four year period that looked especially bad out of context from the middle of my graph, covered up the periods before and after, then tried to use it as a rebuttal of the graph. I'm sure it confirmed your lived experience, but I still think this is a Graph Crime, if not a Graph Church Mortal Sin.
Whatever else you think I'm missing because of my benighted graphophilia, I suspect if you read the post you'll find I considered that too. I promise I'm not just looking at graphs and ignoring possible objections. Instead, I'm looking at the same complex picture you are, but trying to actually follow it where it leads, instead of saying "gotcha" and stopping at the first piece of evidence that supports my own worldview.
Even though I agree lived experience has value, I still think writing posts like these is important, for the reasons I pointed out in If It's Worth Your Time To Lie, It's Worth My Time To Correct It ( https://www.astralcodexten.com/p/if-its-worth-your-time-to-lie-its ). Here's what I wrote about this question in my post:
>>>> "Many people complained that by talking about crime yesterday, I was distracting from the rise in disorder. Probably people will complain today that by talking about littering and graffiti and so on, I’m distracting from some other kind of disorder which is definitely increasing - maybe open-air drug markets, or tent cities, or the boom boxes. That’s fine. But as I said when arguing with you in the comments, I think the following two statements are importantly different:
1. Littering, graffiti, and most violent and property crimes are down, but tent encampments and boom box playing are up. Shoplifting is stable nationally, but that could hide local variation. As some areas gentrify and others worsen, there are shifts in who experiences these problems, and the well-off highly-literate white people who set the national conversation are getting more exposed to them.
2. Crime and disorder are rampant, nobody feels safe anymore, cities are falling apart and the police don’t care, the West has fallen.
My goal isn’t to deny anyone’s lived experience, nor to discount the importance of solving these problems (I support the death penalty for boom box carriers). It’s to push back against a sort of Revolt Of The Public-esque sense that everything is worse than it’s ever been before and society is collapsing and maybe we should take the authoritarian bargain to stop it. On an emotional level, I feel this too - I can’t go downtown without feeling it (one of many reasons I rarely go to SF). But I don’t like feeling omnipresent despair at the impending collapse of everything. Having specific thoughts like house prices are up since the pandemic, so it’s no surprise that there are more homeless people, and more of the usual bad things downstream of homeless people, rather than vague ones like “R.I.P. civilization, 4000 BC - 2026 AD isn’t just more grounded in the evidence. It’s also more compatible with living a normal life. I’m not a pragmatist who thinks you should be allowed to lie or do a biased survey of the evidence in order to live a normal life and escape despair. But I’m also not some kind of weird anti-pragmatist who makes a virtue out of ignoring evidence in order to keep despairing." <<<<<
You have a good point about the car theft graph. I'll add an update.
Now obviously you have no access to my thoughts beyond what I've written here so you'll have to take me on my word that a) I do appreciate your takedowns of conservative bad ideas as well (the ivermectin article was good), and b) I did in fact read all 5,000 words you wrote on this topic. My choice to not include more of your own responses and caveats and impressions from your three pieces was complicated and ultimately very subjective. In the first place I don't think it would have served the goals of my piece, which is already quite long and already quotes you at length. Frankly it would have made it boring. And in the second place, including and re-rebutting your arguments started to feel like this was an attack piece on you which, believe it or not, was not my goal. Yes I'm using you as an example of the kind of thinking I'm criticizing, but it's a widespread phenomenon in my intellectual circle, not something unique to you or to this specific object-level issue. I didn't want to treat you as a punching bag, both because the point I'm trying to drive home isn't really about you (really!) and because of the love I bear you. Specifically, an earlier draft included your quote about finding your own downtown repulsive, but I cut it because it felt like beating up on you without adding anything substantial. I think I gave you a fair shake overall, but given your objection here I'll concede I should have made more of an effort to emphasize that you yourself already wrote about many of these objections, particularly the Target store fiasco. Mea culpa.
Just as obviously, I disagree that this essay is unsuccessful as a response or fails to add something new to the discussion. You talk about being engaged in a balancing act between trusting your eyes and the figures, and claim you're doing a better job of that balancing. My issue is that ultimately even a person perfectly well balanced must look left and right and then choose a side with which to cast their lot. You've chosen to favor the status quo of lax enforcement even though you acknowledge it comes at personal cost to you, and you're using the official figures to justify this policy leaning. If I am misreading your position here and you're in favor of stepping up enforcement and sentencing for the kinds of quality of life crimes I discuss here, I welcome that correction. As for me, I consider myself doing the same balancing act that you are, but I'm casting my lot with what you would probably characterize as authoritarian or draconian enforcement of public order, for the reasons I wrote about at length. This essay is about trusting your own eyes, but you could also say it's about advocating for your own interests. You might think it's the right decision to give up parts of your city for the duration of your kids' childhood, but I don't. Ultimately this is where we differ. Yes, you are performing a balancing act, but to my eyes the written concessions you make to people whose own quality of life has degraded in spite of the graphs' insistence are empty. Without a commitment to do something about it, it's just lip service. What does it matter if you say that you hear and believe the lived experience of people being impacted by crime when you ultimately conclude that sorry, we should do nothing because things aren't actually that bad and are trending in the right direction on a generational timeline according to official figures? My (biased) read is that this is what is animating the remarkable pushback you've gotten on these articles.
One of my readers helpfully shared an article from Compact magazine with related criticisms:
I don't think it describes your own stance very well, but I think it's related, maybe as a cousin or so. What it has in common is a reluctance to take crime seriously for various reasons (I believe yours are substantially different from the ones being ascribed to Democrats in the article), and using the narrative of overall declining as a rhetorical buttress for those policy preferences. I think that's bad, regardless of how successfully you are balanced.
> As for me, I consider myself doing the same balancing act that you are, but I'm casting my lot with what you would probably characterize as authoritarian or draconian enforcement of public order, for the reasons I wrote about at length.
It's still important to calibrate on how much being draconian will actually help reduce crime, and to try to figure which time periods were actually lowest crime to try to get the best policy. If you're overly focused on how much things suck today, you'll probably be overly focused on undoing recent changes, instead of figuring out what policies across many decades are best practices.
Is there actually statistical evidence that national vibes about crime are unusually bad? Or is that view based on vibes/anecdotes about what people think about crime?
This is a sincere question, not a gotcha. I'm persuaded that serious crime is not going up. But I'm not persuaded that people as a whole are any more wrong about crime rates now than they were 10, 20, 30 years ago.
> I do appreciate your takedowns of conservative bad ideas as well (the ivermectin article was good)
Choosing the one article whose thesis was "actually, the conservatives following anecdotal evidence *were* right that ivermectin worked, they were just mistaken as to the circumstances under which it did" is perhaps not the best choice for making your point here. It's kind of supporting the idea that you only like Scott when he's validating conservative intuition.
Just an interjection that should you consider it again, a smart dude wrote a 17 post series on Substack that pretty convincingly refutes Scott’s premises on ivermectin (Scott’s rebuttal in return is weak) And it’s an influential series too, Dr Jay Battycharya endorsed it.
This is the level of snark that has always kept me from respecting Scott Alexander despite the quality of the commentariat he's managed to curate. Particularly the opening paragraphs.
Calling it out is what got me banned from his comments, even though I did so politely.
It is true that Alexander ADDRESSED a number of critiques of his position (though not all of them). It is not true that he successfully rebutted every critique he addressed.
Locale is everything with crime statistics. Well, not everything, because first you need to understand the always present flaws, biases, and limitations of the statistic in question. One person can rightly claim that crime and disorder are no big deal in their life, while another 6 miles away can rightly claim that crime and disorder are worse than they have ever been. Broad averages are for suckers and propagandists. On top of all of this is the fact that things are supposed to be getting better. If we look at life in NYC tenements in the late 1800s, wow, crime is down and life today is good! How does this way of thinking relate to rhetoric around the rights of minorities? You won’t get far with our savior class by pointing out how great things are for blacks or gays today.
One of my many beefs with Progressives is their double standard when it comes to the expectation they have for their neighborhood and public order vs their willingness to abandon the working poor and the just plain poor to conditions that they, the typical well-off Progressive, would never tolerate. I believe that our expectations for public order should not change one bit from zip code to zip code. And yes, that means a lot more police activity in certain areas.
Went off topic a bit. My take is that Kitten’s point is largely correct. Mr. Alexander may have covered a lot of ground in his essays, but I believe he is missing the mark.
There’s one thing I don’t understand: why do Scott and Kitten call a policy that allows dangerous beings to roam near people’s homes “liberal”?
The reason I ask is that lately I’ve been seeing — almost daily — complaints from people in Putin’s Russia, specifically from the Far East. The government there, in a very authoritarian way, has effectively established two rules: (1) tigers must not be harmed, and (2) civilians are not allowed to own firearms.
The result is that the quality of life in those remote Far East communities has dropped to a level that would be hard for an American to even imagine. Tigers mostly kill dogs, but there are also attacks on people. Children walk to school along the same streets where tigers have killed pets. People can’t have guns. And even the police are not allowed to kill the tigers.
Occasionally, after repeated complaints, authorities will capture a tiger and relocate it farther away from populated areas. But the animals often come back.
For example: https://t.me/r_chp/38914 — a tiger in a village, the aftermath. The channel admin writes: “I barely post this kind of thing anymore, even though videos like this show up every day. I’m worried the audience will get desensitized.”
So explain this to me: why is it considered a “liberal” position to prevent a sheriff or a judge from dealing with dangerous beings in whatever way the local residents find appropriate?
To me, this looks like a textbook example of an authoritarian bargain — instead of the local community making the decisions, it’s Scott. Or, in Russia’s case, Putin.
It didn’t occur to me that, in addition to taking care of any potential homeless, the tigers would help clear homes. Isn’t the West heading for a similar situation with wolves, coyotes and bears?
Do you ever test your data, or look for anomalies using standard tools? I only ask because crime data can be tricky. I often find actuaries can be extremely useful in both understanding data sets, and testing them to see how much you can trust them, and see what biases may be present, and how to try to work with them. I know people tend to look down on business professionals, especially in unsexy industries such as insurance, but they are extremely good at analyzing data and knowing the limitation of different data sets. It is literally their job, and if they get it wrong, it comes out of their bonuses, or worse, they lose their jobs.
This is simply a suggestion. I live on the south side of Chicago, and although some crime stats "feel right," such as murder rates and car thefts, many "quality of life"-related crime stats are obviously under-counting. It is a b^tch to extract a good approximation of the real data in this sort of thing, but not impossible.
I mean no harm, and i am not criticizing you, but drawing conclusions from summary statistics can be dangerous, even if it is usually a great starting point (or a reasonable ending point if the data is good).
Scott, graphs influence my perception of crime far less than my motorcycle being stolen this last year, and all the times thieves have ripped open and ripped off the packages on my doorstep
When Biden/London Breed were in office my motorcycles were safe, but my car window got smashed
I’d say that I’ve been personally harmed by crime more since Trump got reelected (and Lurie was elected) but at least the persistent riots we had when Biden was president have stopped (which inclines me to believe that the further Left hate Democrats more than they hate Republicans), but now other cities have them.
Sure, I’m old enough (57) to remember when crime was worse than now, but I’m also old enough to remember when there was much less crime and disorder.
You do realize the secondary costs of crime are indeed still increasing as people just go more out of their way to avoid crime. And even if it is falling, people should frankly be far far more mad that the state not only does little to curb the effect of such few prolific criminals, but they actually facilitate their crime
>and I am blindly subscribing to a Church Of Graphs
Well at least you DO understand why not everybody is a Rationalist, good :) And the need for balance. Formerly you have argued even bad numbers are better than no numbers, but that was 10 years ago, and the kind of wisdom aging brings generally makes one less of a Church of Graphs type. I think bad numbers are not useful, because they give a false sense of confidence. Numbers feel truthy, they feel like something is actually being measured, while when you see just words, you understand it is just an opinion. But basically if you checked the numbers and they seem okay, then that is good balancing.
Also, these are the kinds of things numbers are actually useful for, well-defined targets well-measured.
But for example when people use numbers to argue Finland is happier than Italy, lol.
Your goal absolutely is to deny people's lived experience and you should be proud of that! Rationality is exactly saying to people: I know that it feels to you as if X is true but let's check and see because your experience might be misleading.
Sure, there is a trivial sense in which you aren't saying to anyone: that thing which happened to you didn't happen to you. But literally no one thinks they are denying that true events are true events. Nor is it true that you aren't insisting that anyone didn't experience what they claim to have experienced -- if you say otherwise I can go make a dummy account and claim to have been running a secret perfect RCT trial for the past 50 years on this subject that shows murders are actually up 1000%.
I'm not so autistic as to not understand the gesture. It's a way of saying: hey I'm trying to respectfully suggest your experience might be a bit misleading not to lower your social status by accusing you of doing something in bad faith. But I don't think it's a great way to do that since it is such a confusing idea.
Some people just want to take the “authoritarian bargain” for whatever reason, and get mad when you deconstruct their rationalizations. Many such cases!
I did not grow up in rationalist circles, and honestly some of that was intentional as it seemed like the adult version of that annoying kid in middle/high school who would "erm actually" vocally, rather than keep his thoughts to himself like a socially well adjusted individual who was smart enough to have recognized the teacher might not always be right.
But as part of it, I never really took the time to understand why *they* call themselves "rationalists". I assume (i still dont know for sure and dont care to) it is because they believe they are rational individuals who use logic and epistemic integrity to derive conclusions about the world.
But me personally, I always find it funny because, in layman's terms, if you say someone is "rationalizing something", it means theyre deluding themselves into affirming their preconceived notions by introducing selective evidence. And I think that is far more accurate for "rationalists" based on my observations.
From the philosophical view, they're the empiricists; the rationalists are the other guys, the continental philosophers.
I tend to have more suspicion of the socially well-adjusted people (they're much more likely to be selling me a bill of goods), but as we've seen above the Internet rationalists can make their own mistakes.
"adult version of that annoying kid in middle/high school who would 'erm actually' vocally, rather than keep his thoughts to himself like a socially well adjusted individual" hit too close to home?
Oh absolutely! I thought that was kind of the point; I'm giving the other side. Rationalists have come up with all kinds of silly ideas, many of which Kitten says himself.
But it doesnt take knowing the orthodoxy or history of whatever your movement is to accurately criticize and assess your inability to reason and successfully conclude on the truth of issues.
> Lol. Okay bro. Yes, your movement "Yeah, I'll stick with Scott here" above.
As in I’ll take his side of the disagreement that the whole article is about?
Did you think I meant I’ll stick with him through thick and thin, and agree with him on everything he says and be a part of any group he identifies as? Lmao think better bro
> If you wonder why your worldview has no purchase broadly, which I am sure you dont wonder
I don’t wonder that, because it’s borderline nonsensical. What would it mean for my worldview to “have purchase”? My worldview is composed of a wide variety of values and beliefs, many of which are very widely or near unanimously held. You have already admitted (and demonstrated) that you don’t have any familiarity with the set of beliefs you are attributing to me, let alone my actual worldview, and you have no idea how much “purchase” any or all of them have.
You’re just saying stuff, presumably annoyed by my calling you out for your uninformed criticisms of an intellectual movement you clearly don’t have enough knowledge to speak about. What you take as personal defensiveness is just a general dislike of confidently asserted nonsense.
I am finished replying to you. My comments herein are sufficient to stand on their own for a rational observer to read and comprehend my positions and rebut and "pre-but" your initial and later responses. And, for what it is worth, at this point I actually think your responses are very funny.
In my own Gnostic interpretation of things: I moved away from the USA 13 years ago. I visit my hometown roughly once a year. It is really truly true that I used to walk from my parents' house to the park with my childhood friends, and there was not litter, tents, and crazy people shouting at me. It is really truly true that the Walgreens did not have all the toothpaste and conditioner behind locked plexiglass three years ago, but does now. One of the things about just visiting a town you're really familiar with every year, but just once a year, is that the snapshots really highlight changes that might elide if you were there every day, experiencing a more gradual change.
But what do I know. I'm a Catholic, not a Graphist.
You write that the Church of Graphs privileges doxa over episteme—I disagree. To me, you and Scott Alexander are not so different. You are both conducting epistemic analyses. The difference is that you use gnosis and doxa in your analyses whereas adherents to the Church of Graphs are apparently forbidden from using gnosis at all.
Maybe the Church of Graphs not only forbids gnosis—it actively attacks it. The institutions of science, medicine, law, governance, and philosophy, the entire name of the game is to prove that your very eyes are lying to you. You don’t get a tenure-track position by concluding that crime does, in fact, hurt economic development of poor areas—you get that cushy job by producing a seemingly-well-reasoned epistemic analysis that concludes that crime is actually good for economic development. Or that poisonous pesticides actually produce better health outcomes in comparison to countries that don't use them. Or that crime is at all time lows. Or that the U.S. actually didn’t play a big role in WWII. Or any number of ridiculous things.
I suppose I don’t understand why some intelligent people seem to go down this path. It’s like they shun their ability to come to their own conclusions about the world. Very odd.
If it so pleases the comment section, I would like to present a case study on the difference between episteme and doxa. This one comes from the slightly less topical case of Somalis in Minnesota defrauding the state.
In 2018, an investigator employed by the state’s childcare program wrote a ten-page email alerting state officials of the extensive fraud he was seeing in the subsidized childcare program. Based on the size of the overall program, a cursory analysis of grant recipients similar to those he had successfully investigated, and the nature of the fraud, the investigator estimated that the true extent of the fraud reached $100M. He admits that this is a WAG (wild-ass guess) because the program requires insufficient documentation to thoroughly audit the recipients and his office lacked the manpower to investigate the daycares they suspected of defrauding the system.
The Minnesota Legislature’s Office of the Auditor wrote a report on the email. It’s relatively short, at about 30 pages. The conclusion? Far short of $100M, the fraud in the program is really only about $4M and the investigator has no support for his claim of two orders of magnitude higher. However, reading past the headline will give you the exact opposite impression, especially when one reads the summaries of the cases prosecuted.
The $4M figure comes from tallying up the amounts prosecuted in a court of law across about a dozen or so cases. The evidence in these prosecutions comes from videotaping the entrances to the daycares for a week or two and then comparing the headcount to the handwritten sign-in sheets the daycare provides for reimbursements. In these cases, investigators would find an average of 56 children show up daily vs. claimed figures of over 100, all of which were reimbursed. And so the daycare operator would be prosecuted for the $150k that the state could prove during the recorded two-weeks.
Did the daycare only defraud the state during the two weeks investigators happened to be watching? Of course not! The $150k prosecuted in that case is a fraction of what the daycare has stolen. $150k is all that the state could prove beyond a reasonable doubt in a court of law. The legislator’s auditor used this strict, formal definition to arrive at and debunk the investigator’s informal estimate. But the layperson will almost certainly agree that the investigator’s estimate is much closer to the truth than the auditor’s.
It was Scott Alexander's review of _Seeing Like A State_ that introduced me to the concept of _legibility_. It seems dreadfully ironic that he is falling into such a clear legibility trap here.
I've had this same experience when reading Lyman Stone. Great writer, but an unabashed graph-head.
My cousin is a sociologist with a specialty on how survey questions and reporting mechanisms impact study outputs. Even in seemingly politically unbiased scenarios, there are all sorts of weird confounding factors that can give you an inaccurate picture. Like you said, they are useful, but not holy writ.
Excellent essay. My childhood Church shuttered due to dwindling attendance. The local diocese is now letting the city use the land for "single occupancy social housing." The result of this charity is dozens of drug addicts and criminals now live across the street from a middle school, the only local skate park, the only local football field, and the baseball field where I played little league.
We aren't just experiencing the blast radius from crime, it is being imposed on us by church and state.
I've been reading Scott since 2015, and the last few months have seen a dramatic decline in the quality of his writing. Having kids slowed his output, which I understood, but he's also clearly developed some pretty bad AI Psychosis (he's even joked about it in some of his recent posts). Now he's regularly citing ChatGPT itself as a source for complex analyses and data.
I wonder how much of the crime post stuff was just him getting sucked into a little bubble by spending (in his words) 30 minutes per day chatting with LLMs. This sort of thing was always his failure mode, but possibly it used to be mitigated by the fact that he had to contact actual humans (and could trust that research he found online, if potentially horribly biased and full of methodological flaws, at least originated in a human brain).
I did not enjoy it. You have basically one clear paragraph towards the end when you say
"What the blank understands that the techno-optimists miss is that our current LLM approach has inherent, structural limitations. These models don't have any actual understanding of the world—they're sophisticated pattern-matching engines trained on internet text. Expecting them to suddenly develop genuine reasoning abilities by making them bigger is like expecting a horse to eventually evolve into a helicopter if you just breed it to be fast enough."
I agree with that, but your writing doesn't add any new information about the subject; in fact, the only citations I can find are the graph about airline cruising speed and the list of previous AI hype bubbles; besides those two, which amount to about a paragraph of actual text, the post consists of you calling people "retarded" and various forms of ugly, with occasional pot shots at "climate 'science'". Even the very last sentence is just another way to say "Scott's ugly and I don't agree with him".
In conclusion, I'm plenty capable of generating hateful drivel on my own. The reason I've read Scott's writing for over a decade now is that, even when I don't agree with him on a particular subject, I've generally come away from his posts having learned something new and concrete about the world. To the extent that his AI psychosis has compromised this in the last few months, the solution is not rambling, unsourced, mean-spirited screeds about how autistic everyone involved is.
Well i believe in biofoundationalism. Therefore, If someone is ugly and dysgenic it will effect the way they view they world and is something I take into consideration when evaluating their opinion. Furthermore, I never claimed to be making unique or original points. These observations about AI have been made by many people before me, that makes AI 2027 even more idiotic because It shows how badly Scott and Daniel fell victim to the Dunning-Kruger effect.
I was a victim of crime in SF. I was visiting abd foolishly left my wetsuit and roadtrip compilation cd's and some other stuff in the car. Overnite it was broken into and the gear stolen. I was told there was no point in reporting it. That my recouse was to look for it at flea markets
If the national statistics show a decrease in crime, but one’s own lived experience shows an increase in crime, then either your life is a large outlier, or there’s something wrong with the statistics.
I can’t really compare as I don’t know what public order was like 30 years ago. In NYC, depictions in media and movies suggest it was much worse in the past? Entire areas of the city you wouldn’t go to for fear of being attacked. There are still places like that today, but quite a lot of the city is safe, albeit with a low hum of insane bums bumping around.
Good article. I can't read new Scott Alexander articles any more, after "Bounded Distrust" (and every time he makes a splash on X I feel no need to change this policy).
Part of what made SlateStarCodex so formative for me is that it constantly probed The Church of Graphs to see whether it was *really* reliable . He did great reporting on the reproducibility crisis! "The Control Group is Out of Control" is a classic! Even on ACX he tore apart ivermectin studies, so it's not like he's a blind believer in any peer-reviewed graph.
Really, Mr. Alexander has his preferred conclusions, like everyone does, and he quietly bends his reasoning to produce them, like most people do. It's just disappointing because he used to write so well about precisely these kind of foibles even among rational, scientific types.
Another post starting with "I liked when Scott punctured liberal orthodoxies, but now that I notice he's also puncturing conservative orthodoxies, I'm having to re-evaluate him".
But they're coming from the same place! The liberal with the "lived experience" that police are commiting genocide against black people (even though statistics say such killings are rare and around expected base rates), and the conservative with the "lived experience" that their friend's nephew got autism at around the same time as getting a vaccine (even though studies say there's no link) are coming from the same place. And I accuse this article of coming from that place too.
I know you think you're smarter than that because you use Greek words and appeal to the James Scott tradition of the fallibility of metrics. You would like to think that *you* are wisely balancing the value of data with a well-honed understanding of its potential failure modes, and *I* am blindly subscribing to a Church Of Graphs.
But sorry, I think I'm the one doing the wise balancing. Yes, I posted a graph, but then I immediately talked about the many ways the graph could be wrong or biased, and discussed each of them and whether they're true. Then I talked about ways why people might have vibes that disagree with the data, and came up with theories about those, and test them. In the end, I end up concluding that some types of crime and disorder are up over certain time periods, and other types are down over the same time period, in ways that I hope don't do much violence either to the consensus interpretation of the data or to most people's lived experiences. Then I wrote another post about the conflict between graphs and lived experience and why I think it's important to look at the graphs even when the lived experience also has value (https://www.astralcodexten.com/p/malicious-streetlight-effects-vs) . I wrote five thousand words over three posts entirely devoted to trying to rebut the accusations you're levying here. All of the counterarguments you're bringing up, I examined as honestly as I could and weighed against other things.
I don't think you read the majority of this, because you still make the same counterarguments that my posts were meant to analyze, as if they're totally new things I never considered. For example, you bring up the "shoplifting reports from one Target" thing as proof that I might be missing something, but I discussed it in the post in a section starting "Could stores be failing to report to police" (CTRL-F "there was an embarrassing incident . . . where two stores briefly changed their reporting policy and nearly doubled the total report number). Then I talk about the alternate collateral source I'm using to try to avoid that problem.
Or: you cite a four year trend in motor vehicle theft that looks pretty bad. But when you zoom out to see the years before and after your selection, you find the overall trend looks like https://www.statista.com/statistics/191216/reported-motor-vehicle-theft-rate-in-the-us-since-1990/ , which is exactly the same as the overall property crime trend that my post analyzed. You just took a four year period that looked especially bad out of context from the middle of my graph, covered up the periods before and after, then tried to use it as a rebuttal of the graph. I'm sure it confirmed your lived experience, but I still think this is a Graph Crime, if not a Graph Church Mortal Sin.
Whatever else you think I'm missing because of my benighted graphophilia, I suspect if you read the post you'll find I considered that too. I promise I'm not just looking at graphs and ignoring possible objections. Instead, I'm looking at the same complex picture you are, but trying to actually follow it where it leads, instead of saying "gotcha" and stopping at the first piece of evidence that supports my own worldview.
Even though I agree lived experience has value, I still think writing posts like these is important, for the reasons I pointed out in If It's Worth Your Time To Lie, It's Worth My Time To Correct It ( https://www.astralcodexten.com/p/if-its-worth-your-time-to-lie-its ). Here's what I wrote about this question in my post:
>>>> "Many people complained that by talking about crime yesterday, I was distracting from the rise in disorder. Probably people will complain today that by talking about littering and graffiti and so on, I’m distracting from some other kind of disorder which is definitely increasing - maybe open-air drug markets, or tent cities, or the boom boxes. That’s fine. But as I said when arguing with you in the comments, I think the following two statements are importantly different:
1. Littering, graffiti, and most violent and property crimes are down, but tent encampments and boom box playing are up. Shoplifting is stable nationally, but that could hide local variation. As some areas gentrify and others worsen, there are shifts in who experiences these problems, and the well-off highly-literate white people who set the national conversation are getting more exposed to them.
2. Crime and disorder are rampant, nobody feels safe anymore, cities are falling apart and the police don’t care, the West has fallen.
My goal isn’t to deny anyone’s lived experience, nor to discount the importance of solving these problems (I support the death penalty for boom box carriers). It’s to push back against a sort of Revolt Of The Public-esque sense that everything is worse than it’s ever been before and society is collapsing and maybe we should take the authoritarian bargain to stop it. On an emotional level, I feel this too - I can’t go downtown without feeling it (one of many reasons I rarely go to SF). But I don’t like feeling omnipresent despair at the impending collapse of everything. Having specific thoughts like house prices are up since the pandemic, so it’s no surprise that there are more homeless people, and more of the usual bad things downstream of homeless people, rather than vague ones like “R.I.P. civilization, 4000 BC - 2026 AD isn’t just more grounded in the evidence. It’s also more compatible with living a normal life. I’m not a pragmatist who thinks you should be allowed to lie or do a biased survey of the evidence in order to live a normal life and escape despair. But I’m also not some kind of weird anti-pragmatist who makes a virtue out of ignoring evidence in order to keep despairing." <<<<<
I don't see your post as having any successful response to this kind of thing, just caricaturing me to accuse me of being part of a "church". As well accuse me of being part of the Church Of Wearing Jackets (see https://slatestarcodex.com/2013/06/13/arguments-from-my-opponent-believes-something/ )
You have a good point about the car theft graph. I'll add an update.
Now obviously you have no access to my thoughts beyond what I've written here so you'll have to take me on my word that a) I do appreciate your takedowns of conservative bad ideas as well (the ivermectin article was good), and b) I did in fact read all 5,000 words you wrote on this topic. My choice to not include more of your own responses and caveats and impressions from your three pieces was complicated and ultimately very subjective. In the first place I don't think it would have served the goals of my piece, which is already quite long and already quotes you at length. Frankly it would have made it boring. And in the second place, including and re-rebutting your arguments started to feel like this was an attack piece on you which, believe it or not, was not my goal. Yes I'm using you as an example of the kind of thinking I'm criticizing, but it's a widespread phenomenon in my intellectual circle, not something unique to you or to this specific object-level issue. I didn't want to treat you as a punching bag, both because the point I'm trying to drive home isn't really about you (really!) and because of the love I bear you. Specifically, an earlier draft included your quote about finding your own downtown repulsive, but I cut it because it felt like beating up on you without adding anything substantial. I think I gave you a fair shake overall, but given your objection here I'll concede I should have made more of an effort to emphasize that you yourself already wrote about many of these objections, particularly the Target store fiasco. Mea culpa.
Just as obviously, I disagree that this essay is unsuccessful as a response or fails to add something new to the discussion. You talk about being engaged in a balancing act between trusting your eyes and the figures, and claim you're doing a better job of that balancing. My issue is that ultimately even a person perfectly well balanced must look left and right and then choose a side with which to cast their lot. You've chosen to favor the status quo of lax enforcement even though you acknowledge it comes at personal cost to you, and you're using the official figures to justify this policy leaning. If I am misreading your position here and you're in favor of stepping up enforcement and sentencing for the kinds of quality of life crimes I discuss here, I welcome that correction. As for me, I consider myself doing the same balancing act that you are, but I'm casting my lot with what you would probably characterize as authoritarian or draconian enforcement of public order, for the reasons I wrote about at length. This essay is about trusting your own eyes, but you could also say it's about advocating for your own interests. You might think it's the right decision to give up parts of your city for the duration of your kids' childhood, but I don't. Ultimately this is where we differ. Yes, you are performing a balancing act, but to my eyes the written concessions you make to people whose own quality of life has degraded in spite of the graphs' insistence are empty. Without a commitment to do something about it, it's just lip service. What does it matter if you say that you hear and believe the lived experience of people being impacted by crime when you ultimately conclude that sorry, we should do nothing because things aren't actually that bad and are trending in the right direction on a generational timeline according to official figures? My (biased) read is that this is what is animating the remarkable pushback you've gotten on these articles.
One of my readers helpfully shared an article from Compact magazine with related criticisms:
https://www.compactmag.com/article/democrats-are-kidding-themselves-on-crime/
I don't think it describes your own stance very well, but I think it's related, maybe as a cousin or so. What it has in common is a reluctance to take crime seriously for various reasons (I believe yours are substantially different from the ones being ascribed to Democrats in the article), and using the narrative of overall declining as a rhetorical buttress for those policy preferences. I think that's bad, regardless of how successfully you are balanced.
> As for me, I consider myself doing the same balancing act that you are, but I'm casting my lot with what you would probably characterize as authoritarian or draconian enforcement of public order, for the reasons I wrote about at length.
It's still important to calibrate on how much being draconian will actually help reduce crime, and to try to figure which time periods were actually lowest crime to try to get the best policy. If you're overly focused on how much things suck today, you'll probably be overly focused on undoing recent changes, instead of figuring out what policies across many decades are best practices.
Is there actually statistical evidence that national vibes about crime are unusually bad? Or is that view based on vibes/anecdotes about what people think about crime?
This is a sincere question, not a gotcha. I'm persuaded that serious crime is not going up. But I'm not persuaded that people as a whole are any more wrong about crime rates now than they were 10, 20, 30 years ago.
> I do appreciate your takedowns of conservative bad ideas as well (the ivermectin article was good)
Choosing the one article whose thesis was "actually, the conservatives following anecdotal evidence *were* right that ivermectin worked, they were just mistaken as to the circumstances under which it did" is perhaps not the best choice for making your point here. It's kind of supporting the idea that you only like Scott when he's validating conservative intuition.
Just an interjection that should you consider it again, a smart dude wrote a 17 post series on Substack that pretty convincingly refutes Scott’s premises on ivermectin (Scott’s rebuttal in return is weak) And it’s an influential series too, Dr Jay Battycharya endorsed it.
https://doyourownresearch.substack.com/p/ivermectin-much-less-than-you-needed
This is the level of snark that has always kept me from respecting Scott Alexander despite the quality of the commentariat he's managed to curate. Particularly the opening paragraphs.
Calling it out is what got me banned from his comments, even though I did so politely.
It is true that Alexander ADDRESSED a number of critiques of his position (though not all of them). It is not true that he successfully rebutted every critique he addressed.
Scott is responding in the same tone as kitten's essay.
Locale is everything with crime statistics. Well, not everything, because first you need to understand the always present flaws, biases, and limitations of the statistic in question. One person can rightly claim that crime and disorder are no big deal in their life, while another 6 miles away can rightly claim that crime and disorder are worse than they have ever been. Broad averages are for suckers and propagandists. On top of all of this is the fact that things are supposed to be getting better. If we look at life in NYC tenements in the late 1800s, wow, crime is down and life today is good! How does this way of thinking relate to rhetoric around the rights of minorities? You won’t get far with our savior class by pointing out how great things are for blacks or gays today.
One of my many beefs with Progressives is their double standard when it comes to the expectation they have for their neighborhood and public order vs their willingness to abandon the working poor and the just plain poor to conditions that they, the typical well-off Progressive, would never tolerate. I believe that our expectations for public order should not change one bit from zip code to zip code. And yes, that means a lot more police activity in certain areas.
Went off topic a bit. My take is that Kitten’s point is largely correct. Mr. Alexander may have covered a lot of ground in his essays, but I believe he is missing the mark.
There’s one thing I don’t understand: why do Scott and Kitten call a policy that allows dangerous beings to roam near people’s homes “liberal”?
The reason I ask is that lately I’ve been seeing — almost daily — complaints from people in Putin’s Russia, specifically from the Far East. The government there, in a very authoritarian way, has effectively established two rules: (1) tigers must not be harmed, and (2) civilians are not allowed to own firearms.
The result is that the quality of life in those remote Far East communities has dropped to a level that would be hard for an American to even imagine. Tigers mostly kill dogs, but there are also attacks on people. Children walk to school along the same streets where tigers have killed pets. People can’t have guns. And even the police are not allowed to kill the tigers.
Occasionally, after repeated complaints, authorities will capture a tiger and relocate it farther away from populated areas. But the animals often come back.
For example: https://t.me/r_chp/38914 — a tiger in a village, the aftermath. The channel admin writes: “I barely post this kind of thing anymore, even though videos like this show up every day. I’m worried the audience will get desensitized.”
So explain this to me: why is it considered a “liberal” position to prevent a sheriff or a judge from dealing with dangerous beings in whatever way the local residents find appropriate?
To me, this looks like a textbook example of an authoritarian bargain — instead of the local community making the decisions, it’s Scott. Or, in Russia’s case, Putin.
Are there homeless people in those places tigers roam free and untouchable?
There are more homes than people in these places, so no.
It didn’t occur to me that, in addition to taking care of any potential homeless, the tigers would help clear homes. Isn’t the West heading for a similar situation with wolves, coyotes and bears?
Do you ever test your data, or look for anomalies using standard tools? I only ask because crime data can be tricky. I often find actuaries can be extremely useful in both understanding data sets, and testing them to see how much you can trust them, and see what biases may be present, and how to try to work with them. I know people tend to look down on business professionals, especially in unsexy industries such as insurance, but they are extremely good at analyzing data and knowing the limitation of different data sets. It is literally their job, and if they get it wrong, it comes out of their bonuses, or worse, they lose their jobs.
This is simply a suggestion. I live on the south side of Chicago, and although some crime stats "feel right," such as murder rates and car thefts, many "quality of life"-related crime stats are obviously under-counting. It is a b^tch to extract a good approximation of the real data in this sort of thing, but not impossible.
I mean no harm, and i am not criticizing you, but drawing conclusions from summary statistics can be dangerous, even if it is usually a great starting point (or a reasonable ending point if the data is good).
Can never ever assume the data is perfect, it’s rarely even good
Scott returns from his ant infested home to settle the score.
Scott, graphs influence my perception of crime far less than my motorcycle being stolen this last year, and all the times thieves have ripped open and ripped off the packages on my doorstep
When Biden/London Breed were in office my motorcycles were safe, but my car window got smashed
I’d say that I’ve been personally harmed by crime more since Trump got reelected (and Lurie was elected) but at least the persistent riots we had when Biden was president have stopped (which inclines me to believe that the further Left hate Democrats more than they hate Republicans), but now other cities have them.
Sure, I’m old enough (57) to remember when crime was worse than now, but I’m also old enough to remember when there was much less crime and disorder.
You do realize the secondary costs of crime are indeed still increasing as people just go more out of their way to avoid crime. And even if it is falling, people should frankly be far far more mad that the state not only does little to curb the effect of such few prolific criminals, but they actually facilitate their crime
>and I am blindly subscribing to a Church Of Graphs
Well at least you DO understand why not everybody is a Rationalist, good :) And the need for balance. Formerly you have argued even bad numbers are better than no numbers, but that was 10 years ago, and the kind of wisdom aging brings generally makes one less of a Church of Graphs type. I think bad numbers are not useful, because they give a false sense of confidence. Numbers feel truthy, they feel like something is actually being measured, while when you see just words, you understand it is just an opinion. But basically if you checked the numbers and they seem okay, then that is good balancing.
Also, these are the kinds of things numbers are actually useful for, well-defined targets well-measured.
But for example when people use numbers to argue Finland is happier than Italy, lol.
Isn’t the boombox guy doing his non-consensual audience a favor by taking their addicted eyes off their wicked screens?
Few have ever written so much while saying so little.
Your goal absolutely is to deny people's lived experience and you should be proud of that! Rationality is exactly saying to people: I know that it feels to you as if X is true but let's check and see because your experience might be misleading.
Sure, there is a trivial sense in which you aren't saying to anyone: that thing which happened to you didn't happen to you. But literally no one thinks they are denying that true events are true events. Nor is it true that you aren't insisting that anyone didn't experience what they claim to have experienced -- if you say otherwise I can go make a dummy account and claim to have been running a secret perfect RCT trial for the past 50 years on this subject that shows murders are actually up 1000%.
I'm not so autistic as to not understand the gesture. It's a way of saying: hey I'm trying to respectfully suggest your experience might be a bit misleading not to lower your social status by accusing you of doing something in bad faith. But I don't think it's a great way to do that since it is such a confusing idea.
Some people just want to take the “authoritarian bargain” for whatever reason, and get mad when you deconstruct their rationalizations. Many such cases!
You can’t or won’t admit that data itself is messy
I did not grow up in rationalist circles, and honestly some of that was intentional as it seemed like the adult version of that annoying kid in middle/high school who would "erm actually" vocally, rather than keep his thoughts to himself like a socially well adjusted individual who was smart enough to have recognized the teacher might not always be right.
But as part of it, I never really took the time to understand why *they* call themselves "rationalists". I assume (i still dont know for sure and dont care to) it is because they believe they are rational individuals who use logic and epistemic integrity to derive conclusions about the world.
But me personally, I always find it funny because, in layman's terms, if you say someone is "rationalizing something", it means theyre deluding themselves into affirming their preconceived notions by introducing selective evidence. And I think that is far more accurate for "rationalists" based on my observations.
It's very simple to understand: we call ourselves the Correctists, because we are the ones who are correct about everything
From the philosophical view, they're the empiricists; the rationalists are the other guys, the continental philosophers.
I tend to have more suspicion of the socially well-adjusted people (they're much more likely to be selling me a bill of goods), but as we've seen above the Internet rationalists can make their own mistakes.
"adult version of that annoying kid in middle/high school who would 'erm actually' vocally, rather than keep his thoughts to himself like a socially well adjusted individual" hit too close to home?
⬇️
Oh absolutely! I thought that was kind of the point; I'm giving the other side. Rationalists have come up with all kinds of silly ideas, many of which Kitten says himself.
> I never really took the time to understand why *they* call themselves "rationalists".
More that they study "rationality" as a topic itself. (For example, the maths behind expected utility maximization)
Yeah, it’s very clear that you’re not familiar.
Thats correct!
But it doesnt take knowing the orthodoxy or history of whatever your movement is to accurately criticize and assess your inability to reason and successfully conclude on the truth of issues.
My movement?
Yes, it does take familiarity to accurately criticise, as your example demonstrates by contrast.
Lol. Okay bro. Yes, your movement "Yeah, I'll stick with Scott here" above.
If you wonder why your worldview has no purchase broadly, which I am sure you dont wonder, this comment section would be relatively reflective!
> Lol. Okay bro. Yes, your movement "Yeah, I'll stick with Scott here" above.
As in I’ll take his side of the disagreement that the whole article is about?
Did you think I meant I’ll stick with him through thick and thin, and agree with him on everything he says and be a part of any group he identifies as? Lmao think better bro
> If you wonder why your worldview has no purchase broadly, which I am sure you dont wonder
I don’t wonder that, because it’s borderline nonsensical. What would it mean for my worldview to “have purchase”? My worldview is composed of a wide variety of values and beliefs, many of which are very widely or near unanimously held. You have already admitted (and demonstrated) that you don’t have any familiarity with the set of beliefs you are attributing to me, let alone my actual worldview, and you have no idea how much “purchase” any or all of them have.
You’re just saying stuff, presumably annoyed by my calling you out for your uninformed criticisms of an intellectual movement you clearly don’t have enough knowledge to speak about. What you take as personal defensiveness is just a general dislike of confidently asserted nonsense.
I am finished replying to you. My comments herein are sufficient to stand on their own for a rational observer to read and comprehend my positions and rebut and "pre-but" your initial and later responses. And, for what it is worth, at this point I actually think your responses are very funny.
Great essay. I really like how you framed this.
In my own Gnostic interpretation of things: I moved away from the USA 13 years ago. I visit my hometown roughly once a year. It is really truly true that I used to walk from my parents' house to the park with my childhood friends, and there was not litter, tents, and crazy people shouting at me. It is really truly true that the Walgreens did not have all the toothpaste and conditioner behind locked plexiglass three years ago, but does now. One of the things about just visiting a town you're really familiar with every year, but just once a year, is that the snapshots really highlight changes that might elide if you were there every day, experiencing a more gradual change.
But what do I know. I'm a Catholic, not a Graphist.
Man, another great piece. Thanks for sharing.
You write that the Church of Graphs privileges doxa over episteme—I disagree. To me, you and Scott Alexander are not so different. You are both conducting epistemic analyses. The difference is that you use gnosis and doxa in your analyses whereas adherents to the Church of Graphs are apparently forbidden from using gnosis at all.
Maybe the Church of Graphs not only forbids gnosis—it actively attacks it. The institutions of science, medicine, law, governance, and philosophy, the entire name of the game is to prove that your very eyes are lying to you. You don’t get a tenure-track position by concluding that crime does, in fact, hurt economic development of poor areas—you get that cushy job by producing a seemingly-well-reasoned epistemic analysis that concludes that crime is actually good for economic development. Or that poisonous pesticides actually produce better health outcomes in comparison to countries that don't use them. Or that crime is at all time lows. Or that the U.S. actually didn’t play a big role in WWII. Or any number of ridiculous things.
I suppose I don’t understand why some intelligent people seem to go down this path. It’s like they shun their ability to come to their own conclusions about the world. Very odd.
If it so pleases the comment section, I would like to present a case study on the difference between episteme and doxa. This one comes from the slightly less topical case of Somalis in Minnesota defrauding the state.
In 2018, an investigator employed by the state’s childcare program wrote a ten-page email alerting state officials of the extensive fraud he was seeing in the subsidized childcare program. Based on the size of the overall program, a cursory analysis of grant recipients similar to those he had successfully investigated, and the nature of the fraud, the investigator estimated that the true extent of the fraud reached $100M. He admits that this is a WAG (wild-ass guess) because the program requires insufficient documentation to thoroughly audit the recipients and his office lacked the manpower to investigate the daycares they suspected of defrauding the system.
The Minnesota Legislature’s Office of the Auditor wrote a report on the email. It’s relatively short, at about 30 pages. The conclusion? Far short of $100M, the fraud in the program is really only about $4M and the investigator has no support for his claim of two orders of magnitude higher. However, reading past the headline will give you the exact opposite impression, especially when one reads the summaries of the cases prosecuted.
The $4M figure comes from tallying up the amounts prosecuted in a court of law across about a dozen or so cases. The evidence in these prosecutions comes from videotaping the entrances to the daycares for a week or two and then comparing the headcount to the handwritten sign-in sheets the daycare provides for reimbursements. In these cases, investigators would find an average of 56 children show up daily vs. claimed figures of over 100, all of which were reimbursed. And so the daycare operator would be prosecuted for the $150k that the state could prove during the recorded two-weeks.
Did the daycare only defraud the state during the two weeks investigators happened to be watching? Of course not! The $150k prosecuted in that case is a fraction of what the daycare has stolen. $150k is all that the state could prove beyond a reasonable doubt in a court of law. The legislator’s auditor used this strict, formal definition to arrive at and debunk the investigator’s informal estimate. But the layperson will almost certainly agree that the investigator’s estimate is much closer to the truth than the auditor’s.
This is why you must always interrogate the data
It was Scott Alexander's review of _Seeing Like A State_ that introduced me to the concept of _legibility_. It seems dreadfully ironic that he is falling into such a clear legibility trap here.
I've had this same experience when reading Lyman Stone. Great writer, but an unabashed graph-head.
My cousin is a sociologist with a specialty on how survey questions and reporting mechanisms impact study outputs. Even in seemingly politically unbiased scenarios, there are all sorts of weird confounding factors that can give you an inaccurate picture. Like you said, they are useful, but not holy writ.
I also like the term spreadsheet-brained but it's practically a slur
I wish more people understood this
Excellent essay. My childhood Church shuttered due to dwindling attendance. The local diocese is now letting the city use the land for "single occupancy social housing." The result of this charity is dozens of drug addicts and criminals now live across the street from a middle school, the only local skate park, the only local football field, and the baseball field where I played little league.
We aren't just experiencing the blast radius from crime, it is being imposed on us by church and state.
I left my walkable neighborhood and 3% mortgage due to crime being up.
A quick Google is basically your article.
The news reports
1.) bars and restaurants are closing and the ones that are left close earlier
2.) this is because crime
...but
The area has faced a "bad rap" due to sporadic, intense, late-night incidents.
Crime is actually down!
There’s a Turing test for BS. It’s imperfect, but it can be honed. It’s called the sniff test. Intuition is underrated.
I've been reading Scott since 2015, and the last few months have seen a dramatic decline in the quality of his writing. Having kids slowed his output, which I understood, but he's also clearly developed some pretty bad AI Psychosis (he's even joked about it in some of his recent posts). Now he's regularly citing ChatGPT itself as a source for complex analyses and data.
I wonder how much of the crime post stuff was just him getting sucked into a little bubble by spending (in his words) 30 minutes per day chatting with LLMs. This sort of thing was always his failure mode, but possibly it used to be mitigated by the fact that he had to contact actual humans (and could trust that research he found online, if potentially horribly biased and full of methodological flaws, at least originated in a human brain).
LLMs are perfect members of the Church of Graphs: trained to favour doxa, and fundamentally incapable of gnosis.
I whole heartedly agree. I think you would like my essay about Scott's AI derangement syndrome.
https://giacomob.substack.com/p/everyone-has-aids?utm_source=share&utm_medium=android&r=7g2da
I did not enjoy it. You have basically one clear paragraph towards the end when you say
"What the blank understands that the techno-optimists miss is that our current LLM approach has inherent, structural limitations. These models don't have any actual understanding of the world—they're sophisticated pattern-matching engines trained on internet text. Expecting them to suddenly develop genuine reasoning abilities by making them bigger is like expecting a horse to eventually evolve into a helicopter if you just breed it to be fast enough."
I agree with that, but your writing doesn't add any new information about the subject; in fact, the only citations I can find are the graph about airline cruising speed and the list of previous AI hype bubbles; besides those two, which amount to about a paragraph of actual text, the post consists of you calling people "retarded" and various forms of ugly, with occasional pot shots at "climate 'science'". Even the very last sentence is just another way to say "Scott's ugly and I don't agree with him".
In conclusion, I'm plenty capable of generating hateful drivel on my own. The reason I've read Scott's writing for over a decade now is that, even when I don't agree with him on a particular subject, I've generally come away from his posts having learned something new and concrete about the world. To the extent that his AI psychosis has compromised this in the last few months, the solution is not rambling, unsourced, mean-spirited screeds about how autistic everyone involved is.
Well i believe in biofoundationalism. Therefore, If someone is ugly and dysgenic it will effect the way they view they world and is something I take into consideration when evaluating their opinion. Furthermore, I never claimed to be making unique or original points. These observations about AI have been made by many people before me, that makes AI 2027 even more idiotic because It shows how badly Scott and Daniel fell victim to the Dunning-Kruger effect.
Shrimp welfare man. I mean, come on. At the end of the day they're autistic bad faith actors.
That one Target store story is really funny.
I was a victim of crime in SF. I was visiting abd foolishly left my wetsuit and roadtrip compilation cd's and some other stuff in the car. Overnite it was broken into and the gear stolen. I was told there was no point in reporting it. That my recouse was to look for it at flea markets
Told by whom?
The people i was visiting who had lived there for years. A lawyer and a neuroscientist.
That's more useful advice than the SFPD will give you.
If the national statistics show a decrease in crime, but one’s own lived experience shows an increase in crime, then either your life is a large outlier, or there’s something wrong with the statistics.
I can’t really compare as I don’t know what public order was like 30 years ago. In NYC, depictions in media and movies suggest it was much worse in the past? Entire areas of the city you wouldn’t go to for fear of being attacked. There are still places like that today, but quite a lot of the city is safe, albeit with a low hum of insane bums bumping around.
I am pretty comfortable with the conclusion that things have been much worse in living memory.
Good article. I can't read new Scott Alexander articles any more, after "Bounded Distrust" (and every time he makes a splash on X I feel no need to change this policy).
Part of what made SlateStarCodex so formative for me is that it constantly probed The Church of Graphs to see whether it was *really* reliable . He did great reporting on the reproducibility crisis! "The Control Group is Out of Control" is a classic! Even on ACX he tore apart ivermectin studies, so it's not like he's a blind believer in any peer-reviewed graph.
Really, Mr. Alexander has his preferred conclusions, like everyone does, and he quietly bends his reasoning to produce them, like most people do. It's just disappointing because he used to write so well about precisely these kind of foibles even among rational, scientific types.
There are lies, damned lies, statistics, data science, and left-wing talking points.
https://principlesvstribes.substack.com/p/behind-the-glass-lies-damned-lies?utm_source=publication-search