A podcast I quite like, that I expect many lesswrongers to also like, is The Answer is Transaction Costs by Mike Munger, an economist at Duke University. He talks about economics primarily using the lens of transaction costs - that is, any cost that doesn't just transfer to the other party. These act as friction on the economy, and work like a tax would if the government burned all tax revenue. Mike often talks about his 3 parts of transaction costs: triangulation, transfer, and trust.
The best episode of that podcast that I've listened to so far is the episode on cancel culture. Here, he proposes that the human motivations behind cancel culture have been around for centuries, but modern communications technology has vastly reduced the frictions required to gang up on someone for what they said. It used to be quite difficult to confirm what someone said: you'd have to ask an eyewitness, who may only be willing to gossip but not to testify in court or be too loyal to tattle, and who couldn't remember the exact words stated. Now, if you make a heretical tweet it can become a global internet phenomenon in an hour. With all the sharers who condemn the original, people can easily attain common knowledge that the tweet is cancel-worthy (meaning they know that everyone knows that, and everyone knows everyone knows, etc.).
Related to the Green cars on Japanese trains, one of the things you're paying for with a higher-class ticket is to be out of the company of lower-class people. That's certainly a reason I sometimes travel first-class on trains in the UK.
I will never understand the idea that it is good and right to arrest people for making edgy jokes in a private WhatsApp group, without anyone in the group even minding, as happens in the UK. How does anyone describe this as anything but utter insanity?
I think it is ignorance rather than insanity. They don't believe the Streisand effect is real, they do believe they can turn society into a big skinner box. Therefore by censoring expression enough they can alter behavior permanently. Censoring even in the most private places is key to this working in that view.
It turns out that TSPI’s polling was outright fraud, with their explanation fully AI generated. It is a shame that this is not treated as criminal.
Isn't this supposed to be TPSI (The Public Sentiment Institute)?
AI has taken over this blog.
I have moved to a schedule of seven posts per week, and I still cannot keep up.
We still refuse to abandon the rest of the world. Who knows when I will get to post some of my huge backlog on education or dating or other such topics. But the monthly is a sacred tradition. We continue.
Bad News
A good reminder that most news is bad news, and chosen because the bad news in question is rare, which is good news, but the pattern overall of choosing this to be news is bad news, and often the bad news is that the bad news was chosen as news and now people are talking badly about it.
Alibaba uses your computer’s audio system, and other tricks, to at least try and assign you a device fingerprint.
Things you cannot buy in America at any reasonable price. Mostly it is impressive how little makes the list, how it feels like corner cases.
The one thing I am envious of is the exterior roller shutters, and the ability to be in actual pure darkness on demand. I would pay quite a bit to get that, if I could integrate it into my apartment.
Wrong answers are a skill issue.
I see this as being both A and B at different times.
It turns out that TSPI’s polling was outright fraud, with their explanation fully AI generated. It is a shame that this is not treated as criminal.
Is it a curse?
That’s true only if you judge relative to yourself. Intelligence, consideration and rizz greatly raise your access to others with intelligence, consideration and rizz, including across such traits. So the question is, are you only thinking in relative terms?
Only about half of all psychology professors said, when asked to pick, that they put social equity concerns before truth, although the rest are overwhelmingly ‘it’s complicated.’ It’s not complicated. There is also a gender divide here, and the OP tries to make it look worse than it is, but neither of those is the story.
I strongly agree with David Perell that it is sad that we used to make it a priority to make a variety of things beautiful, and now we don’t and merely make them functional. The big one is architecture, but this also applies to things like tickets, as per Ted Gioia’s The Banishment of Beauty From Everyday Life.
It’s worse than a crime, it’s a mistake. That beauty buys so much goodwill.
I knew coal plants got people killed but I did not know they also have a lot of radioactivity, or that this is still an active subject of a cover-up.
Nature incinerates what credibility it has left, says ‘the death of Jason Arday must prompt deep reflection.’
Which is the context that means this is not as funny as I was hoping it would be:
This from Jack is a remarkably apt metaphor for the whole situation:
Even among a relatively conscientious group, most people will read private DMs?
Not cool, everyone. I can kind of see arguing the glitch is Musk’s fault, but no.
Focus Only On What Matters
Much failure is about failing to realize what is important. Hindsight sometimes helps, and sometimes it does not.
Woke 1 Was Crazy
Matthew Yglesias points out the arguably worst part of it, which is that even the people who successfully avoided saying absurd or false things were still scared into silence.
Once the reasonable people are running for the hills, everything is f***ed.
During Woke 1, I very much stuck my neck out with regard to Covid-19, but I was very careful about picking any other battles, and mostly did shut up about everything else. Indeed, I don’t say things I don’t believe, but on a number of fronts I still pick my battles and instead stay quiet.
That was and remains, I believe, savvy behavior, as I would not have gotten much benefit out of the risks and costs I would have endured. If everyone picked one big thing and was loud, that would be enough. But yeah, I still do feel weird about it.
Good Advice
These are two key lessons in spending:
Early on you want to focus on lesson one, then later shift to lesson two. Then you strategically and selectively unlearn lesson one.
The reason the green car can be worthwhile is that the marginal value of the green car is vastly higher than the cost of the train. At some point, one hopes, the train costs you approximately nothing.
What’s the best way to take a break?
I buy this, which must be balanced against the cost of the break, and sometimes you really do need an introvert break rather than an extrovert break. They serve different needs.
The question then becomes, detached from what? The phone is not automatically not detached, the same way that being with a person may or may not be detached.
I’ve heard similar versions of this enough that I believe it but I basically don’t get bitten in NYC or Berkeley so it never comes up:
Not at Lighthaven? That is unfortunate for you. You’ll need different norms.
I also do not recommend acting like you are with a bunch of Christian Conservatives, which will get you absolutely destroyed among most groups in blue states. There used to be a ‘one generic code’ you could fall back on, and largely be okay. Mostly now there isn’t, because part of the code of one color is reflective dismissal of the codes of a different color. Tricky.
When choosing careers,
Opportunity Knocks
You could instead be at Lighthaven soon.
Inkhaven 3 is coming. If you are thinking about going, you should probably go.
The thread has examples of some writing that came out of previous editions, and doesn’t include my favorite of the blogs, which is by Paola, here are some of her posts. I agree with Vishal that the biggest problem so far is that a lot of residents haven’t sustained their writing after Inkhaven.
Alas, I will probably not be an advisor, because when we tried this in Inkhaven 1 it turned out that I was a poor fit, although it would not be impossible to convince me to try again if there was demand.
The University of Michigan robotics curriculum is on GitHub. A second away.
While I Cannot Condone This
I sympathize greatly with Scott Alexander falling into ‘I am staying out of the Substack religion debate [proceeds to get sucked into religion debate].’ This reinforces that if you don’t want to get sucked into a debate, it is often wise to not explain why.
Whenever you hear an 80/20 or other distributional statistic, consider changing ‘Just the top [X]% were [Y]% of [Z]’ to ‘The top [X]% were just [Y]% of [Z]’ and see which one seems more appropriate.
I would be very surprised if the top 20% of book readers only read 82% of the books. There is an extreme right tail here, and a lot of people read zero.
I too am an Honest Fake Expression Enjoyer:
What I like a lot less are Dishonest Fake Expressions, where the person is acting and trying to present the smile as if it is real, and does not want you to know it is fake.
Know a guy who owns a truck. Then you can move things, or all the things. The cost is buying the pizza afterwards, plus helping out when it is your turn to help. The system works, and it makes everyone richer in a real sense and builds community, but lowers measured GDP. A lot of the story of vibecession and related topics, where economists say you are richer and poor people insist they are instead poorer, is explained by this.
The account here, Nobunaga, has many similarly good stories about a particular American set of norms around community cooperation, reputation, honesty and virtuous norms. I do not take them to be literally true, but this is not important. They are fables in support of norms. Those that have such norms will vastly overperform measured GDP, and people will be happier. Those that lack such norms will underperform measured GDP, and people will be less happy.
The norms are not always efficient, and can be taken too far. The electrician should not be impossible to reach. When you are in higher demand you should charge more, and so on. Growth and productivity rates suffer. I am generally a fan of more capitalism. But something important and valuable is being lost.
If there is going to be an Unexpected Hanging some day this week, what are your options?
Which in turn, I argue, means you can also test on Friday, because on Friday you will think it would not be a surprise so therefore they can’t give the test, which means it will be a surprise.
Astrology is a hell of a placebo, with people who read horoscopes exhibiting 0.2 standard deviations of self-reported personality movement towards their sign, whereas the effect size for others is exactly zero.
Good News, Everyone
If you want to ensure you can cancel your subscriptions, credit cards do offer some options. You can use X1, Rho or privacy.com to spin up a new card for each subscription. You can use Mercury, which allows you to easily block merchants. And so on. I presume AI will soon be able to handle this for you rather easily either way.
The Apple Vision Pro is a dud with consumers. They did not deliver the killer apps, interface and linked capabilities to justify the price point, charging issues and weight. But technically it is an excellent product, such as improving performance in endoscopic lacrimal surgery while being under 1% of the cost of an endoscopic display tower. The linked study has low sample size, so treat with caution.
The NFL press box food offerings, ranked. I appreciated this, I eat very differently but we are kindred spirits in the ways that count. Amazing how many teams are bad at it.
Ordinary Abundance. Many found this highly moving.
For Your Entertainment
Coyote vs. Acme is the third 5-star movie of 2026, matching 2024 and 2025 combined. Recommended for literal everyone of all ages.
I tried an experiment with turning Letterboxd’s top algorithmic picks for me into a Manifold market. It did not go great. Almost no strong opinions were expressed, and the top pick, The Grand Budapest Hotel, was a bitter disappointment. Perhaps my readers can make better picks. You can add new suggestions to the market, as well.
I agree with Tyler Cowen that it does seem like movies peaked in specifically 1999, as per the speech in The Matrix about that being the pinnacle of your civilization. Then movies got worse in key ways, including too many franchises wasting top budgets and talent. Now movies are getting better again. 2026 has been fantastic.
What about Game of Thrones? We have two narratives fighting to this day, between ‘Season 8 fell off a cliff’ and ‘seasons 5-7 also kind of sucked.’
Memory is blurry by now, but I think the truth is in between. Things clearly were best in the first 4 seasons, top level television. The momentum and some great moments carried us through 5, 6 and largely even 7, and fans kept voting high because they wanted it to be great and the characters and scenes were still often individually great. Then in season 8 it became impossible to pretend.
My general thesis continues to be that it is better for shows to overstay their welcomes, and to sacrifice quality on the margin for quantity. It doesn’t have to get this bad, but unless the story has a clear fixed length, keep it going. Often the later parts that aren’t strictly ‘as good’ are still very good in their own way, and occasionally they age well once you’re over the initial disappointment.
Are you ready for Avengers: Doomsday? Collect them all.
That is asking a lot. I am two movies short. But they’re likely a tough two movies, especially Brave New World.
The state of DVDs, I still find this beautiful even though I have abandoned all use of physical media:
We need movie theaters that show a wider variety of movies, and show things for longer.
I have seen enough empty theaters to know that it would be viable. The opportunity cost of off-peak showings is very low. I also would love a low-cost all-theater rental option for off-peak times.
Or how about letting an individual choose the movie for some modest marginal cost? If on average you sell 5 tickets, and I am willing to buy 5 tickets a week in advance and want to put on Eternal Sunshine of the Spotless Mind, do it and see if anyone else comes.
What has been working great is the Anniversary editions of older films. I am increasingly seeing movies from 20 or 30 years ago having brief runs at AMC, and it is glorious, I need to make it more of a priority to attend.
The craziest part is that they are often ‘excluded from A-list,’ which means they charge you like $20 instead of $0, and yet they sell a lot of tickets. This clearly means there aren’t enough such offerings, and they don’t hold them open long enough.
The addiction economy:
I continue to be a big fan of video games, in moderation or even somewhat more than moderation, provided they are ‘good’ games. None of the three things here are inherently bad. The problem is lack of moderation, or getting trapped in bad versions. The version on your phone will usually be the bad version of all three.
I did a Twitter poll (now lost to time) where I asked which of these three things was not like the others, and got defenders of all three.
Correct toon logic: Sonic is faster than the Road Runner (probably), but if Sonic tried to catch the Roadrunner he could definitely fail and probably blow himself up.
Terminator 2 is a great move on entertainment alone, Jeffrey Ladish finally watched it and asks about how accurate it is on the AI front once you ignore the time travel and the Terminators. The answer is better than you would think, especially the emphasis that the dangerous thing was the computer that went into recursive self-improvement rather than the robots.
The big ‘real’ issue is that Skynet learns and gains power absurdly fast at first, and then stalls out at exactly the level where it can somehow be defeated by a John Connor and a ragtag group of survivors, because otherwise the time loop would not close and also we do not have a movie or franchise. This is the typical Hollywood ‘intelligence stops doing anything, and also the AI’s efforts stop scaling, at exactly the point where the humans can still barely win’ which exists for reasons that should be obvious, both in Hollywood and also in the real debates over AI. People like humans winning, and people like telling human stories.
Please Review This Podcast
Many solid choices by Dean Ball. For me this splits into ‘things that I can confirm are very good’ and ‘things I have not listened to and thus it is hard to say but they have good company.’
I am especially a fan, from this list, of Cognitive Revolution, Conversations with Tyler, Odd Lots, Dwarkesh Podcast, Money Stuff and Complex Systems. Dwarkesh suffers from being ‘too good’ so I end up often reading transcripts instead. I sometimes listen to Lex Fridman and 80,000 Hours, as well as AXRP, but long pod is long and I choose topics.
At some point I will write about at least some of my favorite podcasts. Two not here that I very much enjoy are Wait Wait, Don’t Tell Me and Serious Trouble, and I at least sometimes listen to EconTalk, Clearer Thinking, Win-Win and Plain English.
There are also some that fall into a different category of gaming and entertainment. It can’t all be work stuff. My favorite Magic podcast is All Tings Considered, but that is constrained by my ‘I do not know what the new cards do’ problem.
Gamers Gonna Game Game Game Game Game
You can now report players on Magic the Gathering: Arena. There are five complaints: Inactivity, Hate Speech, Inappropriate Username, Verbal Abuse and Spamming.
Three of these seem silly given the lack of communication channels. But it is very good to be able to report players for Inactivity, also known as stalling, which players do either because they are mad at you, because they can or in the hopes that you will give up or accidentally time out.
Andrea Mengucci proposes ways to help bring back competitive Magic, including more focus on Pauper as a cheap format. I’m ‘out of the game’ in all senses, so it’s hard for me to evaluate the current situation, but the obvious thing is that Magic is huge yet they give you very little to strive for. Every reasonable solution starts with running real tournaments, with strong prizes, and ways to accumulate points, qualify for things and earn money, pride and titles, including room for some people to be Pros off of results rather than streaming. Wizards is not doing this because they don’t care, not because they do not know how. I guess they do not like money.
I Was Promised Flying Self-Driving Cars
On September 1, Waymo opened service in Denver, San Diego and Tampa.
Waymo once again expands its approved service area in Greater San Francisco as well as Los Angeles. Most of this region will be waiting for a while, but this means Waymo will have no trouble finding good uses its cars for a while no matter what happens elsewhere.
I continue to eagerly await Waymo coming to East Bay, given that is where I usually visit. Alas, the odds look bad, and say a majority of the time this won’t happen even in 2027. Complaints are the prices remain too high, which is another way of saying demand exceeds supply.
Tesla asks for permits for 5,000 robotaxis in Las Vegas, was initially given 10 but has now been approved for the full 5,000. Waymo and Uber each got 1,000.
There are some things self-driving cars probably won’t ever do for you.
Waymo gets into an accident that makes the news. As one would expect, it is minor contact and entirely not the Waymo’s fault.
Not to be outdone, Waymo ‘nearly ran over’ a two-year old, said many. This is the framing people go with for ‘actual nothing happened and it was not close.’
‘I literally don’t know what you want from me’ very much applies here. The Waymo did exactly what you would want it to do, complete with displaying ‘pedestrian yield,’ and the complaint seems to be that the father was worried that it might not do that?
Yes, I very much worry that one accident will ruin everything. Eventually there will be one accident.
Government Working
I will never understand the idea that it is good and right to arrest people for making edgy jokes in a private WhatsApp group, without anyone in the group even minding, as happens in the UK. How does anyone describe this as anything but utter insanity?
Flock cameras plus AI is coming, while Flock demos things like ‘how to monitor a protest in real time.’ Have we considered not allowing the creation of a dystopian surveillance network for the government or others to hook up to their AIs?
Mamdani (Again) Fails Economics Forever
I will keep harping on the Mamdani NYC city-run grocery store fiasco, mainly because it is funny.
Here is 115th & Park, where the new store presumably will go:
They keep changing their story because there are zero ways to implement the thing they want that do not end in disaster.
Variously Effective Altruism
Remember what the benchmark is, in two parts.
Part 1: The super rich do things like pay a specialty agency $100,000 per year with a $1 million spending minimum above that, and then spend absurd amounts of money on whims and momentary thrills, and most of all on getting exactly the things they want whenever they want them, and yes they are often jaded. Olivia Ferney, a travel specialist they hire, confirms, as we learn in How To Make A Killing, that you might want to think this means they are miserable, but actually the super rich are happy with the results, thank you very much.
I totally get it, especially the part where you have a person you can call that makes happen whatever you want to happen. Must be nice. But I’d feel sick about it, and not only because I could have given the money away. I sometimes wonder how many extra zeroes I’d need on my net worth before I’d feel entitled to start acting like this. At least two, but less than five.
Part 2:
This, on the other hand, feels completely insane, even if you have Too Much Money. Which Seth Rogen clearly does. Costs are not benefits, and the point of having money is partly to spend it but also to have the peace of mind that it is there when you need it. The value of that extra $10 million spent while rich is a tiny fraction of how much you would miss it if you lived longer than you expect while in any danger of being poor.
Dying with nothing in the bank, or close to nothing, is a huge disaster, far worse for Seth Rogan than dying with $10 million.
That’s even if you discount that you could use the money to actually help people.
So, yeah, I criticize it a lot, but maybe this whole ‘effective altruism’ thing is actually pretty good.
How Sarah Constantin thinks about evaluating grant applications, here condensed with a slight Zvi slant to the explanations, as we think highly similarly here. I endorse basically all of this.
If Hell does not exist, then we do not need to create it. We already have New Jersey.
A highly underrated charity strategy is Give Effective People Money, especially retroactive funding for things you loved, with no strings attached. Also speed kills.
Moral intuitions very different from my own, remember that this was asking ‘what do others do’ not ‘what should people do’:
Robin was as usual Robin Hansoning, as he described reason as ‘argued for by our impressive thinkers’ rather than ‘what the person thinks is right, based on their own reason.’ As you would expect, Robin then argues hard against using your conscience.
A very clear case for Effective Altruism, especially from Patrick McKenzie:
The Lighter Side
Sometimes I have sympathy for people who lose money but if you bought Hunter Biden’s $LAPTOP right before it crashed 98% within an hour of its launch then I do not really know what you were expecting.
Spotted on Letterboxd about La La Land, some people are a tough crowd:
In case anyone was wondering, yes, I will also always think like this.
Proof by counterexample:
Cheat, cheat, cheat, cheat, cheat. IYKYK, if not see here.
Fable argued hard that no, Michigan did not actually lose to The Bye (aka Western Michigan) and that there really was a mysterious extra second on the True Stadium Clock that allowed Michigan to have one more play. Uh huh.