2026-03-30 19:18:37
If a novelist is really good at novelling then whenever something weird/unnecessary/contrived happens in the novel, you KNOW that thing will have plot relevance later, because otherwise it wouldn't be there.
I feel this way about Connie Willis, for example: when someone behaves unrealistically in her novels I know that's going to be a contrivance required for later, because she's too good to have unnecessary things happen unnecessarily.
(I guess the best novelists would make it so that the plot-relevant items aren't contrived either, and where you're in such a flow state while reading them that you don't notice when the big clues get slipped right in front of you, but who among us has such talents?)
This has weird consequences: in some ways, worse novelists are in some ways better to read, because when weird stuff happens it might just be because they don't control their writing super well.
For example, I actually really like Brandon Sanderson, and over time I've come to realise that he's super tightly controlled in his own way, but the first time I read one of his books and the protagonist went on a long unnecessary spiel about guns I remember thinking "this might be a case of literal Chekhov's gun, but also this guy might just be a ridiculous writer who really likes talking about guns, so he's self-inserting his hobby here like those ridiculous land-reform sections in Anna Karenina." As a result, I really couldn't tell whether the information was ever going to be relevant, and this increased my overall surprise in the novel.
By the way, for many years I had this (specious) Did You Know in my head about Chekhov's Gun, namely, that Chekhov originally meant it as a complaint rather than an imperative: the problem with theater, he was saying, is that if you see a gun on the wall in the first act you know (from previous theater experience) that it will go off in the third, so the surprise is ruined, versus in real life where there are tons of random superfluous details everywhere and you never know what is going to matter or not.
I have tried to fact-check this a bunch of times and not only can I not-find support for this claim, I have pretty repeatedly found people saying that what Chekhov meant by his Gun is pretty much what everyone thinks he meant by his gun. [EDIT: commenter Alex kindly confirms!] So I'm going to tentatively claim this thought for myself instead and call it Uri's Gun: the problem with fiction is that if at literally any point the author lingers over a gun, you know it's going to go off eventually, and that ruins some of the surprise.
2026-03-25 19:42:47
If you're in New York, come hear me talk about the weirdly unfamous cure for hiccups this Saturday at Nerd Nite NYC.
Recently I've been thinking about things that I wrote in previous ATVBTs that now seem wrong to me.
In 80/20 Strength Training, I approximately-recommended taking creatine. After a while taking creatine, I started to suspect it was making me more agitated/angry. I googled about this and there were various pages on "why creatine rage is a myth", i.e. pages saying that taking creatine doesn't make you agitated, but also that many other people have had this suspicion, which makes me wonder if it's real after all.
An internet friend recommended trying glycine instead of creatine, I bought some and promptly forgot to take it. I don't have a strong opinion on any of this, but wanted to update you guys in case you started creatine-ing on my account.
In my post on Prescheduling Posts, I basically recommended scheduling blogposts with increasing lags into the future. E.g. if you've written 6 posts, schedule 2 for the next two weeks, then 2 a month apart, then 2 two months apart, so you have over 6 months of runway with pre-scheduled posts.
I now feel more mixed about this – I think it maybe reduces my excitement about writing, I have ideas and I want to tell people about them, and by the time a post comes out 6 months from now I might not be excited about it, and/or I will lose 6 months of possible next-thought-having from the responses to the original post. Also, as my poetry teacher told us, the gods hate hoarders, and grant creativity only to those who trust them to give more later.
There's still some tradeoffs here – one of the most common failure modes for blogs is to not-write for a month or two, then feel like the next post has to be really good to make up for it, then not-write for a year or two until suddenly sending a "sorry I haven't written lately" post, and then never writing again. But basically I can't endorse the original thing I said any more, regardless.
In The Confounder Is Being Popular And Important, I complained about thinkfluencers advertising a life-philosophy that made them happy when (I think) really what made them happy is "finding community + having other people listen to your opinions." A reader pointed out that this is exactly what I was doing in the piece. I guess that doesn't necessarily make it an error but it does make me regret it.
I find most blogs/bloggers/blogposts overly confident, but writing in an un-confident way is annoying. In my head, all ATVBT posts have massive asteriskses, but I don't always include them in writing because (again) it's very annoying. So I have a more general sense of error/regret about a lot of my writing here, beyond these specific errors, I just feel like most of the stuff I write here is regrettably over-stated.
What do YOU think I should be sorry about? Feel free to email me or answer in the comments, as always....
2026-03-23 19:48:53
In New York? Come hear me talk about the weirdly unfamous cure for hiccups this Saturday at Nerd Nite NYC.
Here is a thing that repeatedly bothers me: people will claim that a given political system is biased against a particular political party.
This is not true, because it takes for granted something (the composition of the political coalitions) that is made by man and not by heaven, and which can and will be re-made in future.
It's extremely possible for a political system to be biased for or against a particular population distribution – for example, it can give more representation per-capita to low-population areas than to high-population ones.
That can certainly create a bias towards or against certain types of settlement, e.g. more representation for rural areas than urban ones.
Given history and path-dependency, that can also mean more representation for some regions than others, and given even more history and even more path-dependency that can mean more representation for certain ethnicities or age groups or any other form of social cleavage.
What it can't actually do is bias against any political party, specifically, because political parties are not fixed objects composed inevitably of particular demographics, but flexible coalitions that can (and have) been built and rebuilt over time.
This is weird for us to acknowledge because, at any given time, the parties develop very strong brands. I can feel some of you bristling through the screen, because it really does feel like each political party represents a certain set of coherent beliefs, and we often care passionately about those beliefs. So to claim that our party could just change its policies and attract a different set of voters feels as sacrilegious as saying that a religion could change its core tenents and then appeal to new adherents.
But a political party is not like a religion: it's more like a sports team, which at any given time is composed of a certain set of players, and which in future can trade those players for other players, in order to win.
Again, your particular beliefs or philosophies or priorities can be valid and important, and the beliefs you oppose can be objectionable and bad, and the political system can become de facto biased for or against the kinds of people who share your beliefs or priorities. There just isn't some unbreakable connection between those beliefs and the parties that currently represent them.
I think it's pretty psychologically interesting that at any given time there is an illusion of "naturalness" to the parties: our brains are pattern-matching machines, so they pattern-match each coalition together, finding the commonalities and ignoring the differences until it feels like of course (say) the religious people and businesspeople and the working class are in a party together, so if the electoral system is biased for/against religious people then it is necessarily biased for/against The Party Of Religious People And Businesspeople And The Working Class.
But this is not how things work. Those coalitions have shifted in the past and will shift again in future. If a sports team keeps losing matches it will eventually be taken over by a new manager who changes the team to a different one that can actually win. Similarly, if a political party keeps losing elections, eventually it will be taken over by a political entrepreneur who will reshape it to appeal to a new coalition and win elections again.
And it may still be true that the composition of the new coalition is shaped by the over/underrepresentation of certain geographic or demographic groups, because of bias in the electoral system. And you may still think that's bad, and want to change it. But that bias will now favour a different political party than it did before, because the bias never actually favoured a party, but rather favoured one or more components of the coalition that the party was appealing to.
2026-03-18 19:59:40
Ok so: apparently in the early 1900s there was a chain of automat diners where people could put a nickel in a slot and get a slice of pie (or a mac and cheese, or other such foods) out of a large wall of vending machines.
For another nickel you could get a cup of hot coffee out of the head of a dolphin, and this was a big deal because (if I understand correctly) french press coffee was only previously available in New Orleans at the time.
This all sounds kind of bats, no? I remember people wow-ing about food vending machines in Japan in the 1990s; how is it possible there had already been an entire automated food empire in Philly in 1902?
Did you all know about this? I swear I had never heard of it until I stumbled on it just now, and if you'd described it to me and asked if it could have existed before 1990 I would have said "no"? But instead, it was huge in the 1900s and shut its last store down in 1991.
Despite having stores in only Philadelphia, New York and Baltimore, at one point they were the largest restaurant chain in the US (!?)
There's various details as to why the Horn & Hardart company specifically lost its luster, involving the usual tales of poor succession and bad management. (If you want to know more there's a documentary called The Automat, with Mel Brooks and others – it has great old footage if you're into watching machines). The brand has been bought out of bankruptcy and is now selling coffee pods, inevitably: they claim they're going to resuscitate the actual automats too. That this specific company failed is all well and good, capitalism is one long chain of flowerings and wiltings.
But what I really want to know is: why did the concept of automated diners not succeed, overall? Isn't the long march of history towards more automation, not less? How is it possible that we went from automats in 1900 to non-automats by 2000?
Obviously we still have fast food restaurants that work in various factory-like ways, and obviously the automats also involved tons of human labour (e.g. to put the food into the vending machines in the first place). But it still feels more futuristic to me than e.g. the literal Burger Kings which apparently replaced many of the Horn and Hardarts. What's the economic story here, why did automation go backwards in this industry?
2026-03-16 19:23:02
I'm gonna give this my usual disclaimer for electoral-systems posts: I really enjoy thinking about electoral systems, but I think it's a hobby like doing crosswords, it has only a superficial relationship to the real world. If you actually want to change the way we vote you need to dedicate 7+ years of your life to 1) persuasion and 2) figuring out what normal people actually care about. I think that's very admirable, an extremely important and potentially high-leverage way to affect policy outcomes, because a polity's choice of electoral system determines many of their substantive political outcomes. But personally I just want to write blogposts, so here you go instead.
Basically: I wonder if we should bring back triumvirates. Instead of having 1 president we could have 3. Each one has a six year term, and every 2 years we vote for a new president to join the group.
I see (at least) three issues with modern presidential systems which this might solve:
1) modern presidencies create way more see-sawing than I think is optimal. I love the thermostatic element of democracy, if things go too far one way then (ideally) the electorate will respond by voting in someone to take things back the other way, and we can keep iterating and modulating to get to the right answers. But I want those policy shifts to look like sinusoidal curves, not jagged zigzags, and currently our system is built to switch suddenly from one party to the next.
In practice there is some "policy drag" from the civil service etc, so the new president doesn't actually get to enact their chosen policies immediately, but that's not ideal either. Triumvirates might let the cumulative Will of the People get implemented, but in a smoother way over time.
2) modern presidencies thrust people directly into an incredibly hard job that they can't actually prepare for. You can be Vice President or Governor or whatever before you're president, but they're honestly just not the same job: being The Decider is a whole different thing.
If we had a triumvirate then presumably the most senior president can be the Chief President, and the other two presidents can learn on the job more meaningfully.
3) I honestly hate the feeling that we pick Presidents at a specific moment in time and then give them 4-7 year terms, and that elections are influenced by the economy and the weather and the luck of international relations, and that it often feels like if we had voted 6 months earlier or 6 months later things might have broken the other way.
Voting every couple years would be kind of hellish, I know, and create a near-permanent presidential election cycle, but it would mean that the presidency reflected more of a rolling average of the popular will, which to me seems a better reflection of What The People Want than having the desires of a particular Tuesday determine the leadership of the next four years.
Heck, we don't actually have to stop at tri-umvirating: a quinquevirate might reflect the long-term popular will even better. We could toggle the numbers and the voting cadence, but the biggest step is probably just moving away from a single person to any umvirate at all.
Would this system be bad? How would it be bad? Let me know in the comments.
2026-03-11 19:11:50
As best I can tell, there are truly only five options for how to react if there's a mismatch between what you want & what you have:
I suppose the first two are better if you can manage them, but I also suspect that most people live in one of the last three.