The Best Things I've Read Recently

On EA, religion, philosophy, and miscellaneous

EAish topics

In the AI world, the AI futures project’s plan A has been all the rage recently. It’s AIFP’s plan for how to navigate the race to AI, primarily based around an agreed-upon slowdown with China. It’s an interesting proposal and the kind of thing that more people should be thinking about. For some pushback, see my colleague Tom Davidson’s piece Plan A’s problem with dry tinder.

Speaking of Forethought pieces, over on the Forethought substack I published a piece on the prospects of having AI do philosophy. I make the case that it’s reasonably likely that a superintelligence could solve all the thorny problems in moral philosophy.

Lucius Caviola and Will MacAskill have a piece in the Guardian titled “Could AI be conscious?” The basic argument is pretty simple: experts are sharply divided about AI consciousness. We should thus take it as a serious possibility and plan accordingly, rather than ignore it out of convenience.

Lots of people suggest that there’s something evil or dystopian about AI chip regulation. Scott Alexander (not the baseball player) points out that chip regulations are pretty standard as far as regulations go.

Don’t forget to attend the Amsterdam insect protest!

How do most wild animals die? Insects probably mostly die of parasites, but we don’t really know. Adult mammals mostly die because of human activity, juvenile mammals mostly die of predation. For every other kind of tetrapod (reptiles, amphibians, birds, etc), both adult and juvenile, predation is their main cause of death. And for fish, we have no clue.

Friend of the blog Richard Y Chappell (who I recently met at Manifest) talks about how to seriously entertain weird conclusions without losing your mind. Richard and I also gave a talk together at Manifest talking about analytic philosophy and Rationalism—the pathologies in both and what they can learn from the other.

Silas Abrahamsen has a dialogue about veganism that is one of the best things I have ever read. I laughed out loud multiple times reading it.

My friend Value Locked In has been producing banger after banger on EA topics. In a recent post, he pointed out that a lot of defenses of meat eating are also straightforwardly defenses of slavery.

Abi Olvera wrote a piece about why AI assisted bioweapons are unlikely to cause extinction. I’m a bit more worried than she is, but she nicely presents the skeptical case. Her blog is also in general great. Good discussion in the comments.

Update on the state of the effective giving ecosystem in 2026.

Philosophy of religion and religion

Joe Schmid has a great new paper on a strong new argument for theism. The argument in a sentence: if there’s no God, it’s surprising that the things that produce conscious experience—out of the countless physical arrangements that could, in principle, produce it—happen to arise in our universe. You might, upon noticing that the ingredients in a kitchen happen to be sufficient to produce a souffle, conclude someone designed it to be this way. Upon noticing that the conditions needed to produce consciousness—the source of most of the world’s value—happen to arise in our universe, you might similarly conclude that it had some designer. Joe is very smart, meticulous, and hardworking, so the paper is excellent.

Amos Wollen argues that Baha’ism starts out pretty probable because it’s a pluralist major religion. He argues we’d expect God to want to reach many different kinds of people. That would likely culminate in some kind of pluralist major religion—and Baha’ism is one of the few good candidates for this.

My friend Ethan has been investigating Catholic miracles (not to be confused with my other friend Ethan Muse who has been investigating Catholic miracles). He just wrote his first post on the subject, and it’s characteristically clever. More great things to come from his substack, I predict.

I talked with Robin Collins about his greatest story argument for Christianity. It’s a clever argument, but it has a lot of different controversial moving parts. Was a cool discussion!

On Scott Alexander’s blog, one of the book review participants posted a review of the Book of Abraham. It’s very funny and also pretty good at explaining why Mormonism is so implausible. The central Mormon claim is that God decided to wait 1,800ish years after the death of the apostles to start a Church—and he did it by providing a translation from golden plates to a con artist who demonstrably fabricated translations on multiple occasions; this translation contains descriptions of a vast civilization in the Americas that we have no archeological, genetic, or written records of, suspiciously converges with racist folk-lore, and permits its founder to have many different underage wives, some married to other people!

The classic free-will theodicy claims that God gives us free will to hurt each other because having free will is valuable. But why is there so much natural evil? Dustin Crummett has a paper that gives one possible answer: we’re in a simulation and natural evil is the byproduct of our simulators’ free will.

New sun miracle just dropped: this one in 1991 in Puerto Rico. Thanks to Mark for pointing this out.

Arthur T has a nice post raising some skeptical objections to Zeitoun being miraculous. His piece about Sai Baba was great too!

The best paper on whether inferring a multiverse from fine-tuning commits the inverse gambler’s fallacy! Spoiler: it doesn’t.

Alexander Pruss proves that God exists:

1. Humans are ridiculous.

2. Humans are only ridiculous if there is a being much greater than humans.

3. So, there is a being much greater than humans.

I love Pruss’s blog so much. He just writes down every argument that seems sound to him, even if it’s obviously unconvincing. We just get to peer into the mind of one of the most brilliant philosophers in the world and see whatever he’s thinking about.

Hilarious debate about the Trinity.

Philosophy

Christopher Menzel has a proposal for modifying ZFCU to allow for sets who have too many members to have a specific cardinality. It’s a cool idea and my guess is something in this vicinity is the right view (though due to my minimal knowledge of set theory, I can’t comment too much).

This paper explains how to ditch countable additivity, relativize bets, assign a uniform probability distribution, and avoid Dutch book.

When evaluating the infinite, use hyperreals (at least, in many cases, though things get funky with trying to take a hyperreal value over locations).

Speaking of countable additivity, many of the arguments for it would seem to generalize to the undesirable full additivity. I’m not much moved by the betting argument for countable additivity, because it assumes that if there are an infinite number of bets that are each worth taking, then the sequence is worth taking. We shouldn’t grant this. Infinity individually good things might turn out bad or undefined. More in a footnote.¹

Eliezer Yudkowsky debated functional decision-theory with Kenny Easwaran. I thought Eliezer did a good job. I also recently had an FDT debate and a debate about anthropics.

MacAskill et al argue that under uncertainty, you should mostly write-off causal decision-theory because non-causal effects of actions have the potential to be so much more significant than their direct causal effects. This is right but it makes me sad.

Silas Abrahamsen argues that Taurek’s argument against the numbers counting would, if applied consistently, imply it doesn’t matter how much you benefit people either.

Jessica Taylor has a blog post presenting Dutch books for sleeping beauty halfers. Decisive.

Credences can’t just be betting dispositions because sometimes your betting on a proposition interferes with your credence in it. For instance, my mother does not bet, but no doubt she would bet at high odds on the proposition “I will make at least one bet in my life” (for betting on it would mean she won the bet).

Alan Hajek and Lina Eriksson argue that “degrees of belief” are primitive and don’t reduce to betting behavior. A safe heuristic in these areas: when in doubt, defer to Hajek.

Speaking of Hajek, in his brutal war against frequentism, he has two papers: the first gives fifteen objections to finite frequentism and the second gives fifteen objections to hypothetical frequentism. Fourteen would have been too few, apparently.

Other

Five Ways to Know Yourself gives a good introduction to the kind of mindfulness that majorly boosts well-being. I just started reading it a few days ago, but I found applying the practices majorly boosted my enjoyment of daily life.

Lewis’s A Grief Observed is beautiful and amazing and devastating—very much worth a read. It’s primarily about the death of Lewis’s wife, though it also communicates something of the beautiful love story that ended when she died. Too many amazing parts to quote them all, but one notable bit:

Talk to me about the truth of religion and I’ll listen gladly. Talk to me about the duty of religion and I’ll listen submissively. But don’t come talking to me about the consolations of religion or I shall suspect that you don’t understand.

Another:

For a good wife contains so many persons in herself. What was H. not to me? She was my daughter and my mother, my pupil and my teacher, my subject and my sovereign; and always, holding all these in solution, my trusty comrade, friend, shipmate, fellow-soldier. My mistress; but at the same time all that any man friend (and I have good ones) has ever been to me. Perhaps more. If we had never fallen in love we should have none the less been always together, and created a scandal. That’s what I meant when I once praised her for her ‘masculine virtues.’ But she soon put a stop to that by asking how I’d like to be praised for my feminine ones. It was a good riposte, dear. Yet there was something of the Amazon, something of Penthesileia and Camilla. And you, as well as I, were glad it should be there. You were glad I should recognize it.

Solomon calls his bride Sister. Could a woman be a complete wife unless, for a moment, in one particular mood, a man felt almost inclined to call her Brother?

Every time I read a grief-stricken account of death, the notion that donating to save lives is not obligatory appears deeply repugnant.

Scott once again posts about his kids, and as usual, it’s the cutest and the funniest thing in the world.

I was informed by Ethan Muse that even in my recent piece about Candace Owens being totally nuts, I left out the occult ritual theory. Turns out that Owens thinks that the murder of Charlie Kirk was literally part of an occult ritual and that the people around him are into demons and magic.

Probably it will be possible in near-term space activities to avoid detection. In the long term, when you can detect gravitational waves, this gets a lot harder.

Is Daniel Muñoz an agent of the death of philosophy itself? I’ve been wondering this, so nice to see it finally answered.

David Oks writing amazingly interesting posts about varied topics with amazing detail. He has a piece about why China got rich and India didn’t (TLDR India stayed poor because of limited human capital, but Mao vaporized China’s institutions so thoroughly that traditional kinship structures didn’t keep them poor). He also has pieces about why ambulance rides cost so much, why kinship societies often kill their old after declaring them witches, and more.

Intergalactic warfare is probably defense dominant. Once an intergalactic civilization has been set up, it will probably be very hard to dislodge. Invading fleets are easy to shoot down, projectiles can be shot down with lasers, and lasers fired across galaxies can be rendered moot by moving around.

How long will I love you is a very good song! How long will I love the song? Well, as long as stars are above it. And longer by far.

I’ve enjoyed the books of Dale Allison. But ColdButtonIssues argues that he misstates many crucial points in his books. I haven’t checked this super carefully, but it’s concerning if true.

Crazy plagiarism scandal at Cambridge.

In this series, Tim O’Neill debunks many new atheist myths about history. O’Neill is an atheist, but he’s grown increasingly irritated with his fellow atheists distorting history to fit their agenda. He’s also absolutely hilarious and devastating. Here’s one passage I liked especially:

Carrier, the reflex anti-theist and anti-Christian activist, was having none of this. In a characteristically splenetic piece entitled “No, Tom Holland, It Wasn’t Christian Values That Saved the West”, Carrier dismisses Holland as “another amateur” and a “novelist” with “no degrees in history, and no advanced degrees whatever” and declares categorically “everything he says is false”.

Carrier then congratulates himself, declaring “I’ve already refuted Holland’s entire thesis”, linking to several of his blog posts and book chapters in atheist polemics, all of which dispute various things that … Holland does not actually say.

…

This is typically sloppy stuff from Carrier, given that there is no evidence that the bunnies and eggs of Easter are “pagan” at all, no evidence that Eostre (not “Eastre”) was “German” and the name of this local Anglo-Saxon deity is about the only pagan thing about Easter (see “Easter, Ishtar, Eostre and Eggs” for a summary of the evidence and scholarship on this). As usual, Carrier’s bluster and bravado outweigh his knowledge and competence.

I laughed out loud at “Carrier then congratulates himself,” (and decided to incorporate a similar line in my recent article).

1. Here’s one Dutch book for the person with a non-normalizable probability distribution. Suppose there are infinite people in Hilbert’s hotel—one in each room. First you offer everyone a bet where they lose $1 if they’re in room one but gain 25 cents if they’re in any other room. Then you offer everyone a bet where they lose $1 if they’re in room two but gain 12.5 cents if they’re in any other room. Then you offer everyone a bet where they lose $1 if they’re in room 3 and gain 6.25 cents if they’re in any other room, etc. Each person expects to win from each bet. But in the limit, everyone loses money (about 50 cents, depending on their room number). So is a person who assigns zero or infinitesimal credence to being in each room money pumpable?

No. The fact that each individual bet is worth taking doesn’t mean that the infinite succession of them is worth taking. One case to illustrate this: suppose that you want to maximize the number of balls in a vase. You start with ten balls, labeled 1-10. Then you’re offered a deal: ball 1 will be removed, balls 11-20 are added. Good deal. Then you’re offered another: ball 2 will be removed, balls 21-30 are added. And so on forever. Each individual deal is good for you. But take an infinite number of them and every particular ball is removed. Multiplying infinitely many individually good things isn’t necessarily good.

And it does look like each bet is good individually. After all, each bet being taken by everyone leads to people gaining infinite dollars and only $1.