Defining Conspiracy Theories
They're still stupid.
Epistemic status: Conspiracy theories are mostly pretty dumb, but on priors weird things do sometimes happen, so while I dismiss them all here there’s plausibly a case where at least one thing is not maximally wrong.
Defenders of conspiracy theories often claim that they’re unfairly maligned by hindsight bias. Many conspiracy theories turn out to be true, they claim. It’s just that whenever one turns out to be true it gets eaten up by mainstream science, and everyone then turns around and pretends the real conspiracy theory is not believing in it (while still mocking remaining unproven theories as ridiculous). That people make fun of unicorns or bigfoot as silly conspiracy theories while considering underwater arctic unicorns or hairy ape-men who live deep in the Congo’s forest mountains to be perfectly normal animals feels unfair.
Another argument is that there’s a lot of perfectly normal things that can read as conspiracy theories if you describe them that way. Kosher tags on food products can almost reasonably be described as “a hidden tag marking that the producing corporation paid jewish organizations a secret tax to avoid a jewish boycott”, except for the “hidden” and “secret” parts. Cloud seeding can be described as a government weather control project. If these weren’t public knowledge, they’d sound a lot like conspiracy theories, and it makes sense to assume there’s at least a few projects like that that aren’t public. So really, why shouldn’t conspiracy theories be true?
There’s two answers to this. On the direct level, it’s a basic fallacy - a lot of things might be true, but that doesn’t mean any particular thing should be (most things aren’t!). If you have any specific theory for which you don’t have a large amount of evidence, it’s almost certainly wrong based on priors.
The deeper issue though is that “conspiracy theory” isn’t really a category of theories. It’s a category of style of thinking. The fundamental proposition of conspiracy theorizing is “there are big, secret truths they don’t want you to know". The fact that this is complicated, elaborate and hostile is in itself evidence that this is likely to be true”. And this is just not how things work. Complicated things are unlikely to be true, exponentially so in the amount of described complication1. Conspiracy theorists think it’s important to believe in chemtrails or flat earth or whatever because they’re complicated secret conspiracies being held from you. It’s why they’re typically so long and detailed.
Or in short: A conspiracy theory is any theory that argues it should get a complexity bonus, instead of a complexity penalty, for your credence in it.
This seems like a ridiculous adaptation at first glance, which raises the question of why anyone would get such a stupid bias in the first place. The answer is that it makes perfect sense in hostile information environments.
Imagine you’re playing chess, and the other player moves his pawn forwards. The parsimonious explanation is something like “he just felt like moving that pawn”. The complicated explanation is “he’s moving that pawn as an elaborate plot to gain some hidden advantage THAT HE DOESN’T WANT ME TO KNOW ABOUT!”. The complicated explanation is clearly correct, because we’re in a hostile information environment - the information we receive is deliberately engineered to mislead or manipulate us. It’s correct to look for hidden hostile meanings in every piece of information we get, because the information is being generated by a hostile process that uses its expectations on our own assumptions as part of its move-generating algorithm2.
Or consider the stock market. If you have a simple theory (“Apple makes good phones people like buying, so their stock is going to go up”) it’s almost certainly priced in and wrong. It takes a complicated theory or secret information advantage (“I got a secret satellite picture of the Walmart parking lot showing their truck activity is down, so their stock will probably go down”) to actually beat the market and make money3.
This being adaptive to hostile information environments also explains why conspiracy theories are so disproportionately believed by stupid people. It’s not just that the theories themselves are stupid; it’s that they’re an adaptive behavior for hostile information environments, and stupid people are just much more vulnerable to those. Lower intelligence makes you more likely to be snared by anything from ads trying to convince you to pay too much for stuff you don’t need to charismatic politicians pretending to agree with your exact beliefs4. And on the other side, high intelligence is protective from that - going back to our chess example, if you’re a very good chess player you can easily understand why your opponent chose a certain move, and not spend so much time worrying it’s part of an elaborate trap you can’t see5.
Examples
Now that we have the theoretical model of conspiracy theories, let’s look at examples of “conspiracy theories that turned out to be true”. When talking to conspiracy theorists, There’s three examples I usually hear:
The CIA created modern art as a psyop
The FBI was secretly behind the Manson murders
MKUltra
If we look at these with our conspiracy theorist hat on6, we want to believe complicated versions of these theories requiring conspiracies. If we look at it with our normal person “complexity bad” hat7 on, we expect simple versions of it that don’t seem very scary or conspiratorial. As far as I can tell, here’s what actually happened in each case:
The CIA had some kind of cultural funding for western art styles in the 50s and 60s, as part of a cultural diplomacy effort to show the superiority of free western cultural expression over Soviet art. In typical CIA fashion, this was indirectly funded through an intermediate body (artists would probably feel weird about accepting checks directly from the CIA). This was defunded in 1967 when people suddenly noticed how weird it was for the CIA to be doing this. Definitely on the weirder end of cold war plots but still a fairly simple story. They didn’t invent modern art (or any particular art style), just promoted osme general art funding for the hot styles at the time.
The FBI and local law enforcement had some screwups related to the case including a badly managed raid and a parole check miss. Basically your garden variety issues with probation/parole lapses, weak case follow-through and siloed info between different agencies having trouble working together on a case.
MKUltra was a real thing on the weird end of CIA projects. They experimented with drugs (as well as other shady methods like sleep deprivation or psychological stress) to study interrogation and influence techniques. There was some morally shady stuff involved, like participants who weren’t told what the studies were for or who was conducting them.
The crazy stuff people talk about didn’t happen though. The CIA didn’t develop secret mind control techniques. They weren’t secretly behind any murder cults. They didn’t put LSD in the water supply to control the public. If we just work backwards from “It seems like the CIA would be into studying shady interrogation techniques”, we could have guessed something that looks quite a lot like it should exist even without evidence. But only if we kept our guesses simple, instead of also assuming they were creating murder cults on the side.
Overall, all of these aren’t surprising. If we try to guess a bit more about things in the same vein - intelligence services are probably doing similar things with social media snooping and manipulation they don’t tell us about8. There’s some other things in this vein you could guess at9. But the best way to approach it is to figure out what various groups want from information warfare and what the simplest way for them to get it is. This ends up looking like normal military projects, not weird conspiracies.
…Okay fine, let’s talk about the elephant in the room
Given the timing, I guess we have to talk about the Epstein thing.
Conspiracy theorists are lately having a field day with the Epstein story, claiming it proves once and for all that they’re right to be suspicious of elites.
Well, I’m here to say they’re still dumb. Epstein doesn’t look like a conspiracy, and doesn’t show us much we wouldn’t expect. As far as I can tell, these are the facts:
Epstein was a rich wall street guy who got a lot of money working with wealthy private clients, possibly by scamming them.
He was absurdly prolific at networking, both in recruiting clients for his firm and in funding a crapton of science projects, arts and culture stuff, policy and think tank circles, and basically every intellectual elite pursuit you can think of.
A lot of rich people and intellectuals were happy to get money and connections from him, many even after his reputation got pretty shady.
He did a bunch of sex trafficking and supplying prostitutes for people. At least some rich people were in on it.
He seems to have preferred young women (early twenties or even younger) and often didn’t let their status as minors stop him from trafficking crimes involving them. While there seems to be at least one case of a trafficking victim as young as fourteen, there’s no evidence that any actual pedophilia was involved10.
He probably wasn’t working directly with the CIA or some other intelligence agency. This one is hard to be sure of - Intelligence agencies like having dirt on people and Epstein had a motherlode of dirt on rich and powerful people around the world, so it’d be unsurprising if the CIA tried to get in on using some of that11. But there’s currently no evidence of any such thing actually happening, so our beliefs on it should be lower than our priors of the CIA getting involved in something like this (there’s at least decent chance we’d have seen some evidence by now if they did).
This mostly seems unsurprising to me. Epstein liked money and connections (many people do). A lot of culture and science people liked being offered money and connections, often enough that they didn’t look too hard at the guy offering it. And while I find it offputting, people with a lot of money tend to want to find ways to use it for various forms of prostitution and it’s unsurprising they found someone to enable that at scale.
Does it show any surprises at all? I’m legitimately surprised by just how prolific this one guy was, I would’ve expected this social role to be more fragmented (even among the narrow circle of rich elites he liked hanging out with). And I’m also mildly surprised and disappointed about how many rich people seem to not mind prostitution12. So there is some room here to updating beliefs about the world. But only about baseline human tendencies and market fragmentation, not about the existence of secret pedophilic cabals.
Anyway, this makes a testable prediction - there’s tons of information coming out on the Epstein stuff and so far it’s all lining up with my model. If the conspiracy theorists are right, we should see that change as more information comes out. Whoever ends up less surprised will be, as the kids call it, less wrong.
There’s two relevant measures of complication - the description length of the theory, and the irreducible length. The first is trivial to deal with - if a conspiracy theorist claims there’s hidden green alien spaceships stored in a government basement and it turns out the hidden alien spaceships are actually blue, he’s still mostly right (although do note the oddness of him adding unnecessary details to the theory).
Irreducible complexity is a harder problem. The real issue with most conspiracy theories is that they require a large number of things to happen together to work it all, typically including a large number of people to all work together to keep the secret. For most things I’d describe as actual conspiracy theories, there’s no substantially similar simple version that could work.
I used chess as an example here for simplicity, but chess is an open-information game. hidden-information games like Poker (let alone real-life social or military conflicts) are much more intense and complicated versions of this.
In practice, this is overhyped. Most alpha on the stock market isn’t secret tips or amazingly complicated models (those typically fail), it’s getting access to the best/fastest data available and making good complete models of the known factors affecting it. “Just do the obvious stuff extremely well” is by far the most common source of real life alpha. But you do have to actually do it well, not take a single random fact and make half-assed assumptions based on it.
I think smart people underestimate this risk, stupid people may be better calibrated on this one. Smart people who are used to being able to see through normal-tier sales tactics or ads are caught off guard by elite-level political talent being used against them and tend to jump into enthusiastic support when they should be more careful and analytical. The way some smart people react to talented politicians brings to mind a local chess champion confidently thinking he knows what’s going on on the board when playing against Magnus Carlsen.
In real life very good chess players usually play with other very good players and have to worry about elaborate traps even more than the rest of us. But that’s because of how they choose their opponents, and wouldn’t happen if they just played in their local club with the normal crowd.
Presumably it’s made of tinfoil
It’s very cold outside right now, so even normal people are wearing simple wool hats.
We know China does this with tiktok, for example.
Ryan Macbeth’s talk on how information warfare is fires is a good cover of this sort of thing.
To be clear, this is still a lot of very bad crimes that include abuse against minors, the intention here isn’t to defend him. But the claims of pedophilia are often added to paint Epstein and his elite clients as deviant and unnatural beyond just the normal level of sex crime, and that seems wrong on the facts. My point here is that these are normal (bad) sex crimes that normal (bad) people commit, not a special exotic version unimaginable to normies.
Complexity penalty applies as usual - e.g. if they were involved, it’s more likely they stole data about a few people as kompromat instead of getting in on running the whole operation - but even heavier involvement is at least understandable and not totally implausible.
It doesn’t seem like it' happens at a higher than background rate - plenty of poor people call strip clubs or call girls - but I honestly did expect most smart rich people to be more put off by the whole business.