Pet peeve: a market on X being at 25% does *not* mean "25% of traders think" X.

Expand full comment

I'd be happier thinking about whether our heroes at FTX will be charged with a crime if someone could point to the specific law they were supposed to have broken. I'm not claiming that there isn't one, but you can't charge people without a specific crime. Is moving folks crypto assets to your crypto hedge fund actually against a law? These things are mostly just numbers in a distributed database. If Nintendo nixes my in-game fancy hat that I paid money for I don't think a crime has been committed. So why here?

Expand full comment

"Commit fraud": It is hard not to. There are no healthy people, only under-examined. There are no honest tax-payers, dito. Tax fraud seems inevitable. Sure for lil' me - more likely to anyone handling so many different money streams as Scott. How many tax paid on SBF's subscription-fee? How to deduct the grants? How to be sure the tax-advisor is no cheat? Endless opportunities. - That said, there is no blogger I trust more to be intellectually in the category of "about as honest as a homo sapiens can possibly be". Would make me cry, if wrong. Would not care about other scandals, even if he was ever found out to have ever dated a 17 year old. rofl

Expand full comment


Probably a silly question. My serious answer: because working live gives a thrill that writing just can't quite match. And since I very rarely get to play on ships these days, podcasts are the next best thing. I think it's basically because you're matching your wit and memory, modulo whatever prep you did, with the situation, and coming out on top is great. That may not be your cup of tea, but it very much is mine.

Expand full comment

Wow big props to the 3 people who beat Rachel on predicting whether or not she was in a relationship.

Expand full comment

Trump to Lose is leftover from their own 2020 prediction market? I don't pretend to understand despite having read it twice. https://www.coindesk.com/markets/2022/11/15/analysis-ftxs-trumplose-token-isnt-proof-of-an-ftx-democrat-ukraine-conspiracy/

Expand full comment

> I like the “scandal” framing better than the “fraud” framing - I don’t make many specific claims or handle other people’s money, so I don’t know what it would mean for me to commit fraud.

Yeah, this is not ideal, and we've been talking about alternatives. (e.g. on Eliezer's market here: https://manifold.markets/IsaacKing/will-eliezer-yudkowsky-be-found-to#fZcDA1goloKOsldcEtt2)

Once we're able to decide on something that's relatively robust and unambiguous, I plan to create markets every EA/rationalist public figure that all use the same criteria so they're directly comparable.

Expand full comment

The TRUMPLOSE item was presumably the perpetual futures contract tradable on FTX itself.

Expand full comment
Nov 21, 2022·edited Nov 21, 2022

"...and by the way amazing job by “mr22” who started [the 'SBF convicted of a felony before 2026' market] on October 5"

Hold up, what happened here? Maybe an FTX insider? Or just a crazy hunch?

Expand full comment
Nov 21, 2022·edited Nov 21, 2022

On scandal markets: what about activist short sellers?

In the context of the public stock market, there are activist short sellers, who research public companies, find information that suggests they're going to underperform, take short positions, then publish their research.

It seems like scandal markets would have a similar kind of trader. In this context, the way you find out that somebody is about to have a scandal is that you hire a private detective to follow them around and take pictures. If you discover they're cheating on their wife, you buy the Scandal Yes position on the prediction market. Then, you publish your information, and wait for the market to update on it.

Expand full comment

Surely the main problem with "scandal markets", at least for people you care about, is that it gives everyone an incentive to dig up and publically re-litigate every bad, dubious or mildly sketchy thing that could possibly be pinned on the person?

For instance, the Eleizer Yudkowsky fraud market has already prompted a bunch of discussions about long-past supposed sketchy events that I didn't know about before, and which will now forever colour my picture of him, regardless of whether he actually did anything untoward or not.

Expand full comment

284 mana is more like $2.84 than $284, since $2.84 is what it costs to buy that much mana.

Expand full comment

> I’ll take this chance to say: a lot of media is predicting the death of EA, or a major blow to EA, or something in that category. Not going to happen. The media isn’t good at understanding people who do things for reasons other than PR. But most EAs really believe. Like, really believe<

This is a case study. Id love to know more details:


Expand full comment

The fact that lots of people believe in effective altruism does not mean that the Effective Altruism movement/community will continue to exist. Like, New Atheism is totally dead, but more Americans identify as atheists now than did when New Atheism was a big thing. Likewise, it’s possible that 10 years from now there will be more people giving 10% of their salary to charity, but fewer people “identify[ing] as part of the EA community”.

Expand full comment

On the matter of why Twitter became unprofitable, the short answer is it ramped up its operating expenses faster than its growth in revenue. From 2019-2021 TWTR added around $1.6bn in revenue and $2.5bn in op-ex.

The 2019 earnings number also looks much better than it really was because TWTR received a large tax benefit, though it still would have been nominally profitable without it.

2021 really was a good year for revenue growth, so maybe that spending did something. Though revenue growth was decelerating again in 2022 and op-ex growth was still very high. So this is what Elon figured -- a lot of that spending was wasteful and not doing much to drive revenue growth.

Of course, Twitter being profitable in 2023 just by slashing the op-ex to the bone while also losing some revenue doesn't necessarily make this a good investment for him, at the price he paid. The company's problem all along is that it has struggled to monetize itself in proportion to how big of a name we think it might be. More of a Reddit than a Facebook or Google. It was growing revenue too slowly to be considered much of a growth stock, but it was too unprofitable to be a value stock. That's why it was really just a dog of a stock.

Expand full comment

I would recommend against buying "Yes" on the Jones unblocking one.


Expand full comment

> at 4:50 PM California time, I spent M$1000 in play money (=~ $10 in real money) to manipulate it up to 95%.

Are you sure you spent M$1000? Market shows you having spend M$200.

Expand full comment

"Maybe Manifold should add a polling platform to their service?"

According to my market from a month ago this is a good idea!


Expand full comment

> Notice the second-to-last item, “Trump To Lose”. This sounds like a prediction market bet but I don’t know of any real-money market big enough to absorb $7 million on this question - the closest questions on Polymarket have only $1.2 million and $30,000 total. It could have been on a British betting site, but I don’t know details.

My understanding is that this relates to FTX's in-house prediction markets. The reporting on this is a bit muddled — Yahoo News and Coindesk claim this is from the 2020 election, though they don't explain why it would still be on the books now. My best guess is it's 2024 shares, crypto retail traders are probably net long Trump to win, and FTX+Alameda ended up net short providing liquidity to them.

Expand full comment

Please tell me that that last bit was a joke and you don't actually think insider trading is sinister.

Expand full comment

Re: name of Michael Lewis’s book about SBF: There’s no way it will be called Atlas Rugged. Virtually all Michael Lewis book titles are nouns, often preceded by an adjective and/ or ‘The’: THE BIG SHORT; MONEYBALL; LIAR’S POKER; THE PREMONITION. I think it will be called


Other suggestions?

Expand full comment

Surely if there was a death penalty for participating in EA movement, EA would do the effective thing: disband itself, then create a new movement with the same goals and same people, instead of trying to keep being operational against the tide.

So technically it can lead to the end of EA movement on paper for just the PR reasons, regardless of the the power of faith in EA cause. The chance that EA will die in spirit is, of course, much less likely.

Expand full comment

Surely the prediction markets need to be anonymous for scandal markets to work?

I'm a Hollywood insider. I know all the rumours about Harvey Weinstein before the story breaks, but also know he has enough power to end the career of anyone who speaks out. I have to assume he is keeping an eye on who buys yes on his scandal market and blacklisting them from getting employment?

I feel often in these scandals (including FTX to some extent) the whisper networks say "yeah we all knew but they were incredibly vindictive so we couldn't speak out. Non-anonymous prediction markets is the same issue.

Expand full comment

Is "asking that a scandal market be shut down" a red flag in itself, or justifiable by some non-market-related principle?

Expand full comment

Are there any signs of EA PR pushback in more mainstream news sources? Should there be?

Expand full comment

Something I don’t understand about scandal markets is that they seem to have to contradictory requirements.

First, the market has to be large enough to absorb manipulation, or to even hit a true equilibrium in the first place. That’s why this can only be implemented for public figures: there simply aren’t enough people in the world who care about whether or not I will be involved in a scandal to make an efficient market.

But for scandal markets to be a reliable source of information about a person’s true propensity for scandal, the people participating in the market have to themselves have reliable information. But the vast majority of people making bets about a public figure are going to basically just go off of vibes. High-information betters are going to be drowned out by noise.

This is made worse when the markets have public betting: the people with the best information are quite likely close enough to the subject that publicly betting on a scandal would have real-life ramifications for their relationship with the subject.

Expand full comment

Auger peaked at around $11M open interest for the Trump 2020 election. It is possible that a very substantial portion of that was FTX betting NO at short odds.

Expand full comment
Nov 21, 2022·edited Nov 21, 2022

"The market resolves not to “yes” or “no” but to the percent of FTX US users’ funds that they eventually get back; you make money if you were closer than other traders. "

Is that really so? I clicked through and it says yes/no market, and "yes shares worth 1 if yes wins".

Expand full comment

The Republican bias in the predictit averages is a holdover from those people who bought "Trump to win" in 2016. They were the early miners that made bank in the California Gold Rush, and since then, there has been a deluge of Republican 49ers coming to make their claim, looking for the next big score and many of them will lose their shirt.

Expand full comment

(1) Re: Twitter - FUDGE IT, WHY ARE YOU GUYS MAKING ME DEFEND MUSK? By "you guys" I do not mean Scott, I mean the likes of the guy linked for 'reasons why Twitter might go dark':

"12) Oh shit. You reverted one of them. Now every locked account's tweets are visible to everyone. People might literally get murdered with machetes over their posts. That's not a hypothetical. It's now 9pm. The site is fucked. Who are you going to call?"

That's right, if Elon fires the wrong engineer/engie quits in a flounce-off because xie is not going to work for a fascist, people are going to get murdered with machetes.

I humbly submit that if you are in a situation where you are genuinely likely to get murdered with a machete, things are so dire that a freakin' social media site is the *least* of your problems.


And to this?

"28) Genocide. People use your platform to orchestrate mass murder, the machetes in churches kind. And fast. Lightning fucking fast. You need to be prepared *before*. If you don't have a team who knows how to detect and stop this ASAP, your ass is getting dragged to The Hague"

Then the *best* thing would be for Twitter to go dark forever, rather than "eh, let's keep it running so people have to figure out work-arounds about organising their machete massacre". I sincerely recommend that if you are more worried about "but we neeeds the blue birdie, we does we does!" because you tweet on it so you run scare stories about machete massacres in order to get the new boss to do things the way you like them, you should go stick your head in a basin of cold water to clear it and set your priorities right. If blue birdie is being used by people who would like to use it to plan machete massacres only for your crack team keeping it safe for rainbow socks, then down with it, don't preserve it.

I swear to God, *this* is the kind of messaging making me hope it ends up with Twitter Delenda Est (or at least half the workforce), Elon Triumphans and he is awarded one of these, and I can't stand the guy ordinarily:


All that I'm reading with the squeaking protests makes me think Twitter operates like the old joke about the Vatican: "Your Holiness, how many people work in the Vatican?" "Oh, about half of them".

Twitter profitability: going by this graph, it was always loss-making *except* for 2018 and 2019. The only thing I can think of that happened in 2018 was Donald Trump. Was he the reason for the surge in usage and hence driving revenue? Heck knows, I'm not a chartered accountant:


This is a slightly better story about 2018 profitability:


Expand full comment

I want to see a market: “EA will renounce utilitarianism as a guiding principle by 2030.”

Expand full comment

I wonder if "girlfriend markets" can be used to attract girlfriends like "assassination markets" can attract assassins.

Expand full comment

Thanks for the shout-out Scott!

By the way, the data I collected and code for the analysis are available here:


in case anyone wants to do any further analysis (like a comparison of Manifold to 538).

Also, even though PredictIt did the worst of the three sites, I do want to say something nice about it -- PredictIt is BY FAR the best place to watch election results after the polls close. It's so much better at real-time information processing than the traditional media / pundits (who are not only technically unable to do this information processing, but also have an incentive to draw things out and make it seem like everything is a dead-heat race). Also, 538 doesn't update in real time like this on election night, and I think Manifold probably doesn't have the trading volume yet to do it reliably. So even though my post kinda implicitly criticized PredictIt, I do want to give them a huge compliment on that.

Also, I'm probably going to be writing more about forecasting and prediction markets, and doing more of these data collection and evaluation projects -- including sites I missed this time, like Polymarket. So if anyone's interested in this, please consider subscribing (for free)!

Expand full comment

Trump to lose 2024 is nothing like that deep on Betfair Exchange, so I doubt it was a British betting site.

The whole 2024 winner market is only £1,891,024 deep.

Expand full comment

Here’s one from the midterms that I put on Manifold Markets:


My purpose for creating this was to see if I could create a market that would gain traction on Manifold Markets and be a positive step toward resolving a political controversy by creating higher stakes on the question.

The way it played out is that it quickly converged to “No”, which is what I expected conditional on no response from the Don Buldoc campaign.

How could I have taken other action to maximize engagement on the market and the question in general? The main way I see would have been to try harder to force a response from the campaign. I could have, for example, used Twitter or other social media to press a harder challenge on the campaign to respond. Then the campaign would have faced the choice to either double-down or else be in a situation where the challenge is so visible that non-response could be viewed as ‘evidence of no evidence’

Expand full comment

What if instead of a scandal market you could report scandals to the police. This would be different that reporting a crime in the normal way, you could still do that. But this would be like a "half report". The police would only investigate if they got several reports against a person.

For instance people might be reluctant to tell the police they have been raped, because it's their word against the rapist, low chance of conviction. So they do a half-report. If the police gets two half-report against one person they investigate. Now there is two accusers, higher chance of conviction.

This is sort of what happened with me-too except with me-too the accusers was made aware of each other via the internet. But probably lots of people are reluctant to say they have been raped online.

Expand full comment
User was banned for this comment. Show
Expand full comment

Is Polymarket worth getting into? I've been using Manifold for the past few weeks and enjoying it. Thing is I've never messed with crypto before...

Expand full comment

Ok, I've made a post on Manifold to discuss various types of scandal markets to try to figure out which one works best. I hope you don't mind me that I made 7 different scandal markets on you to experiment with various options. :)


Expand full comment

I wonder if/how having the AI-generated image will significantly affect prediction quality. If art really does affect mood/mentality/cognition, will having some art there change anything? Will the surrealist quality of current AI art make people think more outside the box?

Would love to see Manifold A-B test that at the market level and come up with some fair cross-market metric of "consensus prediction accuracy prior to market close" to see if their images inspired better (or worse!) predictions. And of course make a meta-market about the result.

If the results are that the art significantly affects market quality, the possibilities for further A-B optimization are endless. Would adding additional parameters to each generation text to get the art to look a certain way further change market quality? E.g. Would asking the AI to make images in the style of Renaissance art enhance predictions more than modern art? And of course since the possibilities to test are endless, more meta-markets will inform what gets tested next.

If competition to get the most accurate consensus predictions is what will eventually set different prediction sites apart, they'll want to do some work in this vein on trying to create the optimal environment for accurate predictions

Expand full comment

In 2008 I was at an event where Musk was speaking. In those days, he was going through a public (& very messy divorce, gleefully covered in Gawker), plus the financial crisis had limited his ability to raise money for Tesla and SpaceX. The consensus in the room was that this guy is finished, he's just a two-hit wonder (Zip2 and Paypal).

That was the last time I have trusted the "wisdom" of the crowds when it comes to what Musk is capable of.

Expand full comment

[Disclaimer: I work at Metaculus.] Scott, any reason you embed the Manifold graphs but not the Metaculus one (i.e. like https://www.metaculus.com/questions/embed/13499)?

Also, we also have some questions about a variety of banned Twitter users getting unbanned that predate these current events by months: https://www.metaculus.com/questions/10860/who-will-twitter-unban-before-2023/

Seems like we need to improve Metaculus discover / embedding! Feedback here is welcome.

Expand full comment

> I’ve seen a lot of Twitter takes about how rich well-connected people never get in trouble for this kind of thing, but the markets seem less cynical.

"Never", at least sounds like a bit of an exaggeration. The most well-known fraudster in recent history, Bernie Madoff, got 150 years and died in prison.

It looks like the Enron scandal resulted in a number of guilty pleas and convictions as well, although I can't evaluate to what extent people in charge managed to shift blame onto underlings: https://en.wikipedia.org/wiki/Enron_scandal#Trials

Expand full comment

Might scandal markets make whistleblowers less likely to be believed due to accusations that they have invested in the market (i.e. they're self-paid shills)?

Or what about when Wholesome Family Friendly Celeb with a Dark Past hedges himself by putting down $1,000,000 on him being involved in a scandal so that if you try and cancel them they stand to become many times wealthier?

I think that, at the *very* least, we should get more experience with the failure modes of other prediction markets before we try scandal markets for individuals, public or otherwise. You can't just say oopsie we destroyed somebody's reputation and livelihood because of an economic unknown unknown.

I'm somewhat less concerned about scandal markets regarding large institutions though.

Expand full comment

The thing about some people not being able to conceive of other motivations for doing things than PR resonates with me a lot, and is a really easy and quick way of explaining something I'd previously faltered around trying to nail down for ages.

I saw a really crystallising example of this back around the Australian Open (tennis) this year. Super-quick background for non-tennis-fans: At that point Federer, Djokovic, and Nadal were all tied on 20 slams (four per year, biggest titles in tennis) each, which is way way more than the next best at 14 (and long retired). Federer was basically past it, but Djokovic and Nadal were (and still are) actively going for more. Rivalry between Djokovic and Nadal to be the greatest ever (on this very common metric) is super-intense. Likely whichever gets it will hold it for decades, because tennis is normally quite swingy but those "big 3" had an age of domination which was unprecedented, and there aren't clear candidates to do the same going forwards (though you never know).

So the 2020 AO was a big one symbolically for breaking that three way tie, and just every slam right now is a huge deal for Djokovic and Nadal. But Djokovic got excluded for sort-of-vaccine reasons (he initially was told he met the exemption criteria due to infection some months before (rumoured to be faked, but this has died down and I reckon it was probably real, and this didn't feature at the time in the legal proceedings) but then after they let him in, and a judge ruled he was all cool in one of those disaster live public Zoom hearings, he got deported for "public order" reasons anyway based on alleged anti-vax statements in the past (which were in reality of the extremely weak "well who knows if science or random shit is really real; make up your own mind I guess" sauce)) and Nadal went on to win it. Nadal now leads 22-21, so, you know, it was a big deal as the AO is historically the Novak slam - he's won there more than anywhere else - so who knows, maybe it would be 22-21 the other way if he'd played and who knows what that impact will be come the end of their careers.

So anyway, all that's going on, and my boss at work tells me he just can't understand Djokovic's mindset. This is a common statement because Djokovic makes no sense a lot of the time, but I wasn't sure what I meant and I pressed him. Turns out he couldn't understand *why Djokovic wanted to play the Australian Open at all*, given there was enough commotion about his vaccination status and having to rely on exemption criteria that surely the crowd would largely not be backing him, so why on earth would he want to play? What would be the point of becoming the greatest tennis player in history if the media didn't universally laud you for it because of some health stuff?

I suggested that maybe Djokovic, like most elite sportspeople, was somewhat driven by the idea of *winning* or *having a legacy* or something, and that people will remember who won the most slams ever for decades, but not who got cheered slightly less or questioned over vaccination status at one tournament for very long at all, but he just couldn't understand it. To him, Djokovic was just making a simple error: He didn't realise he was out of favour and that it would be in his short-term PR interest to take the L and go home.

It's particularly nuts for Djokovic because he's *always* been the heel of the tennis world, next to the much more PR-friendly Nadal and Federer. He's always let his results do the talking in the face of an indifferent to hostile media (outside his home country of Serbia where he's basically God - I mean worldwide/western media). I say that as a long-term Federer partisan: Djokovic gets the sharp end of the media/fans a lot.

But this guy, my boss, to him it was just incomprehensible. What an idiot, not to realise he was unpopular with the mainstream press and just go hide away for his own good. What good is being a historically significant, greatest of all time sports icon if you weren't the media's darling of the day? An utterly mad perspective but all too common.b

Expand full comment

Request for future posts that embed Manifold markets: can you include in the text of the post what the market's prediction was at the time you wrote the post? I'm reading this post a mere 16 hours after it was posted and some of the markets have moved quite a bit, such that I did a double-take at some of your text about them (e.g. the Manifold and Polymarket predictions on one of the Twitter questions were no longer as close as they were when you wrote the post). (the embedded Manifold graph widget thing seems to show the current probability, not the one at time of posting)

Expand full comment

If Polymarket etc did worse at predicting senate outcome than regular old Seltzer+, doesn’t that show a severe limitation in the usage of the prediction market?

Expand full comment

>And I’m tired of bad things happening, and then learning there was a “whisper network” of people who knew about it all along but didn’t tell potential victims. It’s unreasonable to expect suspicious to come out and make controversial accusations about powerful people on limited evidence. But a prediction market seems like a good fit for this use case.

How would you feel about alleged victims investing heavily in one of these markets before coming forward with their accusation? Would this make their story less credible?

Conversely, how would you feel about alleged victims NOT investing heavily in one of these markets before coming forward with their accusation? Isn't this essentially them leaving free money on the ground, and thus suspicious?

Expand full comment

>4: A story of financial detective work uncovering sinister market manipulation: the market Will [Manifold co-founder] Austin Chen Get A Girlfriend At Any Point In 2022? resolved YES on October 8th. The fourth-most-successful trader on the market is listed as Rachel Weinberg, who won $284 of play money. Rachel’s profile shows she has created questions like Will My Relationship With Austin Last A Year?, suggesting that she is the girlfriend, and made her fake $284 off insider trading! Congratulations to the happy couple.

Is this a loophole to legalize prostitution? EG you can't pay someone $100 for sex, but can you stake $100 on a market for 'this person will not have sex with me', let them enter the market betting against you, and collect an hour later?

Expand full comment

Isn’t insider trading good for prediction markets, as it permits the market to absorb some insider information and hence makes it more accurate? Mind you, I’m unsure why insider trading on normal stock markets is bad, but defer to the greater expertise of all the economists who think it’s bad.

Expand full comment

It's funny that the football world cup didn't even make the cut to be mentioned in this. Large disparity with the focus in other places. :)

Expand full comment

But If you notified Austin Chen of your attempt at market manipulation in advance, doesn't that mean that he (and his friends) could've immediately bet against you in a way that a normal prediction market wouldn't? If people saw you betting against someone you knew in normal circumstances they might think you have insider information and take you bet a lot more seriously (which could start an information cascade).

Also, don't people who want to avoid a scandal usually have much more to lose (job, status, etc) than their money on the prediction market, making an attempt at market manipulation rational from the perspective of a cost benefit analysis?

Also also, don't sock puppet accounts make the whole site ripe for manipulation?

1: Publicly place a large bet on your own trustworthiness through 2023

2: Through a sock puppet place a smaller bet on your committing fraud in 2024

3: Allow the first bet to shift odds against the second

4: On Jan 1 2024 after collecting your modest reward, commit fraud. Reap rewards of fraud. Then reap rewards from betting on fraud while presumed honest.

EDIT: Manipulation idea taken from here: https://old.reddit.com/r/SneerClub/comments/yv1mta/scott_alexander_solves_getting_scammed_with/iwe63sc/

Expand full comment

An FTX whale that goes by GCR did OTC bets for Trump. He's been talking about it for a while across his 2 Twitter accounts. Here's one tweet where he mentions doing the OTC with Caroline, the CEO of Alameda. https://twitter.com/GCRClassic/status/1591108154775212040

Expand full comment

I enjoyed the statement "Election betting is the bread-and-butter of most prediction markets" -- because of course, the bread-and-butter of betting on future events is sporting events and always has been.

Expand full comment

JFYI, this post keeps crashing on mobile browser for some reason. It refreshes every few seconds despite seeming to have been loaded. Maybe related to embedded markets?

Expand full comment