Comments
Couple of thoughts, kinda long and rambly
This gets discussed occasionally on the Manifold Discord and I wanted to share some skeptical points that one of the top forecasters (Semiotic Rivalry) made there:
- "for me to go >5% on this [authoritarian takeover/coup] i'd have to see them openly disobeying orders from the supreme court or like, the very least should be killing the filibuster"
- "I feel like the lightest first step of the fascist takeover would be to have the VP overrule the senate parliamentarian on what can be permitted to go into a reconciliation bill, which is totally legal and tons of Dems wanted Biden to do, and they failed to even do this"
- Supreme Court is still constraining him, e.g. Trump wasn't allowed to fire Federal Reserve commissioner Lisa Cook
- Revealed preferences suggest people don't actually believe dictatorship/catastrophe to be very likely: they aren't moving abroad or stocking up on guns. (To which people replied that it's not easy to find a good place to live abroad due to economics, language)
This was largely in response to me saying that I find it hard to think through Trump/MAGA military (self-)coup possibilities. Because although military self-coups appear to be rare in consolidated or backsliding democracies, they're not entirely unheard of, and it sure seems like Hegseth, Trump, etc. are working towards this. They are systematically dismantling military guardrails:
- They keep pushing the envelope on deploying military domestically (in conflict with the Posse Comitatus Act), which blurs lines among the population and the military for this
- They've fired a lot of military leadership as well as all Judges Advocate General (JAGs) which generally serve as a constraint on executive overreach.
- The pardoning of war criminals, the broader pardoning of J6'ers and Trump allies, and the commands to engage in war crimes like shooting down a boat of non-combatants (allegedly drug-traffickers)
The general pattern of purging appears to be that Trump/the administration gives an illegal/norm-breaking order which functions as a loyalty test: it forces everyone involved to either comply, step down, or refuse to obey (which tends to get you fired - something that the Supreme Court hasn't been adequately protecting besides the Fed).
The coup form I expect, if it happens, would not be a direct command to military generals, but to order his most loyal militarized groups (e.g. red states National Guard, ICE) to take control of the democratic/election process. Opposing military would then have to coordinate on action, which would be very difficult. The general population could resist en masse (South Korea 2024-style), but so far protests have been small, and in the US there's a vocal and dangerous base supporting Trump. That said, base rates suggest a coup is still very unlikely, and coups are difficult. I don't know what probability I would give it, I'm mainly trying to understand the mechanisms here.
Other thoughts:
- Trump attempted to overthrow an election before
- Orbán is often mentioned as comparison (rightly so), but he was able to amend the Constitution in the first year due to Hungarian law, which is a major difference
- An economic crisis would be a major cause of discontent and the AI boom is really unfortunate
- Protests so far have been small, No Kings Protests of "5 million" was [greatly exaggerated](https://bsky.app/profile/siebepersists.bsky.social/post/3lruu445wgk27) (posted on BlueSky but I'm not at all active there otherwise). I think bigger protests will be necessary (but not sufficient)
- I haven't even talked about AI but it's a wild card, probably would favor Trump. Executives are largely very appeasing (OpenAI, xAI), appeasing (Google, Meta), or softly defiant (Anthropic)
- People think Trump is too old and a unique figure, but I'm not confident that a successor wouldn't be as bad. At some point, they either put a successor on the ballot or Trump himself. A successor could pull away power from Trump and then lose. There's generally a lot of possibility around this to dis-unite any coup-interested faction. However, I find the sentiment that "Trump is uniquely bad and his successor will not have the same power and therefore it's not a concerning scenario" overconfident, and there's plenty of systemic reasons to expect a successor to be pretty bad
Before I quit, I was the #1 trader on Manifold. Those markets listed on the forecasting section are very biased, suffer from issues relating to long term markets, have resolution criteria that are very different from the natural understanding of those questions and have several other flaws. Using them to suggest anything like the odds presented is very misleading.
Thanks for pointing this out, I have a vague sense that Manifold markets can be non-ideal but don't really know why or what these biases you're talking about might be. I added a few more questions from other markets to compensate a bit + moved my disclaimer from the footnotes to the top of the section. These are still among the only public forecasts with actual numbers I know of, so seems better than nothing, right?
Also, fwiw, some well-informed people I know (who's estimates I can't share) have estimates quite similar to these prediction markets.
For long term markets, there is a well known interest rate/opportunity cost problem, which prevent markets that have a low (or high) probability from getting priced correctly. Furthermore, the resolution crtieria for many of these markets are extremely subjective and involve an author claiming that "democracy is over as we know it" when potentially common things would count here.
I also didn't understand it. I researched it, and I think this is the more in-depth explanation:
Long-term markets with very low or very high probabilities aren’t correctly priced because of opportunity-cost dynamic. For example: events that should trade at 2% stay at 5% OR events that should be at 98% stay at 95%. Everything gets pulled toward the middle because the opportunity cost of tying up money for years makes it not worth fixing the mispricing.
For a low-probability event priced at 5%, a trader who thinks the true probability is 1% would have to bet "No" and lock up their money for, say, four years to make a small profit (on a market at 5% they invest 95 cents to maybe win 5 cents profit over four years). This small return is often less than what they could earn from safe interest elsewhere or from betting on shorter-term markets.
The V-dem Metaculus market shows a clean time-series bump following recent Trump events https://www.metaculus.com/questions/14141/us-continuously-a-liberal-democracy/
This market, and metaculus in general, doesn't have anywhere near the same frequency of market quality problems as Manifold. For readers, I'd advise against dismissing the OP's headline concerns as some manifold measurement error (though those are also real enough and I agree with the specific claims made by Marcus).
Marcus, ton of respect for your open-mindedness and prediction ability. Sort of parroting Lintz here but if you have the time, I would greatly appreciate if you could give some insight on how to improve the questions.
I understand that questions pertaining to 2028 and maybe even midterms suffer from long term market issues. So maybe we could create a chain of conditional markets? or at least some intermediate steps that we think are proxies and have a reasonable chance of occurring in the next few months?
Additionally, would you say you have updated your views since this comment chain?
Hi Charlie, for improvements for long-term markets, I dont have great ideas other than large loans/leverage for betting on these questions and giving skilled political forecasters a lot of money to bet on them.
For resolution criteria, I would suggest a market on trump being president on March 1 2029 without a constitutional amendment. I expect the odds for this to be below 5% (well, interest rate problem). I think the constitutional amendment part is critical here. I dont think its undemocratic for Trump to be elected for a 3rd term, so long as proper procedures are followed here and he wins the election fairly.
Other markets i would suggest would be on imprisonment/murder of political opponents and judges. I would suggest markets like "will at least 4 of the following 10 people be imprisoned or murdered by Dec 31 2028", etc.
As for my views since the comment chain, my median scenario is worse than it was back then but also my worst ~2.5% scenarios aren't as bad as they were back then. That is to say, I think things are going badly, worse than I thought they would at the start of the presidency/on election night but also it's not going as badly as my worst case scenarios (which were really bad).
I want to be clear, I'm extremely unhappy with the Trump presidency thus far. It's been awful and I feel for many groups of people who have been negatively affected (including Americans and non-Americans, USAID recipients, undocumented workers, tranns people and more). Ive been sad to see many developments and actions of the Trump administration. I also think there are non-authoritarian/dictatorship scenarios I consider to be quite bad. I think the Trump presidency is noticeably worse, on many dimensions than other administrations, especially on things like respecting norms, rhetoric, honesty, and willingness to go after political opponents/other branches of government. That said, im not seeing the kind of things others talk about in terms of being a couple steps before dictatorship.
"I dont think its undemocratic for Trump to be elected for a 3rd term, so long as proper procedures are followed here and he wins the election fairly."
I can kind of see where you are coming from. I would invite you to consider that sometimes even that sort of thing could be bullshit / tyranny cf. the Enabling Act of 1933.
Also, for resolution criteria:
"Other markets i would suggest would be on imprisonment/murder of political opponents and judges. I would suggest markets like "will at least 4 of the following 10 people be imprisoned or murdered by Dec 31 2028", etc."
Do you think specific targets would generally have been easy enough to call in advance for other autocracies / self coups? That seems non-obvious to me?
Conditional markets compound long term issues (assuming they're collateralized): instead of locking capital into a market, that's locking capital into hoping there's a market
Prediction markets can increase capital efficiency by:
* pegging interest rates (like adjustable-rate mortgages)
* recognizing hedged positions[1][2]
* lowering fees
More reasons for cautious interpretations:
Manifold:
* 'mana is purely a play-money and not exchangeable for sweepcash, money, or any other goods and items'
Polymarket:
* orderbook's actually 2.5% which the interface rounds up to 3%
* Polymarket offers 4% variable interest on that (Kalshi went from 4.05% to 3.25% this year)
Metaculus:
* 500 deaths is closer to previous US standoffs than the US Civil War
* Only question of those >3% as of writing only has 33 forecasters
* All the histograms have odd tails (somewhat dampened by Metaculus reporting medians):
There's advantages to play money, such as that players don't care as much about the time value of money (it's also much easier to start and resolve markets, leading to many more markets)
leads to lower signal-to-noise ratio
and heeding time-value still leads to more (play) capital / pricing power