A big part of the answer for me is something like this Scott Alexander post about the probability of X within your model versus the probability that your model is miscalibrated in a relevant way. Given how shaky our models of the world are, this alone makes it hard for me to push past 99% on many questions, especially those that require predicting human decisions.
Organized religion may not have reduced virtue signaling, but it may have contained it within the church.
As stated, I think this has a bigger vulnerability; B and B* just always answer the question with “yes.”
Remember that this is also used to advance the argument. If A thinks B has such a strategy, A can ask the question in such a way that B’s “yes” helps A’s argument. But sure, there is something weird here.
the dishonest team might want to call one as soon as they think the chance of them convincing a judge is below 50%, because that’s the worst-case win-rate from blind guessing
I also think this is a fatal flaw with the existing two-person-team proposal; you need a system that gives you an epsilon chance of winning with it if you’re using it spuriously.
I have what looks to me like an improvement, but there’s still a vulnerability:
A challenges B by giving a yes-no question as well as a previous round to ask it. B answers, B* answers based on B’s notes up to that point, A wins outright if B and B* answer differently.
(This has the side effect that A* doesn’t need to be involved, and so B can later challenge A. But of course we could get this under any such proposal by having teams larger than two!)
The remaining vulnerability is that A could ask a question that is so abstruse (and irrelevant to the actual debate) that there’s a good chance an honest B and B* will answer it differently. (I’m thinking of more sophisticated versions of “if a tree falls in the forest” questions.)
Correlation looks good all the way except for the three massive outliers at the top of the rankings.
“How to get people to take ideas seriously without serious risk they will go insane along the way” is a very important problem. In retrospect, CFAR should have had this as an explicit priority from the start.
Relatedly, there’s an awkward cursor line in the top-right box for optimism bias.
I believe the corporations in Moral Mazes were mostly in the manufacturing sector. (Your second point applies, though, as a decent explanation for why American manufacturing has been increasingly outcompeted in the last few decades.)
Okay, so another necessary condition for being downstream from an optimizer is being causally downstream. I’m sure there are other conditions, but the claim still feels like an important addition to the conversation.
I’m still skeptical of Circling, but this is exactly the sort of post I want to encourage in general: trying to explain something many readers are skeptical of, while staying within this site’s epistemic standards and going only one inferential step out.
Either (1) or (2) (and some other possibilities) would satisfy my prediction. My prediction is just that, however we do things in 2029, it won’t be by handing each merchant the keys to our entire credit account.
Dammit, dammit, dammit, I meant to condition these all on no human extinction and no superintelligence. Commenting rather than editing because I forget if the time of an edit is visible, and I want it to be clear I didn’t update this based on information from the 2020s.
Okay then, how about higher education as a fraction of GDP?
As with last decade, I’m most confident about boring things, though less optimistic than I’d like to be.
Fewer than 1 billion people (combatants + civilians) will die in wars in the 2020s: 95%
The United States of America will still exist under its current Constitution (with or without new Amendments) and with all of its current states (with or without new states) as of 1/1/30: 93%
Fewer than 10 million people (combatants + civilians) will die in wars in the 2020s: 85%
The median rent per unit in the United States will increase faster than inflation in the 2020s: 80%
The National Popular Vote Interstate Compact will not go into force by 1/1/30: 75%
Human-driven cars will still be street-legal in all major US cities as of 1/1/30: 75%
As of 1/1/30, customers will not make purchases by giving each merchant full access to a non-transaction-specific numeric string (i.e. credit cards as they are today): 70%
Conditional on Pew Research Center releasing a survey on the topic after 1/1/28, their most recent survey by 1/1/30 will show that 60% or fewer of American adults identify as Christian: 70%
Conditional on Pew Research Center releasing a survey on the topic after 1/1/28, their most recent survey by 1/1/30 will show that 33% or more of American adults identify as religiously unaffiliated: 70%
More than half of American adults will use a Facebook product at least once per day in 2029: 60%
Real-time 24-hour news networks will still exist as of 1/1/30, and will average more than 1 million average daily viewers (in the USA) in 2029: 50%
The largest not-currently-existing US tech company (by market cap as of 1/1/30) will not have its primary HQ in the Bay Area: 35%
A song generated entirely by an AI will make one of the Billboard charts: 25%
California will put a new state constitution to a statewide vote: 10%
Re (7), there’s a laughable amount of conjunction on even the first prediction in the chain.
Re: higher education bubble, do you also predict that tuition increases will not outpace inflation?
The Rubik’s Cube one strikes me as much more feasible than the other AI predictions. Look at the dexterity improvements of Boston Dynamics over the last decade and apply that to the current robotic hands, and I think there’s a better than 70% chance you get a Rubik’s Cube-spinning, chopsticks-using robotic hand by 2030.
De-aging is different from a fully CGI character.
My most testable prediction (that the top movies/books/albums would make up a much lower fraction of overall revenue) was completely wrong across all media.
Thank maths for videoconferencing, enabling working from home (at least occasionally) for every major tech company.