I strongly disagree with this. I agree that there are competing considerations (weirdness, unusual dynamics when people are losing money), but I think that the factors pointing to the other side (betting cuts through a lot of bullshit and might be particularly informative in times of crises, bets allow a stronger signal of belief than speech, bets reward those who are correct) still trump them.
Also,
force people to express certainty around their statements
is incorrect, and in fact bets allow people to express shades of uncertainty by changing the odds and the betting amount.
I’m not surprised that you would strongly disagree with this. Sorry if this is a bit blunt, but given that you very recently launched a public forecast on the love life of a community member, without their permission, to the dismay of many people on Twitter, I wouldn’t think you have a particularly calibrated frame of reference as to when we should hold off on bets.
As you might imagine, I also disagree about whether that market was per-se bad. I think that there was a small information cascade about that particular topic because people in the EA community didn’t want to appear mean, & so better epistemic infrastructure would have actually helped.
I ended up removing that market because I got too much pushback on it, including from my boss, but Twitter wasn’t actually up to to hosting a nuanced discussion on the pros and cons, just social opprobrium.
I also strongly disagree with the OP. Giving up betting as an epistemic practice ought to be a non-starter, similar to “let’s take a month-long break from assigning probabilities to things”.
Where did I say anything about “giving up betting as an epistemic practice”? My post specifically listed why I know it’s good epistemic practice.
All I said was that during a crisis that involves the potential loss of money of 100,000 people, constantly suggesting 4-digit bets on public online spaces as soon as two people have the slightest disagreement may not be the most empathetic thing to do.
Thanks for this. I’ll add that I think even in less tense times, there are alternatives that achieve similar clarity while avoiding community culture downsides. Money is a fraught topic, EA has at times been accused of being classist, and people should be able to engage in EA discourse regardless of their relationship with money.
It is certainly worth considering whether the people who can see you betting understand what it is you are doing—that you are trying to get people to make clear predictions instead of to gamble over others’ misfortunes. Good point.
Agreed. Trying to trick people into thinking we don’t bet is bad. Trying to trick people in ways that directly compromise our epistemics (insofar as betting is super useful for epistemics) is even worse. This is an awful idea.
See my comment from 2d ago. I personally find the betting thing obnoxious and alienating, and don’t believe it actually contributes to truth seeking more than it harms (by causing a strong selection effect).
The betting thing is annoying, yeah. It’s not actually a great way to get at the truth either. Either the bet is for trivial sums, in which case it’s hard to take it seriously, or it’s for real money but then both parties have a serious incentive to try to set the most unfavourable odds they can for the other party. I was challenged a few days ago on this very forum to bet on whether or not I thought FTX had committed fraud by transferring customer funds to Alameda (I thought they had, some people wanted to take the other side—turns out I was right but hey!).
I responded that we could bet £10k at evens, to which someone replied “I thought you were more confident than evens”. Well, yes, if we’re going to bet for proper money of course I’m going to try to negotiate the most terrible odds for you I possibly can, why in god’s name why I would name a fair price?????
I think the dynamic you describe where betting for real money “incentives [people] to try to set the most unfavourable odds they can for the other party” is correct, but also good. Two rational actors will only engage in a bet if they believe it is +EV for each of them. In a zero sun bet, this is only possible if each actor assigns sufficiently different probabilities to the event in question. So actually in the case you described about the £10k bet, useful information was conveyed. You demonstrated that you were not sufficiently confident in your claims to offer better odds (ie. you didn’t think offering better odds would be +EV for you), which is fine. Obviously if you aren’t a frequent bettor, you have to be more worried about variance (getting unlucky and losing a bunch of money on a good bet that you’ll never recoup), but in general bets are actually good at teasing out someone’s probability distribution for an event occurring, and the dynamic you describe is an important part of that.
well obviously I could have offered way better odds (probably up to 50-1!) and it would have been +EV for me still, but I didn’t because I wanted to get the best possible price and assumed my opponent might engage in some negotiations (which they did not). So no, no useful information was conveyed other that I though there was a small chance my opponent on the other side of this hypothetical bet might have been dumb enough to take it at that price.
FWIW I think that people (such as myself) who were not willing to bet did so because by the time we came back to the bet, we had already updating to you being more correct. So the bet offer did transmit meaningful information, namely that you were correctly confident before the crowd.
I strongly disagree with this. I agree that there are competing considerations (weirdness, unusual dynamics when people are losing money), but I think that the factors pointing to the other side (betting cuts through a lot of bullshit and might be particularly informative in times of crises, bets allow a stronger signal of belief than speech, bets reward those who are correct) still trump them.
Also,
is incorrect, and in fact bets allow people to express shades of uncertainty by changing the odds and the betting amount.
I’m not surprised that you would strongly disagree with this. Sorry if this is a bit blunt, but given that you very recently launched a public forecast on the love life of a community member, without their permission, to the dismay of many people on Twitter, I wouldn’t think you have a particularly calibrated frame of reference as to when we should hold off on bets.
As you might imagine, I also disagree about whether that market was per-se bad. I think that there was a small information cascade about that particular topic because people in the EA community didn’t want to appear mean, & so better epistemic infrastructure would have actually helped.
I ended up removing that market because I got too much pushback on it, including from my boss, but Twitter wasn’t actually up to to hosting a nuanced discussion on the pros and cons, just social opprobrium.
I also strongly disagree with the OP. Giving up betting as an epistemic practice ought to be a non-starter, similar to “let’s take a month-long break from assigning probabilities to things”.
Where did I say anything about “giving up betting as an epistemic practice”? My post specifically listed why I know it’s good epistemic practice.
All I said was that during a crisis that involves the potential loss of money of 100,000 people, constantly suggesting 4-digit bets on public online spaces as soon as two people have the slightest disagreement may not be the most empathetic thing to do.
Thanks for this. I’ll add that I think even in less tense times, there are alternatives that achieve similar clarity while avoiding community culture downsides. Money is a fraught topic, EA has at times been accused of being classist, and people should be able to engage in EA discourse regardless of their relationship with money.
It is certainly worth considering whether the people who can see you betting understand what it is you are doing—that you are trying to get people to make clear predictions instead of to gamble over others’ misfortunes. Good point.
Guys, what did we just learn about deontological rules? In this case honesty.
“Do what you think is ethical, don’t be constantly changing your behavior for PR concerns” is going to work out to better PR.
This is not PR. It’s behaviour norms inside the community.
It’s proposing a behavior norm on the basis of PR. (Specifically, “the optics”.)
Agreed. Trying to trick people into thinking we don’t bet is bad. Trying to trick people in ways that directly compromise our epistemics (insofar as betting is super useful for epistemics) is even worse. This is an awful idea.
I think one can express real concern and empathy for this and bet play money / small amounts.
See my comment from 2d ago. I personally find the betting thing obnoxious and alienating, and don’t believe it actually contributes to truth seeking more than it harms (by causing a strong selection effect).
The betting thing is annoying, yeah. It’s not actually a great way to get at the truth either. Either the bet is for trivial sums, in which case it’s hard to take it seriously, or it’s for real money but then both parties have a serious incentive to try to set the most unfavourable odds they can for the other party. I was challenged a few days ago on this very forum to bet on whether or not I thought FTX had committed fraud by transferring customer funds to Alameda (I thought they had, some people wanted to take the other side—turns out I was right but hey!).
I responded that we could bet £10k at evens, to which someone replied “I thought you were more confident than evens”. Well, yes, if we’re going to bet for proper money of course I’m going to try to negotiate the most terrible odds for you I possibly can, why in god’s name why I would name a fair price?????
I think the dynamic you describe where betting for real money “incentives [people] to try to set the most unfavourable odds they can for the other party” is correct, but also good. Two rational actors will only engage in a bet if they believe it is +EV for each of them. In a zero sun bet, this is only possible if each actor assigns sufficiently different probabilities to the event in question. So actually in the case you described about the £10k bet, useful information was conveyed. You demonstrated that you were not sufficiently confident in your claims to offer better odds (ie. you didn’t think offering better odds would be +EV for you), which is fine. Obviously if you aren’t a frequent bettor, you have to be more worried about variance (getting unlucky and losing a bunch of money on a good bet that you’ll never recoup), but in general bets are actually good at teasing out someone’s probability distribution for an event occurring, and the dynamic you describe is an important part of that.
well obviously I could have offered way better odds (probably up to 50-1!) and it would have been +EV for me still, but I didn’t because I wanted to get the best possible price and assumed my opponent might engage in some negotiations (which they did not). So no, no useful information was conveyed other that I though there was a small chance my opponent on the other side of this hypothetical bet might have been dumb enough to take it at that price.
FWIW I think that people (such as myself) who were not willing to bet did so because by the time we came back to the bet, we had already updating to you being more correct. So the bet offer did transmit meaningful information, namely that you were correctly confident before the crowd.
I disagree. I sense that most people who would really hate it don’t see it. And it’s a useful tool for pushing for confidence.
Also, in this specific case I’m not sure crypto people mind—they are all betting as well. The FTX markets on polymarket have done great. https://polymarket.com/market/will-ftx-become-insolvent-by-eoy