In February 2020, I made many bets on Covid19, including 11 bets at ten to one odds on if it would cause 10 million deaths worldwide by 2022, as estimated by WHO. WHO has a Q&A page on Covid excess deaths that includes this section:
That is plausible, though that's close to a consequential argument that your bad hoping will induce bad actions, which is the usual "sabotage" concern.
Have been thinking about the criticism you received for this and I wonder whether the following argument might be at the back of peoples' heads when they think about such bets:
(1) If you bet money that p will occur, you create an incentive to hope that p will occur.(2) If p is bad, you will be hoping for a bad thing.(3) Hoping for bad things to happen makes you a bad person.
Therefore, you should not bet that bad things will occur, because you should not create incentives for yourself to become bad.
What are the lessons we should take away from this? How should people change their processes for coming up with estimates for this kind of event? Is this a surprisingly large or small number of people to bet with you? What do you think the correct odds were?
RH: Ok, I'll put my $1K against your $10K. If WHO issues low, high numbers, instead of a middle best estimate, are you wiling to use geometric mean of those two numbers?Steven Spencer
SS: so this is based on their predictions and not the health data collected through the end of 2021? either way i would be fine w/ the geometric mean.
Based on this it seems clear that the WHO was agreed on as independent trusted source. I think Robin was already generous by taking the geometric mean of 5.4 and 18.5 and not of 11.6 and 21.6 (which would be 15.8).
Depends on context and how it's worded. For instance, if you ask how many people did the Germans cause to die in WWI should I include half the ppl who died of Spanish flu since, but for the German's war effort, there wouldn't have been a war and the pandemic would have been much better managed and not spread by war efforts. Indeed, this would mean that the English also caused every war death and many of those flu deaths since, but for their choice to fight there also would have been no war.
Sometimes we are asking for the number of ppl who a group is morally responsible for killing, sometimes the number they were the proximate cause, sometimes we mean the delta in number of deaths caused by one scenario relative to another. Point is that without further context or precisification it's ambiguous.
More importantly, the fact that a bunch of us here wouldn't understand the bets the same way you do seems like pretty incontrovertible evidence that it's at least ambiguous what deaths count as caused by COVID.
Of course, the context surrounding the bet could potentially resolve that ambiguity in one direction or the other but from what you've said here it's ambiguous.
Maybe you count foreseeable deaths from the investigation but if it turned out that, because of the investigation, a cop needed to switch his day off and thus caught his wife having an affair which drove him to kill her we wouldn't normally include that death in the deaths caused by the serial killer.
Sure, literally speaking the serial killer caused that death but it's not how humans normally talk. Indeed, I can easily derive really counterintuitive results if you interpret cause literally: eg that Gutenberg (or whoever you think was a cause of the industrial revolution) caused billions of deaths since, absent industrial revolution, there would be billions less people born and hence billions less deaths.
Indeed, if you start getting super strict about it you could argue that every death since COVID became a worldwide since it no doubt is part of the causal chain leading to every death (without it ppl would have had more/less exercise and had their heart attack at a slightly different time). Maybe you try to fix this by saying you only mean the net difference in deaths between the COVID and non-COVID world's but that's now uncheckable (for all we know without COVID there would have been a nuclear war).
Ok, so it's not literally being part of the causal chain and I think it's clear that you wouldn't win the bet if Xi had decided to start fighting Covid by executing anyone who came down with the virus.
I think this makes it clear that when we talk about how many deaths COVID caused we don't mean what a really literal reading of the words suggests so you have to look at how normal competent English speakers would understand the agreement.
I think it would also satisfy moral symmetries if Robin merely endorsed other people posting publicly when they believed themselves to have won bets against him. I am not sure what you mean by "rude" here. Mostly I don't see people use that term to describe immoral behavior, but confrontational behavior against power gradients. Robin's behavior may well be rude, but so much the better for rudeness.
Unaccountable "cost-benefit analysis" is bullshit, determined by (often coalitional) extractive motives, not by the costs and benefits it purports to measure. Bets are an accountability mechanism.
That's one way to do it. I'm making a general claim that a decisionmaking process without accountability selects for extractive behavior, because in the long run you don't see behaviors that don't pay for themselves. Bets are one way to fix this, and if disparagement of bets does not involve proposing a plausibly superior alternative, then the simplest interpretation is that it's just disparagement of accountability itself.
It's confusing why someone would feel the need to signal immorality like that. As far as I can tell it's a convergent behavior for coalitional actors. Coalitions tend to:- Signal loyalty through synchronized blind spots, and thus synchronized hostility against people challenging their blind spots.- Try to induct others into the coalition.
The common interest all coalitional cognition feels, relative to noncoalitional cognition, is an interest against mechanisms for accountability. You can think of this as a kind of metacoalition of coalitional players.
So political actors will sneer at Robin when they notice him being conspicuously and accountably helpful, trying to invalidate that behavior to him and others, in order to induct more people into generalized opposition to accountability.
That is plausible, though that's close to a consequential argument that your bad hoping will induce bad actions, which is the usual "sabotage" concern.
Have been thinking about the criticism you received for this and I wonder whether the following argument might be at the back of peoples' heads when they think about such bets:
(1) If you bet money that p will occur, you create an incentive to hope that p will occur.(2) If p is bad, you will be hoping for a bad thing.(3) Hoping for bad things to happen makes you a bad person.
Therefore, you should not bet that bad things will occur, because you should not create incentives for yourself to become bad.
What are the lessons we should take away from this? How should people change their processes for coming up with estimates for this kind of event? Is this a surprisingly large or small number of people to bet with you? What do you think the correct odds were?
I hope those who bet will pay up in good faith, as I'm sure Robin would have.
The relevantz tweets are
RH: Ok, I'll put my $1K against your $10K. If WHO issues low, high numbers, instead of a middle best estimate, are you wiling to use geometric mean of those two numbers?Steven Spencer
SS: so this is based on their predictions and not the health data collected through the end of 2021? either way i would be fine w/ the geometric mean.
Based on this it seems clear that the WHO was agreed on as independent trusted source. I think Robin was already generous by taking the geometric mean of 5.4 and 18.5 and not of 11.6 and 21.6 (which would be 15.8).
Yes, what I really should have said is: it's not rude as long as you don't systematically bias posting to favor cases you won.
Tho, thinking about it more, even that might just be merely unpleasant rather thsn rude.
Depends on context and how it's worded. For instance, if you ask how many people did the Germans cause to die in WWI should I include half the ppl who died of Spanish flu since, but for the German's war effort, there wouldn't have been a war and the pandemic would have been much better managed and not spread by war efforts. Indeed, this would mean that the English also caused every war death and many of those flu deaths since, but for their choice to fight there also would have been no war.
Sometimes we are asking for the number of ppl who a group is morally responsible for killing, sometimes the number they were the proximate cause, sometimes we mean the delta in number of deaths caused by one scenario relative to another. Point is that without further context or precisification it's ambiguous.
More importantly, the fact that a bunch of us here wouldn't understand the bets the same way you do seems like pretty incontrovertible evidence that it's at least ambiguous what deaths count as caused by COVID.
Of course, the context surrounding the bet could potentially resolve that ambiguity in one direction or the other but from what you've said here it's ambiguous.
Maybe you count foreseeable deaths from the investigation but if it turned out that, because of the investigation, a cop needed to switch his day off and thus caught his wife having an affair which drove him to kill her we wouldn't normally include that death in the deaths caused by the serial killer.
Sure, literally speaking the serial killer caused that death but it's not how humans normally talk. Indeed, I can easily derive really counterintuitive results if you interpret cause literally: eg that Gutenberg (or whoever you think was a cause of the industrial revolution) caused billions of deaths since, absent industrial revolution, there would be billions less people born and hence billions less deaths.
Indeed, if you start getting super strict about it you could argue that every death since COVID became a worldwide since it no doubt is part of the causal chain leading to every death (without it ppl would have had more/less exercise and had their heart attack at a slightly different time). Maybe you try to fix this by saying you only mean the net difference in deaths between the COVID and non-COVID world's but that's now uncheckable (for all we know without COVID there would have been a nuclear war).
Ok, so it's not literally being part of the causal chain and I think it's clear that you wouldn't win the bet if Xi had decided to start fighting Covid by executing anyone who came down with the virus.
I think this makes it clear that when we talk about how many deaths COVID caused we don't mean what a really literal reading of the words suggests so you have to look at how normal competent English speakers would understand the agreement.
Your philosophy on causation may be sensible, but it isn’t the only one, and it isn’t spelled out in the terms of the bet.
I think it would also satisfy moral symmetries if Robin merely endorsed other people posting publicly when they believed themselves to have won bets against him. I am not sure what you mean by "rude" here. Mostly I don't see people use that term to describe immoral behavior, but confrontational behavior against power gradients. Robin's behavior may well be rude, but so much the better for rudeness.
Unaccountable "cost-benefit analysis" is bullshit, determined by (often coalitional) extractive motives, not by the costs and benefits it purports to measure. Bets are an accountability mechanism.
That's one way to do it. I'm making a general claim that a decisionmaking process without accountability selects for extractive behavior, because in the long run you don't see behaviors that don't pay for themselves. Bets are one way to fix this, and if disparagement of bets does not involve proposing a plausibly superior alternative, then the simplest interpretation is that it's just disparagement of accountability itself.
It's confusing why someone would feel the need to signal immorality like that. As far as I can tell it's a convergent behavior for coalitional actors. Coalitions tend to:- Signal loyalty through synchronized blind spots, and thus synchronized hostility against people challenging their blind spots.- Try to induct others into the coalition.
The common interest all coalitional cognition feels, relative to noncoalitional cognition, is an interest against mechanisms for accountability. You can think of this as a kind of metacoalition of coalitional players.
So political actors will sneer at Robin when they notice him being conspicuously and accountably helpful, trying to invalidate that behavior to him and others, in order to induct more people into generalized opposition to accountability.
People who died as a result of efforts to catch a serial killer are indeed deaths caused by that serial killer.
I don't blog on all my bets, but when I make 11 bets on the same thing, then yes I'd blog on it either way.
I was too harsh. I edited that from my comment.