Obviously it is impossible to bet money on the end of the world.
I think this is neither obvious nor true. There are lots of variants you could do and details you’d need to fill in, but the outline of a simple one would be: “I pay you $X now, and if and when self-driving cars reach mass market without the world having ended, you pay me $Y inflation-adjusted”.
Robin Hanson said, with Eliezer eventually concurring, that “bets like this will just recover interest rates, which give the exchange rate between resources on one date and resources on another date.”
E.g., it’s not impossible to bet money on the end of the world, but it’s impossible to do it in a way substantially different from taking a loan.
Oh, thanks for the pointer. I confess I wish Robin was less terse here.
I’m not sure I even understand the claim, what does it mean to “recover interest rates”? Is Robin claiming any such bet will either
Have payoffs such that [the person receiving money now and paying money later] could just take out a loan at prevailing interest rates to make this bet; or
Have at least one party who is being silly with money?
...oh, I think I get it, and IIUC the idea that fails is different from what I was suggesting.
The idea that fails is that you can make a prediction market from these bets and use it to recover a probability of apocalypse. I agree that won’t work, for the reason given: prices of these bets will be about both [probability of apocalypse] and [the value of money-now versus money-later, conditional on no apocalypse], and you can’t separate those effects.
I don’t think this automatically sinks the simpler idea of: if Alice and Bob disagree about the probability of an apocalypse, they may be able to make a bet that both consider positive-expected-utility. And I don’t think that bet would necessarily just be a combination of available market-rate loans? At least it doesn’t look like anyone is claiming that.
Wow, that may be a genuinely ground breaking application for crypto currencies. e.g. someone with 1000 bitcoins can put them in some form of guaranteed, irreversible, escrow for a million dollars upfront and release date in 2050. If the world ends then the escrow vanishes, if not the lucky better would get it.
(Since you said “ground breaking”, I feel I should maybe be clear that the idea wasn’t original to me and the ground was already broken when I got here. I probably saw it on LW myself.)
Note that 1000 BTC at current prices is like $64 million. In general with this structure (trade crypto for fiat, plus the crypto gets locked until later) I don’t see the incentive on anyone’s side to lock the crypto?
I haven’t thought about this in depth and I assume some people have, but my sense is that these kinds of bets are a lot easier if both parties trust each other to be willing and able to pay up when appropriate. If you don’t trust your counterparty, and demand money in escrow, then their incentive to take part seems minimal or none.
The actual structure, and payout ratio, would probably be set in a much more elaborate way. Maybe some kind of annuity paying out every year the world hasn’t ended yet? Like commit 10 bitcoins every year from a reversible escrow account to the irreversible escrow if the servers still exist or the total balance will be forfeited. Something along those lines, perhaps others would want to take up the project.
Fwiw, and this might just be that you’ve thought about it more and I have and/or have details in mind that you haven’t specified. But it still seems to me that this runs into the problem that at least one party basically won’t have any reason to enter into such a bet, because their potential upside will be locked away long enough to negate its value.
I imagine that was one of the critiques for prediction markets using crypto currencies held in escrow, yet they exist now, and they’re not all scams, so there must be some, non zero, market clearing price.
I don’t think that holds up, because with traditional uses of prediction markets both parties expect to see the end of the bet. If it’s a long term bet then that would increase the expected payoff they require to be willing to bet, but there’s some amount of “money later” that’s worth giving up “money now” in exchange for. So both are willing to lock money up.
With a bet on the end of the world, all of the upside for one party comes from receiving “money now”. There’s no potential “money later” payoff for them, the bet isn’t structured that way. And putting money in escrow means their “money now” vanishes too.
That is, if I receive a million dollars now and pay out two million if the world doesn’t end, then: for now I’m up a million; if the world ends I’m dead; if the world doesn’t end I’m down a million. But if the two million has to go in escrow, the only change is that for now I’m down a million too. So I’m not gonna do this.
I think this is neither obvious nor true. There are lots of variants you could do and details you’d need to fill in, but the outline of a simple one would be: “I pay you $X now, and if and when self-driving cars reach mass market without the world having ended, you pay me $Y inflation-adjusted”.
Robin Hanson said, with Eliezer eventually concurring, that “bets like this will just recover interest rates, which give the exchange rate between resources on one date and resources on another date.”
E.g., it’s not impossible to bet money on the end of the world, but it’s impossible to do it in a way substantially different from taking a loan.
Oh, thanks for the pointer. I confess I wish Robin was less terse here.
I’m not sure I even understand the claim, what does it mean to “recover interest rates”? Is Robin claiming any such bet will either
Have payoffs such that [the person receiving money now and paying money later] could just take out a loan at prevailing interest rates to make this bet; or
Have at least one party who is being silly with money?
...oh, I think I get it, and IIUC the idea that fails is different from what I was suggesting.
The idea that fails is that you can make a prediction market from these bets and use it to recover a probability of apocalypse. I agree that won’t work, for the reason given: prices of these bets will be about both [probability of apocalypse] and [the value of money-now versus money-later, conditional on no apocalypse], and you can’t separate those effects.
I don’t think this automatically sinks the simpler idea of: if Alice and Bob disagree about the probability of an apocalypse, they may be able to make a bet that both consider positive-expected-utility. And I don’t think that bet would necessarily just be a combination of available market-rate loans? At least it doesn’t look like anyone is claiming that.
Wow, that may be a genuinely ground breaking application for crypto currencies. e.g. someone with 1000 bitcoins can put them in some form of guaranteed, irreversible, escrow for a million dollars upfront and release date in 2050. If the world ends then the escrow vanishes, if not the lucky better would get it.
(Since you said “ground breaking”, I feel I should maybe be clear that the idea wasn’t original to me and the ground was already broken when I got here. I probably saw it on LW myself.)
Note that 1000 BTC at current prices is like $64 million. In general with this structure (trade crypto for fiat, plus the crypto gets locked until later) I don’t see the incentive on anyone’s side to lock the crypto?
I haven’t thought about this in depth and I assume some people have, but my sense is that these kinds of bets are a lot easier if both parties trust each other to be willing and able to pay up when appropriate. If you don’t trust your counterparty, and demand money in escrow, then their incentive to take part seems minimal or none.
The actual structure, and payout ratio, would probably be set in a much more elaborate way. Maybe some kind of annuity paying out every year the world hasn’t ended yet? Like commit 10 bitcoins every year from a reversible escrow account to the irreversible escrow if the servers still exist or the total balance will be forfeited. Something along those lines, perhaps others would want to take up the project.
Fwiw, and this might just be that you’ve thought about it more and I have and/or have details in mind that you haven’t specified. But it still seems to me that this runs into the problem that at least one party basically won’t have any reason to enter into such a bet, because their potential upside will be locked away long enough to negate its value.
I imagine that was one of the critiques for prediction markets using crypto currencies held in escrow, yet they exist now, and they’re not all scams, so there must be some, non zero, market clearing price.
I don’t think that holds up, because with traditional uses of prediction markets both parties expect to see the end of the bet. If it’s a long term bet then that would increase the expected payoff they require to be willing to bet, but there’s some amount of “money later” that’s worth giving up “money now” in exchange for. So both are willing to lock money up.
With a bet on the end of the world, all of the upside for one party comes from receiving “money now”. There’s no potential “money later” payoff for them, the bet isn’t structured that way. And putting money in escrow means their “money now” vanishes too.
That is, if I receive a million dollars now and pay out two million if the world doesn’t end, then: for now I’m up a million; if the world ends I’m dead; if the world doesn’t end I’m down a million. But if the two million has to go in escrow, the only change is that for now I’m down a million too. So I’m not gonna do this.