I think Astral Codex Ten had an article about this a while ago, although I can't find it now.
Suppose AGI is coming and it will lead to either 1) everyone dying suddenly 2) everyone becoming rich beyond anything they could imagine today.
In that world, you want to front load personal wealth as much as possible. Don't save for retirement, in fact take out as much loans as possible, etc. You are either partying at the end of the world or fronting yourself some of your future riches.
This is a sentiment I have encountered before, people proudly saying they aren't saving for retirement because AGI. Although coming up with thin rationalizations for not saving for retirement is a time-honored activity, so I am not sure how much of it can really be ascribed to principled AGI investing.
I'm not a finance person and don't know how the details here would work. But conceptually if no-one else realises that the world is going to end in the next 10 years, then you should borrow money on terms that require it be paid back with interest in (say) 20 years. That way you get extra enjoyment from our last years on earth, and you never have to pay the loan back.
I'm trying to understand what the point is of betting on a world that is going to break apart in a millisecond. Like let's say you were right and found the perfect instrument and made a lot of money by betting on the right thing, what purpose would that insane amount of wealth serve if everything ends, including your life.
It depends entirely on the delta between when others also think the world will end, and when the world actually ends. If that delta is a millisecond, then there’s no point. I write as much in Section III I think.
This is a fascinating piece, but I think the true bottleneck is sharper than you make it: there are no counterparties.
Burry could short the housing market because (1) subprime risk was already being mispriced and (2) there were institutions structurally incentivized to write the other side—banks flush with fees, AIG hungry for yield, etc.
With AI collapse, you face an untradeable proposition:
--No resolution mechanism that defines “doom”
--No natural long-AI-survival players who need to hedge
--No institutions structurally compelled to manufacture these instruments
--And no counterparties left to pay if you’re “right”
Catastrophe bonds and CDS worked because their risks were diversifiable, time-resolved, and legally adjudicable. AI doom is systemic, unfalsifiable, and terminal. You can’t bet on extinction any more than you can buy insurance on the heat death of the universe.
There are counterparties that would probably be willing to do something like the following:
[CONTRACT PSEUDOCODE]
I, Rapa-Nui, give you $1,000,000 now.
You, Doomer, give me $50,000,000 on June 7th 2035, collateralized by E. Yudkowsky and Z. Mowshowitz , legally enforceable by the United States and escrowed by Merryl Lynch (or whatever)
[/CONTRACT PSEUDOCODE]
You can "win" that bet when everyone drops dead on June 6th. You get $1,000,000 free USD, and I get nothing. There are version of this where you might pay back small chunks which exponentially approach 50 million as Doom Day approaches. So, on June 5th, I might get a 30 million dollar payout from the obligated parties, but I will not get to enjoy it. The REAL worth of those USD is basically zero despite the fact that it looks like I made the best deal ever right before I drop dead.
This doesn't seem like a trade the doomer should take. By making the trade, the doomer is now on the hook for 50 million, which is held as collateral and cannot be spent.
Presumably, the doomer would rather spend the 50 million than the 1 million from the bet.
I used "collateral" incorrectly in my example. Something like "underwriting" would have been a more appropriate term to use.
The two people underwriting me would still be free to use their capital as they see fit, presumably agreeing with me that they will never be on the hook for it because they will die first.
I think Astral Codex Ten had an article about this a while ago, although I can't find it now.
Suppose AGI is coming and it will lead to either 1) everyone dying suddenly 2) everyone becoming rich beyond anything they could imagine today.
In that world, you want to front load personal wealth as much as possible. Don't save for retirement, in fact take out as much loans as possible, etc. You are either partying at the end of the world or fronting yourself some of your future riches.
This is a sentiment I have encountered before, people proudly saying they aren't saving for retirement because AGI. Although coming up with thin rationalizations for not saving for retirement is a time-honored activity, so I am not sure how much of it can really be ascribed to principled AGI investing.
Brilliant note Rohit . The scenario looks so realistic specially the structure bet .
Why bet on the end of the world?
Who is going to pay out?
How are you going to get and spend the money?
The only sensible thing to do is to try to stop it, isn't it?
Or have I missed something?
Because in between the payout and the actual end of the world there can be quite a chunk of time.
which agi are you using?
I'm not a finance person and don't know how the details here would work. But conceptually if no-one else realises that the world is going to end in the next 10 years, then you should borrow money on terms that require it be paid back with interest in (say) 20 years. That way you get extra enjoyment from our last years on earth, and you never have to pay the loan back.
Agree. The problem here is that the timeframe (20 years) is also unknown. So you can’t get easy large loans. Which makes this complicated.
But yes there are ways around it, which is why I (cheekily) came up with as PADS.
I'm trying to understand what the point is of betting on a world that is going to break apart in a millisecond. Like let's say you were right and found the perfect instrument and made a lot of money by betting on the right thing, what purpose would that insane amount of wealth serve if everything ends, including your life.
It depends entirely on the delta between when others also think the world will end, and when the world actually ends. If that delta is a millisecond, then there’s no point. I write as much in Section III I think.
This is a fascinating piece, but I think the true bottleneck is sharper than you make it: there are no counterparties.
Burry could short the housing market because (1) subprime risk was already being mispriced and (2) there were institutions structurally incentivized to write the other side—banks flush with fees, AIG hungry for yield, etc.
With AI collapse, you face an untradeable proposition:
--No resolution mechanism that defines “doom”
--No natural long-AI-survival players who need to hedge
--No institutions structurally compelled to manufacture these instruments
--And no counterparties left to pay if you’re “right”
Catastrophe bonds and CDS worked because their risks were diversifiable, time-resolved, and legally adjudicable. AI doom is systemic, unfalsifiable, and terminal. You can’t bet on extinction any more than you can buy insurance on the heat death of the universe.
Like I said, you could walk into Goldman and ask for that particular insurance to be priced for you. I think it'll be expensive
There are counterparties that would probably be willing to do something like the following:
[CONTRACT PSEUDOCODE]
I, Rapa-Nui, give you $1,000,000 now.
You, Doomer, give me $50,000,000 on June 7th 2035, collateralized by E. Yudkowsky and Z. Mowshowitz , legally enforceable by the United States and escrowed by Merryl Lynch (or whatever)
[/CONTRACT PSEUDOCODE]
You can "win" that bet when everyone drops dead on June 6th. You get $1,000,000 free USD, and I get nothing. There are version of this where you might pay back small chunks which exponentially approach 50 million as Doom Day approaches. So, on June 5th, I might get a 30 million dollar payout from the obligated parties, but I will not get to enjoy it. The REAL worth of those USD is basically zero despite the fact that it looks like I made the best deal ever right before I drop dead.
I don’t follow. If everyone dies how would the contract be settled?
I write the contract with the belief that no one is going to die, and I am going to get to spend my 50 million.
As with all market transactions, if doom happens, I am punished for my incorrect belief.
This doesn't seem like a trade the doomer should take. By making the trade, the doomer is now on the hook for 50 million, which is held as collateral and cannot be spent.
Presumably, the doomer would rather spend the 50 million than the 1 million from the bet.
I used "collateral" incorrectly in my example. Something like "underwriting" would have been a more appropriate term to use.
The two people underwriting me would still be free to use their capital as they see fit, presumably agreeing with me that they will never be on the hook for it because they will die first.
I see your point.
However the end may come much sooner than many think.
Surely it would be better for us to address the potential ending of the world?
If, then yes. That would be the "bet with reputation and action not money" part of the essay too.