The #SBF (#FTX) conundrum might be solvable by throwing these propositions into the mix:
He simply computed the probability of getting away with financial engineering and deception times the potential increase in well-being (by tossing billions at #EffectiveAltruism causes), and that seemed to him higher than the odds of being caught times {investors and customers’ funds lost plus the huge reputational damage that would inflict to the #EA cause}.
So he pressed the red button and bet the world. And he lost.
It’s not trivial to find the flaw in his reasoning, though.
I think the simplest version of that thought experiment is one where it’s a one-off bet (double-or-nothing this universe, once). I guess the math supports saying yes to that, if the odds of winning are > .5, right?
My intuition for the reason to say no to the bet is that there’s a qualitative difference between existing and not existing.
If I were offered a .51 chance of transforming this universe into one where The Beatles had recorded 20 more songs, vs one where we had never known 20 songs by them, I would say yes. That works for me regardless of the no. of songs. But when it gets to “twice as many songs” vs “no songs at all” (ie, The Beatles never existed for all practical purposes), something changes, and I would reject the bet.
…I think 🙂
@tripu Well, with a one-off bet, it would simply be the question “Do you want to win something more than you want to not loose everything.”
The rest of this you have taken in another direction that i also agree with. “Win a little or lose a little, that’s ok” is something i would also say.
But just the idea that there are consecutive all-or-nothing bets that include betting the base that was there before just does not make sense mathematically. No matter how much one values anything, if it is ok to lose it, you have no expected value if step n takes everything you ever won.
I’m not good at formal math, but i would guess you had to lower the expected value of step n-1 if you do a step n by the probability to loose everything in step n.
Like in step 1, you expect 0.51 * 2 = 1.02 value. But if you actually do two steps, your expectation of getting zero is now 0.49 + (0.49^2). Am i doing this right? It should slowly approach 1 for expecting to get nothing at all.
Still onboard with the “scam scammers to donate to a good cause” thing, though.