Comment by Eric Rowles (eric-rowles) on Yet more UFO Betting: Put Up or Shut Up · 2023-08-09T07:08:04.330Z · LW · GW

When making bets with extreme odds, I think the precision of your estimate becomes a lot more important.

To use a simple model, if I am 10% off with my precision when I think an event has 50% odds, the actual odds might be somewhere between 40-60%, which is fine. But if I think an event has 99.4% odds (the ratio you propose,) the actual odds are in the range of 89.4-100%. (Putting my actual estimate at 94.7% - not so good!)

In order to even get up to actual odds of 99.4%, even if I believed with complete certainty, my belief would need to have a precision within 1.2% just to get even odds. I have a hard time justifying that sort of precision when predicting novel future events, no matter how unlikely I think they are.

Now, I would be willing to bet with lower odds. For example, I'd bet at 20:1 odds that no such evidence appears within 22.5 years.