[extropy-chat] silent night

[Eliezer Yudkowsky]

Hal Finney wrote:
> Robin wrote:
> : I predict the sun will come up tomorrow.  I will keep making this 
> : prediction everyday for the next year.  Go ahead, collect stats on 
> : how well I do.  Betcha I'm right more often than not.
> 
> Eliezer wrote:
> 
>>What exact probability do you assign?  Now you've got me curious.
> 
> Robin wrote:
> : Well for the purposes of the above discussion all I needed was p/(1-p) >> 1
> : But if I have to pick a number I'd say p/(1-p) ~ 10^4 to 10^5
> 
> Eliezer wrote:
> 
>>Interesting.  That was just about exactly my own answer - 99.98% to 
>>99.999%.  Though as we may only make a single bet to sum up our 
>>uncertainty, I would say 99.99%.
> 
> Okay, so let me offer this bet to Eliezer: my $10,000 to your $1,
> that the sun will rise tomorrow.  Here is how I propose to do it.
> I will send you a check for $10,000.  I will trust you not to cash it
> unless the sun doesn't rise tomorrow.  If the sun does rise as usual,
> you will send me back the uncashed check along with a one dollar bill.
> 
> Or, if you want, you can decide to continue the bet each day and then,
> once a week, you can send me $7 if the sun has continued to rise on
> schedule.  The series of bets can be terminated by either party.

A problem is that the probability of my surviving a win, is significantly 
lower than the probability of a win.  Furthermore, if I survive winning 
this bet, the money has less utility to my purpose afterward.

-- 
Eliezer S. Yudkowsky                          http://singinst.org/
Research Fellow, Singularity Institute for Artificial Intelligence