From: Jef Allbright (jef@jefallbright.net)
Date: Fri Jan 27 2006 - 16:08:05 MST
On 1/27/06, Damien Broderick <thespike@satx.rr.com> wrote:
>
> At 11:30 AM 1/27/2006 -0800, Eliezer S. Yudkowsky wrote:
> >H C wrote:
> >>4. Given a coin is flipped 10 times
> >> HHHHHHHHHH
> >>What would you guess your odds of hitting tails are next round?
> >
> >...I think it is legitimate to refuse to answer questions predicated on
> >large improbabilities unless and until the scenario comes up in real
> life.
>
> But that's going to happen roughly once in every 1000 such sequences, so
> it's by no means impossibly rare. (This, after all, is one of your/the
> common arguments against the proffered evidence for psi.)
>
> >Given that a fair coin is flipped 50 times and comes up
> > HHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH
> >what would you guess as your odds of getting tails next round?
> >
> >The answer, of course is:
> > On the given hypothesis, this scenario ain't gonna happen.
>
> Okay, but that wasn't the question posed.
>
>
> I have a lot of appreciation for someone refusing to play highly
improbable "what if" games unless all parties involved are seriously
interested in pursuing some specific objective. For example, I've had
multiple experiences with my teenagers wanting to argue "what ifs" that
weren't congruent with my perception of reality, but where no one really
intended to invest the time and effort to provide sufficient specifications
on the discussion such that it could be meaningful. Without careful
specification, "what ifs" can be used to justify anything.
In the larger context of the real world, the response to the highly
improbable string of Heads would involve a significant effort in search of
experimental error or trickery. In the simple context of the test question,
the answer is simply that history has no effect on subsequent trials and the
expected probability would be 50%.
Meaning is always dependent on context.
- Jef
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:55 MDT