> >> >Spike Jones wrote: not signed up for cryonics yet, estimate chances
> > > of living to see singularity ~60%. spike
> >
> >Eugene.Leitl wrote: I don't see how you can attach a meaningful number
> >to the probability right now, until you assume a fixed date (and
> > that it doesn't kill you, of course).
>
> "Eliezer S. Yudkowsky" wrote:You don't need a fixed date to generate a
> probability, just as your insurance company doesn't need to know the date
> of your death to set a premium.
Wow, cooool Im glad this discussion went down this path. In the
past several months I have been toying with probability distribution
functions to make estimates of the probability of finding a new
record prime number in any given time span. {Shall I change
the subject line?}
Eliezer is right: to estimate a probability for living to see the
singularity, one need not first estimate a date, then estimate
a probability of living that long. Of course Gene, I did not
do that estimate scientifically: I don't understand enough
about AI to do that. However, I do have some comments regarding
probability distributions WRT the next record breaking prime,
which I did calculate.
Turns out, it is easy to show that the next record prime, and all
record primes henceforth for all time will be Mersenne primes,
that class of primes which are of the form 2^n-1. The organized
effort to search Mersenne numbers {GIMPS} has four probability
functions that must be modeled: 1) a linearly increasing number
of participants, 2) the compute capability of the computers is
increasing exponentially, 3) the probability of each mersenne
number being prime is decreasing logarithmically, and 4) the
number of compute cycles required to analyze each candidate
is increasing exponentially.
Each of these factors can be expressed as a probability distribution
function, and the four PDFs can be superimposed to find that the
probability of finding a new record prime on any given day is...
decreasing linearly.
Another way of stating this is that the inverse of the probability of
a new prime on any given day is increasing linearly. For instance,
on 1 June 1999 {the day after the current record holder was
discovered} the probability of discovering a prime on that day
could have been calculated at about 1 in 160. The probability of
a record breaker being discovered tomorrow is about 1 in 640.
Record primes were discovered in 1996, 1997, 1998 and 1999.
The Mersenne Prime list was discussing yesterday that 2000 is
the first year since the GIMPS effort was organized that the record
was not smashed. I had to inform them that according to my
probability distribution analysis, there is about a 63% chance
that we will go hungry in 2001 as well. {8-[ dammit.
> I think my grandparents are going to make it.
I sincerely hope they do, Eleizer. {8-] spike
This archive was generated by hypermail 2b30 : Mon May 28 2001 - 09:56:16 MDT