[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: computer-go: Authenticating the identity of a remote go-playing computer program



> I'm not claiming that a "deterministic" program can't do simulated
> annealing and machine learning and so forth. (I vaguely remember that
> someone else may have claimed that, but I definitely remember that it
> wasn't me.:-) I did my Ph. D. thesis on Monte Carlo simulations of
> quantum mechanics, and I learned a thing or two about PRNGs along the
> way. I even went to some trouble to make my Monte Carlo programs as
> deterministic as possible, in part for debugging. My argument here is
> only about the difficulty of making a program's time consumption
> deterministic.

I didn't  say YOU claimed  this.  In  fact I  assumed you knew better,
because you seem to at least understand  the issues.  You seem bent on
trying to prove to me that programs written in a non-deterministic way
are non-deterministic,  which  isn't very useful.   All I  am claiming
that you can write a strong deterministic program that is not severely
handicapped.  You are not making any points I don't disagree with.


> The point I was trying to make is that if you force US to use your
> time management techniques, WE'll be unhappy. 

Ok, you  will be unhappy,  I believe you.   But it's been  pretty much
agreed  that people will  be unhappy  to  modify their programs to  be
deterministic, that is not at issue.

>                                                 I hope to convince you
> that while ignoring the real time clock may work well for your
> programs, it does not necessary work well for other people's programs.

Ok,  so now you   have identified the  single  point worthy  of debate
because this  is where we disagree.   I maintain that you don't ignore
the concept of TIME, you  just have to adjust your  point of view that
you   somehow just HAVE to  use  a non-deterministic  measurment of it
because it makes you unhappy to do otherwise.  You seem to accept that
it's ok to use deterministic random numbers (psuedo-random), why can't
you  use  a deterministic  measure   of time?   Because  it  makes you
unhappy?   

Don