[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



As we have seen, none of this is unsurmountable.

Don


   From: Heikki Levanto <heikki@xxxxxxxxxxxxxxxxx>

   Steve Pagliarulo <s_pagliarulo@xxxxxxxxxxxxxxxxx> wrote:

   > 1. For tournament purposes don't allow random play. Allow pseudo random play 
   > based on an initial seed (to avoid predictability). For verification, the 
   > seed is disclosed and the program will play identically.

   As someone pointed out, careful time management can easily produce
   non-repeatable play, even without randomness.

This one is trivial to fix without disallowing time management.

A much more difficult problem (however also solvable) is programs that
do calculations during their opponents think  cycle.  This changes the
starting state of the program which is a real pain to deal with if you
want to ensure deterministic behavior.

   If detailed time management has to be disallowed, what do we do about time
   limits? Some programs will play better if given more time, so we have to put
   a limit somewhere, or risk really long tournaments...

This is not a problem, but it makes the verification process time consuming.

Taken all together, we have all agreed that this stuff is not a very 
practical solution, however, it's possible to implement if we really
wanted to (which I don't.)