[computer-go] an idea for a new measure of a computer go program's rank.

[Don Dailey]

On Thu, 2007-01-18 at 20:05 -0800, Ray Tayek wrote:
> 
> yes. i would easily give my opponent *much* more time than a few 
> handicap stones. the effect of time making someone (or thing) play 
> better (or worse) is non-linear and probably only effective over some 
> small range of time and talent.

I think the formula is probably similar to UCT or Chess, but even more
so for humans.   Double the amount of time you have, and significantly
increase the quality of the move.  I don't think this is a limited 
effect over a narrow range of time.

I understand chess better than go, I used to be a tournament player.    
Give me time to think and I can produce moves
of enormously higher quality over tournament time-controls.  I know 
this for a fact.   I seriously doubt it is different for go.

There is one human limitation however - one problem with humans is 
interest - can you stay interested in a game or a move over really 
long periods of time - probably not.   If you could stay 100% focused
like you might be in a tournament game and time could be compressed 
for you - you would  play hundreds of ELO points stronger.   This is
still not nearly enough to overcome someone more than a few hundred
ELO points stronger - that's too much ground to cover.  

It probably is non-linear like you say - even in the more limited game 
of Chess, the curve was amazingly linear (every doubling in time seemed
 to give a fixed amount
of ELO strength improvement)  but this has fallen off in recent years.
Still, the curve is strong and computers continue to climb at pretty
fast rates
as they keep getting faster.    

As far as talent is concerned, some chess experiments seem to indicate
that
programs with more sophisticated evaluation functions (more talent) seem
to improve faster with time.   This was a surprise to me but I believe
it to
be true. 

I think it might work the same with humans - they have more to work with
as
it were.   Show a raw beginner a chess game or a go game, and they don't
know
how to use extra time thinking about it.    This intuition is backed up
by
what happens with world championship games - Grandmasters bring teams of
analysts with them to study the adjourned position in a game while they 
sleep.    They wake up the next morning and get briefed on what was
learned.   Sometimes both sides come to the table and a draw is
immediately
agreed to or a resignation, since both sides have done the analysis.
Parallel processing at it's finest!

So I think strength in humans is very much the same - perhaps even more 
scalable than with computers - subject of course to human frailties of
attention span, sleep time, ability to focus for long periods of time,
etc.


- Don