[computer-go] Positions illustrative of computer stupidity ?

[Chrilly]

> In message <002801c70ecd$e0ab3670$7119da52 at Hansi>, Chrilly
> <c.donninger at wavenet.at> writes
>>The attached position shows the "Kirtag Problem". I have named it after 
>>the Austrian proverb "Man kann nur auf einem Kirtag tanzen" (One can dance 
>>only on one village-party***).
>>More mathematically it is the subgame problem. White to move, what is the 
>>status of the white marked stones. Treated locally, each stone can be 
>>easily saved, but alas, one can dance only on one party.
>
> So, the status of each of those two stones is "unsettled".  Not much of a 
> problem.
>
Oh yes it is one. At least if one writes an Alpha-Beta Searcher and has to 
make a global-board-eval. What is the numeric value of "unsettled?". One can 
argue that in this case the position is not quiete and one should continue 
the search. But in fact there are always some unsettled stones and search 
would explode. So one has to evaluate  "unsettled". An Alpha-Beta search has 
furthermore the tendency to create unsettled positions. Tactically 
interesting moves are given higher priority.
One could argue, if Alpha-Beta has a problem with it, thats the problem of 
Alpha-Beta and not of computer-Go. The method is not suited for Go. But 
thats the real problem of computer-Go. In other games the biggest 
advancements are done by the INTEL/AMD engineers and not by the 
game-programmers. With an infinite fast chip chess programms would be 
"infinite" strong. Most current Go programms would only play infinite fast. 
Its an interesting question if Monte-Carlo programms would also play 
infinite strong.

Chrilly