Author: Peter McKenzie
Date: 13:15:04 02/13/04
Go up one level in this thread
On February 13, 2004 at 12:42:29, Roy Eassa wrote:
>On February 13, 2004 at 12:14:48, Tord Romstad wrote:
>
>>On February 13, 2004 at 11:51:27, Roy Eassa wrote:
>>
>>>On February 13, 2004 at 11:01:41, Tord Romstad wrote:
>>>
>>>
>>>>Go games between strong players of similar strengths are often decided
>>>>by very long, complicated forced lines where both players repeatedly
>>>>have to make deep and precise calculations in order to find the right
>>>>moves. This is what I meant when I claimed that Go games can be very
>>>>tactically complicated.
>>>>
>>>
>>>
>>>You've said that eloquently.
>>
>>Thanks!
>>
>>>I'd like to hear your description of the NON-tactical aspects of Go as
>>>played by professionals.
>>
>>I'm afraid I have to leave that question to stronger Go players than
>>me (are you there, Janosch?). I simply do not understand the strategical
>>aspects of Go well enough to understand what is going on in games between
>>professionals.
>>
>>Tord
>
>
>But having read several Go books, I think it's fair to say that it's steeped in
>almost-mystical language of "shape" and "thickness" and "influence" and many
>other exotic terms, and that there are scores of zen-like proverbs that purport
>to pertain to Go as well as to life.
I have taken up Go relatively recently, and have also dabbled in writing a Go
program so perhaps I can shed some light. I'm not a strong Go player (current
ranking of 11 kyu on KGS server) but perhaps I know enough to make some
comments.
Here are some basic comments on strategy:
- the aim of the game is to take more territory than the opponent, never forget
that!
- if you try to take too much territory then your territory will be easily
invaded/reduced or maybe even completely destroyed
- the other side of the coin is making your territory over secure, in which case
your opponent will easily take more territory than you
- early in the game it is sometimes prudent to 'forget' about territory and just
concentrate on establishing some safe and strong bases ("thickness") on the
board. These bases can later be used to back up raids into opposing territory,
or as a base for building your own territory, or simply to disrupt the
coordination of the opponents stones. In a tactical situation, it is to your
advantage if you can direct the battle towards your strong base.
- small groups of stones which get isolated from the rest of your stones are a
big disadvantage. They may be destroyed by the opponent, or more likely they
will cause you to waste many moves protecting them.
- strategic initiative is quite important. For example a classic strong move is
a move which increases my territory (or potential territory) and at the same
time puts an enemy group in some danger.
>That leads me to two somewhat
>contradictory lines of thought: first, that it really IS much deeper and more
>unsolvable than even chess,
Stategically I think Go is probably deeper than chess. The situation is less
clear tactically.
Both tactics and strategy are human concepts, not intinsic properties of the
actual game. I think chess fully extends the limits of human tactical ability
so in that sense I don't think it is possible for Go to be tactically more
difficult than chess.
Strategically I don't think chess fully tests human ability, and I think Go
tests human strategic ability to a greater extent.
> and second, that it's REALLY ripe for a
>"myth-busting" (a new algorithmic approach that deflates a lot of the mystery by
>defeating dan-level pros the way chess GMs are now often beaten by programs).
I suspect that we are yet to discover an effective search/evaluation paradigm
for playing strong Go. I think it is possible, but the difficulty shouldn't be
underestimated.
The main difficulty in programming Go is that there is no known fast evaluation
function that is anywhere near as reliable as even the simplest chess evaluation
function. Life & death is the main problem here, and the current approach is to
put lots of local search into the evaluation function.
>The point for me is that BOTH of these lines of thought can best be furthered
>if/when there is a standard GUI with lots & lots of people developing engines
>for it, like in chess today with WinBoard or ChessBase/Fritz. Hundreds of smart
>programmers worldwide fiercely competing against each other would be a much
>quicker way to learn more than today's state where there's just a handful of Go
>programmers.
Actually there are quite a few Go programmers, particularly in Asia. Also, I
think the GTP protocol is becoming quite common.
cheers,
Peter
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.