Quoting David Fotland [EMAIL PROTECTED]:
I hope it will get much stronger in the next few weeks as I wring out the
bugs.
It's nice to see Valkyria join. Anyone else?
Yes, I gave it a try and I might have found the reason it has been
playing so poorly on 19x19. My laptop overheat easily
Hi all,
Akihiro's talk has finally been put online at:
http://content.digitalwell.washington.edu/msr/external_release_talks_12_05_2
005/15004/lecture.htm
I hope people find it helpful. Sorry for the delay and the fact that he is
missing from the video - for some reason the cameraman
When I play Go on a Go server, I do not try to remember the board
position. I can always find out what it is by looking at the client
window on my screen.
When a bot plays on a Go server, it does remember the position. This is
something that programs are good at, so it seems reasonable to
As I understand it, gtp is for one way communication. I've heard of this as
an issue when developers try to provide output for the benefit of players
(or bot developer debugging the bot)
There's typically work-arounds that we use to overcome this.
On kgs, to inform the players, the version
What is the proper way to interpret the score and opponent columns?
On Dec 9, 2007 7:30 PM, Don Dailey [EMAIL PROTECTED] wrote:
I just put up the improved hall of fame page.
I'm using the values Rémi suggests and the values look more in line with
CGOS.
Also, FatMan-1 is fixed at 1800
Nick Wedd wrote:
When I play Go on a Go server, I do not try to remember the board
position. I can always find out what it is by looking at the client
window on my screen.
When a bot plays on a Go server, it does remember the position. This
is something that programs are good at, so it
Let me refer you to the bayeselo web site, this is not my work but due
to Rémi Coulom http://remi.coulom.free.fr/.I am simply using his
software to build the table:
http://remi.coulom.free.fr/Bayesian-Elo/
The score is simply how well the program scored against it's totally
Another example I found is the impressive Valkyria program. Version
2.7 won 92% of it's games, more than even the top rated greenpeep0.5.1.
However, the average rating of Valkyria's opponents was only 1722.
This is quite a difference. So Valkyria is rated only compared to
greenpeep
I forgot mention why FEN is flawed. It's because it fails to capture
the complete state of the game. It records en-passant information and
color to move, but it's does not capture position history so you
cannot detect draws due to positional repetition.
In GO, this is probably a more
On Dec 10, 2007 10:07 AM, Don Dailey [EMAIL PROTECTED] wrote:
I forgot mention why FEN is flawed. It's because it fails to capture
the complete state of the game. It records en-passant information and
color to move, but it's does not capture position history so you
cannot detect draws
On Dec 10, 2007 4:35 PM, Álvaro Begué [EMAIL PROTECTED] wrote:
In GO, this is probably a more serious problem.
Yes, there is no such thing as an irreversible move in go.
Well there is the opening move... (unless suicide is legal you can
never recreate the empty board).
I think we are
stuck
On Dec 10, 2007 11:05 AM, Erik van der Werf [EMAIL PROTECTED]
wrote:
On Dec 10, 2007 4:35 PM, Álvaro Begué [EMAIL PROTECTED] wrote:
In GO, this is probably a more serious problem.
Yes, there is no such thing as an irreversible move in go.
Well there is the opening move... (unless suicide
Of course if only simple KO is used, then repetition is not an issue -
you only have to store the simple ko state fen style.
But none of this is any good for a general solution (a simple text based
position notation.)
We could talk about systems for compressing move lists of course but
there
In message
[EMAIL PROTECTED], Álvaro
Begué [EMAIL PROTECTED] writes
It is not my intention to sound confrontational. I really don't know
how other rule sets deal with tricky situations.
For long-cycle repetitions:
Japanese: A repetition lead to no result. The game is replayed.
On Dec 10, 2007 11:48 AM, Don Dailey [EMAIL PROTECTED] wrote:
Of course if only simple KO is used, then repetition is not an issue -
you only have to store the simple ko state fen style.
But none of this is any good for a general solution (a simple text based
position notation.)
We could
Nick Wedd wrote:
Chinese: A player may not repeat a previous board position and
when he does the game counts as half a win to each
player.
According to influentual Chinese professionals, the superko rule is a
fake overridden by the referee ko rules section.
On Dec 10, 2007 11:56 AM, Nick Wedd [EMAIL PROTECTED] wrote:
In message
[EMAIL PROTECTED], Álvaro
Begué [EMAIL PROTECTED] writes
It is not my intention to sound confrontational. I really don't know
how other rule sets deal with tricky situations.
For long-cycle repetitions:
Japanese:
In message [EMAIL PROTECTED], Robert Jasiek [EMAIL PROTECTED]
writes
Nick Wedd wrote:
Chinese: A player may not repeat a previous board position and
when he does the game counts as half a win to each
player.
According to influentual Chinese professionals,
On Dec 10, 2007 5:23 PM, Álvaro Begué [EMAIL PROTECTED] wrote:
On Dec 10, 2007 11:05 AM, Erik van der Werf [EMAIL PROTECTED]
Or simply don't use superko. Normal rules work fine with only some
minimal knowledge of the last move. Long cycles are not an issue
because they may repeat multiple
On Dec 10, 2007 6:07 PM, Álvaro Begué [EMAIL PROTECTED] wrote:
It looks like even under non-superko rules, something special happens if a
position is repeated, which means that a program should know the entire
history of the game, or it may accidentally repeat a previous position, even
if it
Erik van der Werf wrote:
On Dec 10, 2007 6:07 PM, Álvaro Begué [EMAIL PROTECTED] wrote:
It looks like even under non-superko rules, something special happens if a
position is repeated, which means that a program should know the entire
history of the game, or it may accidentally repeat a
Arguing whether method A or method B rates a program more
correctly is really close to arguing how many angels can dance
on the head of a Pin. Ratings, at best, are based on mathematical
models with many simplifying assumptions. Ratings are not reality.
Ratings are not reality.
i think that we can probably say that a rating system
for, say, 19x19 go with komi relative to handicap and
time controls roughly the same for each contest (or not,
you choose!) is anything that turns a set of:
(p1,p2,h,t,r) [player 1, player 2, handicap, time, result]
On 6-dec-07, at 19:29, Don Dailey wrote:
Here is an example of why this works so well and why your greedy
approach is so wrong:
Consider a position where there are 2 groups left that are being
fought over. One of these groups is very large and the other is quite
small.The computer
(p1,p2,h,t,r) [player 1, player 2, handicap, time, result]
i should have said that i mean time here to be the
actual date/time that the contest occurred, since skill
can (and often does) change over time.
also the p1,p2 should be taken to be ordered, so that we
know who was black and who was
Dave Dyer wrote:
Arguing whether method A or method B rates a program more
correctly is really close to arguing how many angels can dance
on the head of a Pin. Ratings, at best, are based on mathematical
models with many simplifying assumptions. Ratings are not reality.
Nobody really
Nobody really believes ratings are 100% right on the money accurate. But
it's silly not to use the most correct method possible. Ratings are a very
useful approximation to reality and you might as well get as close to that
reality as you can.- Don
But then we have to take the amount
On Dec 10, 2007, at 11:53 AM, Edward de Grijs wrote:
Nobody really believes ratings are 100% right on the money
accurate.
But it's silly not to use the most correct method possible. Ratings
are a very useful approximation to reality and you might as
well get
as close to that
Mark Boon wrote:
On 6-dec-07, at 19:29, Don Dailey wrote:
Here is an example of why this works so well and why your greedy
approach is so wrong:
Consider a position where there are 2 groups left that are being
fought over. One of these groups is very large and the other is quite
Nick Wedd wrote:
When I play Go on a Go server, I do not try to remember the board
position. I can always find out what it is by looking at the client
window on my screen.
When a bot plays on a Go server, it does remember the position. This is
something that programs are good at, so it
Don Dailey wrote:
Another example I found is the impressive Valkyria program. Version
2.7 won 92% of it's games, more than even the top rated greenpeep0.5.1.
However, the average rating of Valkyria's opponents was only 1722.
This is quite a difference. So Valkyria is rated only
Hi Don,
On Dec 10, 2007 9:08 PM, Don Dailey [EMAIL PROTECTED] wrote:
/snipped a lot of interesting stuff/
However MC play-outs is not horizon limited like this. It's stupid to
make it greedy because it may notice that winning the big group leads to
a loss every time and that some other
Such guidance has to be fairly subtle, however; it often must take the form
of if he plays here, do this, if there, do that.
Doesn't the search tree provide such functionality?
___
computer-go mailing list
computer-go@computer-go.org
Erik van der Werf wrote:
On Dec 10, 2007 6:48 PM, Don Dailey [EMAIL PROTECTED] wrote:
In Go however, even if the fundamental game is unchanged you may be
playing illegal moves if you are not aware of the superko situation.
And you think superko is part of the fundamental game???
We are working to get SlugGo up again.
Cheers,
David
On 9, Dec 2007, at 9:43 PM, David Fotland wrote:
Anyone else?
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
From: Chris Fant [EMAIL PROTECTED]
Such guidance has to be fairly subtle, however; it often must take
the form
of if he plays here, do this, if there, do that.
Doesn't the search tree provide such functionality?
It does indeed - but there are often forced sequences which can be predicted
It looks like the server is down again. It's too bad since there were so
many strong programs connected.
I hope it comes back up soon.
David
___
computer-go mailing list
computer-go@computer-go.org
37 matches
Mail list logo