Re: [Computer-go] Standard Computer Go Datasets - Proposal

[fotland]

I would only use it if it is licensed for commercial use.
 
David

On Fri, 13 Nov 2015 08:39:20 +, Josef Moudrik  wrote:

  Hello List, 
There has been some debate in science about making the research more
reproducible and open. Recently, I have been thinking about making a
standard public fixed dataset of Go games, mainly to ease comparison of
different methods, to make results more reproducible and maybe free the
authors of the burden of composing a dataset. I think that the current
practice can be improved a lot.
 
Since the success of this endeavor crucially depends on how many
authors use the dataset, I would like to ask You (potential authors) a
few questions:
 
1) Would this be welcomed and used? Would You personally use it? (Am I
not reinventing the wheel?)
 
2) What parameters should the dataset have? The number of dataset
variants (if any) should be in my opinion kept at bare minimum to
reduce "fragmentation".
 
2a) Size: My current view is that at least 2 sizes are necessary: small
(1000-2000 games?) and large dataset (5-6 games). 
2b) Strength & year span: Currently I am thinking about including
modern professional games only (1970-2015)
 
3) Do you have any other comments, requirements for the dataset and ideas?
 
 
Thanks for Your attention,
Kind regards
Josef Moudrik
 

-

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go


___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Frisbee Go Simulation

[fotland]

Attached is a frisbee go game 9x9 between me and a Chines 5-dan
amateur.  50% chance of playing in the intended spot.  When a
connection is required, it is just up to chance who wins the fight.
 It's a little silly, but was a lot of fun to play.
 
David

On Wed, 11 Nov 2015 23:13:51 +0100, "Ingo Althöfer"  wrote:

  Hmm.


>> Would the game end after two unintentional passes?

> Good point. In principle I would say so.

That makes little sense to me.
IMO, the principled rule is that two consecutive intentional passes
end the game.


We should have some test games to see how long a game would be
"typically" stretched by unintended passes.

Ingo.
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go




frisbee.sgf
Description: application/go-sgf
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Mylin Valley The World Computer Weiqi Tournament

[fotland]

I don't have records but I watched three games between Zen and
Dolburam, and in each case Dol Buram won in the middle game fighting by
capturing some large group.  In the final of the elimination
tournament the game server crashed wen the game was about 2/3 finished,
so they played another game from the beginning.
 
Dol Buram in playing a pro now with four stones.  The game is being
streamed at
 
51wq.lianzhong.com/yidongwq/index.html
 
David

On Fri, 13 Nov 2015 13:22:48 +, Aja Huang  wrote:

  Congratulations to Dol Baram!  
I was wondering if Dol Baram's author is reading this list and if he
could kindly give a brief description on his main approaches? From my
observation Dol Baram's style is quite human-like and it reads very
well in life-and-death situations. I suspect Dol Baram combines a
life-and-death solver with the main search.
 
Aja

On Fri, Nov 13, 2015 at 12:30 PM, Rémi Coulom  wrote:

  Thanks Hiroshi. This seems to be a more recent post:

http://51wq.lianzhong.com/Home/NewsDetails?newsID=546=%25e7%2584%25a6%25e7%2582%25b9%25e6%2596%25b0%25e9%2597%25bb

Congratulations to Dol Baram!

Rémi

On 11/13/2015 01:17 PM, Hiroshi Yamashita wrote:

  Hi,

It seems DolBaram won. (from last photo on web)

1st DolBaram
2nd Zen
3rd ManyFaces of Go
4th Ray
http://51wq.lianzhong.com/Home/NewsDetails?newsID=539=%25e7%2584%25a6%25e7%2582%25b9%25e6%2596%25b0%25e9%2597%25bb

Hiroshi Yamashita

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

-

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go


___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Mylin Valley The World Computer Weiqi Tournament

[fotland]

The result of the top four single elimination tournament was the same
order as the preliminary round robin.
 
Ray is quite strong, and only 18 months old.  Many Faces played it
twice, in the round robin, and in the final.  In the final it was
quite far ahead in the middle game, but missed a cut, allowing MF to
make the game close, then MF outplayed it in the endgame and won.
 
David

On Fri, 13 Nov 2015 21:17:23 +0900, Hiroshi Yamashita  wrote:

  Hi,

It seems DolBaram won. (from last photo on web)

1st DolBaram
2nd Zen
3rd ManyFaces of Go
4th Ray
http://51wq.lianzhong.com/Home/NewsDetails?newsID=539=%25e7%2584%25a6%25e7%2582%25b9%25e6%2596%25b0%25e9%2597%25bb

Hiroshi Yamashita

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go


___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Mylin Valley The World Computer Weiqi Tournament

[fotland]

Now there are some friendly games, on fast hardware, one hour each time limit.
 
Yu BIn 9P is playing Dol Baram with 5 stones.  The game is so complex
I can't tell who is winning.
Tang Yi 2P is playing Zen with 5 stones and the game is almost over.
 She is losing.  Zen won by 12 points.
Many Faces is playing a very strong amateur, taking two stones, and is
winning by about 20 points.
 
David

On Fri, 13 Nov 2015 13:22:48 +, Aja Huang  wrote:

  Congratulations to Dol Baram!  
I was wondering if Dol Baram's author is reading this list and if he
could kindly give a brief description on his main approaches? From my
observation Dol Baram's style is quite human-like and it reads very
well in life-and-death situations. I suspect Dol Baram combines a
life-and-death solver with the main search.
 
Aja

On Fri, Nov 13, 2015 at 12:30 PM, Rémi Coulom  wrote:

  Thanks Hiroshi. This seems to be a more recent post:

http://51wq.lianzhong.com/Home/NewsDetails?newsID=546=%25e7%2584%25a6%25e7%2582%25b9%25e6%2596%25b0%25e9%2597%25bb

Congratulations to Dol Baram!

Rémi

On 11/13/2015 01:17 PM, Hiroshi Yamashita wrote:

  Hi,

It seems DolBaram won. (from last photo on web)

1st DolBaram
2nd Zen
3rd ManyFaces of Go
4th Ray
http://51wq.lianzhong.com/Home/NewsDetails?newsID=539=%25e7%2584%25a6%25e7%2582%25b9%25e6%2596%25b0%25e9%2597%25bb

Hiroshi Yamashita

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

-

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go


___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Mylin Valley The World Computer Weiqi Tournament

[fotland]

Yes, it won the six stone game.  There was discussion in the panel 
about how computers don't understand double'll, and it looked like the 
pro was seeing up double kos. 



On Mon, 16 Nov 2015 11:49:25 +0900, Hiroshi Yamashita 
<y...@bd.mbn.or.jp> wrote:
Thank you for the infomation. 


It seems DolBaram lost 4h and 5h games, but won 6h game. 
I meke sgf from web site. 6H game is still playing?
 So I can see only latest position. 


4H game http://www.yss-aya.com/lian_dol_4h.sgf
5H game http://www.yss-aya.com/lian_dol_5h.sgf

http://51wq.lianzhong.com/yidongwq/index.html

Regards,
Hiroshi Yamashita

- Original Message - From: "David Fotland" <fotl...@smart-games.com>
To: <computer-go@computer-go.org>
Sent: Sunday, November 15, 2015 11:15 AM
Subject: Re: [Computer-go] Mylin Valley The World Computer Weiqi Tournament


Yu Bin won his game against Dolbaram.  The second official pro game 
is happening now.   51wq.lianzhong.com/yidongwq



___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go





___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Frisbee Go Simulation

[fotland]

I was thinking reg_genmove.  Make the bot support one way to do it to
make the referee simpler.
 
David

On Thu, 12 Nov 2015 03:22:23 +, Josef Moudrik  wrote:

  Frisbee go sounds fun.How do you plan to use the GTP protocol to
support this? I think that the randomization should be handled by the
server, so the bot needs to get feedback about the move actually
carried out. So maybe
genmove + undo & play
or reg_genmove + play
depending on what do the bots support?
 
It should be fairly easy to modify gogui-twogtp to allow for this, and
imo if you want to promote the frisbee go, this should be done early,
s.t. there is platform for testing.
 
 
Regards,
Josef M.

On Wed, Nov 11, 2015 at 11:15 PM "Ingo Althöfer"  wrote:

  > Oh! You can have a continuous handicap control by giving the
players different epsilons. :)

Right. You have "the same" in human-played Frisbee Go by having arbitrary
distances from which the players have to throw their frisbees. (You may
even change the distance during the game )

Ingo. 
 
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

-

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go


___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Frisbee Go Simulation

[fotland]

Yesterday I modified Many Faces to play Frisbee go an played a few
games with some other people at the Beijing Computer go tournament.
 It's a very strange game.  If there is interest I can make an
installer and make it available for free.
 
Josef Moudrik is also writing a program.
 
David

On Thu, 12 Nov 2015 09:29:50 -0500, John Tromp  wrote:

  On Thu, Nov 12, 2015 at 9:07 AM, Nick Wedd  wrote:

I was thinking about the ko rule for frisbee ko, and realised it leads to
problems.

1. Black takes a ko, White tries to make a ko threat, but accidentally
retakes the ko. What should happen?


This was already covered by having any illegal frisbee landing revert
to a pass.
Btw, it's impossible to make a ko threat neighbouring a ko retake, as
all those points are occupied:(


2. Black takes a ko. White tries to make a ko threat, but fails to make a
valid move. Black tries to make connect the ko, but fails to make a valid
move. May White now (try to) retake the ko?


Being a superko fan myself, the answer is clear: not if it repeats the
position.

regards,
-John
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go


___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Reminder: Handicap-29 Prize

[fotland]

Yes. I’m travelling this week, but when I get home I can look for it. I’m not 
sure I can find the old source backups to make a gtp version.

 

David

 

From: Computer-go [mailto:computer-go-boun...@computer-go.org] On Behalf Of 
Álvaro Begué
Sent: Wednesday, January 4, 2017 4:48 PM
To: computer-go 
Subject: Re: [Computer-go] Reminder: Handicap-29 Prize

 

Would you be willing to make the executable of mfgo1998 available so we can run 
it locally? Or even better, something with the same engine but which speaks GTP?

 

Álvaro.

 

 

 

 

 

On Wed, Jan 4, 2017 at 10:32 PM, "Ingo Althöfer" <3-hirn-ver...@gmx.de 
 > wrote:

Like in every year, the reminder on my Handicap-29 prize:
http://www.althofer.de/handicap-29-prize.html

I think, AlphaGo in its current form would not have a chance to cash in ;-)

Ingo.
___
Computer-go mailing list
Computer-go@computer-go.org  
http://computer-go.org/mailman/listinfo/computer-go

 

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] it's alphago

[fotland]

Competitive with Alpha-go, one developer, not possible. I do think it is 
possible to make a pro level program with one person or a small team. Look at 
Deep Zen and Aya for example. I expect I’ll get there (pro level) with Many 
Faces as well.

 

David

 

From: Computer-go [mailto:computer-go-boun...@computer-go.org] On Behalf Of 
Adrian Petrescu
Sent: Thursday, January 5, 2017 10:36 AM
To: computer-go@computer-go.org
Subject: Re: [Computer-go] it's alphago

 

As an individual? Probably, yes.

 

On Thu, Jan 5, 2017 at 1:34 PM, Xavier Combelle  > wrote:



Le 05/01/2017 à 02:16, Yamato a écrit :
> Yes, it is AlphaGo. I am relieved that DeepMind clarified this.
>
> Honestly I got a little frustrated that many people didn't think that
> was AlphaGo. It was almost clear to me because I know the difficulty of
> developing AlphaGo-like bots.
thanks for this insight, if I understand well developing a bot
competitive with alphago
is nearly an impossible task?

___
Computer-go mailing list
Computer-go@computer-go.org  
http://computer-go.org/mailman/listinfo/computer-go

 

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Reminder: Handicap-29 Prize

[fotland]

That's easy enough. Mfog12 trial is a free download and I can provide a 
registration code for the computer used for any competition to enable the 12 
kyu level.  12 kyu has not full board search, monte carlo or otherwise.

David

-Original Message-
From: Computer-go [mailto:computer-go-boun...@computer-go.org] On Behalf Of 
"Ingo Althöfer"
Sent: Thursday, January 5, 2017 3:31 AM
To: computer-go@computer-go.org
Subject: Re: [Computer-go] Reminder: Handicap-29 Prize

Hi,

a few years ago we had agreed that for winning the bet it would be sufficient 
to beat the non-deterministic 12-kyu level of MFoG 11 (or MFoG 12). This level 
has no Monte Carlo elements ionvolved.

Ingo.
 
 

Gesendet: Donnerstag, 05. Januar 2017 um 04:15 Uhr
Von: fotl...@smart-games.com
An: computer-go@computer-go.org
Betreff: Re: [Computer-go] Reminder: Handicap-29 Prize

Yes. I’m travelling this week, but when I get home I can look for it. I’m not 
sure I can find the old source backups to make a gtp version.
 
David
 
From: Computer-go [mailto:computer-go-boun...@computer-go.org] On Behalf Of 
Álvaro Begué
Sent: Wednesday, January 4, 2017 4:48 PM
To: computer-go 
Subject: Re: [Computer-go] Reminder: Handicap-29 Prize
 

Would you be willing to make the executable of mfgo1998 available so we can run 
it locally? Or even better, something with the same engine but which speaks GTP?

 

Álvaro.

 

 

 

 

 

On Wed, Jan 4, 2017 at 10:32 PM, "Ingo Althöfer" 
<3-hirn-ver...@gmx.de[mailto:3-hirn-ver...@gmx.de]> wrote:
Like in every year, the reminder on my Handicap-29 prize:
http://www.althofer.de/handicap-29-prize.html[http://www.althofer.de/handicap-29-prize.html]

I think, AlphaGo in its current form would not have a chance to cash in ;-)

Ingo.
___
Computer-go mailing list
Computer-go@computer-go.org[mailto:Computer-go@computer-go.org]
http://computer-go.org/mailman/listinfo/computer-go
 
___ Computer-go mailing list 
Computer-go@computer-go.org 
http://computer-go.org/mailman/listinfo/computer-go[http://computer-go.org/mailman/listinfo/computer-go]
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] World AI Go Open

[fotland]

I’d go, but I already have a vacation in Europe planned that week. General 
advice:

 

Take a laptop. Arrive a day early to learn your way around and test, and fix 
any issues. Bring executables on a flash drive in case your computer dies or 
gets lost. Even if you have a setup in the cloud you will need a local laptop. 
Don’t forget cables and power adapters. 

 

Test your remote setup thoroughly before you leave. Check with the organizers 
to see how they are running the tournament. If you can’t support their 
tournament software you might need to enter moves by hand.

 

If you run remotely to your home, test the connection through your router 
firewall from an external network. Make sure you have someone you can call if 
someone goes wrong. I ran one contest on my home server, and there was a power 
failure at my home a few hours before the tournament. The server didn’t reboot 
properly. I had to call my son from Japan, wake him up, and ask him to turn the 
machine on. Next time I’ll have a UPS on the machine and my router.

 

I’ve run remotely to AWS, which works pretty well. I recommend writing down 
step by step what you have to do to get everything going. You will be jet 
lagged and it’s easy to forget something.

 

Once I ran remotely to a machine at Microsoft that wouldn’t accept outside 
connections. I had to write a proxy that ran on a machine in my home that the 
server and my laptop could connect to.

 

Since it’s the first event, there might be some issues. Be patient with the 
organizers. It’s great that they are contributing time and money to computer 
go. Let them know how much you appreciate it.

 

Email me directly if you have any questions.

 

David

 

From: Computer-go [mailto:computer-go-boun...@computer-go.org] On Behalf Of 
Brian Lee
Sent: Thursday, June 29, 2017 7:10 PM
To: computer-go@computer-go.org
Subject: [Computer-go] World AI Go Open

 

Hello everyone,

 

Is anybody else planning to go to this event? 

 

https://www.reddit.com/r/baduk/comments/6ge1ul/the_1st_world_ai_go_open/

 

I'm a first timer to in-person computer go tournaments and am unsure how I 
should deal with logistics. I don't want to pack up my desktop in a suitcase so 
I've been thinking I should get my setup replicated in the cloud. This seems 
tricky though with respect to potential firewall issues, latency issues, etc.

 

Would be happy to hear some advice from more experienced devs!

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Source code (Was: Reducing network size? (Was: AlphaGo Zero))

[fotland]

Sadly, this is GPL v3, so it's not safe for me to look at it.

David

PS even though Robert's posts are slightly off topic for the AlphaGo 
discussion, I respect that he has thought far more deeply than I have about go, 
and I support his inclusion in the list.

-Original Message-
From: Computer-go [mailto:computer-go-boun...@computer-go.org] On Behalf Of 
Gian-Carlo Pascutto
Sent: Tuesday, October 24, 2017 1:02 PM
To: computer-go@computer-go.org
Subject: [Computer-go] Source code (Was: Reducing network size? (Was: AlphaGo 
Zero))

On 23-10-17 10:39, Darren Cook wrote:
>> The source of AlphaGo Zero is really of zero interest (pun intended).
> 
> The source code is the first-hand account of how it works, whereas an 
> academic paper is a second-hand account. So, definitely not zero use.

This should be fairly accurate:

https://github.com/gcp/leela-zero

--
GCP
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Zero performance

[fotland]

The paper describes 20 and 40 block networks, but the section on comparison 
says AlphaGo Zero uses 20 blocks. I think your protobuf describes a 40 block 
network. That's a factor of two 

If you only want pro strength rather than superhuman, you can train for half 
their time.

Your time looks reasonable when calculating the time to generate the 29M games 
at about 10 seconds per move. This is only the time to generate the input data. 
Do you have an estimate of the additional time it takes to do the training? 
It's probably small in comparison, but it might not be.

My plan is to start out with a little supervised learning, since I'm not trying 
to prove a breakthrough. I experimented last year for a few months with 
res-nets for a policy network and there are some things I discovered there that 
probably apply to this network. They should get perhaps a factor of 5 to 10 
speedup. For a commercial program I'll be happy with 7-dan amateur with about 6 
months of training using my two GPUs and sixteen i7 cores. 

David

-Original Message-
From: Computer-go [mailto:computer-go-boun...@computer-go.org] On Behalf Of 
Gian-Carlo Pascutto
Sent: Friday, October 20, 2017 10:45 AM
To: computer-go@computer-go.org
Subject: [Computer-go] Zero performance

I reconstructed the full AlphaGo Zero network in Caffe:
https://sjeng.org/dl/zero.prototxt

I did some performance measurements, with what should be state-of-the-art on 
consumer hardware:

GTX 1080 Ti
NVIDIA-Caffe + CUDA 9 + cuDNN 7
batch size = 8

Memory use is about ~2G. (It's much more for learning, the original minibatch 
size of 32 wouldn't fit on this card!)

Running 2000 iterations takes 93 seconds.

In the AlphaGo paper, they claim 0.4 seconds to do 1600 MCTS simulations, and 
they expand 1 node per visit (if I got it right) so that would be 1600 network 
evaluations as well, or 200 of my iterations.

So it would take me ~9.3s to produce a self-play move, compared to 0.4s for 
them.

I would like to extrapolate how long it will take to reproduce the research, 
but I think I'm missing how many GPUs are in each self-play worker (4 TPU or 64 
GPU or ?), or perhaps the average length of the games.

Let's say the latter is around 200 moves. They generated 29 million games for 
the final result, which means it's going to take me about 1700 years to 
replicate this. I initially estimated 7 years based on the reported 64 GPU vs 1 
GPU, but this seems far worse. Did I miss anything in the calculations above, 
or was it really a *pile* of those 64 GPU machines?

Because the performance on playing seems reasonable (you would be able to 
actually run the MCTS on a consumer machine, and hence end up with a strong 
program), I would be interested in setting up a distributed effort for this. 
But realistically there will be maybe 10 people joining, 80 if we're very lucky 
(looking at Stockfish numbers). That means it'd still take 20 to 170 years.

Someone please tell me I missed a factor of 100 or more somewhere. I'd love to 
be wrong here.

--
GCP
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Crazy Stone is back

[fotland]

I've been on a forced break from go programming since last August, but I hope 
to get back to it soon.

David

-Original Message-
From: Computer-go [mailto:computer-go-boun...@computer-go.org] On Behalf Of 
Rémi Coulom
Sent: Tuesday, February 27, 2018 10:14 PM
To: computer-go@computer-go.org
Subject: [Computer-go] Crazy Stone is back

Hi,

I have just connected the newest version of Crazy Stone to CGOS. It is based on 
the AlphaZero approach. The "Weights" engine were in fact previous experimental 
versions. CrazyStone-18.03 is using time control and pondering instead of a 
fixed number of evaluations per move. So it should be much stronger than 
Weights_31_3200.

Does anybody know who cronus is? It is _extremely_ strong. Its rating is low 
because it has had only weaker opponents, but it is undefeated so far, except 
for one loss on time, and some losses against other versions of itself. It has 
just won two games in a row against Crazy Stone.

I hope the other strong engines will reconnect, too.

R�mi
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

RE: [computer-go] When is Pass the best move?

[David Fotland]

During the main search, pass is always one of the moves searched, and if is
the best move found, and pass is allowed, it will play a pass.  I never
avoid searching pass in the main search since it gives a good lower bound
on the score.  If the search returns pass too early, I'll play the second
best move instead.

Many Faces of Go uses the following rules:

- pass is always allowed if the opponent passed on the previous move
otherwise
- pass is allowed if the move number is at least half the board size (181 on
a full size board) and the pass search found no good moves for the opponent.


Before I do the main search for the color to move, I do a pass search for
the opponent, to find good opponent moves if the color to move passes.
These moves are searched early in the main search.  I'm trying to find
killer moves early.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Don Dailey
 Sent: Sunday, October 22, 2006 5:08 PM
 To: computer-go
 Subject: Re: [computer-go] When is Pass the best move?
 
 
 I'm in a similar situation,  I'm trying to identify classes 
 of moves that I can eliminate in an admissible way - which 
 means the move I am throwing out is either not the best move, 
 or there are other equally good moves.
 
 I know that pass moves can be the best move in seki situations - and
 it's non-trivial how to identify them.   
 
 I'm not good at go and this is a severe handicap for me, but 
 here is what I come up with so far - please add to my list if 
 you know of anything else:
 
   1.  Benson space - I use benson's algorithm to find 
 pass-alive groups 
   and if a small region is enclosed completely by benson
   groups,  there is never a point moving inside of them for either
   color (except in cleanup situations for Chinese rules)  
 
   A region is defined as strings of points that can consist of
   any combination of enemy stones and empty intersections.
 
   A large region can exist inside benson safe groups that allows
   for life - so care must be taken that you identify the correct
   regions.   I don't know if I'm doing it the most efficient way,
   but I'm going by region size.  A region of 7 inside a benson
   group cannot possibly support enemy life.   So moves inside them
   by either color do not improve the position.
 
2. I have a quick and dirty pass rule - I throw out all pass moves 
   in the early part of the game.   I can't prove this rule is 
   admissible but I feel very safe with a rule like, don't allow
   pass if half the points on the board are empty.  
 
   I feel you can probably be a lot less conservative - 
 but if anyone
   knows a way to identify when to start including pass moves in a 
   search in a theoretically sound way - I'm all ears!
 
3. Unfortunately, the eye-filling rule is not admissible other than
   in benson situations.   The eye-filling rule I use has been
   described on this group - don't move to a point surrounded on 
   all sides by stones of the same color - where the opponent 
   doesn't occupy more than 1 diagonal (different if on edge of
   board.)  I may or may not use this rule depending on what I'm 
   trying to do.   
 
4. Don't move to any of the corner points on the first move.
 
 
 Right not I'm working on a perfect solver for 5x5 - and rule 
 1 and 2 are
 the only rules I know that are 100% admissible.   I suspect 
 rules 2 and
 4 are admissible - at least on 5x5 or larger although I 
 cannot back that
 up with any theory.   
 
 Rule 4, the corner rule can probably be generalized - and I 
 think that's
 what a good pattern database might be able to do.I would 
 like to be
 able to build an admissible pattern database of the form that veto's
 specific moves.But the database must be provably correct, 
 not built
 based on a humans intuition that a move is probably not good.
 
 So I'm basically lost here.   my solver is just a step 
 towards the goal
 of a pattern database that can admissibly remove many 
 pointless moves from a tree search. 
   
 
 - Don
 
 
 
 
 
 On Sun, 2006-10-22 at 10:47 -0700, Phil G wrote:
  Does anyone have an example where pass is the best move, and not 
  part of the two passes to end the game? I'm trying to determine if 
  passes should ever be considered in a search for the best 
 move, and if 
  so, how to exclude them until it is really necessary.
  
   
  
  Thanks,
  
   
  
  Phil
  
   
  
  
  ___
  computer-go mailing list
  computer-go@computer-go.org 
  http://www.computer-go.org/mailman/listinfo/computer-go/
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 

___
computer-go mailing list
computer-go@computer-go.org

RE: [computer-go] .. if Monte-Carlo programs would play infinitestrong

[David Fotland]

Shodan players are far, far, from perfect play.  Shodan players have a good
understanding of most basic concepts, and can solve simple tactical problems
during a game, but that's about it.  I'm 3 Dan, and almost every move I make
is a mistake of some kind.  The gap in skill between a shodan and a
professional player is huge.

David

 This is something that should not be neglected because 
 shodan players approach perfect play. Will MC development 
 show some kind of ceiling? Who knows, but it certainly could be.
 
 Jacques.
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Monte-Carlo is the future of 19x19

[David Fotland]

How does monte carlo go do with fights that are trivial to evaluate, but
hard to search?

The attached position (I think from Martin Mueller), has many such fights.
If your program can count liberties correctly, it is easy to evaluate and
choose the best move with 1 ply lookahead.  If you try to do a full board
search for the best move it will take 50 ply or more.  This position has a
large number of big fights where the side to move wins the local fight.

I think smart evaluation beats dumb search, monte carlo or otherwise.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of David Doshay
 Sent: Thursday, November 30, 2006 3:49 PM
 To: computer-go
 Subject: Re: [computer-go] Monte-Carlo is the future of 19x19
 
 
 I think that MC will be useful on 19x19 if a clever way to restrict  
 it to
 sub-game searches can be implemented.
 
 Cheers,
 David
 
 
 
 On 30, Nov 2006, at 1:51 PM, Rémi Coulom wrote:
 
  Chrilly wrote:
 
  I believe that MC  will be the only way to write a GO 
 program in the 
  near future leaving the other stuff in the dust (like Mogo has
  with 9x9
  Monte Carlo Go.)This happened in computer chess several times,
  someone came up with some breakthrough idea,  proved it 
 with actual
  results and everyone else had to play follow the leader 
 to catch up.
 
  Do you think its also the future of 19x19 or only of 9x9 (maybe
  13x13)?
 
  Chrilly
 
  ___
  computer-go mailing list
  computer-go@computer-go.org 
  http://www.computer-go.org/mailman/listinfo/computer-go/
  I am certain it is for 19x19. Just look at the KGS games of Mogo on
  19x19. I played one game against it, and won. I got the feeling it  
  was slightly easier to beat than GNU Go, but that may be because I  
  am used to the way Monte-Carlo programs play. I predict 
 that in one  
  year or two, classical programs will be far behind MC programs on  
  19x19. Maybe it will take less than one year.
 
  Rémi
  ___
  computer-go mailing list
  computer-go@computer-go.org 
  http://www.computer-go.org/mailman/listinfo/computer-go/
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


semeai1.sgf
Description: Binary data
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Monte-Carlo is the future of 19x19

[David Fotland]

 
 It's easy to construct problems that any program cannot 
 handle including
 yours. 

Of course, but understanding fights like the attached ones is essential to
strong play on 19x19.
  
 
 You have to understand that Monte Carlo is not great at 
 tactics, 

I do understand this.  That's my point :)

 
 I can't understand why people think a program has to either 
 search or evaluate.  The only thing any program does is 
 evaluate and that's all
 there is.I used to call my search routine eval() because 
 that's all
 it is. 

I agree with you.  Strong programs have to search, and they also need
knowledge in the evaluation.
All knowledge and no search is just as bad as no knowledge and deep search.

 
 I know you understand this too - I think you are just trying to pick a
 fight.   

I'm not trying to pick a fight.  I was responding to a bunch of people who
think that really fast random search with a stupid evaluation will crush
traditional programs next year.  

Monte carlo has a place in go programs, and is a very useful technique, but
I don’t think it can make a strong program all by itself.

 
 My position is that knowledge engineering is at a steep point of
 diminishing returns. How much improvement do you think 
 you will gain
 with a few more years of adding more patterns?Are you expecting
 major breakthroughs which will allow you to reduce the searching that
 you do now?

I agree with you that knowledge engineering is diminishing returns.  I don’t
think that adding more knowledge to existing programs will make them strong
any time soon.  But there is a lot of simple basic useful knowledge, like
counting liberties, and it seems to me that the monte-carlo enthusiasts are
ignoring this.

My point with the file I attached is not that it's a difficult position.
These fights are incredibly easy if you just add a few dozen lines of code
to count liberties correctly.  To me it's as if a weak chess player says, my
program doesn’t need to understand basic pawn structure evaluation.  It
looks really complicated.  I'll just search faster than you.  There is some
basic knowledge that is not complex and is very useful.

-David

 
 - Don
 
 
 
  David
  
   -Original Message-
   From: [EMAIL PROTECTED]
   [mailto:[EMAIL PROTECTED] On Behalf Of 
 David Doshay
   Sent: Thursday, November 30, 2006 3:49 PM
   To: computer-go
   Subject: Re: [computer-go] Monte-Carlo is the future of 19x19
   
   
   I think that MC will be useful on 19x19 if a clever way 
 to restrict
   it to
   sub-game searches can be implemented.
   
   Cheers,
   David
   
   
   
   On 30, Nov 2006, at 1:51 PM, Rémi Coulom wrote:
   
Chrilly wrote:
   
I believe that MC  will be the only way to write a GO
   program in the
near future leaving the other stuff in the dust (like 
 Mogo has 
with 9x9
Monte Carlo Go.)This happened in computer chess 
 several times,
someone came up with some breakthrough idea,  proved it
   with actual
results and everyone else had to play follow the leader
   to catch up.
   
Do you think its also the future of 19x19 or only of 
 9x9 (maybe 
13x13)?
   
Chrilly
   
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
I am certain it is for 19x19. Just look at the KGS 
 games of Mogo 
on 19x19. I played one game against it, and won. I got 
 the feeling it
was slightly easier to beat than GNU Go, but that may 
 be because I  
am used to the way Monte-Carlo programs play. I predict 
   that in one
year or two, classical programs will be far behind MC 
 programs on
19x19. Maybe it will take less than one year.
   
Rémi
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
   
   ___
   computer-go mailing list
   computer-go@computer-go.org
   http://www.computer-go.org/mailman/listinfo/computer-go/
   
  ___
  computer-go mailing list
  computer-go@computer-go.org 
  http://www.computer-go.org/mailman/listinfo/computer-go/
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Monte-Carlo is the future of 19x19

[David Fotland]

 
 It is not what we said. At least it is not what I meant, and 
 I think it is 
 true for the others.

I was reacting to the two statements below.  I didn't realize that this
opinion was not generally shared by the people developing monte carlo
programs.

 I believe that MC  will be the only way to write a GO program in the 
 near future leaving the other stuff in the dust

 I predict that in one year or 
 two, classical programs will be far behind MC programs on 19x19. Maybe 
 it will take less than one year.

 
 I think there is sometimes a misunderstanding of what Monte-Carlo is. 
 Monte-Carlo is a method to approximate an expectation using a 
 finite sample 
 of randomly drawn points. No more, no less.
 It does not talk about stupidity, especially it does not 
 specify the 
 distribution against which you are computing your 
 expectation. If the distribution is pro players playing 
 against themselves, MC with 3 
 simulations per move and one ply search will crush the best 
 human player by 
 far.

I understand the definition of Monte Carlo.  But when people talk about
Monte Carlo go, they mean programs that evalutate random games, not
professional games.  You are making the same point I made.  What I meant to
say is that using random games and an evaluation that only understands final
scores will not make a strong go program.  There needs to be some knowledge
in the evaluation making the games examined non-random.  There are fights in
19x19 games that need a little knowledge to evaluate.  Random game monte
carlo isn't enough.

David

 
 Sylvain
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [spam probable] Re: [computer-go] Monte-Carlo is the future of19x19

[David Fotland]

What's included in an evaluation? Is each evaluation one random game, or a
set of random games that gives good enough statistics about the value of a
position?

David

 On a P4 3.0Ghz mono processor, the number of evaluations per 
 seconds is in the 
 order of 4500/s in 9x9, 2500 in 13x13 and 1100 in 19x19.

 Sylvain
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] language choices

[David Fotland]

I test on IGS, and I also see a lot of cheating against the computer.

Many Faces does its own scoring and transmits dead stone status.  This
prevents people from not indicating their dead stones.  Many Faces keeps
track of people who escape without finishing a game and won't accept matches
from them.  Almost all match requests will be from weaker players, which
will force the rating down, so after each match, Many Faces makes one match
request to a person one rank stronger.

Even with this there is a bias down in the ratings.  I can start a new
version of Many Faces 3 or more ranks higher and it will win more than half
its games.

Many Faces is currently 13k* on Igs (as ManyFaces).

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 Erik van der Werf
 Sent: Wednesday, December 06, 2006 2:36 AM
 To: computer-go
 Subject: Re: [computer-go] language choices
 
 
 On 12/6/06, Magnus Persson [EMAIL PROTECTED] wrote:
  Note that to get these data I deleted all games where 
 Valkyria lost on 
  points, because close to 100% of those games were not scored 
  correctly. I do not know if it is incompetence or outright 
 cheating, 
  but it happens a lot. Fortunately Valkyria always resigns when it 
  loses so it easy to filter out those game.
 
 I found that humans tend to cheat a lot against computers in 
 the scoring phase. This used to be the case on NNGS (my 
 thesis contains some statistics on that if you're 
 interested), and I don't see any reason why this should now 
 be different on KGS, unless of course if computer players now 
 have the same rights in the scoring phase as humans...
 
 Erik
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS?

[David Fotland]

Do you know how long is the average 19x19 game?  Since territories are much
larger, I think it would take many more moves to get down to single point
eyes.  9x9 has 81 points, but averages 107 moves.  I guess 19x19 would
average 500 or 600 moves.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Don Dailey
 Sent: Monday, December 11, 2006 8:49 PM
 To: computer-go
 Subject: Re: [computer-go] 19x19 CGOS?
 
 
 Did you have a specific size in mind?It seems like a 19x19 server
 would be the natural thing.
 
 I could run the old server until I get the new one finished.  
 
 - Don
 
 
 On Mon, 2006-12-11 at 20:29 -0800, David Doshay wrote:
  A few months ago I suggested a number of stepwise increases 
 in board 
  size to see how the algorithms scaled.  It seems to me 
 having just 2 
  data points does not say enough about how the MC (or any other) 
  algorithm scales, so I wanted to be able to graph some measure of 
  strength against increasing board size.
  
  The responses showed some interest in the normal board 
 sizes, but not 
  enough for me to think it was worth it to host the servers. 
 Several of 
  the program authors indicated that they could only run one 
 size at a 
  time, and I assumed that this meant that having multiple 
 CGOS servers 
  would only detract from the usefulness of the 9x9, so I did 
 not start 
  any other size of server.
  
  If things have changed, I have the resources to be a CGOS-N host.
  
  Cheers,
  David
  
  
  
  On 11, Dec 2006, at 7:53 PM, David Fotland wrote:
  
  
   Hi Don,
  
   Clearly UCT and monte carlo is very well suited to 9x9 go.  It
   works much
   better than the traditional computer go algorithms, and 
 it is much,  
   much
   simpler.
  
   Do you have any plans to set up a CGOS server for 19x19 go?  I'd
   like to see
   how well UCT/MC scales to 19x19 go.  I don't think it will work  
   well at
   19x19, but I'd like to see some experiments to see if I'm wrong.
  
   David
  
  
   ___
   computer-go mailing list
   computer-go@computer-go.org 
   http://www.computer-go.org/mailman/listinfo/computer-go/
  
  ___
  computer-go mailing list
  computer-go@computer-go.org 
  http://www.computer-go.org/mailman/listinfo/computer-go/
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Anchor Player

[David Fotland]

I suggest you use anchorman.  It will be weaker on 19x19, but so will the
other programs.

It lets you get set up quickly.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Don Dailey
 Sent: Tuesday, December 12, 2006 10:48 AM
 To: computer-go
 Subject: [computer-go] Anchor Player
 
 
 If I set up a 19x19 server,  we will need an Anchor player.  
 Here is what I need from an Anchor player:
 
   1.  Non-deterministic - should not play same game every time.
 
   2.  Consistent - plays at the same strength at a level that is not
   based on the power of the hardware.  For instance AnchorMan is
   set to a fixed level that does not depend on time.  Lazarus,
   however, players weaker when other jobs are running on the
   computer - something we don't want in an anchor.
 
   3.  Linux binary - because it runs on the server itself.
 
   4.  Low resource usage - I run AnchorMan on the server at a high
   nice level and it takes less than 1 second per move even if it
   isn't niced.
 
   If the Anchor runs on the server, it must be a good citizen.
 
   5.  Should play as strong as possible given the above constraints.
   If possible it should be in the upper 50-60 percentile - but it
   should not be significantly below median strength.
 
 
 It does not absolutely have to run on the server but it must 
 be heavily available - pretty much 24 hours a day.  It should 
 be a non-changing entity - not something being constantly 
 upgraded - although we could from time to time explicitly 
 upgrade the Anchor player.
 
 It's better if the Anchor player is a known quantity on 9x9, 
 then we could actually assign it the same rating and attempt 
 to extrapolate, but we can do that anyway - not a big deal.
 
 The very best candidate may actually be AnchorMan, a 
 program that may fit all the above criteria.  It's an old 
 fashioned Monte/Carlo program that plays about as well is at 
 can and uses little memory given about 1 second per move - at 
 least on 9x9.  So it doesn't use much resources.
 
 At 19x19 AnchorMan would be weaker.  At this boardsize, 
 AnchorMan would benefit greatly from increased time control 
 but then I'm starting to get away from constraint 4 - low 
 resource usage - unless it was run remotely.
 
 GnuGo is another possibility and has the advantage of being a 
 well known quantity, but Gnugo fails to meet some of the 
 criteria above such as being too deterministic and using 
 heavy resources.
 
 If someone wants to host an Anchor player remotely or has a 
 resource friendly candidate that meets the above criteria, 
 let me know.
 
 - Don
 
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS?

[David Fotland]

Maybe not 10,000, but 1,000 yes.  The current CPUs are 90 nm or 65 nm
process nodes, and labs know how to scale down to 10 nm or so.  That's an
area density increase of 50 to 100.

Look at RAMP,
http://www.eecs.berkeley.edu/BEARS/presentations/06Patterson.ppt a research
project to figure out how to efficiently build and program 1000-cpu small
systems.

David

 Does anybody have any idea where the multi-core CPU designs are  
 taking us the next few decades? The SPS already has 7. Are we going  
 to see 10,000 CPU computers?
 
 Mark
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Scaling to 19x19

[David Fotland]

I'd really like to see how this does against traditional programs.  If Don
sets up
a 19x19 server I'll put Many Faces of Go on it.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 Chaslot G (MICC)
 Sent: Tuesday, December 12, 2006 1:17 AM
 To: computer-go
 Subject: [computer-go] Scaling to 19x19
 
 
 Hi David,
 
 I have scaled my Monte-Carlo program (Mango) to 19x19. 
 My idea is to bias the UCT distribution including big 
 patterns. This seems to work quite well, and to lead to a 
 much more human-like kind of play then pure Monte Carlo. The 
 patterns were learnt automatically from professional games as 
 described in this paper: 
 http://www.math-info.univ-paris5.fr/~bouzy/publications/bouzy-
 chaslot-ci
 g05.pdf
 
 I am currently writing a paper describing the techniques I 
 used to scale on 19x19. You can see Mango's results on KGS 
 (it is 11k at the moment). It is a preliminary version, still 
 slow and with stupid bugs (no shisho management, which leads 
 to funny games sometimes :)).
 
 Cheers,
 
 Guillaume
 
 -Original Message-
 From: [EMAIL PROTECTED]
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 David Fotland
 Sent: Tuesday, December 12, 2006 6:10 AM
 To: [EMAIL PROTECTED]; 'computer-go'
 Subject: RE: [computer-go] 19x19 CGOS?
 
 I'd like to see 19x19.  No one plays the game on any other 
 board size than 19x19, so the other sizes are not very 
 interesting.  The current strong programs are all tuned only 
 for 19x19, and the patterns and strategy are quite different 
 at other board sizes. 
 
 Of course you should keep the 9x9 server running as well, 
 since it's probably easier to tune algorithms on the smaller 
 board size.  You probably want longer games at 19x19.  I'd 
 like to see 30 minutes per side, since that's pretty typical 
 for human games.
 
 -David
 
  -Original Message-
  From: [EMAIL PROTECTED]
  [mailto:[EMAIL PROTECTED] On Behalf Of Don Dailey
  Sent: Monday, December 11, 2006 8:49 PM
  To: computer-go
  Subject: Re: [computer-go] 19x19 CGOS?
  
  
  Did you have a specific size in mind?It seems like a 
 19x19 server
  would be the natural thing.
  
  I could run the old server until I get the new one finished.
  
  - Don
  
  
  On Mon, 2006-12-11 at 20:29 -0800, David Doshay wrote:
   A few months ago I suggested a number of stepwise increases
  in board
   size to see how the algorithms scaled.  It seems to me
  having just 2
   data points does not say enough about how the MC (or any other)
   algorithm scales, so I wanted to be able to graph some measure of 
   strength against increasing board size.
   
   The responses showed some interest in the normal board
  sizes, but not
   enough for me to think it was worth it to host the servers.
  Several of
   the program authors indicated that they could only run one
  size at a
   time, and I assumed that this meant that having multiple
  CGOS servers
   would only detract from the usefulness of the 9x9, so I did
  not start
   any other size of server.
   
   If things have changed, I have the resources to be a CGOS-N host.
   
   Cheers,
   David
   
   
   
   On 11, Dec 2006, at 7:53 PM, David Fotland wrote:
   
   
Hi Don,
   
Clearly UCT and monte carlo is very well suited to 9x9 go.  It 
works much better than the traditional computer go 
 algorithms, and
  it is much,
much
simpler.
   
Do you have any plans to set up a CGOS server for 19x19 
 go?  I'd 
like to see how well UCT/MC scales to 19x19 go.  I 
 don't think it 
will work
well at
19x19, but I'd like to see some experiments to see if I'm wrong.
   
David
   
   
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
   
   ___
   computer-go mailing list
   computer-go@computer-go.org
   http://www.computer-go.org/mailman/listinfo/computer-go/
  
  ___
  computer-go mailing list
  computer-go@computer-go.org
  http://www.computer-go.org/mailman/listinfo/computer-go/
  
 
 
 ___
 computer-go mailing list
 computer-go@computer-go.org
 http://www.computer-go.org/mailman/listinfo/computer-go/
 ___
 computer-go mailing list
 computer-go@computer-go.org
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Anchor Player

[David Fotland]

There is no fixed relationship between ELO and handicap stones.  Stronger
players have less variation in their play, so a handicap stone is worth more
ELO points for a stronger player than a weaker player.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Don Dailey
 Sent: Friday, December 22, 2006 2:04 PM
 To: computer-go
 Subject: Re: [computer-go] Anchor Player
 
 
 So really, what I want to be able to do is:
 
   1. Use the ELO rating system.
   2. Determine how many ELO points 1 stone handicap is worth.
   3.  2 stones are worth
   4.  3 stones are worth,


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] UCT vs MC

[David Fotland]

Thanks.  So it seems that doing as many random games as possible is not the
ideal approach.

In UCT, I suppose the equivalent of the principal variation would be the
path from the root that always visits the child with the highest number of
simulations.  When you make a move with 70,000 simulations, how deep is this
UCT_principal variation?  

I expect that after this many simulations the UCT tree will include every
position in some number of ply from the root.  Deeper in the tree it will
not visit every child.  Typically, how many ply in the UCT tree is full
width?

I'm curious about the full width depth and the principal variation depth to
compare UCT wilth alpha-beta.

It's great to see a new approach to computer go that works so well.  Thanks
for sharing your work.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 [EMAIL PROTECTED]
 Sent: Monday, January 01, 2007 3:56 AM
 To: computer-go
 Subject: Re: [computer-go] UCT vs MC
 
 
 Hello and happy new year!
 
 
  I have some questions about your paper...
 Whouah that's a lot of questions :). I'll try to answer well.
 
 
  I thought that the Monte Carlo evaluation of a position is done by 
  making many random games from that position, and averaging the 
  win/loss result. So the evaluation of a position would be a number 
  between 0 and 1.  I thought several thousand random games would be 
  used for one evaluation.
 No, we consider one evaluation as one simulation, so the 
 evaluation function 
 is a bernoulli random variable. Now you are mainly interested in its 
 expectation, and it is why you can think making a lot of 
 simulations and 
 averaging, trying to approximate the expectation by the 
 empirical average. 
 
  In your paper, say that each UCT leaf node is evaluated by 
 exactly one 
  random game (simulation), with a result of 0 or 1.  Is this true?
 Yes it is true. We try having more that one simulation per 
 leaf node, and: -with the same number of nodes, the 
 improvement is very small; -with the same number of total 
 simulations the level is much weaker.
 
 This seems counter intuitive, but in fact it is not. At each 
 simulation we add 
 a node. So a node often visited will have a lot of 
 descendants, so will 
 average a lot of simulations. The key idea of UCT is that the 
 value in a node 
 is the average of its children's values weigthed by the 
 frequency of visits 
 for its children.
 
  I think
  you say Mogo does 70,000 random games per move.  Does this 
 mean that the
  UCT tree for a move has 70,000 nodes?  When you say 70,000 
 games per move,
  does that mean total game move made, or game per node evaluation?
 That means per MoGo's move. So yes, UCT tree for a move has 
 7 nodes. It is 
 the total number of simulations.
 I use the same count for the CGOS versions of MoGo. 
 MoGo_xxx_10k uses 1 
 simulations for each move, or if you prefer, 1 nodes in 
 its tree. That 
 means that if the CGOS game finished after 40 MoGo moves, 
 then MoGo has 
 computed 400 000 random simulations for the complete game. 
 (There is also a 
 5000 simulations per move version on CGOS).
 
 
  How many simulations (random games with patterns) does Mogo do per 
  second?
 On a P4 3.4Ghz, 4500 in 9x9, 1000 in 19x19.
 
  How do you back up values in the UCT tree?  There are values in the 
  example tree, but I can't see how they are calculated.
 
 As in the UCT algorithm. For each node for the root to the 
 leaf of the current 
 sequence you simply add the 0/1 result to a variable, and 1 
 to the count of 
 the number of simulations.
 
  Your code says that the value is backed up by sum and 
 negation (line 
  26, value := -value).  But I don't see any negative values in your 
  sample tree, or values greater than one.  How do you 
 actually back up 
  values to the root?
 Sorry, it is value := 1-value. Thank you for pointing out the mistake.
 
 
  on page 5 you say that UCB1-TUNED performs better and you use that 
  formula. In the code for the algorithm, you use UCB (line 
 16).  Which 
  is correct?
 
 Since the beginning we used UCB1-TUNED and it performed 
 better. Now with all 
 other improvements, and with a fine parameters tuning the 
 difference is very 
 small. UCB1-TUNED has the advantage that it does not need a 
 parameters tuning 
 to performs well in Go (the famous sqrt(1/10) constant Remi 
 Coulom posted in 
 this list).
 
  In your paper you show win rates against GnuGo of about 
 50%, depending 
  on the parameters.  The current Mogo beats GnuGo over 90%.  What 
  changed?  Are you doing more simulations, or do you have more go 
  knowledge in your patterns?
 
 The results near 50% was with the uniform random simulations. 
 The 80% is with 
 the improved simulations. In the current MoGo there are new 
 improvements not 
 yet published. Currently, against gnugo 3.6 level 8 with 
 7 simulations, 
 the result is 92.5%.
 MoGo which plays on tournaments makes more than 

RE: [computer-go] reign of terror

[David Fotland]

It looks like most of these games are being won in the opening.  Doesn't
mogo have a big UCT opening book?  Is it learning from each game it plays as
well?

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Don Dailey
 Sent: Friday, January 12, 2007 6:41 PM
 To: computer-go
 Subject: [computer-go] reign of terror
 
 
 Someone needs to get their bot on CGOS and end Mogo's reign of terror.
 
 A version of MoGo has achieved a CGOS rating of well over 2300!
 
 
 
 - Don
 
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Can Go be solved???... PLEASE help!

[David Fotland]

I have a strong blockus program.  If anyone wants to set up a server I'll
put it up.  Blockus has 4 players, so there is the issue of cooperation
between several players against one other.  My implementation is just
alpha-beta iterative deepening, transposition table, etc.

David Fotland

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Nick Leaton
 Sent: Sunday, January 14, 2007 3:40 AM
 To: computer-go
 Subject: Re: [computer-go] Can Go be solved???... PLEASE help!
 
 
 Blokus www.blokus.com looks like an interesting challenge 
 that is similar to go, but doesn't have so large a state space.
 
 It has some similarities to go if you are using pattern 
 templates to look for structures.
 
 NIck
 
 On 1/14/07, Richard Delorme [EMAIL PROTECTED] wrote:
  Joshua Shriver a écrit :
   I agree, anyone play othello/Reversi?
  Yes, I do othello programming.
 
   From my understanding it has been solved. Yet when I try to find 
   info
  Othello 6x6 has been solved (and can be easily played perfectly on 
  modern computer), but othello 8x8 is still unsolved, as far 
 as I know; 
  although this can be probably done within a few years of intensive 
  computations. Other board sizes (10x10 and more) are out of 
 reach for 
  several years.
   on reversi computer tournaments they all seemed to die 
 out several 
   years ago.
  There is a small community of othello programmers still 
 active on GGS: 
  telnet://opal.cs.ualberta.ca:5000
 
  --
  Richard
  ___
  computer-go mailing list
  computer-go@computer-go.org 
  http://www.computer-go.org/mailman/listinfo/computer-go/
 
 
 
 -- 
 Nick
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] reign of terror

[David Fotland]

How many simulations per move in the 2300 version?  I think you older
version did about 70,000, and your weaker versions do 3000, 5000, or 1.
 
David

-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Sylvain Gelly
Sent: Sunday, January 14, 2007 2:22 AM
To: computer-go
Subject: Re: [computer-go] reign of terror


Hello,
this version plays on a 4 processors machine, so there is limited access.
For the moment, I let only play limited versions of MoGo on cgos, so that I
can run several on only one machine. This takes little resources and I get
interesting statistics. It will play again if new opponents come. 

Sylvain



2007/1/14, terry mcintyre [EMAIL PROTECTED]: 

The version of MoGo with the 2300+ rating hasn't played since Dec. 28th,
last I checked. What's up with that?

 
Terry McIntyre 



Don: 

 Someone needs to get their bot on CGOS and end Mogo's reign of terror.

 A version of MoGo has achieved a CGOS rating of well over 2300! 






  _  

Bored stiff? http://us.rd.yahoo.com/evt=49935/*http://games.yahoo.com
Loosen up...
Download and play hundreds of games for free
http://us.rd.yahoo.com/evt=49935/*http://games.yahoo.com  on Yahoo! Games.

___
computer-go mailing list
computer-go@computer-go.org 
http://www.computer-go.org/mailman/listinfo/computer-go/




___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] MC Go Effectiveness

[David Fotland]

 light is always a good adagium, see David Fotlands hilarious 
 compression of a joseki library into 12 bits/move, IIRC ;-)
 

More like 10 bits per move to store the joseki DAG, moves, pointers, and
good/bad/trick flags.  I would never do anything like that now, but back
then the entire go program including joseki library had to fit in about 300
KB of memory (DOS) on PCs that typically only had 512 KB to 1 MB total
memory.  The library fit about 50K positions in about 60 KB.

David


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] when to stop searching

[David Fotland]

I would add 4.  The program tries to identify good moves, and only tries 
moves that it thinks might be good.  If it is goal-directed, the good moves
are good for a reason, and if the reason is not satisfied, they are
discarded.

This is the way Many Faces works.  It's very similar to 3, but it's a
different 
way of thinking about the problem (adding good moves rather than deleting
bad ones).

You are correct that this approach is inadmissible, and is self limiting.

I like it because when I add new knowledge I know I'm making the program
stronger.  
I don't like tuning parameters or algorithms and then playing hundreds of
games
to get statistically significant data on the better value.  I'm not saying
my approach
better, just that I prefer it :)

David

 
   3.  selective in the true sense.  Such a program tries to 
   identify bad moves and prune them from the tree, but they
   are pruned permanently.  NO matter how deep or long you 
   search they will never be considered.   
 
 I think true selective programs, unless the pruning 
 criteria is fully admissible,  is self limiting.  You can 
 probably build a strong program but you will be bound 
 strictly to the quality
 of your selective algorithm.   Such an algorithm would play
 imperfectly even on an infinte speed computer.
 
 UCT is admissible - it will ALWAYS find a winning move if you 
 are in a winning position.
 
 - Don
 
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] MoGo

[David Fotland]

Every go book says that to get better you need to see the big picture :)
The big difference between low kyu and high dan players is seeing the big
picture.  Low kyu players are already pretty good at local tactics.  If you
read commentaries you will see a lot of waords about direction of play,
which is a big picture concept, and not so much about tactics.

I agree with you that a big strength of UCT is its ability to see the big
picture.  Older go programs were stronger at local tactics than sam-strength
people, and weaker at big picture.  UCT seems to be stronger at big picture
and weaker at tactics. 

David

 
 Wouldn't that be a hoot?   To get better at GO you will be taught to
 see the big picture - think more like a computer!:-)
 
 - Don
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: Re:[computer-go] MoGo strength

[David Fotland]

I tried a few games against Mogo 9x9 on KGS, and it's not professional
strength, but it is very strong.  When I played fast when I was tired it
beat me every time, and when I made a more careful try, I beat it, but it
wasn't easy.  I'm AGA 3 Dan, KGS 2 kyu, so it seems to be about my strength
or a little stronger.  It's very strange to usually lose by 0.5 point, but
win by capturing every stone.
 
I don't think I could win even one game in 100 against a professional player
at this time limit.
 
David

-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Sylvain Gelly
Sent: Wednesday, April 04, 2007 1:52 AM
To: terry mcintyre; computer-go
Subject: Re: Re:[computer-go] MoGo


Hello Terry,




Sylvain,

Were you aware of this challenge from the American Go Association? The
following is from the latest AGA newsletter; you can send corrections or
replies to [EMAIL PROTECTED]


Yes I was aware, Roy Laird asked me to put MoGo on KGS to play these games,
and it is what I did.
Of course as experts, you should have noticed errors on this newsletter, as
e.g. MoGo developed by the inventors of UCT in hungary :-).

You should also know that we never claimed that MoGo plays 9x9 go near the
level of a professional go player, which is of course false, and even if it
was true should ask for many many experiments, and we would have never say
that.

These games are in some extends interesting, especially for part of you who
can understand them. I can't, so the best I can do is count, and there are
so few to count ;-). 

To answer peigang, the 10 minutes is here because players on KGS usually
don't want to wait, and want not too slow games. For the komi, it may be
right that MoGo wins more often with white than black, I don't know. 

Sylvain


 


GO ONLINE: MoGo -- No-Go, So-So or Uh-Oh?
Go has been called The fruit fly of IT, and for a good reason --
although software engineers have created programs that can defeat the
strongest chess players, the strongest go programs are routinely defeated by
talented children. In fact, go is the lone holdout, the only classic game
that has not yet been solved (so to speak) for the computer. If you wonder
why, the Wikipedia article on computer go
http://w3.listlynx.com/l.php?m=1052s=451912l=aHR0cDovL2VuLndpa2lwZWRpYS5v
cmcvd2lraS9Db21wdXRlcl9nbw%3D%3D  is a good place to start. 
One way to simplify the problem is to work with a smaller board, an
approach followed by Levente Kocsis and Csaba Szepesvari, who are working
together at the Hungarian Academy of Science on a 9x9 program called MoGo.
Their recent claim that MoGo plays 9x9 go near the level of a professional
go player made international news
http://w3.listlynx.com/l.php?m=1052s=451912l=aHR0cDovL2FiY25ld3MuZ28uY29t
L1VTL3dpcmVTdG9yeT9pZD0yODkxNjc5  so we decided to investigate.
Sylvain Gelly, a contributor to the program, clarified the one-armed bandit
strategy, a variation of the ancient Chinese proverb Rich men don't pick
fights. Gelly told the EJ that MoGo tries to maximize its winning
probability. When behind, MoGo will play 'strange' moves to try to catch up,
and when ahead, it will prefer safe moves which secure victory instead of
keeping score. Usually it loses points when ahead, trading profit for
safety, aiming to win by +0.5. To learn more see the Sensei's Library MoGo
http://w3.listlynx.com/l.php?m=1052s=451912l=aHR0cDovL3NlbnNlaXMueG1wLm5l
dC8%2FTW9Hbw%3D%3D  page. 
MoGo has played extensively
http://w3.listlynx.com/l.php?m=1052s=451912l=aHR0cDovL3d3dy5nb2tncy5jb20v
Z2FtZUFyY2hpdmVzLmpzcD91c2VyPW1vZ29ib3QmYW1wO3k9MjAwNyZhbXA7bT0x  on the
Internet but evidence that it plays beyond the mid-kyu level is not
compelling, so we're going to put MoGo to the test. Philip Waldron -- a
solid 6-dan with a current AGA rating of 6.47 who has reviewed go software
for the EJ -- will play a best-of-seven series against MoGo in the computer
go room on KGS. Game times will not be announced in advance, and times will
vary to eliminate the possibility of human interference on the MoGo side.
The results, and possibly a few of the games, will appear in a future EJ, so
stay tuned!
- Roy Laird





  _  

It's here! Your new message!
Get new email alerts
http://us.rd.yahoo.com/evt=49938/*http://tools.search.yahoo.com/toolbar/fea
tures/mail/  with the free Yahoo! Toolbar. 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Sylvain's results

[David Fotland]

No, humans are much weaker on 9x9 than on 19x19.  I'm AGA 3 Dan, and I've
played thousands of 19x19 games, and hundreds of serious 19x19 tournament
games.  I've studied thousands of 19x19 professional games, and have had
dozens of my 19x19 games analyzed by pros.  I think before I tried playing
Mogo, I had played about 3 serious 9x9 games, at a 9x9 tournament at a go
congress about 10 years ago.  So I know almost nothing about 9x9 strategy.
Of course my general tactical knowledge applies to 9x9 boards, but I'm far
stronger at 19x19 opening theory.

If someone with equal tactical ability had studied 9x9 as hard as I've
studied 19x19, he would crush me in 9x9 games.  My 9x9 judgment is much
weaker than my 19x19 judgment due to lack of experience.

9x9 is interesting for computer go since it is much simpler than 19x19, but
people don't play 9x9.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Don Dailey
 Sent: Wednesday, April 11, 2007 11:02 AM
 To: computer-go
 Subject: Re: [computer-go] Sylvain's results
 
 
 On Wed, 2007-04-11 at 17:49 +0100, Jacques Basaldúa wrote:
  BTW. There is another stone in the way of 19x19 computer go. 
  Knowledge. Humans play much stronger and do much stronger judgment 
  than in 9x9.
 
 I think you said this backwards from what you intended.  
 Obviously, humans are closer to perfect play and understand 
 9x9 better than
 19x19.Someone on this group even expressed the opinion that
 professional players are close to perfect at 9x9.   
 
 At 19x19 I'm sure there is a great deal of distance to cover even
 for the very top players.   
 
 - Don
 
 
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

[computer-go] people are weaker at 9x9 go

[David Fotland]

I think I see our misunderstanding. I obviously don’t think that smaller
boards are more challenging.

I do think that people are better players on larger boards.  This is also
fairly obvious, even though 9x9 is a much simpler game.  You can measure how
good someone is at any game by how much he improves with additional
practice.  A beginner improves rapidly, and an experienced player improves
more slowly.  This has nothing to do with how challenging the game is.  It
has everything to do with how much the game has been studied by the
community that plays it, and how much it has been practiced by the
individual player.

No one plays 9x9 go.  There is no literature on 9x9 theory, and almost no
examples of professional play.  The statement that people moved to larger
board sizes over time is a myth, or at least somewhere in prehistory.  19x19
board have been used since the first historical recorded game almost 2000
years ago, and all recorded go theory is on 19x19 boards.  9x9 go is only
used as a teaching aid for beginners for their first few games.

I know I'm a weak 9x9 player because I haven’t played or studied it.  I
learned a lot about 9x9 openings just from the few games I played against
Mogo.  If put the same effort into 9x9 study as I have put to 19x19 study, I
would be a much stronger 9x9 player.

For analogy, look at chess and arimaa.  Arimaa may be a more or less
challenging game than chess, but it is certain that people are weaker at
Arimaa than at chess, just because arimaa is new, so there is a lot that is
unknown about correct strategy.  That's why I thought that my best chance to
win the prize was to do it the first year before people got strong at it.

So, of course 9x9 go is less challenging than 19x19, but also of course
people are weaker at 9x9 go than 19x19 go. 

I agree with you that 19x19 is more challenging for computers an humans.  I
also agree that humans are better able to deal with the extra complexity of
19x19.  I think you are overlooking an add ional factor, that people have
studied 19x19 and have not studied 9x9, so the have more practive at 19x19
(and are relatively stronger). 

David

 -Original Message-
 From: Don Dailey [mailto:[EMAIL PROTECTED] 
 Sent: Wednesday, April 11, 2007 8:52 PM
 To: David Fotland
 Cc: 'computer-go'
 Subject: RE: [computer-go] Sylvain's results
 
 
 On Wed, 2007-04-11 at 17:29 -0700, David Fotland wrote:
  No, humans are much weaker on 9x9 than on 19x19.
 
 With all due respect, that's absurd.  If that were true, then 
 all we would have to do is move to smaller boards if 19x19 
 were not challenging enough.
 
 I read somewhere that the 19x19 board size was the result of
 a lot of experimentation over time.  The
 idea was that bigger boards are more challenging and that
 top players considered the smaller boards too easy.
 
 So I just can't make sense out of why you would believe that
 smaller boards are more challenging.   
 
 
   I'm AGA 3 Dan, and I've
  played thousands of 19x19 games, and hundreds of serious 19x19 
  tournament games.  I've studied thousands of 19x19 
 professional games, 
  and have had dozens of my 19x19 games analyzed by pros.  I think 
  before I tried playing Mogo, I had played about 3 serious 
 9x9 games, 
  at a 9x9 tournament at a go congress about 10 years ago.  So I know 
  almost nothing about 9x9 strategy. Of course my general tactical 
  knowledge applies to 9x9 boards, but I'm far stronger at 
 19x19 opening 
  theory.
 
 Of course small boards are easier for computers too.   Even 
 if you don't
 agree, I believe that bigger boards are much more challenging 
 for both computers and humans.  However, humans are better 
 able to deal with
 the extra complexities.   So if you are playing against a 
 computer, you
 would obviously want a bigger board size.   Humans are weaker 
 at bigger
 board sizes, but computers even more so.
 
 
  If someone with equal tactical ability had studied 9x9 as 
 hard as I've 
  studied 19x19, he would crush me in 9x9 games.  My 9x9 judgment is 
  much weaker than my 19x19 judgment due to lack of experience.
  
  9x9 is interesting for computer go since it is much simpler than 
  19x19, but people don't play 9x9.
 
 9x9 is a good board size for computers.  I'm not really sure 
 if there is anything special about 19x19 (why not 17x17 or 
 21x21?)  perhaps they thought 17x17 was too hard (if it's 
 true that smaller boards are harder for humans.) 
 
 - Don
  
 
 
 
 
 
   On Wed, 2007-04-11 at 17:49 +0100, Jacques Basaldúa wrote:
BTW. There is another stone in the way of 19x19 computer go.
Knowledge. Humans play much stronger and do much 
 stronger judgment 
than in 9x9.
   


___
computer-go mailing list
[EMAIL PROTECTED]
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Sylvain's results

[David Fotland]

 I cannot believe 9x9 is harder than 19x19 and
 I don't care how strong the player is who says that - I don't 
 believe it.
 
 - Don

I don't believe it either :)  Sorry for the misunderstanding.  I was making
a statement about how strong people are at a pair of games, not a statement
about which game is harder.

David


___
computer-go mailing list
[EMAIL PROTECTED]
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Yet another article

[David Fotland]

I threw it at Google translation, which is supposed to be the best machine
translator available now, using only Bayesian statistics with no knowledge.
 
It mangles it, but the meaning comes through.  It sounds like an interesting
article.  It would be nice to have a good translation.
 
David
 
Drosophila's luck and end God regarded everything, which it had made: It was
very good. It became evening and it became morning: the sixth day. Towards.
1,31 If I intend a larger financial transaction, I ask a friendly economist
over the prevailing opinion to this topic. Then I with some security, like
it does not become white and does not depend on it. In my Profession is
still simpler the thing. The prognoses John McCarthy's and its AI-young form
an entropy set. The area of the hopeless solutions. Stanley Kubrick' s HAL
and my Windows PCs have only one together: Sometimes goes nothing at all
more. 1989 published J. Schaeffer and M.Donskoy “Perspectives on Falling
from Grace”. The Confessiones of two successful academic chess programmer.
One betrayed the holy goal of the computer Drosophila by the longing of the
tournament victory. Schaeffers Bekehrungserlebnis was a bitter defeat
against chip test. It became again backdue however with the lady program
Chinnook. 1989 I worked in the European space Technology Centre in
Noordwijk/NL. In the damp and cold winter evenings I felt lonely and had
homesickness. Around this feeling to betäuben I bought the Mephisto Polgar
Brettcomputer of OD Schroeder. Playing against helped also nothing and
seized I one evening the resolution in such a way: I can bisserl games of
chess, I can program, why I do not make a chess program? The goal was, a PC
program, which can take up it with that. I had increased thereby the drug,
became sufficient depressing drums of the rain the pleasant mood music.
However I suffer until today from heavy craze features. Of the McCarthy'
Drosophila did not have I the smallest rope, McCarthy was me only as a
creator of its own programming language a vague term. It would have been
also perfectly all the same me: My goal was to betäuben and strike
homesickness. 1989 rank among the sieved fat years of computer chess. OD
Schroeder lives this very day of the pole percentages of profits. 1989
defeated chip test the first large masters. The first article over this
program carried the title by program „for Designing A single chip
Grandmaster while knowing emergency-hung about chess “. Chip test attained
as Deep Blue world fame. John McCarthy, which had formulated the victory
over the chess world champion once as warming up exercise for the actual
tasks of the AI, reacted in a book review in Science to Deep Blues summit
victory säuerlich. Three feature OF of human chess play acres required by
computers programs when they face of harder problems than chess. Two OF them
were used by early chess programs but were abandoned in substituting
computers power for thought. 1) Human chess players CAN emergency examine
all moves RK every position they think about and therefore must forward
prune the move tree and SELECT the more promising moves for exploration;
early chess programs thus pruned. About of 1969 forward pruning which
eliminated, and computer power which relied upon ton examine all moves. It
larva the programs work more better, because early programs sometimes pruned
good move. Eliminating pruning which possible, because there acres only
about 40 moves in A position. In the game OF Go, there acres UP ton of 361
moves in A position, then even computer forward prune must. McCarthy ignored
obviously 30 years development. Modern programs look for not brute force,
but use dynamic information, the structure of the search tree, in order to
extend and shorten variants both. My chess program Hydra looks for 17-18
sections in a central play position deeply. The shortest variants have a
length of 7, the longest up to 40 Plies. Contrary to the punch card age one
uses today „softly pruning “. Instead of completely cutting a course off,
the search tree is shortened. A search tree reduced by 2 Plies is not many
more expensive to omit than the course at all. The program has however the
chance to recognize incorrect Pruning in the following iterations while the
method publicised by McCarthy does not correct the error with infinite
search depth any longer. In addition these shortened searches produce
further information about the search tree, which can be used in consequence
profitable. Most programs compute only the first 3 courses „brute force “,
those remaining courses are strongly geprunt. The optimal exponential factor
amounts to with alpha beta 6, actually has the programs a factor of 3-4.
Without search extensions the factor would sink under 3. For Fritz
programmers fray to rotten lies therefore the intelligence of a program in
the search algorithm. 

-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of chrilly
Sent: Saturday, May 26, 2007 

RE: [computer-go] 9x9 games wanted and the next big challenge

[David Fotland]

  However GTP was way better than what
 preceded it and yet even the top programmers believed GMP was 
 sent by god and anything else was blasphemy.  

I have to object to this characterization :) GMP was very good at what is
was designed to do, which was to allow people to play using a 1200 baud
modem, before the internet was invented.  This we very useful, since there
were many people in the US that had no local club.  Because the link was
slow and unreliable, GMP was binary, and had reties, and provision for
sending text messages, etc.  Once the internet was invented, GMP was
immediately obsolete, but because so many programs had implemented it, it
became a de-facto standard for computer go tournaments.

I think tournament organizers continued to specify it since it was too hard
for organizers to invent something different.

 
 Even things like time-control systems are very logical in 
 Chess,  but not in Go.  

Time control in go is quite logical if you remember that it is traditional
and was invented before electronics.  You can't do fisher time control by
hand.  I think go puts more emphasis on tradition than go, so things that
were very logical and practical before computers, are still being used.
Even though computers and electronic timers enable better approaches.

The traditional ranking system of Go 
 isn't very rational
 although it's understandable how it evolved.   

The ranking system is also very logical if you remember that it is intended
to be maintained by hand, without electronic assistance.  It's very easy to
track the handicap I use with the people I play with most often, and change
the handicap after a few consecutive wins or losses.  Chess doesn't have a
similar handicap system, so it has to rank based on probability of winning.
Go, instead, adjusts that handicap until the winning probability is 50%.

 It may be that because GO is more of a right brained activity,  it
 appeals more to the emotional, visual type of person.   These kind of
 people are probably a bit more into the culture and history of a game
 than in the pure mathematical game itself.   (There are also chess
 players who love the culture of chess more than the game itself.)  

I think you are right.  I don't see many chess players talking about
creating a beautiful game.
 
 
 Also, Chess has evolved more recently,  there have been 
 fundamental rule
 changes within the last 2 or 3 hundred years I believe.  

Go also has many recent rule changes.  Go was traditionally played without
written rules.  Codifying the traditions is what makes Japanese rules so
complex.  Take a look at the Ing Ko rule sometime :)

David



___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 9x9 games wanted and the next big challenge

[David Fotland]

 I'll bet there have been millions of 9x9 games by very strong 
 players, they are probably just not readily accessible.  

Very unlikely.  I'm a strong player (but not very strong - 3 dan amateur),
and I've played perhaps a dozen 9x9 games with people who were just learning
the rules.  I played in a couple of 9x9 tournaments on the crazy go day at
the go congress (along with 3-d go, hex go, etc).  Most beginners only need
a couple of games on 9x9 before they start paying 19x19.  9x9 go is not very
interesting to strong players, since it's not really go.  I might as well be
playing checkers or 9 men's morris :)

 But with something like CGOS a program like Mogo has bragging 
 rights. It's possible one of the  commercial programs is 
 better than Mogo, or
 perhaps another amateur program is better.   But in most 
 peoples minds,
 Mogo is the best at 9x9 because it was willing to take the 
 risk on CGOS
 (in all likelihood, it really IS the best and few doubt 
 this.)

I can confirm that Mogo is quite a bit stronger than the commercial programs
at 9x9 go.  I'm not very interested in 9x9 go.  Most of the commercial
programs have algorithms that don't scale well down to 9x9, since they are
all designed for 19x19 go.  The 19x19 knowledge that makes them strong does
not apply at 9x9.  Since people don't play 9x9 go, there is no incentive
commercial program authors to make their programs strong at 9x9.

 
 
 Here is what we need in order to achieve a Dan level 19x19 
 player within a couple of years in my opinion:
 
   COMPETITION
 
 Not once a year, but constant.   A very high profile occasional
 competition however is still a great and useful thing to have.  
 
   FEEDBACK
 
 You need to always know where you stand so you can constantly  be goal
 oriented.   Where you stand in relation to others that is.  
 
   STATUS
 
 There must be some kind of recognition, highly visible 
 acknowledgment of
 the pecking order to stimulate and motivate the competitors.  

I agree.  Progress was very swift in the Ing competition, with programs
improving from about 25 kyu to about 5-8 kyu.  Since 2000 the competition
and status has been missing, so progress has stopped, or at least is not
visible.  The algorithms that worked well on a 33 MHz 386 with 0.5 MB memory
are very different from what is possible on today's machines.
 
 
 Once money and status come into the picture big time,  then cheating
 will start to play a major role. 

Cheating did play big role.  Even though Ing and FOST had on-site
tournaments, there was still the issue of reverse engineering the top
programs.

 
 I also have to say that Nick Wedd's monthly tournaments are 
 critically important and unquestionably a big part of the 
 sudden progress in
 computer GO.   I think those tournaments and CGOS complement 
 each other
 in a beautiful way.   Probably more credit goes to Nick Wedd's
 tournaments than CGOS.   Those tournament inspired CGOS and they also
 motivated (in my opinion) a lot of progress in computer chess before
 CGOS was even up and running.But they do complement each other -
 CGOS provides instrumentation that KGS is lacking.  

Nick's tournaments and CGOS have made a huge difference in revitalizing
computer go.  I'd like to see both expanded to 19x19 with 30 minute per
player time limits and some overtime.
 
 
 - Don

-David Fotland


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Who's going to the Gifu Challenge?

[David Fotland]

I was there in 2005, and KCC Igo and Go Intellect were there.
http://www.computer-go.jp/gifu2005/English/

Ogaki is very nice, but a little tricky to get to by train.

I ripped up the full board search and rewrote it last year so I'll only go
if I can get it stable and stronger by then.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Ian Osgood
 Sent: Monday, July 09, 2007 11:01 AM
 To: computer-go
 Subject: [computer-go] Who's going to the Gifu Challenge?
 
 
  From what I can tell, there has not been a clash of the Go titans  
 since the 2003 Gifu Challenge, which had all of KCC Igo, Haruka, Go+ 
 +, Goemate/Handtalk, Many Faces, GNU Go, and Go Intellect
 participating. (This was the last public competition for many of  
 these programs.) It seems with the tuning of MoGo and CrazyStone for  
 the full size board and their recent success at the Olympiad, that  
 there is a chance to knock KCC Igo (sold as Silver Star in Japan)  
 from its four year throne. Are any of the Mogo, CrazyStone, 
 and other  
 professional program authors leaving room in their autumn schedules  
 to travel to Ogaki City, Japan for this year's Gifu Challenge?
 
 Ian
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 9x9 games wanted and the next big challenge

[David Fotland]

 
 I also agree that 9x9 doesn't compare to 19x19.   I disagree that it's
 not interesting.   It would be uninteresting if, for instance, someone
 like you were just as good at the top pro's at 9x9.   It stops being
 interested when it can be mastered.If the top players can always
 play a perfect game, it's not interesting to them, but 
 probably still interesting to me, and to a lesser extent 
 someone like you who would probably be playing close to perfect if the
 pro's were playing perfect.   There would probably be very little
 difference in someone like you and a top pro and if you 
 played a game well enough you might get some wins if you were 
 on the right side of
 komi.  But this all assumes the game is almost played out.   I don't
 think 9x9 is.

I don't mean that 9x9 is trivial.  It's just not very interesting to play
for someone who plays go.

For example, if I started talking about playing chess on a 6x6 board without
rooks and with only 6 pawns, it would still be a nontrivial game, but it's
not a game that serious chess players would want to play much.  If I put up
a server to play this 6x6 chess variant and got a lot of programmers
interested in writing programs for it, it's still nontrival, but still not
very interesting for chess players.  It's a different game.

David


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Go programming as a profession.

[David Fotland]

The list price in Japan is closer to $100 than $39.

This is my Japanese product (AI Igo version 15):
http://www.ifour.co.jp/product/aiigo15/ and you can see it lists for 13,440
yen (about $110).  The other strong programs have similar prices.  His
royalty is more than $1 per copy.

Since Go4++ hasn't participated in a competition recently, it's probably not
#1.  KCC Igo is the best, based on Gifu results.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 Joshua Shriver
 Sent: Tuesday, July 10, 2007 8:53 PM
 To: computer-go
 Subject: [computer-go] Go programming as a profession.
 
 
 Was looking up engines when I came across the Go++ website. 
 Is it still the #1 engine in the world? Most of the titles on 
 the page seem to refer to the late 90's and early 2000's.
 
 Anyway, the one thing that shocked me the most was Over 
 400,000 copies sold in Japan! At $39.95 that's just shy of 
 $16 million. Wow Even at $1 a copy that's almost 1/2 a 
 million dollars.
 
 Is this normal?
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Why are different rule sets?

[David Fotland]

The biggest difference in over the board club play is the scoring procedure.
in territory scoring, prisoners are kept separate, and at the end of the
game prisoners are put back in enemy territory and regions are rearranged to
rectangles and counter.  Counting is pretty fast, and the board position is
partially preserved.
 
In Chinese scoring, prisoners are put back in the bowls with stones to be
played, and at the end of the game one color only is counted, first
territory, then the stones are collected into piles of 10 and counted.  The
position is completely destroyed in the process.  So though Chinese rules
avoid the special cases in Japanese rules, counting mistakes are easier to
make.
 
In AGA rules, counting is done using the territory procedure, but the score
is the same as the Chinese procedure since anyone who passes gives a
prisoner to the opponent.
 
In Ing rules, each player starts with 180 stones exactly.  During play
prisoners are put back in the bowls, then at the end all 181 stones are put
on the board into the territory.  The person with territory left over wins.
 
So there is not much difference in the moves chosen, but the scoring
procedure is radically different.  If the players don't agree on the rules
and one saves prisoner and the other doesn't it will be difficult to count
the score.
 
I think this group has a tendency to think mostly about computer scoring,
and not how games are scored on a wooden go board in a club :)
 
David


 
I'm curious... How does the rule sets affect how people play the game of go?
I personally find territory scoring more interesting.  90% of my reason for
that is because the game ends sooner... I don't have to go filling dame
(open spaces between chains of opposing colors). 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Why are different rule sets?

[David Fotland]

I didn't write this :)  I'm pretty familiar with the differences in rule
sets.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 Robert Jasiek
 Sent: Thursday, July 12, 2007 12:17 PM
 To: computer-go
 Subject: Re: [computer-go] Why are different rule sets?
 
 
 David Fotland wrote:
  I'm curious... How does the rule sets affect how people 
 play the game 
  of go?
 
 Different scoring requires my strategy to be adapted. 
 Different counting 
 leads to different kinds of defensive methods against accidental or 
 cheating errors.
 
  I personally find territory scoring more interesting.  90% of my 
  reason for that is because the game ends sooner...
 
 So how do you ensure that in real games? Do you not use 
 Japanese fill-in 
 counting because for that you would need to make the game 
 longer again 
 by filling the dame?
 
   I don't have to go filling dame
 
 You don't have to (I guess you use some verbal Japanese 
 rules), but how 
 do you count?
 
 --
 robert
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] IEEE Spectrum article by Deep Blue creator

[David Fotland]

Many Faces does not use null move, but does extensive caching of life and
death and other tactical results.

-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of terry mcintyre
Sent: Tuesday, October 02, 2007 9:21 AM
To: computer-go
Subject: Re: [computer-go] IEEE Spectrum article by Deep Blue creator


- Original Message 

From: Don Dailey [EMAIL PROTECTED]


I think [Hsu] is betting on null move proving - but I'm real skeptical that
 it will be effective in Computer Go.   It will indeed reduce the tree
 significantly, but this comes at a qualitative price that is not so bad
 in Chess but is a lot in Go.


Hsu also discusses the gains from caching life-and-death analysis of groups.
I suspect that
this will greatly reduce computational effort, once an efficient mechanism
is implemented.
Existing monte carlo programs cache information about playable/non playable
points; when
augmented with knowledge about life and death, search should more quickly
home in on crucial 
lines of play.

I've been playing against Mogo the last few weeks. It has a very interesting
style of play, and it 
often does quite well in tactical analysis, but sometimes it misses a key
move and fails to kill or
fails to preserve a large group - game over! A good life-and-death cache
would be a definite improvement.

Caching parts of trees works better in Go, since well-defined sections of
the board can sometimes be 
partitioned from the rest of the board. Where such partitions leak, analysis
is likely to be critical; 
for example, ladders and ladder breakers can extend across the board;
invasions often depend on 
cutting points halfway across the board. 


  _  

Building a website is a piece of cake. 
Yahoo! Small Business gives you all
http://us.rd.yahoo.com/evt=48251/*http://smallbusiness.yahoo.com/webhosting
/?p=PASSPORTPLUS the tools to get online.

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] IEEE Spectrum article by Deep Blue creator

[David Fotland]

It's a very difficult problem, but I have something that works reasonably
well.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Hideki Kato
 Sent: Wednesday, October 03, 2007 12:27 AM
 To: computer-go
 Subject: Re: [computer-go] IEEE Spectrum article by Deep Blue creator
 
 
 David Fotland: [EMAIL PROTECTED]:
 Many Faces does not use null move, but does extensive 
 caching of life 
 and death and other tactical results.
 
 The problem of such caching scheme is when and how programs can make 
 correctly the cache be dirty (ie, invalid).  It may be very hard even 
 for human players.
 
 -gg (Hideki)
 
 -Original Message-
 From: [EMAIL PROTECTED]
 [mailto:[EMAIL PROTECTED] On Behalf Of terry 
 mcintyre
 Sent: Tuesday, October 02, 2007 9:21 AM
 To: computer-go
 Subject: Re: [computer-go] IEEE Spectrum article by Deep Blue creator
 
 
 - Original Message 
 
 From: Don Dailey [EMAIL PROTECTED]
 
 
 I think [Hsu] is betting on null move proving - but I'm 
 real skeptical that
  it will be effective in Computer Go.   It will indeed 
 reduce the tree
  significantly, but this comes at a qualitative price that 
 is not so 
 bad  in Chess but is a lot in Go.
 
 
 Hsu also discusses the gains from caching life-and-death analysis of 
 groups. I suspect that this will greatly reduce 
 computational effort, 
 once an efficient mechanism is implemented.
 Existing monte carlo programs cache information about 
 playable/non playable
 points; when
 augmented with knowledge about life and death, search should 
 more quickly
 home in on crucial 
 lines of play.
 
 I've been playing against Mogo the last few weeks. It has a very 
 interesting style of play, and it often does quite well in tactical 
 analysis, but sometimes it misses a key move and fails to kill or
 fails to preserve a large group - game over! A good 
 life-and-death cache
 would be a definite improvement.
 
 Caching parts of trees works better in Go, since 
 well-defined sections 
 of the board can sometimes be partitioned from the rest of 
 the board. 
 Where such partitions leak, analysis is likely to be critical;
 for example, ladders and ladder breakers can extend across the board;
 invasions often depend on 
 cutting points halfway across the board. 
 
 
   _
 
 Building a website is a piece of cake.
 Yahoo! Small Business gives you all
 http://us.rd.yahoo.com/evt=48251/*http://smallbusiness.yahoo
.com/webhosting
/?p=PASSPORTPLUS the tools to get online.
 inline file
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
--
[EMAIL PROTECTED] (Kato)
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] best approach forward

[David Fotland]

I agree. Computer go needs someone who will play large tournaments are
publish results.  I'm also curious how Many Faces would do against Mogo on
19x19 in a long match.  Mogo is much better at endgames, and is much
greedier, but Many Faces is much stronger tactically.  Certainly if there
were ratings it would motivate me to work on Many Faces more.  I think games
against mogo would help me find bugs in Many Faces.

Are you volunteering?  Is the 19x19 cgos still going?  I could put Many
Faces on it if there is any interest.

David

 
 One thing computer chess has had for a very long time and is 
 practically
 absent in Go is a rating list.   It's always been possible to identify
 who the best programs and where they stand relative to any 
 other.  There are agencies that play hundreds of thousands of 
 games constantly to track the progress and build accurate 
 rating lists of the programs running on various hardware.
 
 Such a thing is a tremendous impetus for improvement.  Is 
 there any such
 thing in Go?If there is, I stand corrected.
 
 This can also help us chart the best way forward.  In 
 computer chess if a program is even a little better than the 
 previous best, it is clear
 public knowledge.   Suddenly every one want to know how they did it
 and invariably it is discovered.  It is hard to keep secrets 
 in computer chess and that is a good thing unless you are 
 marketing a program commercially.
 
 Right now we know that Mogo dominates in 9x9.   Without CGOS 
 this would
 be speculation based on who won the last tournament.   But CGOS is not
 the right way although it's a useful tool.There needs to be some
 kind of testing agency that is fair and unbiased, visible, 
 and everything is out in the open.
 
 In computer go the only real instrumentation I am aware of is 
 who won the last tournament.  And if you are commercial and 
 didn't win the last tournament you advertise just the ones 
 you did win and you control the
 perception of the strength of your program that way.But with some
 kind of rating agency you cannot run and hide.  If your 
 program stinks you don't submit it to the agency, but then 
 you lose credibility - you cannot claim anything with any 
 reasonable credibility.
 
 The reason I made this post was that I wondered how good 
 19x19 Mogo is.
  And I don't have a definitive answer.   Is Many Faces a lot 
 better than
 Mogo at 19x19?   Who is the best and by how much?Do some 
 program run
 better on certain hardware?Does anyone have a precise 
 answer that is
 more than a feeling, hunch, or anecdotal (subjective) evidence?
 
 
 Anyway, I would propose the following experiment:
 
   1. Test Mogo at 19x19 against one of the strong commercial programs.
   2. Test at many levels.
   3. Test the scalability of Mogo and other strong commercial 
 programs.
   4. Extrapolate.
 
 There are many ways to do this,  but if the question is to 
 build some POWERFUL hardware to create a Dan level program 
 then we can extrapolate the result of hardware with software.
 
 We don't have to test against humans if our sole purpose is 
 to see which way to proceed.  Just test Mogo against programs 
 such as Go++ Many Faces and others at incredibly deep levels 
 and see where they stand.
 
 By the way, how programs do in tournaments doesn't cut it.  
 You need thousands of games unless your program is 
 significantly dominant to be able to say it's better with 
 reasonably high certainty.
 
 - - Don
 
 
 
 
 
 
 
 
 
 
 
 
 -BEGIN PGP SIGNATURE-
 Version: GnuPG v1.4.6 (GNU/Linux)
 Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
 
 iD4DBQFHDjiuDsOllbwnSikRAuhXAJiRlNzmemhaTgNbZKesyVzENN70AKDT1HR9
 nLbhKOW1nOMUSBY70jy0cg==
 =SFrB
 -END PGP SIGNATURE- 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] best approach forward

[David Fotland]

At least for Many Faces 11, if you run it at the top level, it will play the
same no matter what hardware you use since the search parameters are fixed.

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Don Dailey
 Sent: Thursday, October 11, 2007 10:33 AM
 To: computer-go
 Subject: Re: [computer-go] best approach forward
 
 
 -BEGIN PGP SIGNED MESSAGE-
 Hash: SHA1
 
 David Fotland wrote:
  I agree. Computer go needs someone who will play large 
 tournaments are 
  publish results.  I'm also curious how Many Faces would do against 
  Mogo on 19x19 in a long match.  Mogo is much better at 
 endgames, and 
  is much greedier, but Many Faces is much stronger tactically.  
  Certainly if there were ratings it would motivate me to 
 work on Many 
  Faces more.  I think games against mogo would help me find bugs in 
  Many Faces.
  
  Are you volunteering?  Is the 19x19 cgos still going?  I could put 
  Many Faces on it if there is any interest.
  
  David
 
 I believe Many Faces is probably stronger than Mogo but I don't know
 that this has been proven.   But that's my point, I don't think anyone
 really knows for sure where any of the top programs really 
 stand unless they know due to private testing on their own or 
 anecdotal claims.
 
 But what I had in mind in some kind of ratings agency where 
 the conditions are controlled and everything is completely open.
 
 Here is what is required:
 
   1. Someone with at least 2 equal DEDICATED computers plus a server.
   2. Someone willing to do the work.
   3. Software to manage the testing.
 
 Anyone wishing to get on the rating list would have to 
 submit a binary or executable of some kind to the testing agency.
 
 Some kind of restriction concerning which programs get 
 tested.  With a limited resource testing agency you can't 
 have hundreds of programs and versions being tested.
 
 Of course something like CGOS is simpler, but there is no way 
 to verify
 what is being run and what hardware it is being run on.   There is
 nothing to stop you from running Many Faces or some other 
 program and claiming it is your program or running on an 8 
 processor system and
 claiming it's 1 processor.   (although it often becomes obvious after
 some time.  I suspect that if someone got Many Faces running 
 you or someone else would be able to detect this sooner or 
 later if you were paying attention.)
 
 Perhaps something like CGOS would be required to qualify.   Get your
 program running well on CGOS, after so many games and a good 
 rating you
 earn the right to submit a program to the agency.Or perhaps the
 agency evaluates any submitted programs and decides which 
 ones get rated.
 
 The computers don't have to be equivalent either.  But each 
 machine/software combination has to be considered a separate identity.
 Example:
 
1. Many Faces version X running on core 2 duo e6700
2. Many Faces quad version X running on xyz
 
 and so on.
 
 Unfortunately, this does require some dedication on someones 
 part and a
 bit of trust.   All games should be public and so should the testing
 methodology, etc.
 
 My point is that this probably won't happen in computer Go 
 but it happened long ago in computer chess.
 
 - - Don
 
 -BEGIN PGP SIGNATURE-
 Version: GnuPG v1.4.6 (GNU/Linux)
 Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
 
 iD4DBQFHDl48DsOllbwnSikRAmUCAJi5IoVkDQUrfHxPlEf4hNZh0OPTAJ9VlGoC
 +vQLi+uYD/w6u+soc5Z3bQ==
 =ylOv
 -END PGP SIGNATURE- 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Former Deep Blue Research working on Go

[David Fotland]

I already have experimented with the 9x9 server with an anonymous name :)
The results have aged off the server, but I think it had a rating between
1750 and 1850.  So I had working GTP code about 8 months ago.  I'll give it
a try today on 9x9 to see if it still works.

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Don Dailey
 Sent: Thursday, October 11, 2007 12:22 PM
 To: computer-go
 Subject: Re: [computer-go] Former Deep Blue Research working on Go
 
 
 -BEGIN PGP SIGNED MESSAGE-
 Hash: SHA1
 
 Good common sense answer.   I agree that this could be settled.
 
 I'll go ahead and help Chris Fant set up a the server which 
 he will administer.
 
 Meanwhile, can you experiment with the 9x9 server just to see 
 if you can
 get it working on CGOS?You can use any anonymous name.
 
 - - Don
 
 
 David Fotland wrote:
  It's because strong players play strong moves, and the program has 
  knowledge about the strong moves.  When Mogo plays an 
 unconventional 
  move, Many Faces has less knowledge, and is more likely to do 
  something really stupid. People are more able to respond 
 well to odd moves.
  
  9x9 is a different case, since mogo plays nearly perfectly once the 
  opening is done, unless there is a rare tactic that falls 
 outside the 
  uct tree so the monte carlo doesn't see it.  In 19x19 middle games, 
  mogo is still relying on the monte carlo playouts rather 
 than the uct 
  tree, so it is more sensitive to tactics.  I've watched it 
 play 19x19, 
  and it plays greedy for territory while leaving many weaknesses.  A 
  human will focus on the weaknesses and find some deep tactics to 
  exploit them.  Many Faces won't do this since it expects 
 the opponent 
  to play the honest move and not leave this kind of weakness.
  
  But the only way to settle this is to do some experiments.  I could 
  certainly be wrong.  If we have a mogo-many faces match on 
 19x19 cgos, 
  and we also have them play for ratings against people on 
 kgs, it would 
  settle it.
  
  David
 -BEGIN PGP SIGNATURE-
 Version: GnuPG v1.4.6 (GNU/Linux)
 Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
 
 iD8DBQFHDnfuDsOllbwnSikRAjIxAKDMtn/IC7ybKC40Gc73k93y5zkOxACg4qoT
 JOZy56ZHYDPqyno9XMqLhuk=
 =9Kou
 -END PGP SIGNATURE- 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Former Deep Blue Research working on Go

[David Fotland]

Then they are stronger than many face against people.  I think Many Faces
would be around 4k to 6k.

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Eric Boesch
 Sent: Thursday, October 11, 2007 1:50 PM
 To: computer-go
 Subject: Re: [computer-go] Former Deep Blue Research working on Go
 
 
 On 10/11/07, David Fotland [EMAIL PROTECTED] wrote:
  But the only way to settle this is to do some experiments.  I could 
  certainly be wrong.  If we have a mogo-many faces match on 
 19x19 cgos, 
  and we also have them play for ratings against people on 
 kgs, it would 
  settle it.
 
 Mogobot1 and mogobot2 are rated 2k and 3k, respectively, on 
 KGS. CrazyStone is rated 2k. All of these numbers are with 
 moderate time controls (not the 15 minute sudden death time 
 controls that became a subject of controversy).
 
 There was also KCConGui, running KCC Igo, that played for a 
 while on KGS. I don't know whether it was an official bot, or 
 whether its departure had anything to do with its lopsided 
 losing record against CrazyStone. The KCConGui page notes 
 that KCC Igo won the Gifu Challenge four years in a row, most 
 recently against sparse competition, but the best claim to 
 the computer go throne belongs to Steenvreter, for edging out 
 Mogo and CrazyStone in the stronger ICGA tournament. 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] best approach forward

[David Fotland]

I looked at all the games and scored them with Many Faces and I agree with
the result, 14 wins, 5 losses, and one unfinished early.  It looks like
Crazy stone is stronger than any traditional program at 19x19.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Ian Osgood
 Sent: Friday, October 12, 2007 11:54 AM
 To: computer-go
 Subject: Re: [computer-go] best approach forward
 
 
 
 On Oct 11, 2007, at 11:01 AM, Rémi Coulom wrote:
 
  In case nobody noticed, Crazy Stone won a match against KCC Igo
  this summer, with 15 wins and 4 losses. The match was organized by  
  Hiroshi Yamashita. The games can be found in the KGS archives.
  
 http://www.gokgs.com/gameArchives.jsp?user=kcconguiyear=2007month=7
 
  Rémi
 
 Are you certain of the result? Two of the games I examined have  
 scores that don't take into account removal of dead stones.  (W+66.5  
 should be B+20, and W+48.5 looks like a 1 pt game.)  This makes me  
 wonder if some of the other losses and unfinished games (by score or  
 time) are actually wins for KCC Igo on the board. I would certainly  
 like to see the twenty games of this match validated, corrected,  
 commented, and preserved for posterity.
 
 If this is valid, it is quite an achievement! I've updated my own  
 ranking estimates accordingly:
 
http://senseis.xmp.net/?GoPlayingPrograms%2FDiscussion#toc8
 
 Ian
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS

[David Fotland]

10 minutes per side should be enough for Many Faces 11.  Version 11 has
fixed search limits, and only does time management if it runs low on time.
It can usually play a game in 10 minutes on the computer I'll use.  It will
be slower against Mogo since the games are longer and there might me more
unsettled situations to read.  If you do add more time, 15 or 20 minutes per
side should be enough.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Chris Fant
 Sent: Tuesday, October 23, 2007 1:27 PM
 To: computer-go
 Subject: Re: [computer-go] 19x19 CGOS
 
 
 I oppose more time per side.
 
 On 10/23/07, Christoph Birk [EMAIL PROTECTED] wrote:
  On Tue, 23 Oct 2007, Olivier Teytaud wrote:
   http://www.lri.fr/~teytaud/cgosStandings.html
  
   If someone wants to test it, the port is 6919 on machine 
   pc5-120.lri.fr. 10 minutes per side. But only try it if 
 you want to 
   take risks, it is almost surely not stable yet, and the 
 connection 
   might be refused for an unknown reason :-)
 
  Am really curious to see MFGO, Crazystone and Mogo play at 
 19x19. But 
  I suggest allowing more time, at least 20 minutes per side.
 
  Christoph
 
  ___
  computer-go mailing list
  computer-go@computer-go.org 
  http://www.computer-go.org/mailman/listinfo/computer-go/
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS

[David Fotland]

I just tried it, but I can't connect.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 Christoph Birk
 Sent: Tuesday, October 23, 2007 1:16 PM
 To: [EMAIL PROTECTED]; computer-go
 Subject: Re: [computer-go] 19x19 CGOS
 
 
 On Tue, 23 Oct 2007, Olivier Teytaud wrote:
  http://www.lri.fr/~teytaud/cgosStandings.html
 
  If someone wants to test it, the port is 6919 on machine 
  pc5-120.lri.fr. 10 minutes per side. But only try it if you want to 
  take risks, it is almost surely not stable yet, and the connection 
  might be refused for an unknown reason :-)
 
 Am really curious to see MFGO, Crazystone and Mogo play at 
 19x19. But I suggest allowing more time, at least 20 minutes per side.
 
 Christoph
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] XML alternatives to SGF

[David Fotland]

GMP (go modem protocol) was invented for direct computer to computer play
using a 300 baud modem, before the internet existed.  It was used in
tournaments since it was easy to connect up serial ports to emulate modems.

GTP solves a completely different problem, of go engines communicating with
a GUI or referee.  GMP is no longer interesting since no one does direct
computer to computer connection using modems today.

GTP may replace GMP as a tournament method, but it is in no way a
replacement for GMP.  GMP is simply obsolete, since the problem it solves no
longer exists.

Replacing sgf with xml is very different, since they do the same thing.

David


 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 Jeff Nowakowski
 Sent: Tuesday, October 23, 2007 5:57 AM
 To: computer-go
 Subject: Re: [computer-go] XML alternatives to SGF
 
 
 On Tue, 2007-10-23 at 08:42 -0400, Don Dailey wrote:
  GTP pretty much replace GMP.A lot of resistance because 
 GMP was the
  defacto standard at the time.   It would have been foolish 
 to insist on
  being backwards compatible.
 
 GTP was a huge change in protocol with clear benefits.  
 What's being quibbled over now is minor change in the 
 coordinate system at the cost of breaking all existing tools, 
 with the exception of a couple that have implemented this 
 incompatible change.  The benefit does not outweight the cost.
 
 -Jeff
 
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] XML alternatives to SGF

[David Fotland]

 
There is a standard file format like PGN for Go, that uses standard go
notation.  It's the Ishi Press Go Format used by the original Many Faces of
go, and still supported by Many Faces.  You might still find files out there
with a .go extension.  It was invented before sgf, but the go commuinity
preferred sgf, I think because the files were smaller.  
 
I think Anders supports using Korshelt notation in sgf (E4, etc, with no I).
 
David
 
   I'd just like to see adoption of a standard file format for go that is
human readable.  To me, that requires standard algebraic notation such as
E4.I'd definitely prefer to see an sgf-like or PGN-like variant over
XML.  I can't yet commoent on JSON or YAML. 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] XML alternatives to SGF

[David Fotland]

Anders Kierulf, who created the sgf standard for the go program he wrote for
his Ph.D thesis in 1990.  Search for Smart Go to see his current go
program.

-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Jason House
Sent: Thursday, October 25, 2007 11:45 AM
To: computer-go
Subject: Re: [computer-go] XML alternatives to SGF




On 10/25/07, David Fotland [EMAIL PROTECTED] wrote: 


I think Anders supports using Korshelt notation in sgf (E4, etc, with no I).



Who or what is Anders?


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS

[David Fotland]

most computer-computer tournaments have used 1 hour per side, and did 5 or 6
rounds over 1 1/2 days.

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 David Fotland
 Sent: Thursday, October 25, 2007 1:04 PM
 To: 'computer-go'
 Subject: RE: [computer-go] 19x19 CGOS
 
 
 I have no problem with longer time controls.  Many Faces 11 
 was tuned to play in about 45 minutes on hardware available 
 in 2000.  It won't take advantage of any extra time given.  
 The global search is 1 ply with quiescence, and always will 
 always complete, and the local search sizes are fixed at 
 something like 200 nodes per search.
 
 David
 
  -Original Message-
  From: [EMAIL PROTECTED]
  [mailto:[EMAIL PROTECTED] On Behalf Of Don Dailey
  Sent: Thursday, October 25, 2007 11:53 AM
  To: computer-go
  Subject: Re: [computer-go] 19x19 CGOS
  
  
  Hi David,
  
  I argue that the matches should be longer,  perhaps 30 minutes per
  side.They should more closely resemble  time controls used in a
  serious competition.
  
  Here is the reason I say that.One could argue that with 
  computers it
  doesn't matter,  they do not need to be constrained as much
  by our sense of time - they do not feel pressure or get 
  rattled if they play too fast
  and they don't get bored or lose focus if they play too 
 slow. I've
  argued that way myself many times. 
  
  However, the choice of time control, in my estimation,  has a
  good chance of influencing the outcome, especially if we view 
  this as a test of a strong commercial program versus a new 
  experimental technology, which I think it is.  Mogo is a 
  program that clearly performs better
  with more time.I suspect that MFGO is a program that is close to
  optimal at 10 or 15 minutes. I can't say that for sure,  
  perhaps you
  can give us your insights on that.
  
  In such a case what is fair depends on the point of view of the
  observer.   If  someone wanted to see Mogo dominate such a match he
  would consider short time controls unfair and the 
 opposite would be
  true if one wanted to see Many Faces win. Of course I could be
  wrong,  perhaps Many Faces is the one that would benefit more
  from extra time - but I'm working from the  assumption that 
  Mogo would benefit the most based on my own knowledge of how 
  UCT works.
  
  Regardless of the time control used another issue is the
  selection of hardware.  Doubling the computer power 
  effectively doubles the programs
  thinking time.
  
  Having considered all of these issues,  and also taking into
  consideration that this is a contest of sorts,  it makes 
  sense that we should testing  at a level that simulates or at 
  least approaches serious
  computer chess time-controls. Certainly no faster than 
 30 minutes
  per side.These are levels at which most humans will take 
  the results
  seriously.
  
  In addition to this,  it makes sense to know what hardware and what
  time-setting is being used.   Many programs on CGOS were set to play
  very fast, often indicated their level in the name of the
  program something like mogo4k or something similar.
  
  So if we set a liberal time control on CGOS 19x19  we could
  publish the
  identify of the players and draw conclusion based on that. Mogo
  could be tested at several levels and/or hardware 
  configurations and so could Many Faces.  It's not difficult 
  to set up a rotating script for
  logging off one bot and starting up another. (By the way, 
  the right
  way to do this is to select the bot RANDOMLY,  not to 
 rotate back and
  forth.)
  
  The server does report the time each side spent calculating
  in the SGF files, although it's not reported on the web 
  sites, so this is useful
  information if we are considering the scalability of 
 programs.  My
  feeling is that there is likely to be a crossover point - 
  that MFGO will win at time-controls faster than this and Mogo 
  will win at time-controls
  slower than this.That point may be beyond what we can 
 test, or it
  may be testable on the CGOS server soon.
  
  By the way,  I would probably argue for longer than 30
  minutes per side,  but for a server like CGOS that would 
  involve a long wait between
  matches.   
  
  Anyway, that's my 2 cents.
  
  - Don
  
  
  
  
  David Fotland wrote:
   10 minutes per side should be enough for Many Faces 11.  
 Version 11
   has fixed search limits, and only does time management if 
  it runs low
   on time. It can usually play a game in 10 minutes on the
  computer I'll
   use.  It will be slower against Mogo since the games are 
 longer and
   there might me more unsettled situations to read.  If you 
  do add more
   time, 15 or 20 minutes per side should be enough.
  
   David
  
 
   -Original Message-
   From: [EMAIL PROTECTED]
   [mailto:[EMAIL PROTECTED] On Behalf Of
  Chris Fant
   Sent: Tuesday, October 23, 2007 1:27 PM

RE: [computer-go] 19x19 CGOS

[David Fotland]

I puton Many Faces version 11, but it might not be playing at fill strength.
It ouwld be nice if I can click on a game to see the sgf record.  right now
it gives an error.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 Olivier Teytaud
 Sent: Friday, October 26, 2007 9:15 AM
 To: computer-go
 Subject: [computer-go] 19x19 CGOS
 
 
 The cgos 19x19 server is seemingly ok,
 the port 6919 is now opened for all the universe.
 
 The name of the machine is cgos.lri.fr (and not 
 pc5-120.lri.fr as previously).
 
 The port is 6919. It is 19x19, 10 minutes per side for 
 testing; I will move to something longer later (depending on 
 what people prefer, I'll do a weighted average of durations 
 suggested on the mailing list :-) ).
 
 http://www.lri.fr/~teytaud/cgosStandings.html
 
 Unfortunately, I'll be away from my email
 from tomorrow to wednesday and will not be able to
 correct the troubles that people will almost surely find
 in this installation; sorry for that.
 The installation is a bit complicated in order to avoid 
 troubles due to the firewall and I am almost sure that some 
 troubles will appear very soon :-)
 
 All comments welcome (in particular in the next hours as I am 
 still close to my computer a few hours :-) ). [EMAIL PROTECTED]
 
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS

[David Fotland]

no, I never got the viewer to work for me.  

I was too conservative with time control so Many Faces is only playing at
level 8 (of 10), and finishing its games in 2 or 3 minutes.  But it's
winning them all, so I guess I should prefer short time limits :)

Since Many Faces was originally written for a 12 MHz x286, it works pretty
well at very short time limits.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Don Dailey
 Sent: Friday, October 26, 2007 6:30 PM
 To: computer-go
 Cc: [EMAIL PROTECTED]
 Subject: Re: [computer-go] 19x19 CGOS
 
 
 Are you able to watch the games in the viewer ok?I am watching one
 of your games right now.
 
 
 - Don
 
 
 
 David Fotland wrote:
  I puton Many Faces version 11, but it might not be playing at fill 
  strength. It ouwld be nice if I can click on a game to see the sgf 
  record.  right now it gives an error.
 
  David
 

  -Original Message-
  From: [EMAIL PROTECTED]
  [mailto:[EMAIL PROTECTED] On Behalf Of 
  Olivier Teytaud
  Sent: Friday, October 26, 2007 9:15 AM
  To: computer-go
  Subject: [computer-go] 19x19 CGOS
 
 
  The cgos 19x19 server is seemingly ok,
  the port 6919 is now opened for all the universe.
 
  The name of the machine is cgos.lri.fr (and not
  pc5-120.lri.fr as previously).
 
  The port is 6919. It is 19x19, 10 minutes per side for
  testing; I will move to something longer later (depending on 
  what people prefer, I'll do a weighted average of durations 
  suggested on the mailing list :-) ).
 
  http://www.lri.fr/~teytaud/cgosStandings.html
 
  Unfortunately, I'll be away from my email
  from tomorrow to wednesday and will not be able to
  correct the troubles that people will almost surely find
  in this installation; sorry for that.
  The installation is a bit complicated in order to avoid
  troubles due to the firewall and I am almost sure that some 
  troubles will appear very soon :-)
 
  All comments welcome (in particular in the next hours as I am
  still close to my computer a few hours :-) ). 
 [EMAIL PROTECTED]
 
 
  ___
  computer-go mailing list
  computer-go@computer-go.org
  http://www.computer-go.org/mailman/listinfo/computer-go/
 
  
 
 
  ___
  computer-go mailing list
  computer-go@computer-go.org 
  http://www.computer-go.org/mailman/listinfo/computer-go/
 

 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS

[David Fotland]

Thanks.  It works for me now.

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Don Dailey
 Sent: Friday, October 26, 2007 6:58 PM
 To: computer-go
 Subject: Re: [computer-go] 19x19 CGOS
 
 
 As far as I know the viewer works just fine.  
 
 Has anyone else tried the windows viewer on the new 19x19 
 site? I haven't tried it with windows, but you must pass the 
 site and port number to the viewer from the command line like this:
 
cgosview.exe  cgos.lri.fr  6919
 
 
 The viewer is a really nice way to look at games.  A 3rd 
 argument will let you view a specific game number:
 
cgosview.exe  cgos.lri.fr  6919  777  
(view game 777)
 
 
 - Don
 
 
 
 David Fotland wrote:
  no, I never got the viewer to work for me.
 
  I was too conservative with time control so Many Faces is 
 only playing 
  at level 8 (of 10), and finishing its games in 2 or 3 minutes.  But 
  it's winning them all, so I guess I should prefer short 
 time limits :)
 
  Since Many Faces was originally written for a 12 MHz x286, it works 
  pretty well at very short time limits.
 
  David
 

  -Original Message-
  From: [EMAIL PROTECTED]
  [mailto:[EMAIL PROTECTED] On Behalf Of 
 Don Dailey
  Sent: Friday, October 26, 2007 6:30 PM
  To: computer-go
  Cc: [EMAIL PROTECTED]
  Subject: Re: [computer-go] 19x19 CGOS
 
 
  Are you able to watch the games in the viewer ok?I am 
 watching one
  of your games right now.
 
 
  - Don
 
 
 
  David Fotland wrote:
  
  I puton Many Faces version 11, but it might not be playing at fill
  strength. It ouwld be nice if I can click on a game to 
 see the sgf 
  record.  right now it gives an error.
 
  David
 


  -Original Message-
  From: [EMAIL PROTECTED]
  [mailto:[EMAIL PROTECTED] On Behalf Of
  Olivier Teytaud
  Sent: Friday, October 26, 2007 9:15 AM
  To: computer-go
  Subject: [computer-go] 19x19 CGOS
 
 
  The cgos 19x19 server is seemingly ok,
  the port 6919 is now opened for all the universe.
 
  The name of the machine is cgos.lri.fr (and not 
 pc5-120.lri.fr as 
  previously).
 
  The port is 6919. It is 19x19, 10 minutes per side for 
 testing; I 
  will move to something longer later (depending on what people 
  prefer, I'll do a weighted average of durations suggested on the 
  mailing list :-) ).
 
  http://www.lri.fr/~teytaud/cgosStandings.html
 
  Unfortunately, I'll be away from my email
  from tomorrow to wednesday and will not be able to
  correct the troubles that people will almost surely find in this 
  installation; sorry for that. The installation is a bit 
 complicated 
  in order to avoid troubles due to the firewall and I am 
 almost sure 
  that some troubles will appear very soon :-)
 
  All comments welcome (in particular in the next hours as 
 I am still 
  close to my computer a few hours :-) ).
  
  [EMAIL PROTECTED]
  
  ___
  computer-go mailing list
  computer-go@computer-go.org 
  http://www.computer-go.org/mailman/listinfo/computer-go/
 
  
  
  ___
  computer-go mailing list
  computer-go@computer-go.org
  http://www.computer-go.org/mailman/listinfo/computer-go/
 


  ___
  computer-go mailing list
  computer-go@computer-go.org
  http://www.computer-go.org/mailman/listinfo/computer-go/
 
  
 
 
  ___
  computer-go mailing list
  computer-go@computer-go.org 
  http://www.computer-go.org/mailman/listinfo/computer-go/
 

 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS

[David Fotland]

10 minutes is slightly too fast for Many Faces full strength.  It plays most
of the game at level 10, then drops down.

Also, the gnugo 10 that's fixed at 1800 doesn't remove dead stones, so the
score is often wrong.

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Don Dailey
 Sent: Friday, October 26, 2007 6:24 PM
 To: computer-go
 Cc: [EMAIL PROTECTED]
 Subject: Re: [computer-go] 19x19 CGOS
 
 
 Olivier needs to put a .htaccess file in the SGF directory 
 that looks like this:
 
 -[ snip ]---
 AddType application/x-go-sgf sgf
 -[ snip ]-
 
 - Don
 
 
 
 
 David Fotland wrote:
  I puton Many Faces version 11, but it might not be playing at fill 
  strength. It ouwld be nice if I can click on a game to see the sgf 
  record.  right now it gives an error.
 
  David
 

  -Original Message-
  From: [EMAIL PROTECTED]
  [mailto:[EMAIL PROTECTED] On Behalf Of 
  Olivier Teytaud
  Sent: Friday, October 26, 2007 9:15 AM
  To: computer-go
  Subject: [computer-go] 19x19 CGOS
 
 
  The cgos 19x19 server is seemingly ok,
  the port 6919 is now opened for all the universe.
 
  The name of the machine is cgos.lri.fr (and not
  pc5-120.lri.fr as previously).
 
  The port is 6919. It is 19x19, 10 minutes per side for
  testing; I will move to something longer later (depending on 
  what people prefer, I'll do a weighted average of durations 
  suggested on the mailing list :-) ).
 
  http://www.lri.fr/~teytaud/cgosStandings.html
 
  Unfortunately, I'll be away from my email
  from tomorrow to wednesday and will not be able to
  correct the troubles that people will almost surely find
  in this installation; sorry for that.
  The installation is a bit complicated in order to avoid
  troubles due to the firewall and I am almost sure that some 
  troubles will appear very soon :-)
 
  All comments welcome (in particular in the next hours as I am
  still close to my computer a few hours :-) ). 
 [EMAIL PROTECTED]
 
 
  ___
  computer-go mailing list
  computer-go@computer-go.org
  http://www.computer-go.org/mailman/listinfo/computer-go/
 
  
 
 
  ___
  computer-go mailing list
  computer-go@computer-go.org 
  http://www.computer-go.org/mailman/listinfo/computer-go/
 

 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS

[David Fotland]

How about leaving gnugo 10 at 1800 and let gungo level 0 float for a while.
See what rating gnugo level 0 gets, then lock it there as an anchor.

If these two programs aren't 600 points apart and you anchor them that way
it will prevent the rating system from stabilizing.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 Olivier Teytaud
 Sent: Friday, October 26, 2007 11:37 PM
 To: computer-go
 Subject: Re: [computer-go] 19x19 CGOS
 
 
 The anchors are:
 
 /usr/games/gnugo --mode gtp --score aftermath 
 --capture-all-dead --chinese-rules --level 0
 
 /usr/games/gnugo --mode gtp --score aftermath 
 --capture-all-dead --chinese-rules --level 10
 
 The numbers (1200 and 1800) are arbitrary; all suggestions 
 welcome, as for the command-line above. I have a trouble with 
 the positional superko, I'll check that soon.
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS

[David Fotland]

At 10 minute time limits Many Faces rated over 2000 and was top of the list.
At 30 minutes it's 1650.  Many Faces 11 was tuned for the machines in the
1990s, and clearly it needs work for modern machines.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 David Fotland
 Sent: Friday, October 26, 2007 6:13 PM
 To: [EMAIL PROTECTED]; 'computer-go'
 Subject: RE: [computer-go] 19x19 CGOS
 
 
 I puton Many Faces version 11, but it might not be playing at 
 fill strength. It ouwld be nice if I can click on a game to 
 see the sgf record.  right now it gives an error.
 
 David
 
  -Original Message-
  From: [EMAIL PROTECTED]
  [mailto:[EMAIL PROTECTED] On Behalf Of 
  Olivier Teytaud
  Sent: Friday, October 26, 2007 9:15 AM
  To: computer-go
  Subject: [computer-go] 19x19 CGOS
  
  
  The cgos 19x19 server is seemingly ok,
  the port 6919 is now opened for all the universe.
  
  The name of the machine is cgos.lri.fr (and not
  pc5-120.lri.fr as previously).
  
  The port is 6919. It is 19x19, 10 minutes per side for
  testing; I will move to something longer later (depending on 
  what people prefer, I'll do a weighted average of durations 
  suggested on the mailing list :-) ).
  
  http://www.lri.fr/~teytaud/cgosStandings.html
  
  Unfortunately, I'll be away from my email
  from tomorrow to wednesday and will not be able to
  correct the troubles that people will almost surely find
  in this installation; sorry for that.
  The installation is a bit complicated in order to avoid
  troubles due to the firewall and I am almost sure that some 
  troubles will appear very soon :-)
  
  All comments welcome (in particular in the next hours as I am
  still close to my computer a few hours :-) ). 
 [EMAIL PROTECTED]
  
  
  ___
  computer-go mailing list
  computer-go@computer-go.org
  http://www.computer-go.org/mailman/listinfo/computer-go/
  
 
 
 ___
 computer-go mailing list
 computer-go@computer-go.org
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS

[David Fotland]

ver 11 does 1 ply search with quiescence so there is no way to crank it up.
Ver 12 uses full board alpha beta, but it's too buggy right now to put on
cgos.  if this server stays up for a while, I'll use it for testing of ver
12.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Ray Tayek
 Sent: Saturday, October 27, 2007 11:05 AM
 To: computer-go
 Subject: RE: [computer-go] 19x19 CGOS
 
 
 At 09:53 AM 10/27/2007, you wrote:
 At 10 minute time limits Many Faces rated over 2000 and was 
 top of the 
 list. At 30 minutes it's 1650.  Many Faces 11 was tuned for the 
 machines in the 1990s, and clearly it needs work for modern machines.
 
 i have a copy of 11. is there any way to crank it up other than level 
 10. maybe a config file somewhere? have you considered a highly 
 configurable version 12 for some of us on the list?
 
 thanks
 
 ---
 vice-chair http://ocjug.org/
 
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS

[David Fotland]

NO, it's because gnugo got stronger with longer time limits.  When the time
limit got longer Many Faces started taking 1 minute instead of 5 minutes, so
there may be a bug in Many Faces GTP interface time control.

DAvid

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 Christoph Birk
 Sent: Saturday, October 27, 2007 12:07 PM
 To: computer-go
 Subject: Re: [computer-go] 19x19 CGOS
 
 
 On Oct 27, 2007, at 9:53 AM, David Fotland wrote:
  At 10 minute time limits Many Faces rated over 2000 and was top of
  the list.
  At 30 minutes it's 1650.  Many Faces 11 was tuned for the machines  
  in the
  1990s, and clearly it needs work for modern machines.
 
 I don't understand that. The anchor does not take advantage of the  
 time-limit
 change. I always uses about 3 minutes. It probably means that the  
 2000 rating
 was a fluke.
 
 Christoph
 
 
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS

[David Fotland]

Because gnugo has time control and when time is short it adjusts the level
down between moves.  I think with th 30 minute control it is staying at
level 10 the whole game.

I just found a time control bug in Many Faces, and it's been playing at
level 3.  It should get stronger soon :)


 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 Christoph Birk
 Sent: Saturday, October 27, 2007 3:51 PM
 To: computer-go
 Subject: Re: [computer-go] 19x19 CGOS
 
 
 On Oct 27, 2007, at 3:17 PM, David Fotland wrote:
  NO, it's because gnugo got stronger with longer time limits.
 
 Did it? I thought the anchor (gnugo-level-10) plays just 
 that, at level10. How would it get stronger?
 
  When the time
  limit got longer Many Faces started taking 1 minute instead of 5
  minutes, so
  there may be a bug in Many Faces GTP interface time control.
 
 That might be the explanation.
 
 Christoph
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS

[David Fotland]

You're right.  the problem was Many Faces was playing at level 3 instead of
10.  I fixed it and now Many Faces is taking 5 minutes per game rather than
1 minute.  It's rating should come back up now.

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 Christoph Birk
 Sent: Saturday, October 27, 2007 4:50 PM
 To: computer-go
 Subject: Re: [computer-go] 19x19 CGOS
 
 
 On Oct 27, 2007, at 3:59 PM, David Fotland wrote:
 
  Because gnugo has time control and when time is short it adjusts
  the level
  down between moves.  I think with th 30 minute control it is  
  staying at
  level 10 the whole game.
 
 But even now it is only using 3 minutes ... it was not short of time  
 even
 during 10 minute games.
 
 Christoph
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS

[David Fotland]

Would anyone be interested in a highly configurable version 11 with gtp
interface?

Version 11 has a set of parameters that control the searching that I can
easily read from a file.  

/* LEVELS:1  2  3  4  5  6  7  8   9  10  */
int maxmoves[NUMLEVELS] =   /* maximum number of moves to try on full board
*/
{  0, 2, 4, 5, 6, 8,10,12, 15, 20, 28 };  /* lots, so in endgame can
look at lots of moves */
int maxvariations[NUMLEVELS] =  /* max number of leafs per move tried */
{  0, 1, 1, 2, 2, 3, 4, 5, 7,  10,  13 };
char maxscorebrdepth[NUMLEVELS] =   /* max depth for any branches in
getscore scorebestmove */
{  0, 0, 0, 1, 1, 2, 2, 2, 2,  3,  3 }; 
char maxscoredepth[NUMLEVELS] = /* max depth for getscore */
{  0, 0, 0, 1, 1, 2, 2, 3, 3,  3,  4 };
int maxlifecalls[NUMLEVELS] =  /* total evaluations, should be around
maxmoves*maxvariations */
{  0, 5, 9,13,20,30,45,65,95,200,400 };

/* LEVELS:1  2  3  4  5  6  7  8   9  10  */
unsigned char taclibs[NUMLEVELS] = /* max liberties in a tactical fight
*/
{  0, 2, 2, 3, 3, 3, 3, 4, 4,  4,  4 };
unsigned char eyetaclibs[NUMLEVELS] = /* max liberties for eye diagonal
*/
{  0, 2, 2, 2, 2, 2, 3, 3, 3,  3,  3 };
int cancapsize[NUMLEVELS] = /* size of search in canbecaptured */
{  0, 7,10,15,20,30,40,60,80,110,150 };
unsigned char eyecapsize[NUMLEVELS] = /* size of search for eyes diags */
{  0, 2, 3, 4, 5,10, 15,20,25,30, 40 };
unsigned char eyecapdepth[NUMLEVELS] = /* depth of search for eyes diags */
{  0, 2, 2, 3, 3, 4, 5, 6, 6,  6,  6 };
unsigned char conncapsize[NUMLEVELS] = /* size of search for connections */
{  0, 4, 6, 8,10, 20,30,40,55,80,100 };
unsigned char conncapdepth[NUMLEVELS] = /* depth of search for connections
*/
{  0, 2, 3, 4, 5, 7, 8, 9,10, 12, 14 };
char mvmost[NUMLEVELS] =/* number of moves considered for ladder at each
ply */
{  0, 1, 1, 2, 2, 2, 2, 3, 3,  3,  3 };
char eyemost[NUMLEVELS] =/* number of moves considered for ladder at
each ply */
{  0, 1, 1, 1, 2, 2, 2, 2, 2,  3,  3 };
char connmost[NUMLEVELS] =/* number of moves considered for ladder at
each ply */
{  0, 1, 1, 1, 2, 2, 2, 2, 2,  3,  3 };
int maxbranchdepth[NUMLEVELS] = /* maximum depth for branches in tactical
move tree (unless move values are close) */
{  0, 1, 2, 2, 3, 3, 4, 4, 4,  4,  4 };
int maxtacdiff[NUMLEVELS] =  /* maximum difference between best tac move and
this move*/
{  0,16,16,16,32,64,64,96,120,180,250 };
int mintacval[NUMLEVELS] =   /* minimum value move has to be considered
tacticaly */
{  0, 0, 0, 0, 0,-10,-10, -10,-16,-20,-31 };
int numpotmoves[NUMLEVELS] =   /* Number of moves to read for adpot() to
capture group */
{  0, 0, 0, 0, 0, 0, 1, 1, 2, 2, 2 };


/* LEVELS:1  2  3  4  5  6  7  8   9  10 */
int maxjosvariations[NUMLEVELS] =  /* max number of joseki variations -
endpoints per first level joseki move */
{  0, 1, 1, 1, 1, 1, 2, 3, 3,  4,  6 }; 
int maxpatvariations[NUMLEVELS] =  /* max number of pattern variations per
move */
{  0, 1, 1, 1, 2, 2, 2, 3, 4,  5,  6 }; 
int maxjosbranches[NUMLEVELS] =  /* max number of joseki variations per move
at depth 1 */
{  0, 1, 1, 1, 1, 1, 2, 2, 2,  2,  3 }; 
unsigned char mdist[NUMLEVELS] =  /* distance to radiate influence from
live groups */
{  0, 7, 8, 9, 10, 11, 11, 12, 12,  13,  13 };


/* Fights: no fight reading below level 5 */
/* LEVELS:1  2  3  4  5  6  7  8   9  10 */
int maxfightbranches[NUMLEVELS] =  /* max number of fight variations per
move */
{  0, 1, 1, 1, 1, 1, 2, 2, 2,  3,  3 }; 
char maxfightdepth[NUMLEVELS] = /* max depth for reading fight */
{  0, 0, 0, 1, 1, 2, 3, 4, 6,  7,  9 };
int maxfightbrdepth[NUMLEVELS] = /* max depth for branches in reading fight
*/
{  0, 0, 0, 1, 1, 1, 2, 3, 4,  5,  5 };
int maxsemdiff[NUMLEVELS] =  /* maximum difference between best semeai move
and this move*/
{  0, 8,16,24,32,40,50,60,80,90,100 };


  
  i have a copy of 11. is there any way to crank it up other 
 than level
  10. maybe a config file somewhere? have you considered a highly 
  configurable version 12 for some of us on the list?
  


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS

[David Fotland]

I added a copy of Many Faces of Go running at level 1 (with almost no
search) to add some variety for the weak programs.  This version looks at
the top 2 suggestions from the move generator, does a 1 ply search without
quiescence, does a full board evaluation for each, and picks the best one.
Late in the game it includes a pass move in the search, so it does 3
evaluations rather than 2.

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of 
 Olivier Teytaud
 Sent: Friday, October 26, 2007 9:15 AM
 To: computer-go
 Subject: [computer-go] 19x19 CGOS
 
 
 The cgos 19x19 server is seemingly ok,
 the port 6919 is now opened for all the universe.
 
 The name of the machine is cgos.lri.fr (and not 
 pc5-120.lri.fr as previously).
 
 The port is 6919. It is 19x19, 10 minutes per side for 
 testing; I will move to something longer later (depending on 
 what people prefer, I'll do a weighted average of durations 
 suggested on the mailing list :-) ).
 
 http://www.lri.fr/~teytaud/cgosStandings.html
 
 Unfortunately, I'll be away from my email
 from tomorrow to wednesday and will not be able to
 correct the troubles that people will almost surely find
 in this installation; sorry for that.
 The installation is a bit complicated in order to avoid 
 troubles due to the firewall and I am almost sure that some 
 troubles will appear very soon :-)
 
 All comments welcome (in particular in the next hours as I am 
 still close to my computer a few hours :-) ). [EMAIL PROTECTED]
 
 
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS

[David Fotland]

I'm working on MFGO 12 and I'd like 30 minutes so I can test against a
variety of programs at tournament time limits.

I don't need hundreds of games to tune, since my program is knowledge based.
I'm not just changing parameters and seeing what happens.  I'm looking for
bad moves and adding knowledge.

David

 
 About fairness, as classical programs including GNU Go and ManyFaces 
 need about ten minutes for their best performace, why do you give 
 other (Monte Carlo) programs thirty minutes?
 
 I argue ten or fifteen minutes setting is enough and better for 
 many developers than thirty minutes.
 
 -Hideki


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] 19x19 CGOS (CS vs MFG)

[David Fotland]

Oops, I forgot to tell it to randomize.  I'll restart it with random turned
on. 

David

 -Original Message-
 From: [EMAIL PROTECTED] 
 [mailto:[EMAIL PROTECTED] On Behalf Of Rémi Coulom
 Sent: Sunday, October 28, 2007 1:39 PM
 To: computer-go
 Subject: Re: [computer-go] 19x19 CGOS (CS vs MFG)
 
 
 Edward de Grijs wrote:
  The CrazyStone row has dissapeared because not enough
  games were played, so there will be a larger standard 
 deviation around 
  those values (I expect a 1 sigma value of about 50 elo. It would be 
  interesting to incluse those numbers on every row (Don?))
 
 Uncertainty about the rating is much more. Also I stopped 
 CS-8-26-10k-1CPU, because it was deterministic, and so is 
 Many Faces. So 
 they were playing the same game again and again. I have now 
 connected a 
 parallel version, running on two cores, which makes it random.
 
 Rémi
 ___
 computer-go mailing list
 computer-go@computer-go.org 
 http://www.computer-go.org/mailman/listinfo/computer-go/
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

[computer-go] CGOS 19x19 problem

[David Fotland]

mfgo-11 hung last night.  It never responded to a genmove request.  The
client was still running, but the 19x19 web page says mfgo-11 is not
conencted.

I killed the client, and restarted it, bu the server says: Error you are
already logged on.  Closing connection when the client responds to
username with mfgo-11.

Please let me know when this is fixed so I can put mfgo-11 back on the 19x19
server.

David


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

[computer-go] cgos 19x19 hung

[David Fotland]

It looks like CGOS 19x19 is hung.  The game in progress are not changing,
and the cgos viewer can't get the moves of the current games.  When I
connect I get a message that Im connected, but no messages about the time of
the next round.

Can someone restart it please?

David


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Computer Go tournaments - various

[David Fotland]

 
 Ian Osgood wrote:
  What boggles my mind is the lack of participation in these events
 from
  commercial players like KCC Igo, Haruka, Go4++, Handtalk, and Many
 Faces.
 It's because these programs will get killed by the top Monte Carlo
 programs.   It's risky competing when your reputation is involved.
 In
 fact, it's better not to compete than to compete and score poorly.
 

I'd compete, but I didn't know about this one.  Many Faces 11 is tuned for
much slower computers, so it's not competitive.  My latest version is
stronger, but unfinished.

David

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] New engine? From a Chess programmer perspective.

[David Fotland]

It's not clear if you are talking about professional Dan level or Amateur
Dan level.  I've played the top 9x9 programs at 9x9, and so have several
other amateur Dan players, and I think we all agree that the top 9x9
programs have reached amateur Dan level.  I don't think these programs are
as strong as professional Dan players.

At 19x19 their strength is much less clear.  I don't think they are quite to
amateur Dan yet.

David Fotland

 -Original Message-
 From: [EMAIL PROTECTED] [mailto:computer-go-
 [EMAIL PROTECTED] On Behalf Of Robert Jasiek
 Sent: Tuesday, December 04, 2007 7:54 AM
 To: computer-go
 Subject: Re: [computer-go] New engine? From a Chess programmer
 perspective.
 
 Don Dailey wrote:
  Just a few years ago it was widely held that computers will not reach
  Dan level in my lifetime even in 9x9 Go.When it happened in 9x9
  go,  it was not accepted - the day it happened passed us by and
 nobody
  noticed it. It's probably still not common knowledge and it will
  take time for it to be generally believed.
 
 What is the basis of the claim that a program has reached a certain
 human rank level?
 
 There should be systematic tests. Let it play against many humans. Let
 it enter human tournaments. Use a meaningful evaluation context.
 
 For some such evaluation, let me refer to a useful handicap system for
 9x9, which has been used in some European 9x9 Championships: For the
 first 10 rank differences (0, 1,.., 9) decrease the komi from 6.5 for
 an
 even game by 1 point per extra rank. (Komi can become negative.) - OC,
 I
   prefer to see even games. OTOH, until the program rank is well known,
 it may be suitable to let a simgle human (the more humans the better)
 play until the handicap becomes stable.
 
 --
 robert jasiek
 ___
 computer-go mailing list
 computer-go@computer-go.org
 http://www.computer-go.org/mailman/listinfo/computer-go/

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] programs at the US Go Congress

[David Fotland]

I organized the go congress computer go tournaments for many years.

Ask the AGA congress organizers for prize money.  The congress has a big
pool of prize money.  Several years I convinced the organizers to put up
$500 to $1000 in prize money.  Usually the congress budget isn't set until
right before it starts, so they never committed to prize money until the
start of the congress.

Ask the congress organizer about machines.  The congresses are at colleges,
which will have computer labs.  Sometimes they are willing to let us use the
computer lab for a couple of days.  Most participants brought their own
machine in the more recent years, after laptops became inexpensive, so you
might not need computers.

I used to run the tournament on Monday and Tuesday, since Wednesday is a day
off, and an early tournament lets the go players concentrate on the congress
human tournaments after the go tournament is over.  I ran a round robin or
double round robin.  That way you can start games as soon as opponents are
ready and the tournament is over much quicker.

I suggest publishing rules ahead of time to avoid conflicts, and appoint
someone to be tournament director in case of disputes.  You should probably
plan to support both modem protocol and gtp.  Gtp requires a referee
computer, so you will have to bring one extra.

I plan to go to the European congress in Sweden, which should have a
computer competition.  I think it conflicts with the US congress, so I don't
think I'll be there.

David  

Here is the rules and announcement from the last tournament I organized:

1996 UNITED STATES COMPUTER GO CHAMPIONSHIP
.S -2
.R

This announces the 1996 U.S. Computer Go Competition.

The 1996 US Computer Go Championship will take place at the 16th annual
US Go Congress, on July 21, 22, 23rd, at John Carrol University, in
Cleveland, Ohio. 

This is an excellent opportunity to meet and compete with others interested 
in Computer Go.

A plaque or trophy and title of US Computer Go Champion will be awarded
to the winner.  There is no cash prize.  This tournament is not affiliated
with the World Computer Go Congress, although the rules and format are
very similar.

In 1988 and 1989 Acer ran the US Preliminaries to the World Computer
Go Congress at the US Go Congress.  In 1990 they changed their
procedure to have a mail in preliminary in Taiwan.  The United States
Computer Go Championship was organized to ensure that there continues
to be an annual computer go competition in North America.  The informal
discussions and contacts during the tournament help increase the strength
of all the programs.  This will be the seventh United States Computer Go
Championship.  Last year 4 programs participated.  Typically 5 to 7 
programs compete.


.
ENTERING THE CONTEST:
.
You must register for the US Go Congress to enter the Computer Go
Competition.

Please contact David Fotland as soon as possible if you plan to participate.
There is no penalty for withdrawing from participation later.  I prefer
early notice of participation, but will accept new participants up
to the day before the competition.

For additional information on the Computer Go contest, contact:

David Fotland
4863 Capistrano Ave
San Jose Ca 95129-1031
(408)985-1236
E-mail: [EMAIL PROTECTED]

RULES

Contestants must provide their own computer, which must be present at the 
contest site.  All transportation costs and risks will be borne by the
contestant.  There will be a locked room for the contest, but the
Congress will not be responsible for the safety of any equipment left there.

This year there may be some computer hardware available, but at this time
I can't promise anything.

Any Go program can participate in the contest, but only US programs whose
author(s) are US residents are eligible for the title of US Computer Go
Champion.  The title of North American Open Computer Go Champion will be
awarded to the program that finishes first, no matter where it is from.
Each program can only be registered once.  

Play is governed by the SST laws of Go (GOE) published by the Ing Chang Chi
Wei Chi Educational foundation.  In summary, the score is territory plus
live stones (prisoners don't count), suicide of more than one stone
is allowed, repetition is forbidden, surrounded territory in a seki 
counts as points.  Black plays first and gives 8 points komi at the end of
the 
game.  If a program makes an illegal move it loses.  
Mirror Go is not allowed past move number 60.
Any rule disputes will be settled by the tournament referee.

There will be a Go board between the computers which will have the official
game position.  If a move is entered incorrectly the problem may be fixed
while the clock is running.  If the participant is unable to correct his
program to agree with the position on the Go board, the participant loses.

Programs should
be able to remove dead stones from the screen or demonstrate that
the dead stones are recognized at the end of the game.  If both programs

[computer-go] crazystone, mogo, go4++, greenpeep, valkyra or other strong programs on cgos19x19?

[David Fotland]

I'm working on Many Faces of Go 12 engine, which is an alpha-beta searcher,
and it's strong enough now I'd like to some stronger competition on 19x19
CGOS to test against.

Does anyone want to put up some strong programs?  I know everyone prefers to
work on 9x9 since it's simpler, but 9x9 is not go, and is only used in the
real world to play the first couple of games with beginners to teach them
the rules :)  So how about giving 19x19 a try?

-David Fotland


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] crazystone, mogo, go4++, greenpeep, valkyra or other strong programs on cgos19x19?

[David Fotland]

It's a bug.  If the move generator returns no moves at an internal node in
the main search, it passes.  Mogo keeps playing on after there are no points
left.  During the search after the forced response to mogo's move, there are
no moves generated.  I don’t allow a pass in an internal node, so the search
returns an error and it passes.  It will be fixed in the next version (along
with many other bugs).  The Alpha-beta code barely works.  You can see how
weak the first few buggy versions very if you look at gnugo's crosstable.

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?

David

 -Original Message-
 From: [EMAIL PROTECTED] [mailto:computer-go-
 [EMAIL PROTECTED] On Behalf Of Chris Fant
 Sent: Sunday, December 09, 2007 5:47 PM
 To: computer-go
 Subject: Re: [computer-go] crazystone, mogo, go4++, greenpeep, valkyra
 or other strong programs on cgos19x19?
 
 What happened to Many Faces in game 6670?  Did it simply not care
 about all that territory because it already saw the game as a loss?
 
 
 On Dec 9, 2007 2:21 PM, Chris Fant [EMAIL PROTECTED] wrote:
   OK, I have connected Crazy Stone.
  
   Rémi
 
 
  And it's doing very well.
 
 ___
 computer-go mailing list
 computer-go@computer-go.org
 http://www.computer-go.org/mailman/listinfo/computer-go/

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

[computer-go] CGOS 19x19 down?

[David Fotland]

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
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Where and How to Test the Strong Programs?

[David Fotland]

I think AGA and KGS are pretty close.  AGA is a real rating system in that
ratings are earned in sanctioned tournaments so they are not disrupted by
casual games.  http://www.usgo.org/ratings/default.html  

European ratings (also from tournaments) are perhaps 2 stones tougher.  Many
think they are more reasonable, since most feel the top of the amateur
rating should be 7 dan or lower, and top AGA ratings are higher than that.
Top pros that have participated in AGA tournaments have ratings about 10
dan, and there are many amateur 8 dans.

Japanese ratings are less tough.  Japanese amateur ratings can be purchased,
with a test, so there has been inflation, and I don't think there is a
national rating organization like in USA and Europe. 

Korean and Chinese are very tough, since they think a 1 dan amateur should
be close to professional strength.

So, I'm AGA 3 dan, but I would have a tough time playing as a 1 dan in
Europe, and I play at 4 dan in clubs in Japan.

I tried playing in a club as 3 dan in China once, and got totally crushed.

My preference would be a scale that is fixed at the top, with 9 dan pros at
9 dan.  This would put 1 dan pros at about 7 dan and top amateurs at 6 dan,
with a few 7 dans.  This is pretty close to the European scale.  Top
tournament pros almost never lose to pro 1 dans, so there are 300 or more
ELO points between the top amateurs and the top professionals.

So perhaps top human play is 3500 or more on the cgos 19x19 scale.  That's
12 ranks above 2000, with the higher ranks having more ELO points per rank.

David

 -Original Message-
 From: [EMAIL PROTECTED] [mailto:computer-go-
 [EMAIL PROTECTED] On Behalf Of Don Dailey
 Sent: Tuesday, December 11, 2007 6:37 PM
 To: computer-go
 Subject: Re: [computer-go] Where and How to Test the Strong Programs?
 
 I feel that we probably need several more players to have much
 accuracy,  but I don't mind starting the best educated guess we can
 muster - it can be modified at a later time.
 
 How do AGA ratings compare to other systems?   Is any particular system
 considered (defacto or otherwise) more of a standard than some other?
 
 How do AGA ratings compare to KGS?
 
 - Don
 
 
 
 Christoph Birk wrote:
  It looks like my (3k AGA) CGOS rating (tast-3k) is converging around
  2000 ELO. That gives us a zero-point but we need at least one more
  rated player (better more) to get the scale.
  If you would like to use my GUI please contact me by private email
  at ccbirk at gmail dot com.
 
  Christoph
  ___
  computer-go mailing list
  computer-go@computer-go.org
  http://www.computer-go.org/mailman/listinfo/computer-go/
 
 ___
 computer-go mailing list
 computer-go@computer-go.org
 http://www.computer-go.org/mailman/listinfo/computer-go/

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Microsoft Research Lectures: Akihiro Kishimoto

[David Fotland]

This is awful for such a simple problem.  Many Faces' static evaluation
function sees that the white group is unsettled, and
the life/death search finds the B2 killing move in one node (since after B2
the group is dead with no further search, and the move generator returns B2
as the first candidate move).  It spends another 100 nodes proving that D1
and D3 lead to a ko, and the other moves don't work.

David


  For example current version(not released) goes trought 162438 nodes
 before
  it proofs black B2 kills(without any ordering help).
 
  . . . . . . .
  . . . . . . .
  B B B B B . .
  w w w . B . .
  . . . w B . .
  . . w . B . .
  A B C D E F G
 


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Non-global UCT

[David Fotland]

Many Faces' life and death search is best first and probability based, but I
don't use UCT to select moves.  I select the move that has the highest
probability of changing the value of the root (from success to fail or vice
versa).  I don't use MC to evaluate the endpoints.  I look forward down one
line until the result is clear (alive or dead), but I follow the best move
from the move generator, not random moves.

 

David

 

From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of
[EMAIL PROTECTED]
Sent: Wednesday, December 12, 2007 3:11 PM
To: computer-go@computer-go.org
Subject: Re: [computer-go] Non-global UCT

 


Sure, I use MC UCT to test for absolute or conditional life and death of a
string. I think it works well.

-Original Message-
From: Chris Fant [EMAIL PROTECTED]
To: computer-go computer-go@computer-go.org
Sent: Wed, 12 Dec 2007 6:03 pm
Subject: [computer-go] Non-global UCT

Is anyone using UCT (or similar) for non-global searches such as
connectivity/tactics?
___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
  _  


More new features than ever. Check out the new AIM(R) Mail
http://o.aolcdn.com/cdn.webmail.aol.com/mailtour/aol/en-us/text.htm?ncid=ai
mcmp000501 !

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Non-global UCT

[David Fotland]

It's quite different from PN.  PN expands a leaf node one ply and backs up
values to the root.  I play a line as many ply as needed until I get a high
confidence evaluation of win or lose.  In this sense I am doing something
like UCT with nonrandom play outs.  PN typically doesn't use move ordering
information.  I have a good move generator that sorts candidate moves by
probability of winning, so I always pick the highest probability unevaluated
child to expand.

 

Say I want to do a life and death search on a group to kill it.  The move
generator suggests 5 moves.  Make the best one, and evaluate.  If the status
is still unclear, I call the generator to make move to live.  I pick the
best and make it, etc.  Say I go 8 ply deep and find that the group dies.

 

Now I have one line of play that works.   To pick the next node to expand I
want to try a node that can change to root from win to loss.  So I only have
to look at opponent moves.  I pick the move that has the highest probability
of changing the root to a loss (which can be at any depth), and play out
another line until I have a stable result.

 

This models the way I read life and death problems.

 

David

 

From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Jason House
Sent: Thursday, December 13, 2007 5:50 AM
To: computer-go
Subject: Re: [computer-go] Non-global UCT

 

 

On Dec 12, 2007 10:19 PM, David Fotland [EMAIL PROTECTED] wrote:

Many Faces' life and death search is best first and probability based, but I
don't use UCT to select moves.  I select the move that has the highest
probability of changing the value of the root (from success to fail or vice
versa).  I don't use MC to evaluate the endpoints.  I look forward down one
line until the result is clear (alive or dead), but I follow the best move
from the move generator, not random moves.


Out of curiosity, how does that compare to proof number?  Proof number uses
a heuristic like for this outcome to occur, N child nodes must have outcome
X.  It picks the path with the  smallest N. 

 

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] RE: Microsoft Research Lectures: Akihiro Kishimoto

[David Fotland]

Many faces still finds the correct move on the first trial, but now it takes
74 nodes to prove the first move works, rather than one node.
It looks at a total of 114 nodes to prove that no other move works.

David

 -Original Message-
 From: [EMAIL PROTECTED] [mailto:computer-go-
 [EMAIL PROTECTED] On Behalf Of Harri Salakoski
 Sent: Wednesday, December 12, 2007 9:04 PM
 To: computer-go
 Subject: Re: [computer-go] RE: Microsoft Research Lectures: Akihiro
 Kishimoto
 
 5 ...
 4 WW.
 3 BBB..W.
 2 ...B.W.
 1 ..B..W.
   ABCDEFG
 
 Ha, I give blacks more room to play, and shape database did not find
 it, or
 maybe I used it wrong :).
 But this search find B2 now in 1258437 nodes :| which is quite much and
 takes couple seconds.
 Don't know but really hope it goes better so it comes more useful.
 
 t. Harri
 
 - Original Message -
 From: Dave Dyer [EMAIL PROTECTED]
 To: computer-go computer-go@computer-go.org
 Sent: Thursday, December 13, 2007 5:52 AM
 Subject: [computer-go] RE: Microsoft Research Lectures: Akihiro
 Kishimoto
 
 
  My tsumego applet determines without a search that black can kill,
  and white might live if he moves first.
  http://www.andromeda.com/people/ddyer/go/shape/ShapeApplet.html
 
  A table lookup is a little better than searching 162438 nodes :)
 
 
 
   For example current version(not released) goes trought 162438
 nodes
  before
   it proofs black B2 kills(without any ordering help).
 
   . . . . . . .
   . . . . . . .
   B B B B B . .
   w w w . B . .
   . . . w B . .
   . . w . B . .
   A B C D E F G
 
 
  ___
  computer-go mailing list
  computer-go@computer-go.org
  http://www.computer-go.org/mailman/listinfo/computer-go/
 
 ___
 computer-go mailing list
 computer-go@computer-go.org
 http://www.computer-go.org/mailman/listinfo/computer-go/

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

[computer-go] CGOS 19x19 down

[David Fotland]

It looks like CGOS 19x19 is down again.

-David

 -Original Message-
 From: [EMAIL PROTECTED] [mailto:computer-go-
 [EMAIL PROTECTED] On Behalf Of Don Dailey
 Sent: Thursday, December 13, 2007 9:47 AM
 To: Don Dailey
 Cc: computer-go
 Subject: Re: [computer-go] Where and How to Test the Strong Programs?
 
 I'm going to estimate that 100 ELO is roughly 1 rank based on this:
 
  http://en.wikipedia.org/wiki/Go_ranks_and_ratings
 
 This may not hold for 9x9.If a 1 kyu beats a 2kyu about 64% of the
 time in an even game at 19x19,  it doesn't imply that he will do the
 same at 9x9,  but until I have a reason to believe differently I will
 set this as our first guess in the formula.We can always come back
 and modify this later.
 
 We may be able to borrow KGS data of well established players playing
 9x9 games against each other to estimate this. Would anyone like to
 volunteer to do this?
 
 So as it stands, our formula is:
 
   3ky = 1942 cgos_elo
 
dan =  (your_elo - 2142) / 100
kyu  = -dan + 1
 
 1 Dan  =  2242 cgos
 1 Kyu   =  2142 cgos
 2 Kyu   =  2042 cgos
 3 Kyu   =  1942 cgos
 
 The strongest bot on CGOS all time list seems to be  greenpeep0.5.1
 http://cgos.boardspace.net/9x9/cross/greenpeep0.5.1.html with a
 rating
 of 2621.
 
 This implies it is almost equal to a 5 Dan player - which doesn't sound
 right to me.However,  this could be fluky since it is as at the
 extreme end of the scale.  It would be great if this same program
 could play some strong humans at the equivalent time control on KGS at
 9x9 and we could adjust the difference between ranks accordingly.
 
 I suspect there is more than 100 ELO between ranks at 9x9.
 
 - Don
 
 
 
 Don Dailey wrote:
  Christoph,
 
  Your bayeselo rating is 1942 on CGOS.  I compiled a table that
 has
  all players with 50 games or more which can be found here:
 
http://cgos.boardspace.net/9x9/hof2.html
 
 
  - Don
 
 
 
 
 
 
 
 
 
 
  Christoph Birk wrote:
 
  On Tue, 11 Dec 2007, Don Dailey wrote:
 
  Christoph,
  Let me know when you are finished,  what name you are playing under
 and
  I will do the bayeselo thing to get a better figure.
 
  I am playing using the 'tast-3k' account. Right, now I have 71 games
  and a rating of 1979 ELO.
 
 
   Also, I can
  throw out any games that were irregular if you can identify them,
 such
  as if  a match started when you were not looking or your interface
 got
  glitchy or something.
 
  Since I added the GUI I lost no games due to software problems.
  Only a few won games lost to human stupidity :-)
 
  I will take a break over the holidays, maybe playing a few more
  games in the new year, but I guess for my purposes a zero-point
 3k-AGA ~=~ 2000 CGOS-ELO
  is close enough. Unless we get some other (AGA or KGS) rated
  players it not make sense to get a more precise rating for the
  scale.
 
  Christoph
  ___
  computer-go mailing list
  computer-go@computer-go.org
  http://www.computer-go.org/mailman/listinfo/computer-go/
 
 
 
 
 ___
 computer-go mailing list
 computer-go@computer-go.org
 http://www.computer-go.org/mailman/listinfo/computer-go/

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Where and How to Test the Strong Programs?

[David Fotland]

Isn't Greenpeep an alpha-beta searcher, not UCT/MC?

Since Go ranks are based an handicap stones, and 100 ELO points implies a
particular winning percentage, it would be an unlikely coincidence if 1 rank
is 100 ELO points.  Any web site that claims this must be wrong :) and
should have little credibility.

David


 
 The strongest bot on CGOS all time list seems to be  greenpeep0.5.1
 http://cgos.boardspace.net/9x9/cross/greenpeep0.5.1.html with a
 rating
 of 2621.
 
 This implies it is almost equal to a 5 Dan player - which doesn't sound
 right to me.However,  this could be fluky since it is as at the
 extreme end of the scale.  It would be great if this same program
 could play some strong humans at the equivalent time control on KGS at
 9x9 and we could adjust the difference between ranks accordingly.
 
 I suspect there is more than 100 ELO between ranks at 9x9.
 
 - Don
 
 
 
 Don Dailey wrote:
  Christoph,
 
  Your bayeselo rating is 1942 on CGOS.  I compiled a table that
 has
  all players with 50 games or more which can be found here:
 
http://cgos.boardspace.net/9x9/hof2.html
 
 
  - Don
 
 
 
 
 
 
 
 
 
 
  Christoph Birk wrote:
 
  On Tue, 11 Dec 2007, Don Dailey wrote:
 
  Christoph,
  Let me know when you are finished,  what name you are playing under
 and
  I will do the bayeselo thing to get a better figure.
 
  I am playing using the 'tast-3k' account. Right, now I have 71 games
  and a rating of 1979 ELO.
 
 
   Also, I can
  throw out any games that were irregular if you can identify them,
 such
  as if  a match started when you were not looking or your interface
 got
  glitchy or something.
 
  Since I added the GUI I lost no games due to software problems.
  Only a few won games lost to human stupidity :-)
 
  I will take a break over the holidays, maybe playing a few more
  games in the new year, but I guess for my purposes a zero-point
 3k-AGA ~=~ 2000 CGOS-ELO
  is close enough. Unless we get some other (AGA or KGS) rated
  players it not make sense to get a more precise rating for the
  scale.
 
  Christoph
  ___
  computer-go mailing list
  computer-go@computer-go.org
  http://www.computer-go.org/mailman/listinfo/computer-go/
 
 
 
 
 ___
 computer-go mailing list
 computer-go@computer-go.org
 http://www.computer-go.org/mailman/listinfo/computer-go/

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] MC-UCT and tactical information

[David Fotland]

I think Martin Mueller published an improvement to benson's algorithm that
is also proved correct.

 

David

 

From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of John Fan
Sent: Thursday, December 13, 2007 1:36 PM
To: computer-go
Subject: Re: [computer-go] MC-UCT and tactical information

 

Yes, StoneGrid only uses Benson's algorithm.

On Dec 13, 2007 4:30 PM, [EMAIL PROTECTED] wrote:

That's a strong program, and interesting information. For clarity, I assume
that you mean something like Benson's algorithm, while my intended meaning
was alive assuming perfect play. Both are relevant, we just need to keep
them sorted out. 

- Dave Hillis




-Original Message-
From: John Fan [EMAIL PROTECTED]
To: computer-go computer-go@computer-go.org
Sent: Thu, 13 Dec 2007 3:42 pm
Subject: Re: [computer-go] MC-UCT and tactical information

My program StoneGrid calculates unconditional life and death at every move,
in the UCT Tree and in the random playout. I think it helps on its strength
a little bit, especially in the end game. In the begining of the game, seems
to be completely useless. It is slow. But it makes the random playout
slightly shorter. The random playout is about 80 moves per game. I do not
have a comparison if the unconditional life and death is disabled, since it
is not easy to do in the current data structure. 

  _  

More new features than ever. Check out the new AIM(R) Mail
http://o.aolcdn.com/cdn.webmail.aol.com/mailtour/aol/en-us/text.htm?ncid=ai
mcmp000501 !


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

 

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] MC-UCT and tactical information

[David Fotland]

 From: Jason House [EMAIL PROTECTED]




 I've done some dabbling (thought experiments) with  how I'd like to cache
search results and I'm not yet happy with any of them.  Not taking into
account miai and such logic could   lead to excessive storage bloat.  I'd
love to enter a discussion talking just about how to store cached LD search
results. 


Many Faces does life and death search at the root before the main search.
It typically allocates a few hundred nodes to life and death search.  Since
the search is best-first, it keeps the search trees from move to move.
Later searches can extend earlier ones.  The trees are small, so it doesn't
cost much to keep them.

 

During evaluation tactical search results are cached, but the search tree is
not.  During a main search to find the move to play, Many Faces does a few
million nodes of tactical search, so it's too much to cache.

 

David

 

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Hall of fame for CGOS

[David Fotland]

Many Faces does on-line learning of Fuseki, Joseki, and half-board patterns.

David

 -Original Message-
 From: [EMAIL PROTECTED] [mailto:computer-go-
 [EMAIL PROTECTED] On Behalf Of Gunnar Farnebäck
 Sent: Friday, December 14, 2007 1:28 PM
 To: computer-go
 Subject: Re: [computer-go] Hall of fame for CGOS
 
 Don Dailey wrote:
  Also, even though we can ask people to never change their program
 unless
  they give it a new login name,  we can't enforce that, nor is it
  reasonable to try. I might have a program with an on-line
 learning
  algorithm which improves itself over time - it would be unreasonable
 to
  ask me not to put that on CGOS.
 
 MonteGNU is doing on-line learning of its fuseki database.
 
 /Gunnar
 ___
 computer-go mailing list
 computer-go@computer-go.org
 http://www.computer-go.org/mailman/listinfo/computer-go/

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

[computer-go] RE: several million nodes

[David Fotland]

When MF evaluates a position it does local tactical search to see if blocks
of stones can be captured.  It does this for every block with 3 or fewer
liberties, and for points at the diagonals of eyes, and to see if
connections are solid by trying to cut and doing a search to see if the
cutting stone can be captured.  It often does more than one local search for
the same block, varying the side to move first and the ko status.

This local search is why the evaluation function is so slow.  MF evaluates
about 40 to 200 full board positions per second during the global alpha-beta
search.  The local search examines over 100K positions per second, so a 30
second search examines 3 or 4 million positions in local search.

Here is search output for a typical early endgame position (move 175).  This
is on one processor of a 2.3 GHz Core Duo.

First it spends 0.7 seconds in the life/death search, evaluating 101
positions.
Then it does 5 iterations of alpha-beta search, on 20 candidate moves, using
about 34 seconds.
It does a total of 5548 full board evaluations (164 per second).  
The local tactical search examines 3845717 positions, 24% for eyes, 10% for
connections, 19% for blocks.

Search 24-48 seconds (1 max evals),   0.7 secs in life reading level 5,
life reading 101 life()
New best 1870 (strat 1090)m9  m9 
New best 1943 (strat 988)j9  j9 
Iteration 1 complete 20 moves in  1.11 secs.  Total evals 154, search life()
53, 
Final value is 1943 for j9 : j9 
New best 1818 (strat 988)j9  j9 j8 o11 
New best 1885 (strat 1125)m9  m9 l6 
Iteration 2 complete 20 moves in  2.57 secs.  Total evals 412, search life()
311, 
Final value is 1885 for m9 : m9 l6 
New best 1820 (strat 1125)m9  m9 l6 e13 f13 
Iteration 3 complete 20 moves in  5.10 secs.  Total evals 879, search life()
778, 
Final value is 1820 for m9 : m9 l6 e13 f13 
New best 1820 (strat 1125)m9  m9 l6 e13 f13 
Iteration 4 complete 20 moves in 11.14 secs.  Total evals 1932, search
life() 1831, 
Final value is 1820 for m9 : m9 l6 e13 f13 
New best 1870 (strat 1125)m9  m9 l6 e13 f13 k6 
Iteration 5 complete 20 moves in 33.74 secs.  Total evals 5548, search
life() 5447, 
Final value is 1870 for m9 : m9 l6 e13 f13 k6 
Pass val -5.6, best val 37.4
 33.7 secs. life() 5548 (164/s), searchevals 1613, Nodes 2056 ( 61/s), full
gen 319, Q gen 1483, Moves 3246( 96/s), tree nodes 207/207, list
10211/10211, tac nodes 3845717 (113971/s), evalopj 107868 (  3%), life
3171808 ( 82%) [ fixgralive 3508037 ( 91%) miai 817026 ( 21%) tvpot 603326 (
16%), group 742975 ( 19%), conn 376456 ( 10%), eye 922580 ( 24%) ]


David

 -Original Message-
 From: Peter Christopher [mailto:[EMAIL PROTECTED]
 Sent: Saturday, December 15, 2007 3:18 PM
 To: [EMAIL PROTECTED]
 Subject: several million nodes
 
 Hi David,
 Thanks for you generosity in sharing your experience and perspective
 on the computer-go list.  I noticed your statement that MFG does
 several million nodes of search.  I assume this is per move.  I see
 what you mean that this would be uncachable.  It seems really hard for
 me to believe, is that really correct?  Maybe I don't understand what
 is meant by a node - is it the creation of a representation of a
 global or local situation with some form of valuation metric applied?
 Do several million nodes really get some valuation metric applied to
 all of them?
 
 Sorry to send off-list; my email address almost always gets bounced
 from the list for some reason (but I can read the response there).
 Peter

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] MC-UCT and tactical information

[David Fotland]

Not version 12, but I expect I'll have to implement multiprocessing some
day.  Since the number of cores will double every 2 years, in 20 years we'll
have PCs with over 1000 CPUs :)

David

 -Original Message-
 From: [EMAIL PROTECTED] [mailto:computer-go-
 [EMAIL PROTECTED] On Behalf Of Chris Fant
 Sent: Sunday, December 16, 2007 7:23 PM
 To: computer-go
 Subject: Re: [computer-go] MC-UCT and tactical information
 
 On Dec 14, 2007 2:29 PM, David Fotland [EMAIL PROTECTED] wrote:
   Many Faces does life and death search at the root before the main
 search.
  It typically allocates a few hundred nodes to life and death search.
 Since
  the search is best-first, it keeps the search trees from move to
 move.
  Later searches can extend earlier ones.  The trees are small, so it
 doesn't
  cost much to keep them.
 
  During evaluation tactical search results are cached, but the search
 tree is
  not.  During a main search to find the move to play, Many Faces does
 a few
  million nodes of tactical search, so it's too much to cache.
 
  David
 
 
 Dave, will Many Faces 12 be able to take advantage of multiple
 processors?
 ___
 computer-go mailing list
 computer-go@computer-go.org
 http://www.computer-go.org/mailman/listinfo/computer-go/

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] rotate board

[David Fotland]

I only use 2 random numbers per point, one for black and one for white.  I
xor another random number indicating the side to move.

 

David

 

  _  

From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Álvaro Begué
Sent: Wednesday, December 19, 2007 4:15 AM
To: computer-go
Subject: Re: [computer-go] rotate board

 

Say you represent the content of each point with 0 for empty, 1 for black
and 2 for white. Start by creating a table of 19x19x3 random 64-bit numbers.

unsigned long long zobrist_table[19][19][3];

...
unsigned long long zobrist_key=0; 

for(int row=0;row19;++row){
  for(int col=0;col19;++col){
int point_content = board[row][col];
zobrist_key ^= zobrist_table[row][col][point_content];
  }
}

The result is the zobrist key. In practice, you would make the zobrist key
part of your representation of the board, and when you modify the board, you
just incrementally update the zobrist key. Just remember that when you
change the content of a point, 
new_zobrist_key = old_zobrist_key ^
zobrist_table[row][col][old_point_content] ^
zobrist_table[row][col][new_point_content];

Get that working first and then reread Jacques's post.

Álvaro.




___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] FooBar

[David Fotland]

Hi Don,

I never heard of this technique before.  Are there any more you can share?

ManyFaces12 uses:

-Iterative deepening, with hash table
-Zero-window search (beta is alpha+1, and research when fail high)
-Null move (reduce depth by one, only try null when beta is not infinite,
only one null move allowed per variation)
-Try null first, then the best move from the previous iteration (from the
hash table), then moves in the order suggested by the move generator.  The
move generator gives a pretty good ordering since it estimates the number of
points gained by the move.
-Full ply extension when there is one forced move
-Fractional ply extensions

I plan to add:
-2 killer moves, then history heuristic
-This pruning idea from Don.
-Internal iterative deepening
-Some kind of iterative widening
-narrow search window based on result of previous iteration

Today I search 20 to 40 moves at the root, then 10 moves per ply during the
main search, then a tapering quiescence search with no maximum depth.

I find developing search code less satisfying then working on the evaluation
function.  If I add knowledge to the evaluation function I know the program
is stronger.  When I make a change to the search parameters I can only test
to see what happened.  It's a much more trial-and-error process.

Still, I have to do it because global search makes the program much, much
stronger.

David


-Original Message-
[EMAIL PROTECTED] On Behalf Of Don Dailey


The various versions I'm testing are selective, I use a technique
similar to that used in modern chess programs of aborting the search
at a given node after only a few children nodes have been tried when
the expectation remains below alpha.  

There are many potential tricks that can speed up the search greatly
and I've barely scratched the surface, but I have done a couple of
simple basic things that appear to be working very well.



___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] Alpha-beta and computer go.

[David Fotland]

DF: thanks for the link to this new (to me technique).  I'll implement it
soon.


In my alpha/beta searcher a very simple minded version is actually
working very well - it will take weeks though to test and refine
this.  The only thing I am doing now is:

   At any given node ...
 If  less than 4 ply remaining 
 if move is not a capture  
if at least 5 moves have been tried ...
 do not search the subtree.

DF: how many ply is your usual search?  I'm getting 3 to 6 ply in the main
search, but the quiescence search often adds another 3 to 8 ply.  The Q
search has almost all the nodes.

Right now I use weak patterns to veto moves from the tree.If a
move is below a given rating I don't even try it if I am on the last 4
ply of the search

DF: patterns are really important.  I don't have bad move patterns since
there are almost no absolutely bad moves.  I use patterns to generate moves,
and only try moves that are suggested by a pattern.  Weak moves are just
moves that have no pattern to suggest them.
 
What seems to be the case in GO is that you can be a lot more
liberal about pruning moves without hurting the search.   

DF: yes, this is true.  There are usually a small number of tactical moves,
based on life/death or local shapes (say to defend the border of a
territory), and a lot of quiet moves with very similar values.  If you find
one or more tactical moves you can prune all the quiet moves.  If there are
no big tactical moves, then there will usually be many good quiet moves, so
as long as one is in the set of moves you try you will get a similar result.

DF: for example, go books make a big deal about where to extend along the
side, or when to play in one corner or another, but the difference between
these various moves is usually only a few points.

David Fotland wrote:
 Hi Don,

 I never heard of this technique before.  Are there any more you can share?
   
Since you are using hash tables,  I assume you are aware of ETC 
(Enhanced Transposition Cutoffs.)  ?

For anyone unaware -  it works like this:

1.  If a hash table entry does not exist for a given move.
2.  Do a quick search of the children - in the hopes of finding a
cutoff.

At interior nodes it's cheap and it will provide a cutoff sometimes when
a direct cutoff is not in the table.   If it does not provide a cutoff
it might be used to narrow the alpha beta window.

DF: this is what I meant by internal iterative deepening.  When there is
no hash table entry, I do a iterative deepening search from this position
rather than just depending on the move generator for move ordering.  Do you
mean that I should do a less deep search?

 ManyFaces12 uses:

 -Iterative deepening, with hash table
   

Do you have parity issues in ManyFaces?This is a problem in GO
with a MC evaluation function.   An odd ply search seems to return a
substantially higher score than an even ply search,  and it seems like
you cannot compare an odd to an even ply search.  This affects the
hash table implementation signficantly because you cannot take a cutoff
based on scores of a different parity. (at least not without a
slop margin.)

DF: I don't have parity issues.  The evaluation function is tuned to avoid
the problem (so it's not just a simple sum of the ownership of each point on
the board).


Do you know about MTD(F) ?  Although hardly any programs use it,  it
is a win.   All searches are zero width and you have a driver that
zero's in on the correct score  so it's rather like a best first
search.   You must have hash tables for this to work but it's quite
effective and it has always given me about 10 - 15 % over even PVS
search.   

DF: I know about MTD(F), but I prefer PVS since it's simpler.  I want
something I can debug :)

 -Null move (reduce depth by one, only try null when beta is not infinite,
 only one null move allowed per variation)
   
How is this working? My sense of this is that it would work well in
GO if your program is not overly dependent on life and death via global
searching.In other words, if it has good life and death in the
evaluation function.

DF: Null move works really well for me.  My main search move generator is
very expensive (because it has to do a full life/death evaluation), so null
move has the extra benefit of avoiding move generation.

I also use the killer heuristic which seems very effective.I haven't
really gotten into heavy tuning - I suspect my move ordering is far from
optimal but I will get into this more when I implement hash tables.

DF: Killer should work well since a big tactic is not affected much by other
moves on the board.  My move generator usually finds the big tactics, so it
won't help me as much.

 -Full ply extension when there is one forced move
 -Fractional ply extensions
   
I will experiment with extensions for atari moves and perhaps captures.

I want to identify serious atari and capture moves from unimportant
ones

[computer-go] Odd results on 19x19

[David Fotland]

The styles of CS (CS-9-17-10k-1CPU), MFGO (mfgo12exp-15), and GNUGO
(gnugo3.7.10_10) are different, and it's generating some odd results.

Many Faces beats GnuGo 70%.  There are not many games, but this is
consistent with over 100 test games I've run.
CS beats GnuGo 55%.  Over 100 games played.
CS beats Many Faces 90%.  Only 20 games, but consistent with earlier
results.

If we look at results against GnuGo, Many Faces seems stronger than CS, but
in games against CS, Many Faces is much weaker.

Many Faces plays a fighting style, and CS plays a territorial style, but I'm
still surprised at the difference.

David

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

[computer-go] How to get more participation in 19x19 CGOS?

[David Fotland]

I think there are two reasons there are not more programs on 19x19 CGOS:

1) The anchor, Gnugo, is quite strong, Many Faces 12 is stronger, and
CrazyStone is much stronger.  Since the programs playing are so strong, it
is demoralizing for a new program to lose so often.  Without weaker
competition, it is hard to get accurate ratings for new, weaker programs.

2) The rounds take almost an hour, so it takes much too long to get enough
games to see how your program is doing.  In my local testing, I use 10 or 15
minutes per side.  I like to see 50 to 100 test games get any confidence
that a new version is stronger.  I prefer to get a test run complete in
under a day.

I propose the following changes:

I'll put up 3 weaker versions of Many Faces 12, so there is competition at
lower ratings.  These versions are quite fast, using only a few seconds for
a full game.  This will provide some stable opponents for the weaker
programs.

I think the time limits for CGOS19x19 should be reduced to 10 or 15 minutes
per side.  This is enough to test programs, and it's still a reasonable time
limit for games against people.  Since programs that search scale with
additional time, the relative ratings of these programs should be similar at
10 minutes per game and 30 minutes per game.

-David


___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] How to get more participation in 19x19 CGOS?

[David Fotland]

Then 15 minutes should be good.  We want the anchor to play at the same
strength as before.

David

 -Original Message-
 From: [EMAIL PROTECTED] [mailto:computer-go-
 [EMAIL PROTECTED] On Behalf Of Alain Baeckeroot
 Sent: Tuesday, January 08, 2008 12:40 PM
 To: computer-go
 Subject: Re: [computer-go] How to get more participation in 19x19 CGOS?
 
 Le mardi 8 janvier 2008, Don Dailey a écrit :
 ...
  On 19x19 it might be 30 minutes per side like we have now,  with 5
  minute games for the fast time control.We would probably have to
  work it out so that program like gnugo would be able to handle the
 fast
  time control at their standard settings.
 
 
 At level 10, gnugo might need more than 10 min for some games (with
 lots
 of possible ataris or kos).
 At level 0 (with tons of really ugly moves, and only 2 stones weaker
 on kgs)
  5 min should be ok.
 
 I think adding handicap to 19x19 is really a needed feature.
 Alain
 
 ___
 computer-go mailing list
 computer-go@computer-go.org
 http://www.computer-go.org/mailman/listinfo/computer-go/

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

RE: [computer-go] How to get more participation in 19x19 CGOS?

[David Fotland]

It was Doug Larson's idea.  Something like:
Pull a single stone out of Atari once
Play in a liberty of the group with fewest liberties that has 5 or fewer
liberties
Play random

 I also am thinking about building a really fast and weak playing bot.
 Something that is similar to the rule based program someone suggested
 to
 you, but with a few simple 3x3 patterns to supplement it.Do you
 remember what we called that program? It played by simple rules
 such
 as attacking the group with the least liberties, etc.We names it
 somewhat after the person who suggested the idea.
 
 
 - Don
 
 
 
 
 
 David Fotland wrote:
  I think there are two reasons there are not more programs on 19x19
 CGOS:
 
  1) The anchor, Gnugo, is quite strong, Many Faces 12 is stronger, and
  CrazyStone is much stronger.  Since the programs playing are so
 strong, it
  is demoralizing for a new program to lose so often.  Without weaker
  competition, it is hard to get accurate ratings for new, weaker
 programs.
 
  2) The rounds take almost an hour, so it takes much too long to get
 enough
  games to see how your program is doing.  In my local testing, I use
 10 or 15
  minutes per side.  I like to see 50 to 100 test games get any
 confidence
  that a new version is stronger.  I prefer to get a test run complete
 in
  under a day.
 
  I propose the following changes:
 
  I'll put up 3 weaker versions of Many Faces 12, so there is
 competition at
  lower ratings.  These versions are quite fast, using only a few
 seconds for
  a full game.  This will provide some stable opponents for the weaker
  programs.
 
  I think the time limits for CGOS19x19 should be reduced to 10 or 15
 minutes
  per side.  This is enough to test programs, and it's still a
 reasonable time
  limit for games against people.  Since programs that search scale
 with
  additional time, the relative ratings of these programs should be
 similar at
  10 minutes per game and 30 minutes per game.
 
  -David
 
 
  ___
  computer-go mailing list
  computer-go@computer-go.org
  http://www.computer-go.org/mailman/listinfo/computer-go/
 
 
 ___
 computer-go mailing list
 computer-go@computer-go.org
 http://www.computer-go.org/mailman/listinfo/computer-go/

___
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/