Ongoing GO programming?

[Eelco de Groot]

What is the strongest GO (commercial or free) program currently? Is there any in development possibly by members of this forum?

Hi guys,

Just read about an interesting development by Gian-Carlo Pascutto, author of Deep Sjeng and Stoofvlees:



Go-program Leela showing expected territory gains

Geschrieben von Gian-Carlo Pascutto am 29. November 2007 20:32:

Hi all,

next to preliminary work on Deep Sjeng 3.x, I have been trying
Sjeng ideas in another game. And they have seem to have worked
rather well

Behold Leela - the Go program!

http://www.sjeng.org/leela

Leela is among the strongest of the current Go programs and won the
last public tournament - the KGS Computer Go tournament in November.

Feel free to try it, the 9x9 and 13x13 version is free and includes a rated
game mode that is ideal for people learning to play.

Now, I will have to experiment with some Leela ideas in Deep Sjeng as well...

--
GCP

Leela has a nice clean interface, there is no .NET required as far as I can see and no 30 days limit, and great game in the free 9x9 version for a beginner like me!

Eelco

[IanO]

• 1. MoGo
  2. CrazyStone
  3. KCC Igo (Silver Star)
  4. Haruka
  5. Go4++
  6. HandTalk (Goemate)
  7. GNU Go
  8. Many Faces of Go
  9. Go Intellect
  =10. Wulu, Aya, Katsunari, Go Ahead

Notes:

CrazyStone is by Rémi Coulom, author of The Crazy Bishop. Other than him, there has been little successful cross-pollination between computer Go and computer Chess circles.

Ian

Yesterday, CrazyStone won three different tournaments, defeating MoGo in each of them. So I now would put CrazyStone at the top of the list as the world's strongest Go program.

Leela is another addition to the list of strong Monte Carlo players from chess program authors. Deep Sjeng's GCP has created his own Go program, appropriately named after the one-eyed character on Futurama, which has done well in the last few months. I would place it somewhere below CrazyStone/MoGo and above GnuGo.

(Re: Mango. Sorry, its 19x19 results are sparse, and have been mediocre at best.)

Ian

[mschribr]

Yesterday, CrazyStone won three different tournaments, defeating MoGo in each of them. So I now would put CrazyStone at the top of the list as the world's strongest Go program.

I think it would be more interesting to see how well the best program does against humans? Has Crazy Stone played in any human tournaments?
I have heard some say the best program plays at 1 dan. This would put the best program better than the average human player. Is Crazy Stone past 1 dan?
Mark

[Rémi Coulom]

I think it would be more interesting to see how well the best program does against humans? Has Crazy Stone played in any human tournaments?
I have heard some say the best program plays at 1 dan. This would put the best program better than the average human player. Is Crazy Stone past 1 dan?
Mark

Crazy Stone has played against humans on KGS. Its rating there is 1k.
http://www.gokgs.com/graphPage.jsp?user=CrazyStone
That is not based on many games, though. KGS 1d level will be reached very soon.

On 9x9, Crazy Stone is much stronger than 1d.

Rémi

[mschribr]

Crazy Stone has played against humans on KGS. Its rating there is 1k.
http://www.gokgs.com/graphPage.jsp?user=CrazyStone
That is not based on many games, though. KGS 1d level will be reached very soon.

Hi Rémi,
I see a fast rise and fall in April and then a fast rise in December. Why the spikes in ratings? Is this the process of developing new code? You sound confident about reaching 1d. Looks like you almost had 1d in April then a big drop. Why are you so confident about reaching 1d? When do you think you will reach 2d?
Mark

[Rémi Coulom]

I see a fast rise and fall in April and then a fast rise in December. Why the spikes in ratings? Is this the process of developing new code? You sound confident about reaching 1d. Looks like you almost had 1d in April then a big drop. Why are you so confident about reaching 1d? When do you think you will reach 2d?
Mark

The rating before april was completely meaningless, because based on too few games. The rating varies all the time, but it does not mean that games were played. On KGS, rating evaluation changes all the time, even if the program does not play. I connected it recently with an improved algorithm, and it made a big jump.

Rémi

[mschribr]

I connected it recently with an improved algorithm, and it made a big jump.

Crazy Stone played a 5dan in the UEC Cup. Why such a strong player? Would a 1d have been more evenly matched and more interesting? Can you give your thoughts on this game? Where can we find the game?
Mark

[IanO]

Crazy Stone played a 5dan in the UEC Cup. Why such a strong player? Would a 1d have been more evenly matched and more interesting? Can you give your thoughts on this game? Where can we find the game?
Mark

This was an exhibition game; Mr. Sagawa was on hand. You can get an impression of the flow of the game from photos taken by Hiroshi Yamashita, the author of Aya, starting here:

http://www.yss-aya.com/photo/20071202ue ... T2172.html