Catalin Taranu against computer Mogo


	Home / MyStuff	Go Library	Forum	Articles	Blogs	Tournaments	Users

Catalin Taranu against computer Mogo

Bookmark | < Prev | Next >

On 22nd May 2009, Catalin Taranu and computer Mogo faced for the second time for a 9x9 board series. Last year, Catalin won by 2-1, all games won by white!

The komi was the same, 7.5 points and the time limit 30 minutes. This time, the organisers from Rennes proposed a 4 games series.

MoGo was developed by the french National research institute in Data processing and Automatics (INRIA). More info here: http://www.espace-sciences.org/jsp/fiche_pagelibre.jsp?STNAV=STRU04&RUBNAV=&CODE=52957396&LANGUE=0&RH=ACCUEIL

This time Catalin won by 3-1 after leading with 3-0, so we can say he improvised in the last game :) Here are the files:

Game 1 (SGF file)
Game 2 (SGF file)
Game 3 (SGF file)
Game 4 (SGF file)

UPDATED:

Catalin's comments on the games:

The match itself started with a relatively easy win for me with black. As komi is quite large, 7.5 points, i was determined to use the strong point of human mind in comparison with a machine : inventivity and flexibility. Since Mogo made an early mistake, such tactics were no longer neccesary and i won after a peaceful yose play. One interesting thing to notice about Mogo, it plays quite well when it has still a chance to win, but starts to make tricky moves when the game is completely lost, an attitude quite similar to human desperation when in hopeless positions.

The second game i gained some confidence and i was also worried about the public not getting enough show. I felt also a lot of personal curiosity about the reading strength of Mogo, so i challenged a difficult fight right from the start. As i played white, it was a risky thing to do, but i needed to measure my opponent.
I think Mogo played astonishingly well throughout the fight and at the end if i played normal yose i would lose. So i started a series of tesujis aimed at living in the black territory or get a better yose through a sacrifice strategy. In the end i won, but i felt that this second game was the most spectacular of all the series.

The third game was also an easy win, because i used my idea of inventivity. In order to win Mogo should have been also very imaginative, but this was too much for a machine. Without enough territory, it challenged a semeai fight that it lost in the end.
For the fourth game , knowing i already won the match i lost some concentration and a few blunders were enough to lose the game. Still, it was very nice to see how well Mogo fought , with computer precision in reading that show much promise of this program in the future.

I do believe that the final score of 3-1 was a good result, defending human superiority in the game of Go against machine, but also showing the undeniable progress of computer Go programs.

Share it:

Comments

User	Post
benutzu	5/28/2009 10:14:45 PM Updated with Catalin's comments on the games.

Please login to post comments