Editor's note:  This month we are fortunate to have a treat. Albert Silver has put the final match of the 2003 World Championships into GNU, and had several experts offer their commentaries. I've gone through a few of the games, and will be going through the rest of them in the next couple of weeks and adding my own comments as I go along. —Kit Woolsey

In July, when the finals of the World Championship 2003 took place in Monte Carlo, GamesGrid relayed the match on their server as a courtesy to its members. It wasn’t complete at first, and watching or recording the entire match also required being present at the right time. I had it all though, and had analyzed it with GNU 0.14 at its best settings. That’s how this project first came to existence, intended to be as simple as could be: Using GNU’s excellent HTML export functions to just share the results with GammonLine’s readers. I told Kit this, and he welcomed the idea, though he didn’t answer my suggestion to add a comment or two of his own. A day later, GammonVillage published their file of the match all analyzed by Snowie 4 3-ply. Since there was still time, I began running rollouts wherever the bots disagreed in order to add a comment on just this. That is the nature of the Bots commentary that can be seen. It is merely a comparison of Snowie 4’s 3-ply analysis with GNU’s equivalent 2-ply* analysis. These rollouts revealed several surprises such as both bots and the player choosing the wrong move, and not always by a small margin.

Seeing some of the positions, and receiving questions by some, whom I had shown the match to, along the lines “Is that correct? I don’t understand it. There must be some kind of mistake.”, it was clear expert commentary would really be welcome for this, in order to provide an explanation beyond the percentages of wins and gammons. I began e-mailing experts I knew by sight from GammonLine’s forum. I expected extremely limited luck and figured that even one commented position would be a great boon, making the whole commented match idea far more interesting. The answers I got were overwhelming and suddenly I had enough promised assistance to cover the entire match.

Providing comprehensive commentary of entire sections of the match are (in order of appearance) Casper van der Tak, Chuck Bower, Alex Zamanian, and Tom Suzanski. The names of the first three should need no special introduction to GammonLine readers: they tirelessly share their understanding and knowledge either in the forum or via articles of their own. Tom Suzanski is a master player who now teaches at Gamesgrid, and having had classes with him myself, I was confident he could add very lucid explanations. Unfortunately, despite accepting, he was leaving on holiday two days later, so his comments will only be included shortly. I had also asked a neighbor, Christian Toth, who had played professionally on the international circuit for 8 years, whether he’d add anything of his own. He was now retired for the past 3-4 years, but anyone who had played professionally for so many years surely knew what it took, and this was a player who had studied with the likes of Paul Magriel, Nack Ballard, and Jerry Grandell. To my ever increasing surprise, he also accepted, and readers will find his original and perceptive comments within as well.       

All this already promised to keep me very busy coordinating and gathering the material, yet a new and not insignificant problem cropped up: the game score. It bore several differences with the one published at GammonVillage, begging the question: Who was right? I questioned Ken Arnold about it, the owner of Gamesgrid, and he swore by the score of his site as did Liz Barker, who had provided GammonVillage with its score. What now? In game 1, a change from a 5-4 instead of a 6-5 led to almost no change at all, neither in the decisions nor the bot evaluations, but in game 10, one different play was followed by a flurry of ‘blunders’ according to the bot. This was reduced when using GammonVillage’s score so I corrected it. Then, in game 12, it was the reverse: GammonVillage’s score showed Tissona to have made a couple of horrendous blunders due to a difference in the game score. In the Gamesgrid file a different move was given and the blunders magically disappeared. Finally, in game 18, the score I had showed a very bizarre move followed by a series of crazy blunders. Here, GammonVillage’s score showed a different move after which the moves made more sense. In all cases I favored the move that led to the least number of errors, presuming a problem in the recording as opposed to presuming a sudden leave of wits from the players. This was the most significant match of the year, so the reasoning seemed sensible.

I’d like to profusely thank all the commentators for taking so much painstaking effort to share their knowledge of the game with the readers. It goes without saying, that this great commented match could not have been possible without them. Finally, as a feather in this article’s cap, Kit has decided to add his own, so the readers are in for a real treat. Enjoy!

Albert Silver

* The difference in Snowie’s 3-ply and GNU’s 2-ply isn’t in the depth but in the way they begin counting plies: Snowie begins with 1-ply and GNU with 0-ply, hence they are effectively the same.

To get to the match, click here: [First Game]