By Robert Wachtel
The second part of Robert Wachtel’s review of the Denmark vs. the World match, which drew most of the attention in the last Nordic Open backgammon tournament. Click to read the first part, with
introduction to the teams’ members, the Nordic Open and backgammon in Denmark generally.
Through some unaccountable miscalculation on Mr. Rice’s part, your correspondent was not offered a spot on the World team — but no worries. I arrived in Copenhagen a few days before the start of the backgammon tournament and got a chance to observe the World’s preparations and participate in some training matches with them.
Everything seemed in order. The standard of play in the practice sessions was very high. Each team member (as befitted his or her reputation) played almost flawlessly, with the full team in consultation recording error rates of around 1.5 in 7 point matches. The only cloud on the horizon was the obscure possibility, as demonstrated by Stick, that in the speed matches no one was quite prepared for the damn-the-torpedoes backgame strategy of an opponent who was angling just to make you lose on time.
The big showdown began one day before the tournament proper. There was an abundance of media coverage, in the form of live video feeds, accompanying Snowie video feeds, and beautifully crafted interviews and commentary — all supplied courtesy of the British outfit Mind Engagement Media, a group that had evolved from last year’s WSOB production team.
Speed & 1-Point Backgammon Matches – Denmark Leads
The Danes played the speed matches straight, as did the World; and the first lesson we learned is that there is a lot of luck in backgammon. The pairings were Hansen vs. Falafel and Holm vs. Michi, each pair to play three 5-point speed matches. The difficult conditions (each player had ten seconds per move, with a reserve of two minutes) made for some uneven play. Mistakes on both sides: but in the end, the Danes won every match! They led 6-0.
Next, came the one-point match competition. These cubeless matches (also called double-match-points, or dmps) are the simplest form of backgammon, and most world-class competitors can play them virtually error-free. In the Stick-Mads Andersen faceoff, both players performed at an awesome error rate of around 1.0 over seven matches. But, Andersen won six and lost only one! On the basis of no skill differential, Denmark had opened a huge lead.
A strong showing by prodigy Matt Geier began the World’s comeback, but it was too little and too late. At least Geier was not able to boast, by the end of the competition (as he had earlier), that he had scored more points than the rest of the team combined!
Within the complex format that Holm had conceived, these combined results left Denmark with a 35-29 lead in a race to 43 (think 6-0 to 14) in the final 8 by 8 consultation match.
And there we learned our second lesson: great (bot-trained) minds think alike. Only a few decisions caused the full team the slightest dissension. For example: in the first game of the match, the World gave an aggressive but proper (given the match score) double.
The Danes took, but the game turned ugly for the World. Eventually it faced a nasty-looking recube:
Everyone’s a Critic
The computers say this is quite an easy take, basically because of the match score: if you pass, you are down 8-0 to 15. But the position is ugly, and no one on the World team spoke up for the take. And so the World did pass, an decision that unleashed a torrent of recrimination and abstruse criticism for weeks from the armchair quarterbacks who post on Stick’s web forum. “Did the team calculate gammon fractions?’ asked one critic. Who was keeping the Kleinman count?” another demanded. “Which MET (match equity table) were you using?” another wanted to know, Woolsey’s or R-K?” “I’m wondering,” another posted, “whether this position is an example where estimating recube vig via effective gammons, as I explained [elsewhere on the forum] works better than the traditional method of guessing an intermediate point between live and dead cubes.”
Whew! Talk about being under a microscope. It was refreshing to hear Matt Geier, the most “theoretical” player of the lot, confess that, in several cases, he had no idea what the parameters were that the critics had cited — nor would he have known how to use them had he known what they meant!
Later in the match, with its back even more against the wall, the World team was faced with another disgusting-looking cube:
This time there was some dissension in the ranks. Falafel, the playing captain (Stick was the organizational captain) looked at his position and made a face as if someone had laid a three-day dead fish on the table in front of him. “We can’t take this,” he said, “There’s no way we can win.” But world #1 Mochy disagreed. “We have to gamble some time,” he insisted. We must try it. This might be our last chance.” Falafel relented at last, the World took … and lost. But was it really a take?
The World about to make the fatal pass at 1-7 to 15 in the full consultation match. From left to right: Stick Rice, Katja Spillum, Matt Geier, Falafel (seated), Mochy, Fernando Braconi (seated), Michy
It depends on what bot you are using and what MET that bot is using. Snowie rollouts say pass, Extremegammon rollouts say small take, GNU rollouts say easy take. Since Snowie is now considered the weakest of the bots, the odds are that it is a take. Which brings us to lesson number 3: tournament backgammon is a tough game indeed!
Funny, there was no discussion in the forum of how the team got this one right …