A Backgammon Swiss: Efficiency

Continuing with the FEPS analysis that compares the 32-player Swiss backgammon tourney I ran on Saturday with a format commonly used in past events. Now let’s consider how the two compare in the item of efficiency. There are two main aspects of efficiency to consider: the time needed to run the tourney, and the time and attention of the tournament director.

As noted in the last post, the Swiss tourney required five rounds of play as opposed to seven for the main-and-consolation elimination tourney. But it’s not necessarily the case that the Swiss will take less time. In running a Swiss, you have to wait for each round to be complete, or at least nearly complete, before you can generate the pairings for the next round, while with a bracketed tourney, the rounds can and do overlap.

For backgammon tourneys, some experienced tournament directors, as a rule of thumb, estimate the time needed for one round of play as 11 minutes times the number of points in a match. A round with five-point matches thus takes about 55 minutes, while a round of seven-point matches takes 77 minutes. But Swiss rounds must be spaced not according to an average match length, but according to some estimate of the time needed for the longest match in the round. For my tourney, I planned on 75 minute rounds, though the matches were only five points. Similarly, some backgammon tourneys using a modified Swiss system for nine point matches allow at least two hours for each round.

The need to schedule rounds according to the longest expected matches rather than the average match gives Swiss tournaments a distinctive rhythm. Most of the time, when their last match is not unusually long, players will have some down time between rounds. An early finisher might know, for example, that there will be enough time to grab a bite to eat, or even for a short nap.

In bracketed tournaments, there is less down time between rounds, but there still is some. But because players don’t know when their next round will be called they experience the delay differently. They usually can’t leave to eat or to nap because they may, at any moment, have to spring back into play. Thus, while there is less down time between rounds in a bracketed tourney, I think players generally dislike the gaps in play more. Instead of having some time that they know they can use to relax, they have to wait in readiness. That makes playing a bracketed tourney somewhat more anxious and exhausting for many players.

The length of the longest match is more variable for some games than for others. In modern chess tournaments, the time controls are such that the length of the longest games in a round is fairly predictable. If, for example, the time controls are g/100, d/10, all of the longest games will end together after about 220 minutes. 200 of those minutes are the reserve time used by each player, and another ten minutes or so is used by the player in the form of the ten seconds per move that’s allowed before reserve time starts running. Even an extraordinarily long game seldom takes more than 100 moves, and the cumulative delay time used by each player is seldom more than ten minutes.

For backgammon, the length of the longest match in a round is more variable. For a five point match, the time control (expressed as it is for chess tourneys) is g/10, d/12. That means that only 20 minutes of the match length comes from the fixed reserve time. But the number of individual moves in a backgammon match is highly variable. A match in which two or three long back games are played might run to 500 moves, and because the dice need to be thrown for each move, more of the 12 second delay time will actually be needed.

Even allowing 75 minutes for a five-pointer, I had one round that lasted just over 100 minutes. Ordinarily, I would have paired the next round based on a tentative result, but that might have upset the fortunate symmetry of having exactly 32 entrants. Fortunately, the next round took only 60 minutes, so I was able to reclaim some of the lost time.

For my backgammon Swiss, I judged it necessary to guard against slow play. Time clocks are usually “preferred” in this sort of backgammon tourney, meaning that while either player may insist on one, they may agree not to use a clock. The director retains the right to impose a clock on a match that’s holding things up, but by that time the damage has already been done. I made clocks mandatory. I also announced some unusually strict rules to speed play, including requiring all breaks in play to be approved by the director. I was a bit worried about these, but the rules were received in good humor, and I didn’t have to actually enforce any of them.

And, most significantly, I specified five-point rounds. The usual practice for a bracketed tourney of this type would be to play seven-point rounds in the upper bracket and five pointers in the lower bracket. Now, for most players this was a sacrifice worth making. They were willing to play somewhat shorter rounds because they knew they were going to play several of them. But the same skillful players who didn’t much value the five-round guarantee because they expected to do well enough to play five rounds any way missed the extra two points. A longer match is not only more backgammon for them, but also enhances fairness (C) by making upsets a bit less likely. I’ll discuss the fairness aspects of the Swiss in the next post.

In the end, I was satisfied with the flow of the tourney. More players played longer, but I didn’t have to contend with a few straggling lower-bracket matches at the end of the day. There was time enough to give everyone a decent dinner break before the evening event began. Again, far more players were happy with the format than not. But it’s worth noting that those who were less happy were the most skillful players, and these folks also tend to be include those who serve as tournament directors at other events.

A final note, then, about the other resource that’s differentially consumed by a Swiss tourney – the attention of the director. At small backgammon tourneys like this one, a bracketed tourney practically runs itself, so that the director can, and usually does, play the event.

There’s more for the director to do in a Swiss. I was running three different programs on my laptop: WinTD, the Swiss pairing software; Excel, which I needed to keep track of the prize money, as WinTD inflexibly applies prize money rules that seem sensible to chess players, but don’t work at all for backgammon; and a small app I used to generate name badges and lottery tickets. (The function of the lottery tickets will be described in a future post.) WinTD, in particular, is a program that with a significant learning curve – I had run more than a dozen simulated tourneys in preparation, but that was only just enough to teach me what I needed to know.

Even had the mechanics of running the tourney been less demanding, I would have chosen not to play. Running an unfamiliar format I wanted to avoid any perceived conflict of interest, and also to try to present as a calm, steady hand in control. But I don’t see myself being comfortable simultaneously running and playing a Swiss tourney any time soon. This is a shame. I want to introduce a true Swiss format to backgammon at least in part because it’s a format I’d really like to play. But at least I get to blog about it.