This post will complete the FEPS analysis of my recent backgammon tourney, run according to the Swiss system, by considering the fourth goal of tournament design: spectacle. Spectacle is not the glory of the Swiss system.
Whatever its other virtues and defects, the Swiss system is not good at catering to the needs of spectators. It is not good at producing high-stakes loser-goes-home matches. And it is not even good at producing a clear winner. There are good reasons why the Swiss system is all but unknown in professional sports.
My five-round Swiss for 32 players yielded much more orderly results than it would have with different numbers of players or rounds. As it was, every round had even numbers of players in each score group. After one round, there were 16 1-0 players, and 16 0-1 players. And the distribution for later rounds went thus: (8 2-0, 16 1-1, 8 0-2); (4 3-0, 12 2-1, 12 1-2, 4 0-3); (2 4-0, 8 3-1, 12 2-2, 8 1-2, 2 0-4); and (1 5-0, 5 4-1, 10 3-2, 10 2-3, 5 1-4, 1 0-5). The (1 5-0, 5 4-1) part means that there was one clear winner, and five clear runners-up.
What would have happened if there had been four rounds, or six? After four rounds, there would have been two players tied for first, and no sensible way to separate them.
If there had been six rounds, the result depends heavily upon whether the one 5-0 player defeated which every 4-1 player they drew in the sixth round. Perhaps there would still be a clear winner. But if the 4-1 player won that match, there would have been a three-way tie between players all having 5-1 records.
And consider what happens if there would have been just one additional player. With 33 entries, there would have been 17 1-0 records going into the second round, the 16 first-round winners, and the one bye. Thus, for the second round, there would be eight matches between 1-0 players, but one player from this score group would have to be paired against one of the 0-1 players. If that player loses, then there are 8 2-0 players going into the third round. But if that player wins, the top score group has 9 players, one of whom will have to “play down”. If the player who plays down wins in every round, there will be two perfect records at the end of the fifth round. And as the number of entries rises further above 32, so does the likelihood of an ambiguous result.
Swiss system tourneys are highly prone to producing ties. In part, the response to this is for players and tournament organizers to become comfortable with ambiguous results. And it part it has been the development of various tie-break systems. The WinTD program I used to run the recent tourney implements 15 different tie-break systems. Since many of these won’t break particular ties, they’re used in sequence.
There are some backgammon tournaments that purportedly use a Swiss format, but most of these actually use some sort of hybrid between Swiss pairings and a knock-out elimination format. The Swiss part of the tourney is actually a Swiss qualifying tourney.
Round robin tournaments have a similar propensity for producing ties, and thus have spawned similarly complex rules for breaking them. And round robins are also used as part of a hybrid system. A sensible way to look at many professional sports is that a season is a hybrid between a round robin qualifying tourney and an elimination playoff for the championship.
At first blush, the Swiss system would seem to be indicated for any league in which individual teams do not play enough matches to make up a round robin. But that’s not what one sees happening in professional sports, where spectacle is the principal goal. For example, that the Big 10 conference (which is essentially a professional sports league despite its claims to amateurism) needs to select a winner from among 14 football teams using only nine games for each team. It could use Swiss pairings to decide which four teams, each year, a team does not play. With winners playing winners and losers playing losers, there would be fewer blowout games. But, for commercial reasons, there needs to be a high-stakes championship game. And the season needs to be know in advance. Whether or not anyone outside the state of Indiana cares about it, Indiana University against Purdue needs to be the last game of the season.
The Swiss system is well established as the primary format for chess tournaments, and it is reasonably common in a few other games, like bridge. But it simply doesn’t provide what professional sports leagues need. And, possibly because it is not used by the professional leagues that most people are familiar with, it is not commonly used in any number of other contexts for which it is better suited.