This was a sidebar to an article I wrote for Bridge Today magazine.

WBF Open Pairs Final – Lille 1998

The implementation of the movement for the Open Pairs final in Lille was fatally flawed. Midway through the first session some pairs found themselves matched against opponents they had met earlier in the session. While the players knew little or nothing about the movement we had been told that each of the 72 pairs would play against each other pair exactly once, so each pair faced with a rematch knew something was wrong and summoned the director.

At this point the directors stopped the game and attempted to ascertain what had gone wrong. They discovered the source of the problem, repaired it, and issued new assignments from which we were to continue play. Unfortunately the new starting assignments also led to some rematches – I don’t think this could have been helped. Each pair with a rematch was told to sit the round out – I for one welcomed the break.

What did happen? With four years to prepare, how did things go so disastrously wrong, and what could have been done to prevent it?

Let’s start with the movement. There was nothing wrong with it as designed. It was a complete 36 table Howell, with each pair scheduled to meet every other pair over the course of 71 rounds. Each round half the pairs would move up one table and half the pairs down one table, except for one stationary pair. A pair moving upwards would continue until they reached table 36, when they would reverse directions – pairs moving downwards would reverse directions at table 1. In order to properly balance the movement pairs would switch from NS to EW every few rounds, seemingly at random. If not for this the movement would appear straightforward, and it’s easy to construct a spreadsheet showing where each pair is assigned each round. In order that the players should know where to sit custom guide cards had been prepared and placed on each table, as in any Howell movement, telling the NS and the EW pair where their next assignment was, both table number and direction. Since the event was scored barometer style, with the same boards in play at every table in a round, the movement of the boards was not an issue.

The problem was that two of these guide cards had been made out incorrectly. Unfortunately the mistakes were complementary, so that no two pairs were assigned to the same table and direction. Instead two pairs were shuttling back and forth between two tables while the rest of the pairs moved around them. If only one guide card had been in error, or if the mistakes were not complementary, the problem would have been discovered after the first round. Instead it was not discovered until it resulted in a few rematches, and having been in effect for several rounds it would inexorably result in more rematches over the course of the event.

The second through fourth sessions proceeded smoothly, but there was some confusion in the viewgraph room during the fifth session. Pairs who had been expected to play on viewgraph had been given sitouts. The spectators in Lille never found out, nor did I until some days after the event, but this was just a consequence of the problem in the first round, this time manifesting at the other end of the movement. The directors knew about it and expected it, as did the pairs with sitouts, but for one reason or another this news never reached the viewgraph commentators.

How could all this have been prevented? More important, how can we keep it from recurring?

My first thought was testing. In the software business we’ve learned that testing is of vital import, and that it’s all the more important with a custom product. It would have been impractical, though, to find 144 players and ask them to step through the movement, noting whom they’d already met. This turns out to be unnecessary to catch the problem at hand. Instead a director could have started at table 1EW and followed the directions on each guide card until he returned to table 1, having visited each table in turn. This would have quickly turned up the problem.

A second solution is publicity. While the movement was not intentionally secret it was not publicized either. Had the players known what to expect they would have questioned the movement earlier, likely at the first change. One might think that the pairs moving between two tables should have known that there was no way they could play against every other pair if the movement continued that way. They had no reason to suspect a flawed movement, however, and they no doubt were, properly, concentrating on playing their best, had all the pairs known the basics of the movement then any pair skipping two tables would know that something was wrong.

A third solution would be for the scorers to note that the results coming in were not the matchups expected, and to report the same to the directors. I do not know why this didn’t happen. This at least was a possibility, since in the finals we had at last eliminated the traveling scoreslips that were present in all the other pair games, in favor of score tickets as are standard in the ACBL. I’d like to think that this was because the WBF understood that travelers not only mean the scores take longer to come out, but also that they slow down the game and make it less pleasant. Instead I expect it was because travelers are of no use in a game that is scored barometer style!

A fourth solution was implemented, belatedly, halfway through the second session. Pair numbers were assigned and guide cards were issued to all the pairs, telling them their exact table assignment each round. For some reason only one was issued to each pair – was there a paper shortage? Also the cards did not show the pair number of ones opponents. Still they were much better than nothing, and had they been present from the start they would likely have helped to nip the problem in the bud.

Where I could I have tried to explain and offer suggestions rather than to criticize. When assessing a WBF tournament one must realize that most of the work is performed by volunteers, without whom there would be no tournament at all. Still, I do take the game seriously, and I believe a thing worth doing is worth doing well. I can only hope any criticism will be taken in the constructive spirit in which it is offered.

Return to Adam Wildavsky’s Bridge Laws page