I’m planning to play Viking Classic backgammon tournament in Minnesota next weekend, and contacted the director to see if he needed any brackets. He allowed as how he thought he had things well in hand because he’d found what he needed on printyourbrackets.com.
After some back and forth, he’s decided to let me draw some brackets for the tourney. The stuff he’d found on PYB was pretty awful.
I’m not a big fan of what I’ve seen on PYB in the past. In Getting the Drops Right I looked and one of their brackets, a 16DE, and found that it had a relatively subtle error. They didn’t have any drops that might cause a repeat pairing where that was completely avoidable, but the hadn’t done what they could to reduce the number of repeats overall. I showed how this had a very modest negative effect on fairness (C).
The ones I looked at this time were much worse. Bad enough that I think I need to strengthen my caution against that particular source of brackets.
The format for the main events in Minnesota is single elimination, with all the losers of every match before the final dropping into a progressive consolation. That format avoids most repeats because the last two repeat-prone matches in a full DE aren’t played – you don’t drop from the upper bracket final, and you don’t reunite the brackets by having the lower champion play the upper champion.
The consolation makes the lower bracket, in effect, its own tournament, with a trophy and a prize fund of its own. It’s a popular format because it avoids one of the besetting problems of a full DE – the problem of having the players who do well in the upper bracket waiting around for the lower bracket to catch up. There are nearly as many matches played, but the tourney is two full rounds shorter.
Eliminating those two matches eliminates about two-thirds of the repeats that would happen in a well-drawn bracket – the only matches that can cause repeats should be the final and the two semi-finals of the lower. But that wasn’t the case with the the PYB brackets. Both of them caused four – count em, four – new matches to risk repeats.
I can’t tell you what this means in terms of fairness because I haven’t bothered to run simulations – life is too short to spend much of it worrying about other people’s mistakes. And besides, the fairness (C) statistic is not defined for consolations. (I’m mulling whether and how to redefine fairness (C) for this and other reasons). But I shouldn’t have to run a simulation to convince you that making four extra matches repeatable is a silly drafting mistake.
There are a number of other features in the PYB brackets that I didn’t care for. I thought the typography was poor. The match numbering system is oddly unhelpful, with a single sequence winding back and forth between the two brackets. The layout was confusing, and you would have had to make an enormous print to make some of the lines long enough to write a name on. And somehow it happened that when you downloaded your free bracket for printing, some of the drops mysteriously vanished.
These matters are right out there in the open, and if the PYB brackets don’t offend your taste, I won’t quarrel with you. But the bad drops are another matter. Unless you look systematically (I explain how to do this in Getting the Drops Right, Part II), you’re not going to know there’s a problem. You may never know that you could have avoided some of the repeated pairings that people complain to you about.
In the next few days, I’ll post some of the new brackets I draw for the Minnesota tournament on my printable brackets page. In the mean time, have a care before you rely on printyourbrackets.com.