Oy, these are some long threads.
Anyway, I'm here because Dizzy asked me to chime in on the bracket issue. I'll try not to go too in-depth on how I make brackets since Dizzy did a pretty good summary already in
this post. But briefly, my preferred method for making brackets for larger tournaments is this:
1. Identify the people considered to be "top" seeds, ideally about 6%-13% of the total entrants (e.g. 4-8 people for 64 entrants).
2. Identify the people considered to be "second" seeds, again about 6%-13% of the total entrants.
3. Enter the seeds into the bracket/pools so that all seeds are spread apart from each other and each pool or bracket section has equal numbers of top and second seeds. Account for regional seeding if necessary.
4. Randomize all other entrants into the bracket, obeying regional seeding.
When it comes to seeding, however, one does not need to be super-precise about number of seeds and how to order them. After all, any ranking system is pretty much just going to be an estimate. Sometimes there's a definite 1 and 2 seeds among the top players, in which case they can certainly be separated in the bracket. There need not be hard numbers of people in each tier of seeds either. At a recent tournament I ran, there was a clearly defined top 3, then a group of 5 players that were a level below. Rather than pick the "best" player from among the 5 second-tier seeds to promote to the top level, I just made the bracket so that one quarter of the bracket had two randomly-chosen second seeds, whereas the other three sections of the bracket had one top seed and one second seed.
Even when there are ranking systems in place that could ostensibly be used for seeding (like the MK rankings), I would not strictly adhere to them, as there is always a large amount of variation and fluctuation. For example, somebody may be ranked very low, but if that person always seems to get the 1 seed first round every tournament, then they haven't faced a sufficient variety of competition to determine their true skill level.
However, that doesn't mean you can't use them at all. In fact, if you would like to use them for tournament purposes, I would recommend using the method of tournament making used on the professional tennis tours (ATP and WTA). The singles competitions the major tennis tournaments have 128 entrants, and each tour has ranking lists that go far beyond that number. However, only the top 32 in the rankings are seeded, with the rest of the players randomized. The 33rd person in the rankings could just as easily get Roger Federer first round as he could the 128th-best player. Moreover, only the 1st and 2nd seeds have hard positions in the bracket. The 3 and 4 seeds are guaranteed not to meet the top 2 until the semis at the earliest. However, it is randomly determined whether the 4 seed will be in the 1-seed's half or in the 2-seed's half (with 3 going in the other half). Then the 5-8 seeds are randomly assigned to each quarter of the bracket, then the 9-12 seeds, and so on. Had seeds held to form in the most recent tennis major (the US Open), the quarterfinals would have been, from top to bottom, 1 vs. 6, 3 vs. 5, 4 vs. 8, and 2 vs. 7.
The key to making fair brackets, however, is randomness. If no human purposely makes the bracket (i.e. a computer randomizes it or names are drawn out of a hat), then the amount of say they have in its structure is drastically reduced--even if they picked the seeds. I used to use a series of randomly-generated numbers to make my brackets. I've since written a program that randomly assigns players to a bracket (while accounting for skill and regional seeding). It is difficult to claim such unbias if no randomization method is used. One may try to assign things arbitrarily, but "arbitrary" implies that a human still is aware of the affect of their seemingly-random decisions. Something that is truly random should have no human involvement whatsoever. As Dizzy said, it is even more ideal if the person making the bracket is not taking part in the tournament, but I understand that this may not always be possible to achieve. If the assignment of entrants is randomized, any such bias is drastically mitigated. (Though if for some reason no options for randomization are available, then yes, the job of making a bracket should be given to a person with as little connection to the tournament and scene as possible as they might provide the best arbitrarily-made bracket.)
Anyway, sorry for being long-winded and rambling (I did write this at 3:30 am), but that's my take on how brackets should be done. Feel free to ask any questions. I may not get back to you with an answer immediately as I don't check this site that frequently, but I will check back a couple times this week.
Sent you a message.
