Cool idea you've introduced the reaver. I have a suggestion for it.
I used to play a good amount of Magic the Gathering back in the day. A major part of competitive Magic is a format called "limited". One section of limited is a draft. I just c+p'ed from wiki for the explanation:
In a "Booster Draft", several players (usually eight) are seated around a table and each player is given three booster packs. Each player opens a pack, selects a card from it, and passes the remaining cards to the next player. Each player then selects one of the remaining cards from the pack he or she just received, and passes the remaining cards again. This continues until all of the cards are depleted. Players pass left for the first and third packs, and right for the second. Players then build decks out of any of the cards that they selected during the drafting. Talking, signaling, and showing cards is forbidden during the drafting process.
I'd imagine the best way to work this for pokemon, is for each person to be given one pack only. The total number in each pack, along with the ratio of OU to UU could be decided upon by the Tournament Organizer, however for ease of this explanation let's say the packs would be 6OU and 10UU. The draft consists of six players for the purposes of this example.
Each player would start with a unique pack, pick one pokemon, and pass it to the next player. Each pack would go around the table twice (each person would pick from a pack twice), leaving 4 pokemon in each pack to be discarded. Then each player would have 12 pokemon to choose from to create a team. I feel that this would give enough freedom, and allow players to build functional teams. The pokemon would not be limited to a specific set, and you could do whatever you wanted with that pokemon. This method of drafting is also more "fair" to everyone involved than the other previously mentioned non-bid drafts. To add further consistency and fairness, you could create the packs through a random number generator and usage statistics. A pack could be created as follows:
Ou #1: 1st to third in usage
Ou #2: fourth to 10th in usage
Ou #3: 10th to 15th in usage
Ou #4-6: 15th and lower of OU
UU #1: 1st to third in usage
UU #2: fourth to 8th in usage
UU #3: 9th to 15th in usage
UU #4: 15th to 20th in usage
UU #5-7: 20th to 35th in usage
UU #8-10: 35th and lower of UU
Again, the numbers are just examples I quickly came up with that are put there in order to demonstrate the idea. The same principle could be used for a draft game of only OU, only UU, or even Ubers. It doesn't have to be done by usage, however I see it as a fair and effective way of randomly making each of the packs "good".
Another added bonus of this method is that an added element of strategy is introduced in the drafting portion. If the player that is passing you packs is passing you packs chock full of amazing walls, he probably is not trying to make a stall team, and you might be wise to pick up some of those walls, since it is likely he will continue passing more in future packs.
The logistics of such a tournament would be the stopping block (as it always is). It would probably be wise to start small, no more than 32 people perhaps even 16, so that logs could be submitted in order to curb cheating. Relying on good faith would make it so much easier to run, but could be a problem. I foresee the problem of cheating being this idea's breaking point, and any suggestions as to how to solve it would be welcome.
You'd probably want y pools of x people being supervised by y people. Each person supervising a pool would be responsible for keeping track of each pack in a pool, and subtracting a pokemon each time someone picked one. After pools you'd set it up as a regular bracket. Since trying to do this through pm over the course of a few days/a week is fraught with complications, I feel that the best "venue" for a draft tournament like this would be something similar to the late night tournament. You would set a time that it will take place in the tournament thread a few days before, and then take applicants for the tournament at the set time. That way it can all be done over shoddy, in a much quicker and more time efficient manner. Another method is to have the draft on Saturday, and the actual match on Sunday, to avoid the block of dedicated time required being too long. People being inactive could be a problem, however if you made sure to get two or three subs per pool I feel good about its chances of working. I'd be more than willing to help out with such a tournament if anyone is interested.
If anyone has any thoughts on the cheating aspect of this type, or the Rochester type of draft whether they be solutions or further issues, please share them.