Scheduling:
I recommend an aggressively planned tournament schedule. That is, we have 2-4 specific dates (weekends, probably), and very specific times. Teams should be expected to play on those dates and times.
The gaps between matches needs to be large enough that 3 90-minute games can be played, as well as with a gap in-between. I think everyone can agree that if multiple games run over 90 minutes then they have amnesty regarding appearing to the next match late.
I think that we should aim for at least two matches a day, and at a decent time for the entire world (so 11 AM EST would probably be a good time to aim for since that's afternoon for Euro, not-godforsaken-early-morning for California, but it is also late night for Australia, though it would be hard for Aussie players a ton).
I think we should do a double-elimination format. It makes the tournament a little longer and more intresting. I suspect we'll be seeing 6-12 teams so if we do single-elimination then it will be too short.
Demigod Selection:
1) Unrestricted - I think this is better for smaller tournies since demigod restrictions are a way to "even out" the competition and make T2 teams have a better chance against T1 teams, and, naturally, in a smaller tourny less T2 teams will exist so it's not important.
2) "Draft Pick" - Each Demigod can be played only once in a series of three matches. I think this makes for some really interesting matchups (especially "budgeting" your UB and Rook), but it inserts a LOT of meta-game into the game itself. I think this way will result in a lot of mirror matches, just a series of 3-4 different mirrors over and over in the same order. Third games in a series might be a bit weird as well. This also punished T2 teams a lot.
3) Restricted - We draft up a slightly modified set of demigod choices. I think that we should add Ereb-Oak-TB to the allowed list of teams. We could remove Sed-Occ-TB as well due to their overly prevalent participation last tourney, and if we do a 4th we might be able to allow them again depending on how it works out.