The current picking structure rewards strategic thinking and having a good robot. It also gives robots with a lower qualification average a chance to get on the wining alliance, especially if they fill a strategic niche. The way alliance selection loops back around gives a bit of balance. This seems like a nice balance of rewarding teams with high QA but also giving other bots a chance.
The reason waterloo in particular is often such a 1st+2nd seed lockout is because of how incredibly good the best two bots are. Waterloo is an extreme example; dallas regional has a lot of amazing teams, but even there 1st+2nd seed lost to 3rd+11th seed.
On average, 1st alliance is probably the most common winner, and most likely 2nd seed is the most common 1st pick. However, waterloo is not the rule. My team won greater DC this year picking the 9th seeded robot as 1st pick (623). Our 2nd pick (4541) was picked for doing a fantastic job filling a strategic niche, canburglarizing. In the finals we didn't face the 2nd alliance, but the 8th alliance. It seems arbitrary to punish the 2nd seeded team. Sometimes the 2nd seeded team only is there by a narrow margin, or maybe had better alliances in qualifications. Throwing games would definitely happen.
