Honestly, if you start seeding people then you would be obliged to run #1 against #16 the first round, right? I mean, that's how brackets work. It's not fair to run #1 against #2 the first round either, if you look at it from their point of view.
The problem is we seed to make the run groups, but then we go and make things random within the run groups. That doesn't work. It needs to be all ranked or not at all - otherwise you end up with messed up pairings.
What would be wrong with #1 vs. #2 first round (and #63 vs. #64)? If we are going the bracket way as you suggest, than it really should be #1 vs. #64 (AM vs.HS) - which would be aweful. Isn't #1 vs. #2 a lot more fun than #1 vs. #64? Keep in mind this is meant to be fun - sure, if we were doing an NCAA type championship that was all about avoiding any upsets we would want #1 vs #64, but this is meant to be fun. For that, the design should make pairings closely matched. Or, do we just want this to be fun for #1 (or as we ran it last year #1, #17, #33, and #49)
For some reason everyone seems to think it is important to split up into 4 run groups based on PAX, but then say PAX is not really important so mix it up within the run groups - this is not a consistant design. Inconsistant designs lead to bad pairings and arbitrary splits that favor the tops of each run group and disfavor the bottoms - which is why some people are complaining so vocally. Either make it totally random and screw the lower classes or use PAX (or another good measure) all the way through the pairings and keep the competition close and exciting. Those are the only valid designs - anything else is flawed from the start.