Comp is BS. I think it's a terrible idea to punish people for using the armies and units they like. On the other hand, current 40K game balance is a hot mess. For the sake of fairness something has to be done about the dominant armies being played by dominant players so that the guy who really loves his Dark Eldar has an outside shot. The win rates of armies that include Eldar and Tau over ones that don't is pretty disheartening.
Seeding is a pretty good approach to providing decisive tournament results when you can't have the mathematically required number of rounds to pick a true winner. The problem with seeding is that 40K tournaments are so different in terms of terrain and scenarios or just the variance of 40K as a rules system. It's hard to get a very good sense of how good a players is outside of the extremes of the scale. The best players rise to the top, the worst fall to the bottom, but there are lots of good players that can fall in between due to pairing and luck. The solution so far has been to have two day tournaments where the first is essentially a seed for the second. That approach worked pretty well for 5th edition, but with the skill multiplier effect the power armies have this edition, we're seeing second day top brackets looking pretty homogeneous in terms of army selection.
My solution: mirror matches for first round. Pair the first rounds according to primary/ally choice. We get to cut half of the power army players out of contention right off the bat while still rewarding player skill. This cuts the skill multiplier effect of army selection out of the results for the first round. It incentives diversity in codex selection while not penalizing the band wagon jumpers out rite. You're going to lose to a better player with same army eventually in the tournament. Best that it happens earlier before you knock out decent players hamstrung by poor GW balancing.