Given the same accuracy, scores in each higher level of difficulty are much higher than the one before it. Even if you're playing a No DD version of the song and therefore seeing all of the notes even on easy difficulty. That continues on in to Master difficulty, which is all the notes PLUS master mode (so you can't see them). If you get a 100% accuracy on Hard and then Master, the Master score will be significantly higher than the Hard one. That all makes sense and is fair, but Score Attack scores in general are a bit flaky. Not exactly arbitrary, but the algorithm that gives you a score rewards some things that aren't necessarily objectively "better". For one thing, getting a long streak increases your score multiplier, on up to x99. If you get a long streak and then collect points at that high multiplier, you can easily get a *much* higher score than a run where you have fewer overall misses, but they are spread out evenly through the song. In that way, Score Attack scores tend to favor getting clusters of misses, preferably at the very end or very beginning of a song. Also, you get more points for being "Perfect!"ly on time for notes, or at least "good". That is sensible, but on the other hand is subject to the vagaries of different people having different amounts of input lag, etc. So, that makes Score Attack scores a decent way to judge your own progress compared to past scores, but arguably less useful in comparing your scores to the scores of other people (who might have an advantage in a nearly lag-free setup, or a handicap from an older system or poor input lag setup). So, for Championship purposes, it was decided that precise accuracy is a better, more objective way to rank participants. It is just a bit confusing, because Score Attack is the only way to get a highly precise accuracy rating -- down to the hundredth of a percent compared to Learn A Song truncating down to an integer value for accuracy. That makes Score Attack better for getting a precise accuracy rating, even though the actual score itself is only a last-resort sort of tiebreaker. I think that is the best way of doing things for the Championship, although it does create a couple of weird edge cases. First, Score Attack has the problem of "strikeouts" if you get a lot of misses in 3 sections, which is annoying. I wish it would let you keep playing even if you "strike out" -- invalidating the score would be fine, but just give us a non-sugarcoated overall accuracy value as usual at the end instead of cutting the song short. Playing a NDD version of the song on Easy can help with that problem -- NDD makes it so that even "easy" has all of the notes (and therefore gives a legit accuracy), but it is significantly more forgiving about how bad you have to screw up to get a strike / strikeout. And second, the rules rank a 100% run on hard (or easy NDD) above a 99% or even 99.9% run in Master. Arguably, there could or even should be some consideration granted to very high (but not perfect) accuracy Master Mode runs. On the other hand, it would be hard to nail down a way to objectively do that; plus there is still a bit of a concession in that a 100% Master run will have the same accuracy and streak as a 100% run on hard, but will definitely win out on the last tie-breaker by having a higher score. Plus I believe Rodman gives some points for Master mode runs on his multi-week leaderboard. ...OK, done rambling. Hope that explains the preference for Score Attack but NOT actual SA Scores, and wasn't too long and boring. :)