For one thing, framing your accuracy in percentage instead of the obscure "centipawn loss" makes understanding how well you played much easier. BTW I've compared their analysis on a number of games, and it seemed like lichess is much tougher on you in what it defines as a blunder.
I disagree, ACPL is a well defined system, while nobody really knows what the accuracy means or how it's defined.
1 CPL, in short, is the loss of 1/100th of a pawn. Take the average of that and you get ACPL. So a ACPL of 100 means that you on average lost one pawn on every move (of course this isn't literal) that you played.
Interesting, I find the exact opposite. Measuring things in terms of pawns give me a feel of how far behind I was from optimal play, whereas with the percentage, it's quite abstract. What does 70% accuracy mean?
100% is only best moves (my record is 99.8%), anything above 90 is awesome, etc. I'm not sure what 15 ACPL means, but if you tell me 95% perfect it's more intuitive to me personally. But it might be because I'm not a very good and sophisticated player...
I think the percentage give you a sense of familiarity. You're used to be graded 0-100 in exams and such. However, that grade is actually not that useful or intuitive as you may think, since you don't know how your game is scored. You don't understand the metric. For instance, it could mean the number of moves you got perfectly divided by the total number of moves in the game. It could be the number of moves you lost at most 30 centipawns by the number of moves. It can mean anything, the point is just that it's a black box and no one truly understands that metric outside of chess.com. Whereas average centipawn loss is a tried metric which everyone uses, except for chess.com.
