Completely agree. If Tyson Clabo is expecting chip help on the outside from an RB, and the RB doesn't show up, all it looks like to us is Clabo whiffed completely and got his QB killed (again). He gets a negative grade on his PFF report card while the RB gets off scot free.
Having said that, for most plays I do think an intelligent football observer can determine the assignment for most players, even without knowing their playbook. As a result, an assessment of each of the players relative to their assumed assignment is, IMO, going to be fairly accurate. No, it's not perfect and yes, there is some subjective judgement going on. But there's definitely value in it, and over the course of a season, I think the inaccuracies average out - and what you're left with is a pretty good overall ranking. I have no issue with calling PFF the best NFL-wide grading system out there not only because it tries to use the data from each play (as opposed to a few highlight reel plays that many so-called "analysts" base their opinions on), but because they also try to standardize their approach across all the games and teams. Unless we can get a bunch of ex-coaches to do the grading for PFF, I'm not sure how we can improve it much (maybe normalizing a player's grade based on the ranking of the player they're playing against?). It's not perfect but it's pretty good.