Agree about Joe Montana as the best ever QB of all time, and we’ll never have a statistical formula to clearly rank the elite at every position. I would say Dan Marino in 1984 had a great pair of receivers in Mark Duper and Mark Clayton to throw to, and center Dwight Stephenson was better than anyone on the 49ers offensive line. Marino with an elite RB like Roger Craig would have been scary.

Other reasons for the concentration of good seasons on the last decade or so would be the pass-heavy nature of the game. Even w/ normalization for schedule length, any system that uses attempts/plays will favor recent players. You can’t adjust for this in a linear manner because there was too much variance in terms of how pass heavy teams were in earlier eras. The penalty for lower QB attempts/game early on is somewhat mitigated by another form of variance anyway: efficiency. If you’re comparing against the league avg. for a given year, players weren’t as tightly bunched, so that gives guys like Luckman and Graham an opportunity to stand out regardless.

One other thing: it’s been brought up by others before, but Manning’s 2004 season looks even more statistically freakish when you consider the time he sat on the bench in blowouts (7 quarters) compared to Brady (3 quarters) and Marino (roughly 2, if I remember correctly). The final regular season game v DEN in particular where Manning played a token series really put a dent in his numbers. If we were to assume that he would have shredded DEN the same way he did the following week in the playoffs had he been unleashed in a meaningless final regular season game, he could have been looking at a season value of roughly 2630.