The problem is that football is the ultimate team-oriented game, which in turn makes individual statistics less meaningful and relevant than they are in other sports. With the evolution of the internet and computers, more and more statistics are readily available compared to generations past, and current sport fans crave those stats - as do talking heads and written media types for their professions.
Blend in the fact that the NFL has by far become the nation's most popular sport, and you have the perfect storm for a stat-driven site such as PFF to not only become popular, to be given a free pass by far too many to question the validity of their data.
Exhibit A: Earlier in this thread Monson provides an example of Brady throwing an incomplete pass, missing his receiver by 6-7 yards. Immediately he diagnoses the blame as being solely on Brady, without ever considering any other possibility (e.g., the receiver ran a bad route and was in the incorrect spot).
Whether that is due to lack of training (never having worked as a coach or scout at any level), laziness, marketing (a desire to avoid pointing out a potential error in their stats, which would discredit their 'research'), the bottom line is that the data is corrupt - and therefore if not worthless, at minimum very suspect.
Exhibit B: Many of their stats rely far too much on information that they can not possibly know from watching a televised game, and others are simply too subjective. For example quarterbacks are rated on throwing under pressure, but conversely quarterbacks who sense the pressure sooner and react accordingly do not get that credit. A quarterback who maneuvers in the pocket to find a passing lane gets no credit, but a quarterback who has a pass tipped at the line gets that incomplete pass removed from his accuracy percentage. A quarterback who makes the mistake of throwing into double coverage is not chastised, but gets extra points if that pass is completed.
Exhibit C: A few years ago PFF's stats came to the conclusion that David Garrard was a better quarterback than Tom Brady; Stylez G. White was a better DE than Mario Williams and Jared Allen; Lamarr Woodley was better than DeMarcus Ware; Mike Vrabel (with KC) and Tamba Hali were better than Elvis Dumervil; Sione Pouha was a better DT than Vince Wilfork; Gary Guyton was a better LB than Jerod Mayo; and Jonathan Stewart, Justin Forsett, Jason Snelling, Ladell Betts, and Brian Leonard were all better running backs than Adrian Peterson.
Those type of rankings should have sent the developers of the site back to the drawing board, but instead they steadfastly defended them as being impeccable and beyond reproach. That stubbornness on PFF's part to defend their analysis rather than to even consider the possibility of the methodology being flawed makes me seriously doubt the validity of any of their research.