Lewis hasn't been as good as hoped, but he hasn't been nearly as awful as folks insinuate. He's been a C- player through 3 games. In terms of the power run game, he's not the problem, because he's made 23 of 27 pulls. In fact, that's been his strongest contribution.
He's had troubles as a pass blocker, but those are exaggerated by PFF's 1 or 0 system for pass rush events. For example, Lewis was partially responsible for the sack by Ray (Q2, 1:06). If you look at that play again with either the broadcast or all-22, you'll notice it's actually the front-side pressure from Miller that triggers the sack. Lewis is slowly beaten outside by Ray, but Flacco is taken down only after he's flushed forward in the pocket.
In terms of that play, I could see an argument to giving Hurst the entire sack, because it was his pressure that compromised the pocket, forcing Flacco to move and pull down the ball. However, since Lewis also allowed his man to get to Flacco right around the 3 second mark and Ray actually took down Flacco, I gave Lewis 2/3 and Hurst 1/3 (the same charge for just a pressure). In retrospect, I think a half charge to each would have been more fair, but I brought this up to illustrate a significant problem with PFF OL scoring...they don't have partial charges for pass rush events. That adds a big element of inadequacy to their system, since they are only able to ascribe either 1 or 0 for any player on a given event.
It's funny, because this offseason, PFF did some regression analysis to fine tune the value of various pass rush events surrendered. They posted an article with a 4-decimal place valuation of each. I applaud them for work to more accurately value events, but it won't help them much without more accurate count of events.
In mathematical terms, the result of an operation on 2 numbers A and B is only as accurate as the input with the fewest significant digits.
In common sense terms, don't tell me that sack was worth 2.0427 expected points when your enumeration of the sacks is off by 33% or 50%.
