Boston chose not to re-sign Jason Bay, in part, based on his below average defensive metrics. Well, the problem with UZR is that no one really knows how to calculate it; in the latest version, Bay's an above average outfielder.
The theme of Theo Epstein's offseason was "run prevention," which emphasizes pitching and defense. That focus on defense was a big reason the Sox didn't play ball with Bay, since he was clearly an inferior defensive player.
You could tell from the naked eye, and the advanced statistics backed it up. Bay's UZR was -13.8, which, to oversimplify everything, means he cost his team 13.8 runs last season.
Well, back up. Mitchel Lichtman, the creator of UZR, is still fiddling with the formula. In the latest update, more park factors have been taken into account, including the quirks of Fenway's left and center fields. And lo and behold, with the current formula, Bay's UZR last season jumps to +1.9.
We're not blasting the Red Sox for not-resigning Bay. For one, they don't use UZR, but rather their own proprietary formula. And by any metric, Mike Cameron is a much better fielder than Bay (Jacoby Ellsbury, on the other hand, is one of the worst).
We are saying that it's foolish to jump aboard the Sabermetrics bandwagon (like one notable Boston Sports Fella did) while they're still working out all the kinks. Bay's 2009 play didn't change since last week, but by UZR's reckoning, he prevented 16 more runs than we thought. And that number will change again with the next update. That's the definition of a flawed stat.
UZR owes Jason Bay an apology [Boston Herald]
UZR Updates! [FanGraphs]