the variance happens because the sample's so small. most analysts agree that you'll need nearly 2-3 years of defensive data for UZR to be a good reliable indicator of defensive value. i know it's not entirely analogous, but looking at AVG right now, i have trouble believing Jason Bartlett is a .370 hitter. But does that help? We already know that AVG is a pretty bad measure of the offensive value of a player. For the record, I think Bartlett is hitting closer to .350, but your point remains. Then again, if you're saying you also don't think Bartlett is a .930 OPS player...well, I don't know what to say. To continue using the Bradley example, looking at his OPS this year you could conclude that OPS is garbage using similar logic used to criticize UZR(not that all the criticisms on that front are illogical or not valid). Because fielding has been so subjective for so long we like to think of it as a static ability. Jim Edmonds is a great defender, Adam Dunn is a terrible defender, etc. But guys can have fielding slumps just like hitting ones, especially when it comes to stuff like range and cutting off balls in the outfield, that's more difficult for us watching at home to identify.