QUOTE (Jake @ Sep 17, 2013 -> 09:22 PM)
UZR is the absolute best way to evaluate player defense - it is predicated on humans reviewing video with objective (ie, not opinion-based) measures. The belief is that UZR is best evaluated with about three seasons worth of data, though. This is of course true for lots of other statistics; we know that batting average and other conventional stats tend to be unreliable over short periods of time, even single seasons can be predicated on luck or random fluctuations.
So, I look at UZR with smaller sample sizes, but always take it with a grain of salt. One season isn't a bad amount of data, but when one season sticks out as different from the player's other years, you should definitely look upon it skeptically. This is especially true because defense isn't nearly as prone to things like slumps as offense, so in most cases a brief change in UZR is nothing to be alarmed about unless there is a clear cause like injury. Some players will have an upward trajectory in the early parts of their careers and many others will steadily get worse defensively.
I thought it was opinionated. ESPN is also responsible for some of these new stats, for instance "Total QBR" which is a QB rating out of 100 rather than a PASSER rating up to 158.3. I guess Sabermetrics is doing the same.