doesn't UZR require a huge sample size to be accurate?
3 years is the line you hear the most. UZR/150 is particularly tippy at this point due to a few plays having outside influence.
Having said that, the common (un)wisdom about Dunn was he could not field any position, and many pointed to his UZR/150 from last year at 1st to support that view. Live by the small sample, die by it. The other thing that is interesting is that FJB observed that his bad UZR was mostly racked up when he was a spot fill in for Nick, and that once he moved over full time he stopped piling up negative UZR runs.
This all kind of jibes with what we all casually observed - that he is not a historically bad first baseman and his play seemed OK once he was a regular. If he is just a -5 to -10 fielder in terms of talent, he is a 4+ WAR player.
[some of you will take that statement as praise, some will take it as gibberish.]