If it was coming from any old hoohoo on the street, I'd agree completely. However, coming from a well-known, long-running organization with a history of great statistical breakthroughs, you can be pretty confidant that they aren't making things up or haphazardly slapping a stat together. They update the WARP formula yearly to incorporate new information and improve upon the old formula. They've been tweaking WARP for years.
Another test for stats is "Does what the stat say match up well with a) perception and b) other stats" and for WARP the answer is a qualified "Yes" to both a and b. Should it be used exclusively? No. Should it be completely ignored? No.
WARP3 for the first player: 10.7.285 AVG, 31 HR, 75 RBI
.296 AVG, 30 HR, 94 RBI
Who's better?
WARP3 for the second: 11.1
This one surprised me but upon further examination, I see why. Mickey Mantle is the first player - 1959. 2007 Jimmy Rollins is the second. While Mantle was easily the better hitter (.321 EQA to Rollins' .297) and both were plus defenders (14 FRAA for Mantle, and 9 FRAA for Rollins), Rollins played every single game, and that has a ton of value, which is why Rollins was more valuable in 2007 than Mantle was in 1959.
The first player is Matt Cain in 2007. The second player is Ernie Shore in 1916. And this is a prime example of why a) won-loss record is useless in evaluating players and b) simple stats, while okay for comparing players in the same season, are utterly useless for a total evaluation of players because they give zero context.7-16, 3.65 ERA, 163 K in 200 IP
16-10, 2.63 ERA, 62 K in 225.2 IP
Who's better?
Cain's 3.65 ERA in 2007 is better than Shore's 2.63 ERA 90 years earlier. Matt Cain's WARP3 was 6.6. Ernie Shore in 1916 was 2.2.