I'm not "in love" with WAR, but there is a strong correlation between team WAR and team wins. Last year in the NL, 12 of the 15 teams had a differential between WAR and wins between 46 and 52. That is a pretty narrow range. Obviously the error is greater with only 15% of the season gone by, but there is no denying the correlation. I wouldn't question the metric because I saw an outlier.
Actually, though, what was faulty was my mathematical interpretation. If a replacement level team wins 49 games over a year, that would be about 7 games through the 24 that were played when andyb reported the numbers. Six games above replacement level is 13 wins, and the Phillies had won 15. That's not really an outlier at all; the Phillies are only slightly outperforming their WAR.