For those who can't understand without a visual
The graph clearly shows one class spiking way above the standard deviation for all classes. No matter what system you happen to be examining, that sort of sudden change in recorded values is indicative of something not working as intended.
I'll admit right here and now that I don't know how raidbots data collection works but...
Wasn't this around the tie that the Seal of Truth glitch gave ret paladins 3908314970 dps?
An unrelated bug massively inflated Ret DPS for about half a day, and that's what's causing the weirdness in that 14-day moving average. Note that the period affected is 14 days, the same as the moving average window and that changing it to a one-week sample period reduces the period affected to 7 days.
The other poster is (hopefully unintentionally) using very bad data to make a point. That data has nothing whatsoever to do with Gurthalak and the topic at hand.
A better graph to use is:
... which shows 75th percentile DPS over a longer period. You can see the rise of Ret up to Jan-1, and then a slow fall followed by a very gradual rise again. Geared, well-played Rets are only recently doing better than they were on Jan-1.
I'm sympathising more with Krinu's position now - we've done this to death and we're not saying anything new now. I started reacting to the Warrior earlier, but on reflection, I'm comfortable knowing that random people on the Internet disagree with me on fairly esoteric points of game balance.