My wife pointed these out to me – MMO designers who like metrics will probably find themselves mesmerised as well.

This blog runs metrics on the Armory to determine the class makeup of teams at different Arena ratings. (For the uninitiated, 1500 is an average, starter rating, whereas 1800 is considered good and 2200 is considered wtfpwnage). As an example, you can see that Druids and Warlocks own in 2v2, whereas locks are average in 5v5 and druids are considered kinda crappy.

They are very fascinating to compare to these metrics that a Blizzard designer released on the boards. These metrics, unlike the other set, are somewhat normalized, meaning they factor in participation from all people playing a class (in theory, meaning that warriors will score lower because many raid-specialist prot warriors will choose to take alts into the arena instead)

In the thread, Blizzard hints that they are starting to use data such as this to balance the game. Which is not unsurprising – anytime a game has multiple formats, designers do their best to balance with all formats in mind. As an example, Magic has normal, extended, draft, legacy (play with any cards you want!). They do the best they can to support all of them, but at the end of the day, the ‘normal’ game is becomes the tiebreaker. In MMOs, PvP often becomes the tiebreaker, as it is the source of the most direct competition, as well as the most vocal complaints when there are flaws.

But there are problems, of course. As the Elitist Jerks thread points out, top teams tend to follow one of only a few templates (as most freeform build-your-own-ability game systems tend to do), and of those, shockingly few of those has a spot for a hunter (especially shocking when considered hunters are the most played class). Does this mean the hunter needs a buff? Or does it need some sort of synergy tweak to make them play nicely with others?