[jameshart]

It's a practice, but as an English speaker I have the privilege of making my own corpus up as I go along. When I say I find I prefer 'most wicked' to 'wickedest', I'm expressing the subjective opinion that I think I would fall in the 22 part of that corpus, not the 20 - if anything I wrote about relative wickedness were to make it into the COCA corpus that number would go up to 23.

But that would be true even if I were in the minority. The corpus is just going to reflect usage, and my usage is as valid as any other, so... while it provides some sort of peer validation, it doesn't answer the question of whether one form or another is 'more correct', only whether one is 'more common'. Or commoner, maybe. No, I'm going to stick with more common.