LogFAQs > #965029211

VintageGin posted...

Since you're essentially just looking at how "common" each quality is, I think it's probably a relatively objective way to measure things. Of course, you'd have to discount the fact that categorization is something that people are generally bad at (including those who built the tool).

The categorization is the problem since it's fully subjective, so the derived stats aren't objective at all. You can tweak the numbers however you like in order to create the conclusion you want.

They probably think it balances out by getting a bunch of different people's opinions in there to get a consensus on the numbers, but that's just design-by-committee, which is the most intensely biased method possible.