[JohnFen]

> I wonder what kind of evidence might actually be used.

IANAL, but perhaps the dividing line isn't whether or not the term is commonly used as a generic one, but whether people are not generally aware that it's a brand name. I wasn't aware that "Velcro" was a brand name for a long time, for instance.

I think the majority of people using "google" generically are also fully aware that "Google" is a brand name.

So the evidence would be some sort of study showing that people think it's really generic?