This is calssic Chicken & Egg problem. You won't find much evidence of it
in use because there's not much support for it. You won't get support for
it until there's evidence of it in use.
The "image type fallback" I proposed was specifically to address this
issue. It wasn't about WebP specifically, but about the idea that it's
fundamentally a *smart thing* to allow for a mechanism that chooses
whatever file-format the current environment happens to support. Because
that's the only way to break the chicken-egg cycle problem.
Right now it's binary: either the browser supports the format and you see
a picture, or it doesn't and you don't. That's not tollerable and therefor
no-one risks using the new format. Which makes any new format unattractive
to implementers.
-Matt