But he's directly refusing to make --no-doc the default, which is the whole point of the request.
People are complaining about having to install a .gemrc to get sensible behavior. Making the file a few characters shorter doesn't make it less of a burden.
Personally, I can switch to a browser window, google for a gem, find its docs, and read them online in less time than it takes for ri to return its usually incomplete result or ask for brain-dead disambiguation. Ruby would have a better reputation if ri simply didn't exist.
That's the sensible behavior if you plan on reading local documentation.
I can't say I've ever used ri, although I do browse local rdoc, sometimes. I know plenty of developers that hadn't heard of "gem server" until I mentioned it. And this is completely discounting production environments, where the documentation is almost certainly a waste of deploy time.
Approximately 60% of all questions on #ruby-lang could have been answered by ri, just for the core and stdlib. It's a shame it doesn't get used more.
It's an orthogonal problem, though. The main issue with doc generation is that it is really, really slow (compared to the rest of the install process). If it can be sped up – patches are surely welcome – it'll just leave the envs that flat-out don't need the docs, ever, and those can use the new --no-doc.
People are complaining about having to install a .gemrc to get sensible behavior. Making the file a few characters shorter doesn't make it less of a burden.