Is the dichotomy better described as popular vs. useful?

No, the dichotomy is better described as r-selection (worse-is-better) versus K-selection (the right thing)[0].

For a practical example of worse-is-better software becoming popular, look at the World Wide Web in relationship to Project Xanadu[1]. For a practical example of The Right Thing software being popular, look at Python. The argument is broadly that New Jersey worse-is-better produces software more likely to become successful than the MIT right-thing approach. That doesn't mean that worse-is-better is always more successful, or that the MIT approach can't produce successful software, or that there's not a time and/or place for both approaches.

[0]: http://en.wikipedia.org/wiki/R/K_selection_theory [1]: http://en.wikipedia.org/wiki/Project_Xanadu

No, I don't think that captures the ideas in Richard Gabriel's essays. Simple vs. The Right Thing are the two sides, where the right thing is a complete and polished piece of software full of features, what a typical young engineer would create if given lots of time and resources to do things "the right way".

I think he calls it "worse" as a way of emphasizing that e better course of action may not feel fully responsible, as in the PC-losering example.

> PC-losering

Has any OS other than ITS ever fully addressed this?

It's mentioned in the paper "EROS: a fast capability system" (http://www.eros-os.org/papers/sosp99-eros-preprint.ps): they write "This design is similar to that adopted in Fluke and MIT’s ITS system" and cite the ITS reference manual for the latter.

(I remembered reading that paragraph since before that I had only ever seen the technique mentioned in Gabriel's essay :) ).

Well, my point is this: If only three or four OSes have ever implemented it, none of which have been very successful, how important could it possibly be? Do other OSes have ways of making the problem irrelevant without actually fixing it like ITS, Fluke, and EROS did?