Firefox succumbs to GNOME disease

Concerning the first assumption, which you framed as "possibly the only really widely used Free application", it's simply too broad to measure. The problem could be defined and refined to fit the assumption or to deny the assumption. For example, Apache is free and used by many millions of people, but maybe that doesn't fit what you had in mind for "application". Mplayer? Gaim? GCC (without which it might be that neither Linux nor Firefox would exist)? It's just too loose to make a count.

Refining the question more won't help because you would still be left with the old problem of counting who actually uses what. Even counting hits by Firefox on websites wouldn't help because (a) some people deliberately change the user agent and (b) that restricts the software you're counting to only browsers. These counting problems leave you with nothing to measure except the amount of publicity surrounding a particular piece of software, which does not evaluate the effect of the configurability of the user interface.

The other part, "[t]he proof is in the results, right?", ignores the effects of marketing. I probably messed up by assuming on top of your assumption (we're looping!) that you were talking about the proof of user interface design being shown by how many people voluntarily use some piece of software.

Firefox received a great deal of publicity by groups and individuals. Even security organizations advised IE users to switch to Firefox. This was not advised because of the user interface. "Take back the Web" icons appeared on a very great many web pages, including my own, linking people to where they could download Firefox. This was due, in part, to security concerns, to concerns about non-standard extensions to Web protocols, to desires by some people to promote freedom, and to a kind of "me too!" attitude we all fall into at one time or another.

The ease of use of Firefox played a role in some downloads, but by no means all downloads or continued use. I would guess that a lot of secondary marketing, that is, people recommending Firefox after switching from IE, came about because of the great relief from all the breakage and ties in IE.

The ease of use of Firefox again played some role in getting that second tier of people to try it. That is probably a good indicator of how well the Firefox developers managed to make a user interface that does what is expected and kept the user interface from being scary. It does not prove that a simple user interface explains the popularity of Firefox nor does it prove that a lack of configurability explains the popularity of Firefox.

I realize you were not using the word "proof" in any strict sense, so please don't take the above as just playing semantics. Consider that it could just as easily be argued that much of Firefox's popularity could be explained by the ease of add-ons. Extensions seem to be extremely popular.

Once you consider marketing, relief from a dangerously broken alternative, familiarity and expected controls of the user interface, and the extensibility of Firefox, the idea that reducing access to configurable features and functions increases the popularity or usability of Firefox just doesn't hold up any longer. It also doesn't hold up when applied to any other software.

Finally:

"As far as usability goes, what metric would *you* use to define success, if not popularity?"

One way is to check feedback in all its forms. For example, if it is _known_ that there are lots of users of a program but there are few bug reports, suggestions or complaints buzzing around concerning the user interface, it could very well be that the developer(s) got it just the way people want and expect it to act.

This actually could argue against Firefox's design. Instead of all those extensions (and the talk about them) being a vote in favor of Firefox as it is, they could be construed as being many votes against those features being left out. The good thing is that Firefox's developers can evaluate those extensions to see which might be worth including by default to serve a great many users without too much "cost", as they put it, to those users who do not want them.

Dang it, I didn't mean to prattle on so much but you asked a question that interests me and I'm a motormouth when started. If you'd flamed or trolled, it would have been easy to just snap back something nasty. I hope some of the ramblings make some kind of sense, for a change.

(Just because I can't resist, here's what I think of one notorious piece of software that came with a very, very simple user interface: SonyBMG)