Please don't take my criticism as an indication that you're headed on the wrong track, Dave. I'm a professional nitpicker, after all. I'm still quite taken with your data!

One additional resource you might consider is indeed.com, a job hunting site where I've personally wasted entirely too much time playing with their tools. (Let's see, at indeed.com/salary I can compare, say, the pay for python and perl programmers in PA. Sure, the published salaries are guaranteed to be on the low side—most jobs say "negotiable" unless they're warning you they cheapskates—but it's a start.) I have no idea what it takes to work with indeed.com's data but they were very nice folks and sent me a charming e-mail message after I once mentioned them in passing. (Now, maybe they'll have reason to do so again.)

Another possible resource is realrates.com, which was started by my old buddy Janet Ruhl. (She and I were co-sysops of the CompuServe Computer Consultant's forum in the late 80s and early 90s. Lovely lady.) I'm not sure how up-to-date the surrounding info is (I think Janet's attention turned away from it a few years back) but people keep updating the consulting and salary rates so it may still be useful.

One question, however, is what you mean by "popular." If you looked at popularity by the number of lines of code all-time which are still in production, I dare say that COBOL would win out, based on the language's age as well as its wordiness. I wouldn't expect most of it to be reflected online in any way simply because most of that code is and always was proprietary. And it would be unlikely to show up in job ads because most such code is now labeled as "legacy" and thus its maintenance is outsourced or given to people who have committed an unforgivable sin. Still, there's a big financal company in New Jersey who just pulled my brother-in-law out of retirement (at, my sister tells me smugly, an impressive consulting rate) because he knows all 600 of the company's COBOL programs from the inside. Popular? Heck no. Used? Sure.

Most of us think of popularity within some time frame—whether that's "what's in fashion" (or likely to become so) or "currently being used." That is possible to ask of developers, and it's among the things that Evans Data does regularly. (In full-disclosure mode: I've been the analyst on some of their reports over the last five years. And I've known the company owner since 1980.) One methodology difficulty is that very few developers write in only one language; most use a mix of several. So if you program 5% of the time in, say, Ruby, does that make you a Ruby programmer? We could argue about it over a couple of beers (especially if you're buying), but I'd say Yes, because anything non-zero implies skill even if it's at the bottom. Is that popularity, though? Is it reason enough to advise someone, "You'd better get ready for this new language on the horizon... start training the development staff now!"? That's harder to say. Which, of course, means it's a lot more fun to think about.

Dave,

I appreciate the effort you put into your study, but I have to say that part of your approach seriously skews the results against Microsoft languages. Much of the discussion around the .Net languages takes place in online communities focused on those languages, rather than on the sites you included in your study. The same can be said for open source development using .Net languages. Finally, many .Net developers avoid sites like Slashdot because they do not want to be subjected to the serious Microsoft-bashing that is so prevalent on those sites. I don't know what the answer is - including those Microsoft-focused sites could also skew the results - but it would be worthwhile to investigate alternatives if you update the study in the future.

Ed

Perhaps we shouldn't be using the word "popularity" here, as what we mean is "most common" or "used most often."

Because you don't have to like to write in a language to write in it.