February's TIOBE index has #Fortran as the 11th most popular language, marking its 12th consecutive month in the top 20, beating languages like #Rust, #R, #MATLAB and #Julia. About time to do away with that "Fortran is ancient/dead/obsolete" myth?
@samharrison7 Their index reported C falling 50% down over two years and then shooting 200% in the next few years.
What’s more likely – that the slow-moving ubiquitous language that still loves its 1989 version had a dramatic collapse followed by a magnificent rebirth, or that nothing changed, and TIOBE is so noisy and inaccurate that they have +/- 50% error bars, making their numbers displayed with 0.01% precision completely meaningless?
@samharrison7 Also they estimate that Visual Fox Pro, a niche language discontinued 16 years ago, is more popular than TypeScript. There is no useful data in TIOBE. It’s a horoscope for programmers.
@robinm@kornel Horoscope would imply completely made up. I agree that TIOBE isn't a pinpoint accurate ranking, but as a very rough guide to language popularity, it's not too bad. And, unlike a horoscope, is actually based on data.
@samharrison7 They measure counts of results in search engines, which clearly fluctuates with changes to ranking algorithms and content crawling more than actual language popularity. They’re being dishonest by not showing error bars on the measurements, and discussing fraction of a percent changes they don’t have precision for.
@samharrison7 They imply they measure current popularity and number of users, but they measure phrase mentions (posting “I’d never use x language” counts towards usage) in an aggregated set of documents spanning all of time (which is why dead langs are overrepresented). They also have a problem that disambiguations for language names don’t match common usage (“c” would overcount, but “the typescript programming language” undercounts).
It’s a very noisy data that doesn’t mean what it claims.
@samharrison7 and as such it works like a horoscope – people use it to confirm their beliefs, or are entertained when it says something unexpected, but it’s not real data, and only seems real because people selectively look where it happen to be right, and disregard all the counter-examples as unimportant.
Add comment