Hacker News new | past | comments | ask | show | jobs | submit login
TIOBE language popularity absurdity (accidentalidad.wordpress.com)
43 points by fogus on Apr 8, 2011 | hide | past | web | favorite | 16 comments

Anecdotal: There are several shops in town that use Progress 4GL. I used to work at one of them. It is a definite niche language language, I had never heard of it until I worked there.

However, if you are a customer of a (smaller) bank or credit union there's a good chance that you are indirectly using this technology.

That said, I'm not sure what this rant is about? Is he mad that scala/groovy are more niche?

You wouldn't find (any?) Progress 4GL on github because it's a commercial language. You have to pay for it. But it generates millions of dollars of sales and is in use in hundreds of companies (at some level).

I find it really hard to understand why TIOBE keeps being brought up as an authorative source when time and time again people demonstrate its inaccuracy. Could it be just because there's no serious alternative?

Has anyone considered a more serious method of comparing programming language popularity?

As the guy behind http://langpop.com I wonder too. I don't think my results are perfect, but I think they're a bit better.

In any case though, no survey of this type is going to be entirely 'accurate'. To be 'accurate', you'd have to follow around millions of programmers for a few years and see what they're reading, writing and talking about, at home, at work, and with friends. So I do the best with the data I can get.

Another thing to consider is what 'popular' means. You could consider both 'velocity' (current use) and 'acceleration' (increase in usage). Cobol for instance, is widely used in some environments, but has no acceleration. Java has a lot of the former, and still has quite a bit of use in new projects. Something like Ruby is not widely deployed, but more popular for new web projects.

I think your answer is right, which leads to the rather shocking conclusion that something that is without alternative can never be useless or braindead enough to be ignored. It's like a piece of dog shit in the snow.

TIOBE is not particularly accurate, but that doesn't mean that counting Github projects is better.

For example, I program in SUN's Forte's 4GL TOOL programming language (sadly, our legacy system is built on that), it's a dead language, but it probably counts for the other 4GL searches or results.

And I doubt you'll find it on Github.

Same for the local Genexus 4GL .

Edit: it's also English-biased. Genexus 4GL returns over 77.000 results, much more than listed. This blog post seems worse researched than TIOBE's index.

Genexus the Uruguayan Genexus? I've never thought this will be ever mentioned in HN :D

Yep, that one. It was actually a keynote by Nicolás Jodal (co-CEO of Artech) that directed me here !!!

I wonder why it doesn't gain traction in the US, but I digress..

It's not a tool that would be liked by the HN community, as hackers like flexibility and under-the-hood access.

Github is still hugely biased towards Ruby, which is one reason I still haven't included it in http://www.langpop.com

How is it any different from Google Code (which you've included) which is heavily biased towards C?

Google's Code search is heavily biased because there's a crapload of C code out there, which makes sense, given that it's been around for nearly 40 years and is used as the basis for... well, pretty much everything, including many other programming languages.

Github's bias is based on the early adopter Ruby guys climbing on board quickly.

Some languages have a community developed site for open source and popular mailing lists or forums for questions. So GitHub and StackExchange by themselves are a poor measure of language popularity.

I've commented on here and other forums about the absurdity of Tiobe being taken seriously. He changed the rules dropping one language nearly forty places. I emailed him pointing out that following his new rules several other languages would also drop. He replied to the effect it was his site and he would do as he pleased.

Surely a site can be created that uses somewhat objective measures based on jobs available, blog posts and tweet mentions etc to arrive at a more accurate composite score.

If a language in Tiobe drops from, say, 25 to 65 in a single month when the number of search engines used increases to 7, then Tiobe's being gamed. You'd think language implementers would have better things to do than create webpages with circuits of links.

How about this: Any language popularity index that doesn't include COBOL is a joke.

This is an interesting requirement because COBOL is used everywhere the kinds of people who make these indexes aren't looking. Finding ways to include it in a reasonable fashion will likely uncover uses of other languages that really are obscure, and that would be interesting.

Exactly. There are COBOL jobs that go unfilled every year. The local college started teaching it again (I think using COBOL.net of all things) because the COBOL programmers are all retiring.

Seems to me TIOBE is a good tool to show your boss to convince him/her that it's ok to use Ruby b/c it's gaining in popularity. (I wouldn't do that of course, I would instead state that obviously Python is more popular and we need to use that :-) )

Registration is open for Startup School 2019. Classes start July 22nd.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact