Ask HN: Why are programming lang ranks on StackOverflow and Tiobe so divergent?

[satishgupta]

Most popular languages according to StackOverflow 2020 Survey: JavaScript, and C is way down below.

https://insights.stackoverflow.com/survey/2020#technology-programming-scripting-and-markup-languages-all-respondents

While in TIOBE index, which is based on search engine traffic, C just overtook Java. These two have been the top two languages since 2002.

https://www.tiobe.com/tiobe-index/

How that can be explained?

Note that I am asking which language is better. I am trying to understand why these two are so different. Because how what programmers tell in a survey be so different from what they google?

[davidhbolton]

Another excellent source is language subreddit subscriber numbers. Go to https://www.reddit.com/r/programming/wiki/faq#wiki_what_lang...

Mind you the numbers for each language get out of date very quickly. Visit the actual subreddit to see the true numbers.

As far as I'm concerned Tiobe is at odds with everybody else and should be regarded as junk.

[rskar]

TIOBE tracks the trends (or at least hopes to) of people looking to learn/re-learn/familiarize/re-familiarize themselves about whatever programming languages for whatever reasons. That's certainly the most one can really glean by the stats on searches of +"<language> programming". People who are "googling" so are not necessarily just starting out in the field, or learning in school, or researching to decide what language to use in some greenfield project. What's also likely is they are seeking employment someplace or recently got work on somebody's (perhaps legacy) software; i.e. they're searching for immediate professional interest.

StackOverflow attracts a population of programmers interested in regularly helping out other fellow programmers navigate through whatever arcana of tools and libraries and frameworks and languages etc. It will mostly be this set that takes the survey. It also seems that it is the web/mobile technologists that predominate, which is understandable considering how big a deal web/mobile is. There seem to be so many new&shiny things in that web/mobile sphere, year after year ("Web 2.0" is so early 2000's), and that's certain to keep the arcana wheel going.

Fun factoid: Stack Overflow was founded by Microsoft alumnus Joel Spolsky and developer/blogger Jeff Atwood. Spolsky, while Program Manager on the Excel team, designed Excel Basic and drove Microsoft's Visual Basic for Applications strategy. Atwood says: "I was weaned as a software developer on ... Microsoft BASIC in the 80's... I continued on the PC with Visual Basic 3.0 and Windows 3.1 in the early 90's... I am now quite comfortable in VB.NET or C#, despite the evils of case sensitivity." (https://blog.codinghorror.com/about-me/). Spolsky still seems proud of Visual Basic: "I am always saying 'I could do that in a weekend in Visual Basic' when developers tell me some feature is going to take a year." (https://www.joelonsoftware.com/2018/04/06/the-stack-overflow...). Visual Basic has long been the "most dreaded" language on Stack Overflow surveys, and in 2019 explicitly got dropped from that year's survey. The very fact that VB.NET manages to be in the top 10 of TIOBE, and VB clinging to the top 20 of TIOBE, has often been given as immediate evidence of why TIOBE should not be taken seriously.

[matt_s]

StackOverflow tends to attract new developers and newer language frameworks simply because it is a Q&A site.

There are scores of developers out there that don't need beginner level help with their language or use a proprietary language where the help they get is via professional services training, courses, etc. Think whatever SAS or Oracle ERP Apps use under the hood, as an example.

These surveys only purpose is marketing/research for the owner of the survey.

[pmontra]

One is a survey in which people have to participate actively (so there is a bias in favor of the languages with more vocal users) and the other one is a sample of an indicator (of popularity? of complexity? other?)

I'm not surprised that they give different results. You can average the two rankings and get something maybe closer to the real popularity.

You can also rank languages by number of repositories on github and get a ranking with a different bias.

[satishgupta → pmontra]

That is a good point.

I looked at GitHub: https://madnight.github.io/githut/#/pull_requests/2020/2

And that matches with survey. How that can be reconciled? What are the biases here?

[pmontra → satishgupta]

I might be wrong but there are only public repositories in that graph. Closed source private projects are not there. The open source libraries they use do.

[s1t5]

The Tiobe ranking looks just plain wrong. It would be interesting to know how exactly they're calculating and interpreting search engine traffic.

[Someone → s1t5]

They don’t. They use search engines to do searches and count hits. See https://www.tiobe.com/tiobe-index/programming-languages-defi...

Their method means that, in general, their numbers will trail changes in programming language popularity. For example, if everybody stopped using JavaScript tomorrow, there still would be millions of hits on +”JavaScript programming” for years to come.

Also, their methodology overrates languages many people wrote about or those that trigger more questions online. I would guess it underrates COBOL, for example.