Please Stop Citing Tiobe | Hacker News Reader

Please Stop Citing Tiobe

blog.nindalf.com

54 pointsnindalf17 comments

Hey everyone, I noticed several times over the years people (mis)using TIOBE to support whatever their argument was. Each time someone in the thread would explain the various shortcomings of TIOBE and why we shouldn't use it.

I decided to write up the issues so we could just point them towards this link instead.

Thanks for writing and sharing this

I do not doubt your central claim (that the ranking is derived from merely counting search results, as lampooned by the xkcd example), but I'm curious how are you sure of this, or where is it documented?

The details of the calculations are in https://www.tiobe.com/tiobe-index/programminglanguages_defin...

oh my that is daft (thanks)

Use RedMonk instead: https://redmonk.com/sogrady/2022/03/28/language-rankings-1-2...

The inclusion of Delphi, a basically dead language, always made it obvious that Tiobe was terribly inaccurate.

I actually checked indeed for jobs and found around three, nationwide.

There are tons of references to Delphi on google and indeed, but they are almost all for other companies named Delphi, like the auto parts maker. Hardly any for the language.

Tiobe isn't even trying.

Redmonk doesn't mention Delphi!

Probably listed under "Pascal" which it does feature.

TIOBE was manipulated by the Delphi community - http://delphi.org/2008/10/delphi-keeps-climbing/

That’s why it’s so high even today.

It's more that TIOBE is hugely controversial in the Delphi community because they won't count Delphi. There have been multiple campaigns to get the language represented "accurately" (for an inaccurate index.) No matter how inaccurate TIOBE is, if it doesn't add its own measurements which are applicable to a specific language, even by comparison with its own data it becomes inaccurate.

Not sure what beef the TIOBE people have against Delphi. But what you saw 14 years ago was community action: these days it's moved to community protest... or, like this article, just ignoring TIOBE altogether.

What do you mean "won't count Delphi"? It's obviously severely over counting Delphi.

A quick check of indeed shows one job with Delphi as a nice to have within 100 miles of San Jose. For python, 11000. Ruby, 559.

"Delphi programming" on google, the string that the delphi has been trying to push, gets 127,000 hits on google, "python programming" gets 17,100,000

In computer science, Amazon shows 600 delphi books, with the top book having a sales rank (lower is better) of 1,600,000. Python, too many to count (says over 10000) top sales rank 1000. Ruby, just below Delphi in Tiobe, over 10000, sales rank 800,000

So, basically: no google hits, no book sales and no jobs. How is that number 16?

Just delusional. They justified their manipulation of a useless index by convincing themselves they were just preventing undercounting. It’s BS.

What’s worse is that the language clearly failed to provide any compelling reason to use it and died a slow death. This manipulation was evidently their best attempt at showing the language was worthwhile.

https://madnight.github.io/githut/#/pull_requests/2022/1

No Delphi in top 50.

I wish https://www.openhub.net was still maintained. I like their project stats, and they have overall language comparison by number of contributors:

https://www.openhub.net/languages/compare?measure=contributo... (except the data point for an incomplete month is garbage…)

Whatever, management will keep looking into it when doing hiring decisions.

Maybe Java and C aren't showing up in search results because everyone already knows how to do it.

Don’t confuse search results (the number near the search bar) with Google trends (the number of people searching).

Bizarrely, TIOBE uses the first one.

Reply on news.ycombinator.com