TIOBE is as mindless as counting search results for [programming language] + "programming" in Google and Q&A sites. This is the cheesiest method of assessing the popularity of a progrmaming language. Perhaps Pascal is low in ranking since it is a rather consistent and logical language, so that less questions arise compared to the more "popular" languages.
Even than, when I search for something I either search for "lazarus [question]", "fpc [question]" and "delphi [question]". None of them would count into tiobe because I don't add the term "programming" to it.
The only thing you can somehow tell from tiobe is how many new people are interested in a certain language, as it only evaluates the most general search term possible. E.g. when I try to find out what for exampel scala is, I search for "scala programming language" or when I want to learn it I search for: "Scala programming tutorial" or something like this.
People who are already using the language usually don't search for this, and from what I have read, TIOBE does not intend to capture anything else, so ranking of newcommers is more or less the goal.
Combine this with the fact that people often search not for "objectpascal programming language" but rather "lazarus ide" or something like this.
TIOBE is therefore inherintly biased against programming languages, where the search term does not include the name of the language (like for lazarus). It also heavily favors things where you basically have to add "programming" to get any meaningful results. E.g. searching for "go" in your favourite search engine will, if you are not already a programmer) most likely not grant you any information about the programming language.
That said, there is of cause also the way people start using a language. As I said TIOBE only measures how relevant a language is for newcommers, not according to specific questions. If you now start learning Pascal at your Uni or School (where it is still quite popular, at least here in Germany), you will more or less directly start with very specific questions, because you have a set of tasks to solve. There is no need for a general overview, because you are already told what to use and do.
Certainly Object Pascal today does not have the popularity of 20 years ago. However in Tiobe index it was always in the top 20 until last month. Three months ago he was in position #13. Is it credible that a language loses 10 positions in 3 months? Is it realistic that Scratch is more relevant than Delphi, Scala, Rust, Kotlin, Cobol, Fortran and Lisp? After 35 years in the IT industry I can say that this is completely wrong. I'm sorry, but Tiobe is losing credibility to my eyes.
I don't think TIOBE is nessecarily "bad", but you need to understand what is measured. First it does not measure the current state, but the newcomers. So it may give a hint how the state will be 5 years or so from now, but even thats a strech. It can't be used to make any statements about current popularity of a language. And that Delphi or Lazarus is not the hot topic everyone who wants to get into programming searches for is rather uncontroversial. On the other hand, nearly everone who wants to get into programming knows at least something about python.
And even if you are looking at newcomers, it also has a lot of caveats. If you want to assess the popularity of a language, there is a very simple way, look at the download figures. The most current version of Lazarus for windows has on sourceforge currently >2k downloads per week, a few years ago this was only around 1.5k (I don't exactly know when this was, but I remember times downloading it when the number there war around 1.5-1.6k). Lazarus is getting more popular.
Also measuring the popularity is not an easy task in it of itself. For example you could take a look at languages used on Github:
https://githut.info/ But this only measures open-source projects that use Github which is therefore heavily biased towards highly accessible general purpose languages (which is why for example C is below C++) and Pascal doesn't even make it into the top 50. Evaluating search terms is also not easy, for example when searching for go the term usually used by the community to avoid confusion is "golang". So any Index trying to capture the current usage by search terms, needs an indepth understanding of the terms the community uses. Going by what buisnesses use, is as missleading as the Github sources, just in the opposite direction, against open-source projects.
After all, all measures have their pros and cons and using one of them is like using any tool, if you don't understand what it does, its useless.
Weak argument. So GNU C++, Visual C++, C++ Builder, Watcom C++ and Think/Symantec C++ have separate wiki pages and still they all count under one C++ Tiobe index.
[...]
Let's just look into Think/Symantec C++ at https://en.wikipedia.org/wiki/THINK_C. As you can see it has originally extended C before it implemented a subset of C++. Do you think that they didn't call it C++ just because at first it supported only single inheritance? Of course they called it C++. And they had the right to call it. What strikes me is the fact that you have much better chances to compile the same visual application under Delphi and Lazarus/FreePascal then to compile some application with GNU and Microsoft C++ compiler, and yet both C++ compilers are in the single Tiobe index while that is not the case with Object Pascal and Delphi.
I have only one word for Tiobe's consistency in separation of Pascal, Object Pascal and Delphi - and that is discrimination.
Bad comparison, C and C++ are standardized. If you have standard C++ it will compile on VC++, G++, Intel C++, etc. and run the same. Sure everyone can call his C++ dialect C++, but there is the term ISO-C++ which can only referr to standard compliant C++. Vc++, G++, Intel C++, etc. are all ISO-compliant up to C++17. There is a standardized Pascal dialect, which no one uses.
On the other hand is the Object-Pascal dialect the FPC implements and Delphi incompatible. Look no further than function pointers. While I think that splitting those two indices up with the reasoning of wikipedia articles is kinda stupid, and after all the differences are rather minor, these are still two different dialect, which is not true for two ISO-C++ compilers, which implement the same language and can only differ in parts where the standard is undefined.
A better comparison there would be Lisp, as every software using Lisp (such as emacs) uses it's own dialect (as Lisp for itself is just S-Expressions). As TIOBE simply has one category LISP, even though LISP is only a family of language dialects, I think that the only consistent way would be to also have one entry for ObjectPascal (or even Pascal as the more general family)