Fortran-lang is now second in Google

Here is how it looks like for me in private browsing:

My screen is not big enough to show the full link, but you can see that we are second in Google now! This means we are on the right track.

Now if you see the questions that “people also ask”, you can see that we still have a lot of work to do. I tried searching for other languages and Google does not show this box about what people also ask.


That’s a great achievement!

But don’t forget you can only say that “we are second in the English Google”. A “Fortran” search in the French google does not show, not even in the first results pages… Same situation as two months ago: Promoting Fortran in non-English languages

A question: is it possible to appear at the first position? or will Google always show the Wikipedia page first? In my Google, is appearing first, and Wikipedia second. Hence it seems possible. But Wikipedia will always be either first result or at the top right of the page in a framework. So Reviewing all "Fortran" Wikipedia pages can also be important for Fortran.

Another question: how are the “People also ask” questions chosen? By an algorithm (or a human being)? Have we an idea of the criteria it uses? We should search information about that tool.
Interestingly, and happily, in the French Google the questions are much more positive (translated with DeepL):

Other questions asked:
How to write a program in Fortran ?
How to install Fortran?
How to compile and run a Fortran program under Linux?

I really don’t know why there is such a difference!


Thank you for your many contributions to the Fortran community, both current and prospective. I reread the fortran-lang landing page and have a few suggestions:

(1) It’s very hard for a language without a free implementation to remain viable. I think g95 and gfortran have been important to keeping Fortran viable. Could you change “Mature and battle-tested compilers and libraries” to “Mature and battle-tested compilers and libraries, both open source and commercial”?

(2) Many of the terms in the Natively Parallel section could benefit from links, such as Coarrays, teams, events, and collective subroutines. But maybe you want to keep the landing page uncluttered?

(3) Since Fortran is ranked 30th or lower in popularity in programming languages, it is important that it :“plays well” with many more popular languages, such as C, Python, Matlab, and R. Could you add a section called “Embeddable” discussing this?


(4) Unlike some other popular languages, Fortran is standardized. Does that merit a paragraph? Standardization encourages the existence of multiple implementations on many platforms and increases the long-term viability of code. Fortran is not going to lose the PRINT statement (and invalidate tens of millions of lines of code) even though it is redundant with WRITE.


Many people try to understand how the “People Also Ask” section works:

In my opinion yes, because for people used to program in non-standardized languages, it has a lot of consequences. There is no Fortran 3.2 or Fortran 3.2.5, like Python. That’s very different. The compilers take years to implement a new standard. It can be very weird for beginners… And there is so many compilers…

Maybe a trojan horse :carousel_horse: would be to create a FAQ page on the site (there is a FAQ on the homepage but with only three questions):

How to get your content featured in PAAs
According to Search Engine Land, there are four ways to get your law firm content featured in the answer box or PAA:
* Create complete questions and answers.
* Use plain language
* Avoid any sales pitches.
* Add Q&A schema
In other words, you need to develop content that answers popular questions about your topic or service in short, plain language or as a bullet list. In addition, Google has heavily hinted about the importance of structured data (schema) — which are tags added to a website’s HTML to improve the way search engines read and display a page in SERPs — by basing many of its new search features on structured data. As the space for organic search results continues to shrink, it will become increasingly important for law firm marketers to develop content that answers a search question as quickly and clearly as possible. If you don’t, someone else will reap the benefits.

Source: “People Also Ask” is Another Way to Dominate Search - The Rainmaker Institute

I agree and this would be an easy PR. Do you mind submitting it?

I agree, though we should only link once we have quality tutorials for these features on our own Learn page. We shouldn’t link off-site.

This would be a good addition. Perhaps even say something like “If your programming language can interface C, it can interface Fortran.”

I think we could address this in the first question of the FAQ.


Two other interesting links about “People also ask” in Google:

The second one is giving some clues about the functioning of that feature. As far as I understand, the engine is collecting questions/answers on the web and putting them in a Knowledge Base. Knowledge graphs and deep learning are then used to provide answers in PAA.

A page with Questions and Answer in a clear format can hence be collected by the search engine and enrich its knowledge base. I don’t know if questions asked in a Discourse can be collected.

Note also that I read somewhere that PAA questions could tend to be redundant: that’s clearly the case in the English Google. The four questions are very similar: if it was only one question it would not be a problem. One could say that the important thing is that the answers bring the good message, but repeating four times the same message has clearly a negative impact.

1 Like

Here are the rankings that I get:

  • Google: #2
  • Bing: #1
  • DuckDuckGo: #1
  • SearchEncrypt: #1

Almost there. :slight_smile:


From France:

  • Bing: #2 (or #1 after the French Wikipedia page in a frame),
  • DuckDuckGo: #2 after the French Wikipedia page,
  • SearchEncrypt: #2 after the French Wikipedia page (if I don’t count an Amazon ad at first place).

Still non existing in the 10 first pages of Google… where the French and English Wikipedia pages are #1. Google seems to be the most localized search engine: the first site in English (except Wikipedia) appears only on the third page and is the GNU Fortran page. Then I can see Intel Fortran Compiler, then on the 5th page,,,, etc.
I really don’t understand why does not appear even in the 20 first pages of Google… Most of the other cited sites have probably no pages in French but they finally appear. It’s very mysterious…

It could be interesting to have those rankings from other countries on our little planet to see how search engines, and especially Google, are functioning.


Localized Google from Germany gives me as #52, most notable English Fortran sides on the way are:


Google is very problematic, and unhappily it is the dominant engine. Even when modifying its settings to show both national and English languages results, there is not much difference.
I remember a (far away) time when it was possible to use the English Google in other countries, using the url, but it seems not possible anymore. You are automatically localized and served with local “food”…

With Bing, it seems easier to break out the language bubble.

Tell us the Google’s behavior in your country.

1 Like

Nice, indeed, it is first now in duckduckgo for me also. Impressive.

@vmagnin, @awvwgk if you are willing to help me port to, then we can put a banner at the top of each page there and redirect people to fortran-lang, and thus we get up in your results, as I can see is doing better there. :slight_smile:

In terms of priorities, I personally believe the goal (for me) is to deliver in the English / U.S. part of the web first (and we are well on the way!) and only later worry about the rest of the world. I personally do not want to spread too thin. However, if you feel strongly about helping in France or Germany, then let’s finish Draft: Internationalization for fortran-lang by awvwgk · Pull Request #201 · fortran-lang/ · GitHub. I can help with testing, I just don’t have time right now to implement it myself.

I follow now the same approach with LFortran, in the last couple months I have been laser focused on simply compiling enough of Fortran, you can follow our progress here, we are making very good progress:

Everything else that I really want will follow once we deliver on the core: interactive usage, automatic interfaces and translations to other languages, good formatting capabilities, language server for VSCode, custom very compilation backends (x86, arm, …), good debugging / strict / pedantic modes, etc. People will help with all these (or I can do it later), once we deliver on the core, which is just “boring” compilation of people’s projects.

In the same way, we will eventually deliver in all countries with fortran-lang. It just does not happen over night, and I personally believe we have to first deliver in English/U.S., and if we do, the rest will follow.

1 Like

What is your plan to port to I of course agree with you that the most important thing is always to work on the project. And good work will imply a better visibility.

Unhappily, in our countries even is in pages where few people will look… Only the Wikipedia pages seem able to help find outside English speaking countries, at least in Europe.

Personally, I prefer the 2021 Fortran to the 1998 Fortran, but I prefer the 1998 Google to the current one, for many reasons. And here we have two more reasons to be unsatisfied:

  • The English Google IA was fed with all the prejudices one can have about Fortran, and that’s a well known problem with IA,
  • and outside of English speaking countries it has become more difficult to find the interesting technical Fortran webpages in English.

Prejudices and bubbles… Hum, it really sounds characteristic of the present time… This is worrying.

1 Like

My plan is to simply port all the content into tutorials / guides at fortran-lang, and then just put a banner at each page (that we port) to the new content at fortran-lang. I just don’t have time at the very moment, but I will get to this after I deliver on LFortran. But if anyone goes ahead with porting it I will help.

Yes, I agree with your analysis, and I think the best way to fight prejudice and bubbles is with good solid work, as we have been doing. In the long run, that always wins, and I don’t worry about the short run, as I personally can’t do anything about how search engines rank things. I do agree with some of the older Google advises not to worry about how they rank things and “simply” provide good content.

1 Like

@certik @awvwgk
I have seen that the source of is ReStructuredText. A first step could be to put Markdown files into the Pandoc seems to do a good job translating .rst into .md. We would just have to care with a few characters or local links (but I will look at pandoc options to see if those little problems can be fixed automatically).

Is your objective to restructure the whole text, which seems to be mainly contained into four files, into smaller tutorials?

1 Like

Any way that can be approved by the community and merged. I think these two pages can be ported to Markdown and approved relatively easily, as I think they don’t have many things that would be controversial:

The FAQ might be more tricky, as it is a little more opinionated:

So we might just do it part by part and gather feedback on each question / answer.

Finally, the gotchas I think will go through, but I got some feedback regarding the wording of the floating point constants recommendations:

Well, we didn’t really move forward on this matter since I declared failure on the linked patch. I believe we arrived at the conclusion that we would have to rework from ground up to even support translations. It’s a lot of work.

I’m still willing to invest my time into this project, because I believe there is so much good to gain from opening up to the rest of the world and I got just a little glimpse of it when working on this front. Still, I won’t be able to pull it off alone by myself.


OK, I will work on porting them to Markdown using Pandoc, as perfectly as possible.

I guess we should now rather discuss there: Transfer content · Issue #79 · fortran-lang/ · GitHub