I was pondering similar thoughts. Will LLM assistants ossify our current programming languages? My limited testing seems to show LLM assistants do well the more popular the language is (more data in its training), so is the hurdle for adoption of something new going to get even higher?
In an alternate universe, if LLM only had object oriented code to train on, would anyone push programming forward in other styles?
LLMs could also ease adoption of new languages by making hiring less of a barrier to using something more niche. It becomes easier for someone to hit the ground running and be productive even if they still need to put the time in to become an expert.
Instead I find myself more concerned with which virtual machine or compiler tool chain the language operates against. Does it need to ship with a VM or does it compile to a binary? Do I want garbage collection for this project?
Maybe in that way the decision moves up an abstaction layer the same way we largely moved away from assembly languages and caring about specific processor features.
I totally expected JavaScript to get the 2nd spot but looks like TypeScript pulled the votes away. I personally consider JavaScript and TypeScript to be close enough for their numbers to be added up.
As a backend dev (mostly working in fintech) I feel weirdly unable to find a target language to move to.
After working with Node and Ruby for a while I really miss a static type system. - Typescript was limited by its option to allow non strictness.
Nothing catches my eye, as it’s either Java/.Net and its enterprisey companies or Go, which might not be old but feels like it is, by design. Rust sounds fun, but its usecases don’t align much with my background.
Do you think those languages are scratching the itch? If not maybe we need to pick a suitable underdog and champion its use. Clojure comes to mind. Or Unison. Something where immutability shines.
It's hard to find good data sources for this, especially that StackOverflow is in decline[1].
IEEE's methodology[2] is sensible given what's possible, but the data sources are all flawed in some ways (that don't necessarily cancel each other out). The number of search results reported by Google is the most volatile indirect proxy signal. Search results include everything mentioning the query, without promising it being a fair representation of 2025. People using a language rarely refer to it literally as the "X programming language", and it's a stretch to count all publicity as a "top language" publicity.
TIOBE uses this method too, and has the audacity to display it as a popularity with two decimal places, but their historical data shows that the "popularity" of C has dropped by half over two years, and then doubled next year. Meanwhile, C didn't budge at all. This method has a +/- 50% error margin.
Yep. And the sources are too often self-reinforcing and self-referential.
Use the "right"/better tool from the toolbox, the tool you know best, and/or the tool that the customer wants and/or makes the most money. This might include Ada[0] or COBOL[1]. Or FORTH[2] or Lua[3]. Popularity isn't a measure of much of anything apart from SEO.
Given how mediocre LLMs are, I don't see this happening anytime soon... but I think a "better" LLM (that puts the "language" into large language model) can seamlessly translate between programming languages.
Right now, it's apparent to me that LLMs are mostly tuned in the programming space for what n-gate would call "webshit", but I think it is a clear (to me) evolutionary step towards getting much better "porting" ability in LLMs.
I don't think that is in the priority list of the LLM companies. But I think it would be a real economic boon: certainly there is a backlog of code/systems that needs to be "modernized" in enterprises, so there is a market.
Ultimately I wonder if an LLM can be engineered to represent code in an intermediate form that is language-independent to a large extent, and "render" it to the desired language/platform when requested.
I'd like to see some clarity in these stats, it can't just be me that finds it hard to believe that there are significantly more Python jobs than Java. I wonder if job listings are saying "Python, C++" or something, so that's a point for Python, even though, the job is < 1% Python just for test rigs or something.
The methodology involves search hits, Stackoverflow, conference and journal articles.
In all of these Python is artificially over-represented. Search hits and Stackoverflow questions represent beginners who are force fed Python in university or in expensive Python consultancy sessions. Journal articles are full of "AI" topics, which use Python.
Python is not used in any application on my machine apart from OS package managers. Python is used in web back ends, but is replaced by Go. "AI" is the only real stronghold due to inertia and marketing.
Like the TIOBE index, the results of this so called survey are meaningless and of no relevance to the jobs market.
I’m skeptical. There are more people writing PHP and Ruby than HTML? And HTML is a programming language? Those two very surprising results make me doubt the others.
Elixir behind OCaml? Possible, I guess, but I know of several large Elixir shops and I haven’t heard much of OCaml in a while.
Java at #2? Is it really still being used much for new code in this day and age? Or is its popularity mostly due to so much legacy code out there to be maintained?
I wish there were less programming languages because every library needs to be rewritten as many times as there are languages, a combinatorial waste of time.
Now that CoffeeScript is gone I would like to see all Ruby become Python.
Anything short of a big AI winter is not going to move Python from its top spot. As Python has become the first choice of output code for LLMs its dominance is only going to grow.
Arduino is a company that likes to make fundamentally incompatible things, add their branding and claim that they are the big technology and not AVR, C++ etc.
OTOH, if you have a programmer that's been only using the Arduino IDE and its libraries, they probably haven't learned any C++ beyond loops and procedures and other atoms. It's technically C++ underneath, but a tiny subset in practice.
Oh, and actual corporations and labs do use them, though. It's really a dead simple way to automate some electronics with literally off the shelf tools. Almost any alternative will be extremely niche, require invasive SDKs and probably either cost 100 times more or need shipping overseas or both. Much like Raspberry Pi that you can get in your corner store and find ready code to go while "professional" SBC's are behind crazy hurdles with outdated sw and no documentation.
74 comments
[ 4.5 ms ] story [ 92.5 ms ] threadIn an alternate universe, if LLM only had object oriented code to train on, would anyone push programming forward in other styles?
Instead I find myself more concerned with which virtual machine or compiler tool chain the language operates against. Does it need to ship with a VM or does it compile to a binary? Do I want garbage collection for this project?
Maybe in that way the decision moves up an abstaction layer the same way we largely moved away from assembly languages and caring about specific processor features.
After working with Node and Ruby for a while I really miss a static type system. - Typescript was limited by its option to allow non strictness.
Nothing catches my eye, as it’s either Java/.Net and its enterprisey companies or Go, which might not be old but feels like it is, by design. Rust sounds fun, but its usecases don’t align much with my background.
Any advice?
IEEE's methodology[2] is sensible given what's possible, but the data sources are all flawed in some ways (that don't necessarily cancel each other out). The number of search results reported by Google is the most volatile indirect proxy signal. Search results include everything mentioning the query, without promising it being a fair representation of 2025. People using a language rarely refer to it literally as the "X programming language", and it's a stretch to count all publicity as a "top language" publicity.
TIOBE uses this method too, and has the audacity to display it as a popularity with two decimal places, but their historical data shows that the "popularity" of C has dropped by half over two years, and then doubled next year. Meanwhile, C didn't budge at all. This method has a +/- 50% error margin.
[1]: https://redmonk.com/rstephens/2023/12/14/language-rankings-u... [2]: https://spectrum.ieee.org/top-programming-languages-methodol...
I like this:
https://madnight.github.io/githut/#/pull_requests/2024/1
It gives you a count of public repos on GitHub by language used, going back to 2012.
Use the "right"/better tool from the toolbox, the tool you know best, and/or the tool that the customer wants and/or makes the most money. This might include Ada[0] or COBOL[1]. Or FORTH[2] or Lua[3]. Popularity isn't a measure of much of anything apart from SEO.
0. https://www2.seas.gwu.edu/~mfeldman/ada-project-summary.html
1. https://theirstack.com/en/technology/cobol
2. https://dl.acm.org/doi/pdf/10.1145/360271.360272
3. https://www.freebsd.org/releases/12.0R/relnotes/#boot-loader
Right now, it's apparent to me that LLMs are mostly tuned in the programming space for what n-gate would call "webshit", but I think it is a clear (to me) evolutionary step towards getting much better "porting" ability in LLMs.
I don't think that is in the priority list of the LLM companies. But I think it would be a real economic boon: certainly there is a backlog of code/systems that needs to be "modernized" in enterprises, so there is a market.
Ultimately I wonder if an LLM can be engineered to represent code in an intermediate form that is language-independent to a large extent, and "render" it to the desired language/platform when requested.
In all of these Python is artificially over-represented. Search hits and Stackoverflow questions represent beginners who are force fed Python in university or in expensive Python consultancy sessions. Journal articles are full of "AI" topics, which use Python.
Python is not used in any application on my machine apart from OS package managers. Python is used in web back ends, but is replaced by Go. "AI" is the only real stronghold due to inertia and marketing.
Like the TIOBE index, the results of this so called survey are meaningless and of no relevance to the jobs market.
Elixir behind OCaml? Possible, I guess, but I know of several large Elixir shops and I haven’t heard much of OCaml in a while.
Is it something to do with frameworks like React?
Now that CoffeeScript is gone I would like to see all Ruby become Python.
OTOH, if you have a programmer that's been only using the Arduino IDE and its libraries, they probably haven't learned any C++ beyond loops and procedures and other atoms. It's technically C++ underneath, but a tiny subset in practice.
Oh, and actual corporations and labs do use them, though. It's really a dead simple way to automate some electronics with literally off the shelf tools. Almost any alternative will be extremely niche, require invasive SDKs and probably either cost 100 times more or need shipping overseas or both. Much like Raspberry Pi that you can get in your corner store and find ready code to go while "professional" SBC's are behind crazy hurdles with outdated sw and no documentation.