Largely closed, proprietary services. Not saying it wouldn’t be broadly accurate, but it’s a real shame that the open source options are not being represented.
Other search engines. All of the VCSs that are hosted not by Microsoft. Matrix/IRC/XMPP rooms. Discourse & other language forums could be polled for activity.
There’s a host of FOSS-forward project on GNU Savannah, Codeberg, SourceHut, et al. or self-hosted for wanting to own one’s data. A lot of those projects also do not want to be locked into proprietary communications. A lot of the work is volunteer work or sponsored by grants which don’t get corporate job offerings. This also would lend some to think that a language/project is not legitimate because they aren’t listed on proprietary platforms which could cause pressure to mirror or move to these platforms if they want to be ’counted’–which would be awful considering DVCS doesn’t need to be hosted anywhere in particular so why do we need to give corporations more sway/control?
No as I’d said, but with sort of behavior starts a cycle that pushes folks to think The Microsoft GitHub is the only legitimate place to host code, etc.
That said, ever since 2016 slowly we’ve seen multiple projects such as KDE, GNOME, et al. move to self-hosted options. There’s been a good uptick in Codeberg/SourceHut usage. More communities run a Mastodon account now, and IRC is still popular. Including them in the results helps legitimize these as a part of ones developer code/community suite. To start overtaking how entrenched the proprietary tooling is, the alternatives need to be added to the conversation.
I don’t get the “Jobs” methodology[1], it seems very flawed. It says “Shell” is 3 times higher than “Go”. There are almost no “Shell” only jobs, likely it’s just something that they put in the job description for a position in another language. Likewise for SQL, which is the most popular one in the Jobs section, and it’s almost always a secondary language in most companies: take any company and the number of DBAs will be smaller than the number of application programmers.
I don’t get the “Jobs” methodology[1], it seems very flawed. It says “Shell” is 3 times higher than “Go”. There are almost no “Shell” only jobs, likely it’s just something that they put in the job description for a position in another language
That doesn’t make it wrong. If there are 100 jobs that require shell scripting and some other programming language, ten that require Go and JavaScript, and five that require just Go, knowing shell scripting seems more valuable. If you are developing or deploying on any *NIX platform, shell scripting is an essential skill.
It isn’t the only skill that will you need to get the job, but these days most software is written in multiple languages and so knowing only Go (or C++ or Python) isn’t likely to get you a job either.
That’s the kind of thing that makes sense statistically, but does not pass the smell test in reality. The issue is that the way they collected the stats does not rank these languages by importance when they appear together. Maybe they couldn’t, it’s a hard thing to glean from text, but nevertheless it produces an incorrect conclusion. In almost any hiring situation shell scripting is the last thing a candidate will be differentiated by (aside from specific industries).
For instance, in those 100 jobs knowing shell scripting alone would not get a candidate hired, but knowing Go alone would. Which is pretty much what happens in reality: if I meet someone with 5 years Go experience I am extremely confident that person can learn enough shell scripting in a week to help out with the few shell scripts I have, so their value is higher than someone who knows zero Go.
The same goes for other languages as well, simply because shell scripting is not the basis of most companies’ products. In a more general way, if you rank job requirements by importance, you get a better view of what the market actually values.
The issue is that the way they collected the stats does not rank these languages by importance when they appear together
That’s probably hard to do, but I’m also not convinced that it makes much difference overall. It may cause a few things like shell scripting to be ranked a bit higher than they should be but it highlights that a good developer needs to know lots of different languages competently and a small number very well.
In almost any hiring situation Shell scripting is the last thing a candidate will be differentiated by.
Sure. Given an expert C++ programmer who doesn’t know any shell scripting and a moderately competent C++ programmer who does, you’ll hire the first person for a job that is primarily C++ and expect to teach them shell. But given two programmers who seem equally competent in C++, one of whom is comfortable with shell scripting and the other who isn’t, you’ll pick the one who is. Very often, jobs require a handful of different skills and when you’ve got a bunch of candidates at a similar level, the differentiation often comes on the secondary ones. Shell scripting won’t be near the top of the list for any job, but it will be somewhere there for a lot of jobs.
Python is probably ranked a bit too high for the same reason: it’s used for a lot of AI stuff so is very buzzwordy at the moment, but it’s also used as glue in a huge number of places, so it’s the second or third language that you’d need.
I think you’re valuing the importance incorrectly.
Jobs I’ve had that never required shell scripting: 0
Jobs that required more than one person in the team to have more than absolutely basic knowledge: also 0
For Go, you usually end up proper “software”, the only case I can imagine if you’re in Ops and only ever tweak some monitoring checks, but that’s stretching it.
There’s never enough conversation about the non-language skills required. Eg “language x devs get £500k, I’ll learn that” - full picture, they get paid that not for knowledge of the language but because they have three phds and have walked on the moon. The lack of mention of full requirements makes the figures far less meaningful.
Since none of my projects are written in Python or Java and I haven’t touched C++ for years, I find the creation (and purpose) of these “top lists” a bit mysterious every time. How are the “top languages” determined and what do I get out of this information?
(I sincerely hope that the majority of those who want to learn programming don’t base their decision solely on the question of what is the easiest way to find a job. That would be boring).
It’s sad that according to this list there are less jobs for Rust than for example for J or Ada. I also don’t understand why TypeScript is outside top 10.
I like Ada or other languages that I know above a Rust. In my opinion its sad because it means that Rust probably will never have adoption like C++/Java. Less jobs => less people willing to learn, less people willing to learn => less companies will adopt a too/language that they will have problems with hiring people. What I really like in Rust is that it’s really difficult to shot yourself in the foot and performance is pretty good. I believe this one deserve a broader adoption where C++ is still used.
Python and SQL are old languages themselves?
How we construct the rankings:
Largely closed, proprietary services. Not saying it wouldn’t be broadly accurate, but it’s a real shame that the open source options are not being represented.
What are those open source options?
Other search engines. All of the VCSs that are hosted not by Microsoft. Matrix/IRC/XMPP rooms. Discourse & other language forums could be polled for activity.
Why would that be better?
Incidentally, most other search engines are also closed source.
There’s a host of FOSS-forward project on GNU Savannah, Codeberg, SourceHut, et al. or self-hosted for wanting to own one’s data. A lot of those projects also do not want to be locked into proprietary communications. A lot of the work is volunteer work or sponsored by grants which don’t get corporate job offerings. This also would lend some to think that a language/project is not legitimate because they aren’t listed on proprietary platforms which could cause pressure to mirror or move to these platforms if they want to be ’counted’–which would be awful considering DVCS doesn’t need to be hosted anywhere in particular so why do we need to give corporations more sway/control?
Do you think the inclusions of these would have measurably altered the results?
No as I’d said, but with sort of behavior starts a cycle that pushes folks to think The Microsoft GitHub is the only legitimate place to host code, etc.
That said, ever since 2016 slowly we’ve seen multiple projects such as KDE, GNOME, et al. move to self-hosted options. There’s been a good uptick in Codeberg/SourceHut usage. More communities run a Mastodon account now, and IRC is still popular. Including them in the results helps legitimize these as a part of ones developer code/community suite. To start overtaking how entrenched the proprietary tooling is, the alternatives need to be added to the conversation.
I don’t get the “Jobs” methodology[1], it seems very flawed. It says “Shell” is 3 times higher than “Go”. There are almost no “Shell” only jobs, likely it’s just something that they put in the job description for a position in another language. Likewise for SQL, which is the most popular one in the Jobs section, and it’s almost always a secondary language in most companies: take any company and the number of DBAs will be smaller than the number of application programmers.
1 - https://spectrum.ieee.org/top-programming-languages-methodology
That doesn’t make it wrong. If there are 100 jobs that require shell scripting and some other programming language, ten that require Go and JavaScript, and five that require just Go, knowing shell scripting seems more valuable. If you are developing or deploying on any *NIX platform, shell scripting is an essential skill.
It isn’t the only skill that will you need to get the job, but these days most software is written in multiple languages and so knowing only Go (or C++ or Python) isn’t likely to get you a job either.
That’s the kind of thing that makes sense statistically, but does not pass the smell test in reality. The issue is that the way they collected the stats does not rank these languages by importance when they appear together. Maybe they couldn’t, it’s a hard thing to glean from text, but nevertheless it produces an incorrect conclusion. In almost any hiring situation shell scripting is the last thing a candidate will be differentiated by (aside from specific industries).
For instance, in those 100 jobs knowing shell scripting alone would not get a candidate hired, but knowing Go alone would. Which is pretty much what happens in reality: if I meet someone with 5 years Go experience I am extremely confident that person can learn enough shell scripting in a week to help out with the few shell scripts I have, so their value is higher than someone who knows zero Go.
The same goes for other languages as well, simply because shell scripting is not the basis of most companies’ products. In a more general way, if you rank job requirements by importance, you get a better view of what the market actually values.
That’s probably hard to do, but I’m also not convinced that it makes much difference overall. It may cause a few things like shell scripting to be ranked a bit higher than they should be but it highlights that a good developer needs to know lots of different languages competently and a small number very well.
Sure. Given an expert C++ programmer who doesn’t know any shell scripting and a moderately competent C++ programmer who does, you’ll hire the first person for a job that is primarily C++ and expect to teach them shell. But given two programmers who seem equally competent in C++, one of whom is comfortable with shell scripting and the other who isn’t, you’ll pick the one who is. Very often, jobs require a handful of different skills and when you’ve got a bunch of candidates at a similar level, the differentiation often comes on the secondary ones. Shell scripting won’t be near the top of the list for any job, but it will be somewhere there for a lot of jobs.
Python is probably ranked a bit too high for the same reason: it’s used for a lot of AI stuff so is very buzzwordy at the moment, but it’s also used as glue in a huge number of places, so it’s the second or third language that you’d need.
I think you’re valuing the importance incorrectly.
Jobs I’ve had that never required shell scripting: 0
Jobs that required more than one person in the team to have more than absolutely basic knowledge: also 0
For Go, you usually end up proper “software”, the only case I can imagine if you’re in Ops and only ever tweak some monitoring checks, but that’s stretching it.
There’s never enough conversation about the non-language skills required. Eg “language x devs get £500k, I’ll learn that” - full picture, they get paid that not for knowledge of the language but because they have three phds and have walked on the moon. The lack of mention of full requirements makes the figures far less meaningful.
Since none of my projects are written in Python or Java and I haven’t touched C++ for years, I find the creation (and purpose) of these “top lists” a bit mysterious every time. How are the “top languages” determined and what do I get out of this information?
(I sincerely hope that the majority of those who want to learn programming don’t base their decision solely on the question of what is the easiest way to find a job. That would be boring).
In this case their methodology is described at https://spectrum.ieee.org/top-programming-languages-methodology
Thank you. :-)
It’s sad that according to this list there are less jobs for Rust than for example for J or Ada. I also don’t understand why TypeScript is outside top 10.
Why is that sad?
I like Ada or other languages that I know above a Rust. In my opinion its sad because it means that Rust probably will never have adoption like C++/Java. Less jobs => less people willing to learn, less people willing to learn => less companies will adopt a too/language that they will have problems with hiring people. What I really like in Rust is that it’s really difficult to shot yourself in the foot and performance is pretty good. I believe this one deserve a broader adoption where C++ is still used.
Have you considered that popularity is partly a function of age?
Yes, I hope that 20 years from know Rust will run on billion devices ;)