Loosely interpreting developer surveys

2017 Mar 22

Stack Overflow, the single most useful site to software developers out there, has published the results of its 2017 Developer Survey, which you can find here. In it, they also offer some loose interpretation of the results - as they do every year - as they might pertain to popular urban legends, misconceptions about our trade, and so forth.

We can’t be no exception.


72.6% of developers taking the survey identify as a “Web developer” in at least one part of their job(s). Which begs the question, if three quarters of developers on SO are engaged in some way with the web, does it make sense to consider all of them a “web developer” anyway? On the other side of the coin, of course, why should we assume developers are uniformly represented on SO? Perhaps developers who don’t deal with the web are somehow at a position - of luxury, perhaps - to not use SO and not having to learn from it. More on that later.

Years since learning to code

An impressive ~1/5th of developers on SO learned to code 20 or more years ago. About 3/5ths of them learned to code some time in the last 10 years.

Years coding professionally

An even more impressive half of developers on SO are coding professionally for less than 6 years. Newbies outnumber the gurus on SO.


Only 11.4% of developers taking the survey identify as something other than male. Only 25.6% don’t identify as white/european. If only we could ask developers who aren’t using SO why they steer clear.

Educational Attainment

More than three quarters of developers taking the survey have a bachelor’s degree or higher, and yet most of them are newbies (see above), which means they’re coming to SO to learn. I think that speaks volumes of the kind of education we’re receiving.

Importance of education

Is it me or does that look like a normal distribution? You could even say the answers to this question don’t matter at all.

Types of education

90% of developers taking the survey say they’re self-taught. This only piles up on the other questions: our education clearly isn’t what it’s made out to be.

How devs teach themselves

80% of developers on SO say they’re teaching themselves via SO. The only thing that makes me wonder is what are the other 20% being taught from. Documentation? P-uh-lease.

Program as a hobby

75% of respondents program as a hobby, too. It’d be interesting to know why they do so - do they feel like they have things to offer which their jobs can’t or don’t allow them to? Do they write software outside work for legal or other reasons? To better themselves?

Javascript. Basically Javascript developers are on SO, which speaks volumes about JS and its ecosystem. It could basically be sketched as a job requirement: create and learn to use StackOverflow if you want to be successful at this job.

Loved, Dreaded, Wanted

Erlang is dreaded by 60% of the developers taking the survey. In the same ballpark are C, Objective-C, Common Lisp, and R. 45% for Haskell. That’s just baffling.

Loved, Dreaded, Wanted

“Linux Desktop” is most loved, by 70% of respondents. Stockholm syndrome, probably. Why else would you love a reason that drives you to SO. And don’t tell me that’s not true because we’ve already seen how devs identify and what they do in their day jobs..

Developer environments

35% of respondents consider Notepad++ a “developer environment”. Maybe that should be a hint to those trying to quantify productivity.

Top payers

Javascript, the prime reason respondents are on SO in the first place, is among the worst paying jobs.

Jobs in the latest and greatest pay more. Notable exception: Perl (no version distinction).


Only 1 in 3 developers taking the survey work for a software house per se.

They are generally satisfied with their equipment (in general or the eq they are using? I forgot).

Job discovery

Word through the grapevine is still the most popular way to get informed about jobs. Obviously, who the fuck would use LinkedIn?

About 5-6% of respondents spend more than 14 hours a day on job searches. I wonder if they’re looking for a job or they’re posting them.


Everything goes towards the mean (3 on the 1-5 weird scale). To me that just points out a surveyed census which either doesn’t care about the scale, or isn’t actually representative of anything. Bear in mind the vast majority identify as male. This is a section in which generalisations don’t really work, and the full data would be much more illuminating.

Regarding compensation, respondents from countries with extremely low salary (relative to US) are more interested in compensation than benefits. Healthy balance near the middle of the spectrum, and the US just massively overvalue benefits. Probably because they don’t have any, the poor sods.


About half of respondents feel like they’re underpaid. Bear in mind most of them are newbies, as previously established.

Students and starters expect to be paid starting from around $33k. Not very unreasonable for places like Germany, UK, or the US, I’d imagine (I’ve almost zero knowledge of US cost of living). Very unreasonable for most others, especially the EU startup incubators.


Most developers would like to be judged on being on time/budget and customer satisfaction. The select-all-that-apply nature of this question is just confusing the results imo, making all conclusions invalid.

Most developers feel employers should prioritise communication skills when recruiting. Could it be that they feel they would perform better if they could communicate better? Who knows.


More respondents than not advocate for tabs. In Javascript tabs are wanted because of the implication that they reduce file size. I’m not saying the two are related, but they might be.


Reinforce the conclusion that most survey respondents are there because SO solved their (Javascript) problem(s), multiple times a month. To put it succinctly, Javascript is the major reason the rest of us we have such a good resource like StackOverflow.

More interesting stuff to follow once we can mine the full results.

« Past Future »