2016 Stack Overflow Developer Survey Results (opens in new tab)

A very fun fact indeed!

Here are some sources to go with it: http://www.slate.com/blogs/moneybox/2012/05/01/why_big_macs_...

http://latitude.blogs.nytimes.com/2011/11/24/argentinas-big-...

As a South African this hurts. We were ranked number 1 for purchasing power, and I've heard similar statistics from other sources. But the economy is shitting on us hard, and we are all struggling.

The one thing that isn't taken into account is things like tax wastage. I'd say easily 70 to 80% of our tax money is wasted by our inept government. So we have to fork out so much cash for basic things that other countries take for granted. It sort of feels like I pay tax twice. Once to the government which might as well be the money equivalent of /dev/null and another for private services which the municipality/government should be supplying, but what is offered is just too shitty to use.

So we may have the best purchasing power, but we have the least amount of money, so it doesn't help much.

This is not too surprising given their other statistical creativity [0].

[0] http://www.economist.com/node/21548242

The whole big mac index phenomenal is funny because Economics magazine didn't intent to make an index when they started it.

Did you mean to comment in the fast food automation thread?

https://news.ycombinator.com/item?id=11304628

> Might be a loaded question, but doesn't full-stack mean just "web developer"?

Huh, interesting question. I think a "web developer" could focus more on front end or back end; I usually identify as "full stack web developer" to indicate I do both.

However, if I read your question another way: are you asking if "full stack" could apply outside the web realm? Can you be a "full stack"... erm compiler developer? I know compilers generally have a "front end" and a "back end", but I'm not sure if I've heard the terminology used that way. I'd be curious to know if "full stack" is used in other niches of software development than just the web.

I don't think so. I think the self distinction is between developers who work exclusively on the front end or back end of websites vs those who do it all. My guess would be the correlation between the self identification has less to do with developer skill than to team/business size. If you are on a team of two or three people, and changes need to be made to the database, JS, and server-side code, you're going to be working across the full stack in a variety of languages. I don't think this approach lends itself to mastery or specialization, and so if you're like me you will be using SO to find out how to do simple things in JS.

But I see your point, the commentary suggests full-stack developers are more skilled.

It is a matter of self-identification. I take the phrase with a large dose of irony. If someone calls themselves a full-stack developer, I assume they mean "JS/CSS/HTML/<app language>/SQL" and this is especially likely in this case, since it's a SO poll.

But I still want some sense of awareness of the many layers on which web systems rest.

"Oh you're a full-stack developer? Great! In this job interview we will be designing a new TCP retransmission algorithm, configuring a CI/CD pipeline, debugging CPU microcode, negotiating with a law enforcement agency, and testing the FM200 system in the DC"

I think it's more of a distinction with 'front-end developer', who is someone that works exclusively on the front-end - e.g. angular, ember, react, etc. This role barely existed 5+ years ago, and every web developer was more of a full-stack developer. Nowadays you could focus only on the front-end side and not need bother with a back-end ever.

I think there are some big holes in the list of occupations.

Let's say you work in a place with a "microservices" architecture. You write in Java or Ruby or another language. Which occupation are you on this list?

Why is there a "backend web developer" but not "backend developer"?

I always assumed full-stack was backend language plus SQL as well as the web server and server OS. So, the L and A in LAMP. Keep in mind that I started in 1996 or so, and only started hearing "full-stack" in recent years.

e.g., using common options:

  - A front end web developer would be HTML/CSS.
  - A back end web developer would be HTML/CSS/MySQL/PHP (or equivalent).
  - Full-stack all of the above plus Apache and Linux.

Besides front and back end, someone who identifies with full-stack should also be proficient at the data layer. https://en.m.wikipedia.org/wiki/Multitier_architecture

My company makes recruiting software so I talk and deal with many recruiters.

I have to say I severely cringe every time a recruiter says "Rockstar" or "Ninja" (in a job description or ask if I am one). Just imagine using those terms for a talented medical doctor or lawyer. "Microservice" is quickly becoming another new recruiting term that I am beginning to despise. At least "agile" is mostly dead... mostly...

Speaking of which what ever happened to "Scotty", "Rocket scientist" or "Macgyver" (those were the sexy euphemisms back when I started software dev). They were far more apropo.

Don't forget about Android Studio at 13%

http://blog.jetbrains.com/idea/2013/05/intellij-idea-and-and...

Were people only allowed to pick one? If not, I'd assume the total is lower than that since some people probably use more than one of them.

I have used Notepad++, Sublime, Vim, Spacemacs and the Jetbrains suite.

If I want to do a large enough project, the Jetbrains suite really can't be beaten in terms of just getting things done IMO.

Probably speaks to the quality of their product. There are a bajillion free IDEs out there and people are willing to PAY IntelliJ.

That being said, their personal licensing is so affordable that I wouldn't blink to buy it out of my own pocket it if my company didn't pay for it. If you don't need IDEA, I think most of the smaller ones are $50 for a personal license.

I used Eclipse for Java and Komodo for Python/PHP for years. Then Komodo 8 was so bad that I gave PHP Storm a try and liked it so much I gave IDEA a try for Java and never looked back.

Yeah, and a big endorsement for Rust - at the top of the "love" list.

To be fair, JS appears in 'loved' as Node.js.

For Windows, Notepad++ fits nicely for situations where "I need more than Notepad, but less than an IDE". It's a great example of the a well done pieces of free software on Windows. I've tried Emacs, Vim, Atom, Code, and Sublime, and none of them beat N++.

Windows + free + low system impact / short install time. Makes it a goto in a lot of cases on the Windows platform.

I've found myself using Notepad++ for personal notes/to-do lists and Sublime or Pycharm for coding. I could do everything in Sublime, but this division of duties seems to reduce clutter.

Question - are you primarily a windows or mac developer? Because the counter part on mac is sublime text and I see a ton of that!

I see it kind of like this: 10 years ago when I had my first official job post college we needed some tools for interacting with our manufacturing data sets. I opened up Visual Studio and created some Windows applications and installers.

Then we deployed a web service written in Java (because at the time any web server we had was *nix) to create some additional functionality and hooked the desktop apps up to it.

Then we swapped out the desktop apps because too many people were not upgrading and getting all kinds of errors. Or desktop support was getting too many calls because the end user had a borked environment.

This was all many moons ago and happened in the span of 18 months.

I started out as a desktop and server API application developer. Today I haven't written an exe since I left that job but to build UIs over the years I had to learn HTML/CSS/JS. That is, I feel, we got the 'original' full stack developer.

Biased towards users of StackOverflow, not developers in general. Unless you believe the population of developers who took the survey is representative of the entire world of software developers--I'd be willing to bet it's not.

Web is where the new stuff is happening, if you see a new framework or library 90% of the time it is some JS/browser based language. Thus this leads to a larger online community of people trying to make stuff work, and in the end more SO users.

Not to mention web apps being the new big thing.

To further that I think a lot of devs that tinker with things outside of work tend to tinker with web stuff more.

My reaction to CoffeeScript over time:

CoffeeScript is shit, CoffeeScript is ok, CoffeeScript is great, CoffeeScript is ok, CoffeeScript is shit

I would assume it got "downvoted" by JavaScript developers that had to deal with it at some point.

(I personally prefer coffeescript)

Right there with you. I was generally aware (mostly from HN) that it seemed less popular than when I started using it -- but _dread_? Surprising. I really enjoy it.

My personal impression was that Rails has matured and there aren't as many open problems at this point-- there are agreed upon patterns and gems for most things you want to do in creating a web app-- or at least a relevant SO answer that is the first thing that comes up when you google a problem or error message.

I know that isn't exactly what this question was measuring but I'd expect engagement for Ruby in general to fall on SO. Compare that to the current wild wild west of JS frameworks where everyone seems to be doing their own thing.

Well the SO community is definitely the representative of Web Developers community. And since Ruby is mostly Rails, and Rails is a web development stack, this tell a lot.

And I think that is the biggest problem with the Ruby communities in general, NO one cares about Ruby is dying. Not Matz, not DHH.

I noticed the same. Maybe the results are skewed a little bit by the SO audience but I think it's a real trend.

Here I thought it was referring to the Spark programming language.

I read it as: "Man, I better get over my hatred of JavaScript." (Started when building front ends to databases in 1999. Haven't really looked back and that is 100% my fault.

It's also the most common language on GitHub in terms of both # of repositories and total pushes: http://githut.info/. Why is it hard to believe?

Well, anyone who has ever written a webpage with a single script has used it.

I suspect that accounts for the overwhelming majority of developers now. I doubt that it means that majority is also writing the majority of their code in it.

I think it might actually be the case. It's a testament to how valuable it is to have a ubiquitous platform (especially one that's fairly uniform) on which to run your software.

But, sure, SO is extrapolating here.

Older devs probably don't use SO as much, because they are the one who built the great tools. They built the techology stackoverflow was based upon, they didn't need SO when they were developing in the 80s, 90s. they probably don't need it now.

Well, even if there weren't a problem with age bias I would expect the median to be skewed lower. The details indicate that not all respondents are professionals. This could include youths, still in secondary school who are learning to program.

But if the age distribution of people is such that the median age of humans were 27, it shouldn't concern us that the median age of this population of survey respondents is 27. I don't know what the human population distribution is but I took a quick peek at the US census bureau and it seems to indicate much more young people than older. I suppose that shouldn't really be surprising.

That caught my eye as well so I googled past surveys to see how the results have shifted. What was more interesting to me is the 2016 distribution is very comparable to the 2013 distribution. But then again this survey hasn't been going on for that long. https://docs.google.com/spreadsheets/d/16C6ZeNSvd-qIg45J9E-v...

Nope. A lot of people that identify as 'developer' move on to other roles such as team lead or architect or, if they're lucky, retire before they hit 40. There's also the bias of stack overflow's users, of course; I can imagine that experienced devs visit SO less and thus never saw the survery (I know I didn't, and I still occasionally visit it)

Those numbers are for the average SO user, not the average developer. Developers who've been developing for 20 years before SO showed up on the scene are less likely to make SO part of their development routine and less likely to fill out surveys about current programming trends.

In any field undergoing exponential growth, the median experience will be the same as the doubling time.

It wouldn't surprise me if tech doubles every 5 years, so 27 is not unrealistic.

In the past, gender discrepancy was justified just as glibly as ageism is now. People said almost exactly the same nonsense then: women just got married, they moved to other fields, this particular poll is biased (never mind they all show the same trend), they quit to raise children. In the ageism case, someone even ludicrously suggested retirement before 40 accounts for a significant percentage

Ageism is indeed a very serious problem and the same personality types that didn't want to fix bias problems in the past are resisting addressing the age problem with equal vigor today.

probably due to young populations in countries like India.

I wonder if it's actually decreasing in popularity or StackOverflow is just becoming increasingly popular with other languages.

Looking at the number of responses to that question, there are lots more people taking the survey in the last two years!

2013 - 8042

2014 - 6537

2015 - 21982

2016 - 49397

There is probably a correlation to the slow rise of the mac platform and the slow fall of the OS X platform.

http://stackoverflow.com/research/developer-survey-2016#tech...

There are a ton of developers that will choose whatever language they have available and if developers are switching to mac they wont be reaching for C#. Then you have those that have moved on to F#. Objective-C has the same problem for those that have moved to swift. It is weird to think that next year perhaps 1/3 of SO developers will be on OS X. That is such a huge amount.

And in the last decade Linux has absolutely won the server market. It is going to take some time for the clr on Linux project to stop the bleed from projects that moved off c# just to move off of Windows servers.

Link to the full answers about tech from parent: http://stackoverflow.com/research/developer-survey-2016#tech...

Perhaps Microsoft's reasons for their "recent efforts" are becoming more obvious to you...

Well. It didn't decrease as much. Perhaps it's just taking a moment to reverse the momentum.

It might also be partly explained by selection bias among SO's audience. At my first real job there was no other developer there below 40, and a few approaching retirement age (indeed, I was brought on to replace a developer who had retired). But I was the only one who used SO, and really the only one who turned to the internet for answers at all rather than reaching for the books on their shelves.

(And also, our proportion of female developers was way, way higher than 5%.)

Older developers are certainly out there, but they're woefully underrepresented in open source.

Most loved top 5:

1. Rust

2. Swift

3. F#

4. Scala

5. Go

I totally forgot that they even asked that, even more I forgot what my educated guess was.... Shame on me.

It's users vs developers.

1.7% of respondents are WP users (10% x 17%), and only 0.1% of respondents are WP developers.

Ruby is part of the "Losers" in the Trending Tech. But so is Rust that is also the most Loved. The trend is only the number of upvotes in a topic year to year.

But this is always hard to interpret : does it mean that all basic questions have been asked so there is less activity or is it due to a drop in questions because lack of interest ?

Use of Ruby is probably mostly for Rails and while Rails is still cool it's not as big as it was before. There's competition.

Well actually close. If you have 40 years of productive work in you (20 years to 60 then retirement), then yes every 10 years 25% of the workforce does retire.

But I would guess that those 10% of students are heavily skewed to students in their final year of study. Second point is that not all undergraduate courses are 4 years, some are 3 years, so probably closer to 7.5% of students on SO graduate every year.

Which might indicate that the software development field is growing in terms of people active in the field.

*Then again software development is a young persons game, so "retirement" out of the industry might be higher since a lot of people will retire to other fields / management positions.