Ryan and Eira talk with Stack Overflow senior research analyst Erin Yepis about the results of our 2024 Developer Survey, which polled more than 65,000 developers about the tools they use, the technologies they want to learn, their experiences at work, and much more. Erin highlights what the survey reveals about devs’ favorite programming languages (JavaScript, HTML, Python), the rise of Rust, the popularity of embedded technologies (Raspberry Pi, Arduino), developer sentiment around AI, and why tech debt tops the list of developer frustrations.
Read the blog post or dive into the results of our 2024 Developer Survey.
A few highlights to get you started:
Speaking of our developer community, Stack Overflow user Frank earned a Stellar Question badge by wondering How to use C++ in Go.
[intro music plays]
Ryan Donovan Welcome to the Stack Overflow Podcast, a place to talk all things software and technology. I'm Ryan Donovan, and I am joined today by Eira May, and today, survey says, we're going to be discussing the results of the 2024 Stack Overflow Developer Survey. You excited to chat these results, Eira?
Eira May I'm very excited. Definitely one of the highlights of our year is finding out what folks have been working on and what technologies they've been working with, and it just gives us a snapshot of the folks that we're talking to. So we have Erin on the show today to help us talk through that. Erin, do you want to introduce yourself and talk a bit about what you do here at Stack Overflow?
Erin Yepis Yes, I would love to. My name is Erin Yepis. I'm a Senior Research Analyst at Stack Overflow, and I've been here for almost two years. I work with a lot of our internal and external data sources, just trying to understand developers, our developer community, better all the time. The developer survey is of course step one, and there's so much data to look at. Love looking at it, and I also love getting to check on my favorite coding language in the ranks every year. This year, unlike the others, my coding language of choice is falling down in the ranks.
RD Oh no. What is your coding language of choice?
EY R.
RD Of course.
EM Why did you say that in a ‘figures’ voice?
RD Well, she's a data scientist.
EM So typical.
RD Nobody uses it but data scientists.
EM No, that's true. That's fair.
RD And speaking of the tech rankings, this is my fifth year participating in putting together the dev survey. And the tech rankings, the loved, hated, used and abused languages, those are always some of the favorites by people who read it. What was this year's results like?
EY So this year we see a lot of our old favorites coming back, which we like to see. We like change. We like new stuff. We like the flashy new stuff, but some of us like the old reliable stuff too. So our top three programming languages have stayed pretty consistent over the last couple of years, being JavaScript, HTML, and Python. But you mentioned for the most admired and desired, we see the return of Rust coming in as one of the most admired.
RD I know Rust was always a big top when we called it ‘the most loved languages,’ and I think the admired/desired framework helps people understand what these mean. Are there other technologies that were shifting around or is it old favorites?
EY A lot of old favorites, however, with a couple of caveats. So one of the more interesting, I think, shifts that we're seeing, we've seen on Stack Overflow's tag trends as well, is the gradual rise of Python up the ranks. We see, especially if you go to our results site and check on the Learning to Code tab, Python is their top choice for programming language this year. And it was last year too.
RD The other thing is that I think it's professional developers that rank Postgres pretty high, and almost 50 percent of them use it as the most popular one. We're going to be talking to somebody about Postgres in an upcoming episode, so that is an interesting stat to share.
EY So one of the technologies I love seeing on the list, we added it last year, it's back this year. For asynchronous tools, we have Markdown files coming in again as one of the top three technologies that developers prefer. It also ranks very highly for that admired score as well. I also added something new this year, very adjacent to Markdown– because I love Markdown files, obviously. I create a bunch of them in R as well. But there's a tool called Obsidian that one of our coworkers got me wise to, which works like Markdown as well. It’s just creating tabs that you can use all of the Markdown shortcuts. So it made its debut on async tools list this year and is doing pretty well given the long list and that it’s just now ripe. And I only just started learning about it not that long ago too. So I love how there's just these old reliables that show up.
EM What are the new questions that were added this year or anything surprising?
EY So sticking to the technology lists, two new things this year. One of them is a little bit more new than the other one. We added a new section for embedded technologies, and this came directly from feedback from our meta community. So our community members love to be very specific. “Hey, you're grouping this certain technology with all these other ones. It doesn't really belong there.” And so because I know the survey is getting longer and longer each year, I try to be very careful about what we're keeping and what we want to add and see if it can be assumed by other content areas and not just adding just tons of questions. But this seemed worthwhile– we were able to move some of the technologies out of different sections so we have just one big embedded technology list that people got to choose from where we saw Raspberry Pi top of the list. And that was one of the technologies at least that had the brand name recognition so I knew what it was. A lot of them I don't know what they are, so I love that our community could step in and help with vetting that.
RD That's nice that we have the community involvement. I remember complaints about previous ones where people were saying, “This isn't a framework,” or “This isn't a programming language,” or whatever, just getting that nuance. What were some of the other embedded technologies?
EY So we had at the top of the list, Raspberry Pi followed closely by Arduino, which I believe they're both pretty related. They show up again also on the work with/want to work with chord graphs where a lot of the Arduino users want to work with Raspberry Pi. Not so much the other way, but I think that's because there's components that are duplicated. I’m not an embedded developer so I don't exactly understand it, but I do know there's a relationship there. That's why they both scored pretty high as far as popularity. And then right underneath that, pretty close to the Arduino popularity, is GNU GCC, which sounds like I'm reading a barcode out loud.
RD Is it a compiler GCC?
EY That sounds right. So of all the developers that responded to the survey, embedded developers make up 3 percent of our responses– not an insignificant number, considering all the other types of roles that we ask about. So they're coming to Stack Overflow, they know a lot more about this stuff, they're participating in the meta community. I would love to learn more about this, maybe I will. Maybe I'll have some more time now, now that the survey results are finally out in the wild.
RD And if you want to drop us a comment about embedded technologies, I think Erin would appreciate that.
EY Yes.
RD You said there were two new tech questions. Is that right?
EY Yes, the other one is a little less new, but we did switch it up. So on the technology section, we also have the AI technology section. Last year we split it up, so we had “What are the most popular AI search technologies?” and “What are the most popular AI developer tool technologies?” We combined them this year because there's just a lot of overlap and they're changing a lot, so let's just keep them all together. We can monitor this list a little bit more. It's a little bit more organized that way. We didn't see a lot of difference in the results. We have two players that are just dominating the game right now, but these are relatively new technologies, so I would expect that. ChatGPT being at the top of the list, I think it's a little bit more helpful to see it all together now so you can compare them to each other. And I know in the meta comments as well from last year there was a lot of people that were making comments about how ChatGPT is a developer tool as well, that they use it as that. So glad to have less confusion there. And probably with all of the interviews and resources you guys have been working with on new AI technologies, there was a moment there where we thought maybe it was more apropos to talk about the models instead of the names of the technologies, but those are changing just as often as the technology options themselves, so one has a little bit of that marketing behind it, whereas maybe the model names are not as well known or you're not sure which one goes to which tool.
RD We see that too. I think so many tools are both developer tools and search tools at this point in that there's new models all the time. It’s a tough one. Your survey is going to be out of date by the time it's published.
EY Exactly. But it is a nice snapshot in time at least, so we have that. But who knows, maybe next year we'll change our minds about including those models. I don't know. It depends on what the people want.
EM It's a long time to see into the future, especially when we're talking about something like AI and how developers are using AI.
EY I know.
RD Speaking of which, I know we've also been taking developers' temperature on how they feel about AI and how they're using it at work. What sort of things did we ask this year and what were the results?
EY So a couple of the questions that we asked last year for the first time came back this year. Just getting that temperature check on, are you using it, do you like it, and what are the perceived benefits that you see with using these tools? So we found this year that more people are using or interested in using AI tools, but there are still problems. So they're using it, they're getting more familiar with it, and I think with that familiarity they're getting a little bit more familiar with the challenges. So this year, one of the new questions that we asked was, “What are your top challenges that you face at work if you use AI tools with your team or at your company?” And the lack of trust is the top challenge, followed by inability to understand complex code bases. So we also asked another question related to this, asking developers if they felt that AI tools or AI in general was a threat to their positions, and 70 percent agreed that it was not. And I think because they have had a chance to actually interact with it more, actually use it at work, try different flavors of it, whether they're trying some of the new options that are coming out or maybe using an enterprise solution, they are seeing that it's not necessarily– they still have to do a good chunk of work to make it apply to what they're doing.
EM Right. It's not going to step in and take over the job for them.
EY Right. Which I think is surprising since we talk about how it's going to take jobs, I think, every other day.
EM Exactly.
RD A lot of folks do. And I think there are some analyses that show that some developers have lost jobs because of it, but the promise of the AI developer agent isn't entirely there yet. You still have to know how to code to know if this is a good coder.
EY Yes. And so it becomes that you are in control. You're using this tool to help you get a job done. You still have to be the one that's in control.
RD Was there anything that surprised you about the AI stuff?
EY My major surprise was that more developers did not agree that AI was a threat to their jobs. Most of the developer roles for our respondents are back end and full stack developers, so they have pretty complicated jobs. They're not worried about this, especially with new stories coming out. Our recent big CrowdStrike story coming out. The more experience you have, I think the more opportunities for big errors these developers are seeing. So they know that there's just too much going on that AI is not. And again, that top challenge of inability to understand complex code bases they've picked up on as well. So it makes sense, but I also was just expecting there to be a little bit more parity there, more developers being maybe unsure about if AI was a threat. Because that was one of the options too that you could select if you weren't sure. But no, 70 percent said, “No, AI is not a threat to my job.”
EM That's a higher percentage than I would have expected who would feel that level of confidence too. But I think you're right, probably the kinds of roles that we're speaking to has a lot to do with this. If we were asking folks who are front end developers or junior developers, they might have a different response.
RD Yeah. As people are using these more and more, almost a familiarity breeds contempt. Once you find out the actual capabilities of these, you're like, “Oh, okay. They're not going to replace me with this robot just yet.”
EM That is a little bit how I feel as a copywriter. “Oh, I'm good for now.”
EY I definitely think you're good. Every time I use ChatGPT to get the thought starter, I hate the results. I hate it. So I always have to edit them or ask at least 10 more times, “Please fix this. Do this instead.”
RD You mentioned that we had community engagement on getting the survey ready. We also had some community-related questions in the survey. What did we ask and what did we find out on that?
EY We had a couple of new questions that we added to the community section, and one of them was– so in the past we have been asking, “Do you have a Stack Overflow account? Do you participate on Stack Overflow?” But this year we also asked, “What are the benefits that you use Stack Overflow for?” And this is based off of some previous work that our UX team started earlier in the year where they're doing their qualitative research to find out exactly pain points that real users are having on the site. So I had a little bit of help there, a little bit of inspiration, but overwhelmingly the top choice that is selected by actual Stack Overflow users is that they come to the site to save time, to find solutions to code problems best. So for all of the haters out there that say that it takes too long to find answers on Stack Overflow, that's exactly what they're doing. It's nice to have that stat out there.
RD I think that the initial burst of ChatGPT finding answers but then having to verify whether those answers are true leads us right back to us again.
EY And they're complex problems. We know that developers don't trust AI with complex problems. Stack Overflow has an assorted array of complex problems and solutions.
RD Did you see if they were just on Stack Overflow or did they go all over the place on our sites?
EY So we asked them as a multi-select, “Which one of our sites are you familiar with?” Stack Overflow, top choice. 97 percent of people that answered that question said that that's the site that they're visiting, followed by Stack Exchange, added to the list this year is our blog and podcast site. So we have 15 percent of the respondents saying that they visit that site as well, which more people should.
RD We appreciate you.
EY Yes.
EM Howdy, folks.
EY It's a beautiful site. The layout is very enticing. There's lots of great tidbits on there, new stuff all the time.
EM Lots of great writing, I've heard.
EY Lots of great writing, lots of great interviews, lots of great content.
RD All right, all right. So I know we are always adding questions, trying new things. What else was in the survey this year?
EY So another addition that we made for the professional developer series of questions that we have is that we asked developers about their top frustrations at work. We asked them if they were satisfied at their job and what did they find satisfying the most? So this was interesting and it relates to a topic you guys had covered not too long ago. We had 63 percent, the top choice for professional developers for their frustration at work is tech debt. So on one hand we have that, and then on the other hand we asked about what makes you happy at work, and the top choice for those– happy, unhappy, or complacent, somewhere in between– their top choice for all three of those segments was improving code quality and database environments, so working on tech debt.
RD I think when we've talked about it, whether you're able to work on and eliminate tech debt is sort of a canary for a lot of organizations. Because if you're not taking care of your debt, that adds up just like real debt and then you have to declare tech bankruptcy.
EY And I can understand how when you take an afternoon to organize the junk drawer and you feel really good about that even though you didn't really get anything out of that, you still have the same amount of stuff, I can see how that would be the case. It's, on one hand, frustrating, but on the other hand, rewarding.
RD You gain greater control over your environment. And I think there's a lot of tech debt that comes from moving fast, trying to get out the door fast, or optimizing for things that you regret later.
EY I'm not too well-versed in it, but I know something that our developers are proud of and also complain a lot about is referred to as ‘the monolith,’ so it giveth and it taketh away.
RD Yes, the monolith, although monoliths are making a comeback. I keep seeing hot takes that services are out of control, monolith is fine. So it depends on what you're working with now.
EY Exactly. So interestingly enough, related to this or that for what's the preferred, what's the best class solution, I think for next year, especially after listening to and reading through your transcript about your tech debt interview that recently was posted, I'm going to have to update this question so it's a little bit more specific instead of just referring to a big tech debt and maybe breaking this out. So something about monolith versus services, something about refactoring versus rewriting, something about upgrading versus replacing.
RD That would be an interesting one. “Are you working on a monolith or service-oriented architecture?” “Do you have multiple cloud regions?” “What do you optimize your code for?”
EY So to get more specific, and maybe also paying the meta community again and ask them to vet if I am going about this the right way, and obviously probably listen to more of your interviews before then too.
RD Of course. Are there other things you're looking at adding for surveys in the future?
EY I think what we're after, the big goal is just getting better and better at understanding the developers that come to Stack Overflow. And I feel like there's just still big groups but they're not specific enough, and I want to get closer to that. So questions that help me get closer to why you say you're a full stack developer. What does that mean to you? What exactly do you mean by that? What do you like? What distinguishes you from another full stack developer? And not just because you might be born in a different year or live in a different country, something just more characteristic about your job, your work, the type of person you are. I think that's the everlasting challenge.
RD Maybe it's time for some new pulse surveys.
EY Oh, yeah. Great idea. Love that.
RD Do you not love that?
EY I love what that is– what do you call it? It's my version of tech debt. Research debt.
EM Research debt. Survey debt.
EY Survey debt. But after getting these results out for the dev survey, I'm going to need at least 24 hours before I start thinking about another survey.
EM I was going to say, you're just not ready to wrap your brain around a pulse survey.
EY I love the idea about it. I am definitely being sincere, but I see my eyes glaze over because just the survey debt is floating in front of my face.
EM Timing is everything.
EY Timing is everything.
[music plays]
RD Well, it is that time in the show again when we shout out somebody who came on Stack Overflow and shared a little of their curiosity. Today we’re going to shout out a Stellar Question Badge, awarded to Frank for asking, “How to use C++ in Go?” If you, too, are curious, maybe somebody has dropped an answer there. And if you have that answer, go share a little knowledge. I am Ryan Donovan. I edit the blog here at Stack Overflow. You can send us comments, suggestions, feedback at podcast@stackoverflow.com. And if you like what you heard today, please drop a rating and review.
EM And I am Eira May. You can find me online if you want to @EiraMaybe.
EY I am Erin Yepis. I would love it if you could go to Stack Overflow and post some new R questions and R answers. Let’s get those people back. We need them back.
EM Special shout out.
RD All right, everyone. We'll talk to you next time.
[outro music plays]