Slashdot is powered by your submissions, so send in your scoop

 



Forgot your password?
typodupeerror
×
Programming

The Rise and Fall of Software Developer Jobs 64

The demand for software developers has declined sharply from the peak seen in 2021 and 2022, according to independent analysis by job portal Indeed and research firm ADP, reflecting a broader slowdown in high-paying white-collar job opportunities across tech, marketing, and finance sectors. Nick Bunker, an economist at Indeed, identified these positions as the labor market's current weak point. The shift follows a period of intense recruitment during the pandemic, when tech workers could command premium salaries.

ADP Research adds: Employment of software developers in fact has been slowing since 2020, the year pandemic lockdowns first hit the United States. In January 2024, the U.S. employed fewer software developers than it did six years ago. [...]

The ADP Research Institute tracked employees at 6,500 companies, including more than 75,000 software developers and engineers in 10 industries, between January 2018 and January 2024. Using this data, we built an index to track the employment of software developers beginning in January 2018.

Developer employment grew from January 2018 to November 2019, then began to fall. The index dropped sharply in January 2022 (down 4.6 percentage points), May 2022 (down 3.5 percentage points), and January 2023 (down 3.4 percentage points). Despite intermediate increases in August 2021 and October 2022, the developer employment index has been falling since 2020.
This discussion has been archived. No new comments can be posted.

The Rise and Fall of Software Developer Jobs

Comments Filter:
  • Peak Digital. (Score:2, Interesting)

    by Qbertino ( 265505 )

    Said it before, will say it again: The IT frontier has been conquered, it's all industrialization and commodification from here on out. We aren't needed anymore. And then there's AI, but we'd have the same situation withoit it too. The experts predicted this a few years ago and I too saw it coming. Peak Digital. Plain and simple.

    • time for an union!

    • Re:Peak Digital. (Score:5, Interesting)

      by avandesande ( 143899 ) on Friday July 19, 2024 @11:18AM (#64638123) Journal
      I've been hearing about the end of software jobs the last 25 years, yet there seems to be endless amounts of things that need to be done as far as I can see.
      • Re:Peak Digital. (Score:5, Insightful)

        by Brain-Fu ( 1274756 ) on Friday July 19, 2024 @11:31AM (#64638177) Homepage Journal

        Indeed.

        AI is not yet capable of replacing software developers. A lot of really rich people really want it to be, and plenty of marketers are more than happy to exploit that desire to profit on false hope, but the cold hard fact is: AI is not there yet.

        The reason jobs are in the decline is the high interest rates. This was all by design, and it wasn't some secret evil conspiracy or anything. Interest rates were raised to combat the runaway inflation that we were experiencing after the pandemic, and the mechanism by which this was expected to achieve its goal was twofold:
        1. reduce the available money supply (by reducing the number of loans made),
        2. motivate mass layoffs across all industries (by reducing available business loan money). This was expected to move wealth out of the middle classes and back into the upper class, having the effect of less spending money for most people, thus forcing reduced spending and in turn price competition to pull prices down.

        That was the plan. There were additional options that could have softened the blow here, but it would have required a significant curtailment of government deficit spending, which never wins votes, so neither political party is willing to do that.

        So here we are now. The plan worked. Inflation is down, mass layoffs have happened, wealth is being re-concentrated at the top, and software developers have to compete against one another for jobs again, thus driving salaries down. Things are looking good for the upper class.

        Maybe someday AI will actually replace software developers. Until then, this cycle is just continuing along, as it has been doing ever since software development was even a job. Eventually, that pendulum will swing back, as it has done many times before.

        • Re:Peak Digital. (Score:5, Insightful)

          by 14erCleaner ( 745600 ) <FourteenerCleaner@yahoo.com> on Friday July 19, 2024 @12:01PM (#64638301) Homepage Journal
          As someone who came of age around 1980, I have to laugh at the description of the past four years as "runaway inflation". But I digress...

          The US unemployment rate overall is only 4% currently, which is low by historic measures. If there are "mass layoffs", it isn't reflected in the job numbers yet. Maybe AI is the real reason...we'll see how things shake out in the next few years. I can certainly see less demand for code monkeys, since you can generate semi-reasonable code for many situations with a simple prompt.

          On the other hand, when I started my first job in 1981, I expected automatic coding of some sort to eventually replace coders. The programming environments are obviously vastly better now and do a lot for you, but the demand has stayed pretty high for coders for the past 40 years, since computers and software permeate our lives so much. Maybe we've just reached saturation on software demand, since everything that uses electricity already has a CPU in it.

          • Re:Peak Digital. (Score:5, Insightful)

            by leptons ( 891340 ) on Friday July 19, 2024 @12:15PM (#64638353)
            >since you can generate semi-reasonable code for many situations with a simple prompt.

            Sorry, no, nobody that isn't a programmer can cobble together a working system for anything using an AI prompt. Until someone like my non-tech wife or mother can use a prompt to create a fully customized and unique app, AI isn't going to steal programming jobs. The fact is a programmer needs to check what the AI spits out, and a human still needs to know how to put the pieces together. An AI as it is today only regurgitates code it's read, but it doesn't really understand what the code does, and the entire AI field right now is plagued by "garbage in, garbage out". I use AI every day when I code, and it gets it completely wrong 95% of the time. About all it's really good for is a fancy autocomplete, which is only correct about 50% of the time. I am not worried one bit about AI stealing my programming job.
            • quoting myself: "sufficiently developed requirements are indistinguishable from code"
            • by MobyDisk ( 75490 )

              I can certainly see less demand for code monkeys, since you [AI] can generate semi-reasonable code for many situations with a simple prompt... The programming environments are obviously vastly better now and do a lot for you...

              Sorry, no, nobody that isn't a programmer can cobble together a working system for anything using an AI prompt.

              It must be a Slashdot phenomenon that someone starts a post with "Sorry, no" then proceeds to not actually disagree with the previous post. It's like a form of one-upmanship.

              The premise here is that as technology improves, the tools for coding get easier, hence you need fewer programmers. At the very least AI is a tool that is another step on that road. This may explain the drop in programmer demand as much as interest rates do.

              • The premise is wrong though. Coding efficiency just means more software and more complex business systems.
          • by vivian ( 156520 )

            As long as you have businesses that need problems solved, you will need someone that can concisely define the problem and the desired output with enough detail to get the answers you need to help run your business.
            It used to be that you needed a team of business analysts and programmers to do this - analysts to clearly define the nature of the problem and programmers to implement it, and a test team to validate that the results are what you actually want.

            AI might make the programming bit easier, but you ar

          • The U-2 rate you are quoting is known to be BS.

            It doesn't count people who have been unemployed for long.

            It's literally designed to be false.

          • As someone who came of age around 1980, I have to laugh at the description of the past four years as "runaway inflation".

            That is because your own personal bubble has not been popped yet. If you hadn't been living in a bubble, you would be seeing millions of people suffering pretty badly... but that is not a knock against you, it is a knock against the media misinforming you.

            Ask anyone who is starting out without familial assistance: Life is fucking brutal right now, more so than ever before.

    • This could have been said at every stage of the evolution of "technology". And you/re just dead wrong about this being the end of it. From here it's just going to morph into more of an AI world. From here, AI is going to begin to program the humans. It's already begun.

      One must take into account the fact that no matter what a society takes on as a new, the effects of it will hide until an entire generation of people are born into said newness. In 20 years from now, the humanity will likely be gone. Not

    • Weird comment on the day of wide spread software-caused outages.

    • by TJHook3r ( 4699685 ) on Friday July 19, 2024 @12:25PM (#64638395)
      The market must be saturated by software developers, what it really lacks is software engineers. Some software devs I've met are doing little beyond glorified secretarial work - they take an exact spec and code a website to suit. No questions or suggestions. They just do what they're told. And for that they expect to earn a little more than a chartered civil engineer. Market forces and all that but it drives me mad that that actual engineers are valued so lightly (by engineers, I mean anyone who solves problems, constructs workable solutions and pays attention to fitness for purpose - as opposed to hacks)
      • >> They just do what they're told

        I think most people just do what they are told at work. There's little incentive to do otherwise, and you might even be punished. In modern-day scrum shops you will get a set of work tickets at the daily standup meeting and push them to the right through JIRA. Innovation doesn't appear on the agenda.

  • by BringsApples ( 3418089 ) on Friday July 19, 2024 @11:17AM (#64638117)

    I remember once when the mentality was one that programmers would threaten those that they deemed lower than they, with the famous threat, "Go away or I'll replace you with a script..."

    "Well well well how the turntables have.....
    --Michael Scott"
      --BringsApples

    • I remember once when the mentality was one that programmers would threaten those that they deemed lower than they, with the famous threat, "Go away or I'll replace you with a script..."

      "Well well well how the turntables have.....
      --Michael Scott"

        --BringsApples

      I don't think anything really changed there, other than using AI to write the script.

  • Guess it (predictably) went the way of 'everybody should make buggy whips'.

    • by ceoyoyo ( 59147 )

      Incredibly, the world does have a limited need for web forms.

    • by Seven Spirals ( 4924941 ) on Friday July 19, 2024 @01:29PM (#64638641)
      The mainstream is clueless about programming and have been making untenable false predictions for years. First CASE programming was going to kill coding. The next magic bullet was IDEs and code-completion. Now, it's AI. Anyone who uses OpenAI CODEX via some IDE like CoPilot or NeoVIM knows that helps a little but isn't even close to the panacea they claim.

      What I think AI will do is discourage young people from learning in most of the domains that it competes in. Show a kid Midjourney, Dall-E, and Sora and then say "So, you ready to enroll in art classes?" They are going to look at you like you're crazy. The same will soon be true for music. You wanna learn guitar so you can strum next to a campfire or just let the AI do the playing while you put song after song together easily? Right now there are plenty of musicians and artists that have skills they built up pre-AI, and can do things AI cannot (like save in a specific format at super-high resolutions and aspect ratios while staying completely consistent style-wise). However, convincing kids to study and learn stuff AI can do (even with only low to moderate skill like coding) is going to be tough. It was already tough.

      Meanwhile someone still has to code in assembly in many cases or write a compiler. AI can barely write 25 lines of shell script that doesn't fail or C that will compile. I'm talking about the very very best AI like ChatGPT 4.0o or Anthropic Claude Sonnet. Don't lie and say it's already amazing. I'm using it right now and it's wholly unimpressive. Not only that, the interfaces (even in IDEs) are extremely shitty and rarely do what you want the first time unless you have very modest expectations. So, I'm thinking the old crew who still knows how to navigate by the stars is going to need to stay on deck a while.
      • by ctilsie242 ( 4841247 ) on Friday July 19, 2024 @01:51PM (#64638713)

        I find that with programming, AI is something to use to start from something other than a blank slate... but there might be times where debugging AI generated code can take more time and effort than just writing everything from scratch.

        Does AI replace programmers? Maybe it replaces the most clueless. AI, if one can compare it to something, is basically a power tool. You can do more with a cordless drill, but if you don't know what you are doing, you can screw up a lot. It definitely won't replace good devs or people who know how all the stuff works together. However, many managers think they can just get some contractor out of Lower Elbonia to magically write everything via CoPilot... and then get absolute garbage, taking far more time to debug than if they kept their original coding team. AI can be of help, if you know what something does, and can move you in a direction, but if one doesn't know where they are going, AI probably won't help much, maybe even cause more problems.

        • Well said. Sounds like you have actually been using this stuff like I have and you aren't the guy who said that anyone who didn't have ChatGPT 4.0o was going to be like "Muggles" in Harry Potter (no magical talent in a magical world). I have been using this stuff for months, now it and your power drill analogy is a good one. Sometimes it's worth while, other times it just lies or screws up convincingly with astonishing speed.
  • by battingly ( 5065477 ) on Friday July 19, 2024 @11:43AM (#64638235)
    What do you expect with a profession which has as its primary objective the automation of things, including software development itself, so human intervention isn't required?
    • It's more than just that. In a recession, companies just sit back and collect rent. Everything is a subscription now. They don't have to sell anything. More importantly, they don't have to improve anything because no startup is going to get enough funding in this time period to become a serious competitor.

  • But the main thing killing software developer jobs right now is high interest rates are killing us. And frankly any other job you can think of because the way it works is the crank interest rates which causes layoffs and then we all have to take lower pay desperately scrambling to survive and since we've all got less money we spend less and that's what's supposed to control inflation.

    So do you work for a living? Then you have a target on your back from the Federal reserve. Jerome Powell in particular ad
    • by JustNiz ( 692889 )

      "Jerome Powell in particular admitted under oath that he wanted 3.5 million layoffs"

      Let's hope one of them is his.

    • by theshowmecanuck ( 703852 ) on Friday July 19, 2024 @12:26PM (#64638403) Journal
      These are NORMAL interest rates. The interest rates of the past 15 or so years were ABNORMALLY LOW. The software job market was one of unfettered growth from the 1970s to 2008, and even into the 2010s, and that was independent of interest rates. Average interest rates over the last 100 years has been about 5%. In the early 1980s interest rates got up to around 15%. In the early 1990s they peaked at 20+%. And they kept at an average 5% all through the dot com era when software jobs were just screaming busy. And they stayed that way through the dot com bust, AND the recovery. THESE NORMAL INTEREST RATES HAVE FUCKING NOTHING TO DO WITH THIS, IF ANYTHING THEY WILL HELP. Interest rates need to stay like this, chase over-extended assholes out of markets they have no business in, which is what slows the housing market and makes prices more affordable, which will allow people to buy more and grow the economy in a healthy manner.
  • Fixed that for you.
    • Weird how we fired millions of people and now software-based systems are breaking. I am sure that correlation isn't causation in this case.

  • While it's probably somewhat true that we've had some sort of Peak Digital [slashdot.org], one thing we haven't reached yet is peak security [slashdot.org].

    All these security solutions aren't all going to be in hardware (and even so, they still require some level of software), so I just see the "typical" developer jobs transitioning to more security-related positions instead.

  • by MpVpRb ( 1423381 ) on Friday July 19, 2024 @12:52PM (#64638503)

    ...than just one number
    Software developers working on real projects that actually work are doing fine
    Highly speculative projects like the metaverse were doomed from the start

    Software can be roughly divided into two kinds...
    Really useful stuff in science and engineering
    Crap like IOT and "smart" appliances that marketoids seem intent on forcing us to use even though they suck

    Tech used to be tech, actual advances in hardware and software
    In today's distorted world, the word "tech" has been mutated into social media, advertising and ever more desperate attempts by corporations to limit the freedom of customers while feeding them ads and increasingly shitty products

  • by Tony Isaac ( 1301187 ) on Friday July 19, 2024 @01:06PM (#64638561) Homepage

    Based on the linked graph, we are currently at the same level as 2020. Demand spiked around 2022, and has now fallen back again.

    Is this really "the peak" or was the peak a side effect of the pandemic, and we're now back to now where we would have been without the pandemic?

  • by Tony Isaac ( 1301187 ) on Friday July 19, 2024 @01:11PM (#64638587) Homepage

    Over the 35 years of my career, I've seen programming become more and more automated. 4GLs, machine learning, low-code / no-code solutions, and of course, now, AI. All these things have their place, and for the right audience, save a lot of time. But every single one of these automation tools has a common problem: They are great at what they are designed to do, but if you stray outside of that boundary, you are screwed, or at the very least, your job becomes much, much harder than without the tool.

    Consider Entity Framework, for example. It generates SQL code for programmers who don't want to mess with SQL. But if you want to do something outside of what EF is good at, or you need performance-optimized SQL, your just just got a lot, lot harder.

    If you're creating boilerplate code, your job is in danger. If your job involves creating new technology for which the solution isn't yet known, automation tools, including AI, isn't coming for your job.

    • by gweihir ( 88907 )

      If you're creating boilerplate code, your job is in danger. If your job involves creating new technology for which the solution isn't yet known, automation tools, including AI, isn't coming for your job.

      Indeed. But that is in no way a surprise.

    • Over the 35 years of my career, I've seen programming become more and more automated. 4GLs, machine learning, low-code / no-code solutions, and of course, now, AI. All these things have their place, and for the right audience, save a lot of time. But every single one of these automation tools has a common problem: They are great at what they are designed to do, but if you stray outside of that boundary, you are screwed, or at the very least, your job becomes much, much harder than without the tool.

      Yes and no.

      100% agree with something like an ORM, sure.

      AI? Well ... no. It's literally the opposite, it's open ended. "Rewrite this for me using {completely different language}/{different library}/{different strategy}". "Make me an add on that does the following requirements: 1. ...". "Here's the URL for the API documentation, write some sample integration code in php. We'll need the following data elements ... "

      • Yeah, I know, I've used GitHub Copilot quite a bit. It's open-ended all right, but you can *never* trust the code it writes. You can't even be certain that what it produces, will compile, or that the logic will be placed within an actual function body. When asked to modify some code, it often supplies the modification and inserts it right after the original, in such a way that neither will compile. Maybe these are growing pains, but it's a long, long way from being able to just tell it what to write, and it

        • Agreed. The other problem that has caused me a lot of frustration is version differences (typically the old version deprecates something or an API changes in other ways). Asking it what should be changed to upgrade this code, or to rewrite for version n+1, mostly results in a chirpy can-do reply that still uses version n. Telling it it is wrong results in the familiar You're right, sorry about the confusion reply, but I've never been able to break out of the circular hallucinations.
    • If your job involves creating new technology for which the solution isn't yet known, automation tools, including AI, isn't coming for your job.

      Correct; however, you now have a hundred thousand people all claiming they are better than you while clamoring for YOUR job. While less than 1% can actually do your job, employers have to filter through the hundred thousand to find you now.

      AI is dangerous for you even if you are competent and gainfully employed.

      • Competition for jobs has always been a thing, this is not new. And there have been ups and downs in the past, for the "real" jobs. If a hiring company can't tell the difference between a candidate who's a wannabe expert, and a real expert, real experts don't want to work for that company anyway. There are ways in an interview, to distinguish a person with actual skills, and a person who can only talk about skills. Of course, interviews aren't perfect, but for companies that put appropriate effort and skill

  • Employers have figured out that they can replace seasoned developers with fresh college grads who know how to copy/paste from Stack Overflow and then ask people to fix their code if it doesnâ(TM)t work.

    • And honestly, if that's the new reality it's not unreasonable. The entry level grunts can be just barely competent to take a request and turn it into an AI search and assembly job.

      With a bunch of them, you can multiply the productivity of a real programmer many fold. And they can do bug testing when not doing search-copy-paste.

      Of course, there's not much of a career path there when the pyramid is so flat.

  • This is likely in part a result of changes in tax treatment of research and expenditure for software development,

    While the law will cause headaches for many industries, the software industry is particularly impacted. The TCJA added IR.C Sec. 174(c)(3), which explicitly states that any expenses incurred in connection with software development must be treated as an R&E expenditure (capital asset) and therefore amortized.

    Previously, software development expenses were not explicitly under the purview of IRC Sec. 174. Instead, Rev. Proc. 2000-50 provided that software development expenditures âoein many respects so closely resemble the kind of R&E expendituresâ that fall under IRC Sec. 174 that it warranted âoesimilar accounting treatment.â Thus, a taxpayer could elect to deduct 100% of these expenses under IRC Sec. 174, amortize the expenses over a five-year period, or amortize the expenses over a three-year period under IRC Sec. 167(f). However, software development expenses did not fall under the definition of âoeresearch and development expendituresâ under IRC Sec. 174. The TCJA changed this by formally defining software development expenditures as an R&E expenditure subject to IRC Sec. 174.

    The impact of this change on software and technology companies cannot be overstated. Many companies have had their taxable income increase dramatically because they can no longer deduct expenses. In fact, some companies have gone from being unprofitable to profitable and liable for federal and state taxes due to the change.

    https://www.eisneramper.com/in... [eisneramper.com]

Experiments must be reproducible; they should all fail in the same way.

Working...