– Baldur Bjarnason
The full essay version is below the overview and embeds.
You can subscribe to the audio-only podcast feed over on Apple’s Podcast app and other apps if audio is more your speed.
It’s also available as a raw podcast feed if you want to subscribe to it directly in the podcast app of your choice.
People in the software industry don’t seem to be doing too well at the moment. There has been a spate of “I’m leaving my tech job” videos – some because they’ve been laid off, some leaving preemptively – but the interesting action has been taking place in the comment sections, replies, and discussion forums. The replies are where it’s at. The videos themselves vary in quality.
Facebook’s aborted “AI coder” transition: Mark Zuckerberg had a bold plan to replace Meta staff with AI. Here’s how it imploded.
The drop in new students in Computer Science in Iceland. (The drop was 30%, not 40% as I said in the video/podcast.) Nýnemum í tölvunarfræði og tengdum greinum fækkaði um þriðjung milli ára
We’re not doing well in the software industry.
We’ve all sensed it. Things are a bit off.
It’s not just the “AI” financial bubble, or the debates that have been taking place – arguments, more properly – or the ongoing shift in public sentiment regarding the software industry.
I’m talking the fact that people who work in software – specifically developers and user experience designers – are not doing great.
I keep coming across videos of people talking about burning out, being unable to work, and in some cases fairly serious mental health issues. And, obviously, people are getting fired. But often in these videos, those who are doing videos and announcing they’ve been laid off mention that they were thinking about quitting anyway and are now planning on leaving tech entirely.
It’s not just senior people who might have memories and a fondness for how things used to be – stress mixed with nostalgia. It’s also students. I’ve found several videos of students who announced that they had switched majors, had been studying user experience design, but now decided to switch to something else.
With a few notable exceptions, these generally aren’t people who are against “AI” – or, at least, they weren’t against “AI” until very recently.
These aren’t people who disliked the technology from the start, who have had very severe criticisms about the way it’s made, the way it’s forced on people, or have severe doubts about the long-term benefits of applying this technology to our industry.
They seem to have all embraced “AI” at some point and assumed that what they were told was true, that this was the future. “This is where things are going.” This is what software development will be like, now and forever.
After all that, one of the reasons why quite a few of them are leaving, is that they believe this promise. When they’re told that this is the future of the industry, they accept that statement as true. They’ve tried it; understand it and think to themselves: “I see your point, but this isn’t fun any more. This isn’t enjoyable. This is bad for me.”
And decide to leave.
There are a number of systemic factors for why this is happening to them, for why the software industry is an increasingly inhospitable work environment.
What software developers are facing is a case of “incentives”.
Namely, managerial and executive incentives, specifically. The industry’s incentives for those who run the industry give them a strong motivation to try to undercut labour and their workforce at every turn.
And they have been giving it their honest best.
Facebook attempted, for example, to replace a number of their software developers directly with AI tools.
They had to dial that down because the technology simply isn’t capable of what they believed it could do.
But the intent has always been there to reduce the headcount of the industry. A number of incentives for managers and executives that are lined up to push them in that direction.
The fewer people you have working for you, the fewer stock options you have to pay out, the less resistance there will be to watering down the product, collaborating with the military-industrial complex, or helping authoritarian governments control and murder their citizens. Over the past two decades, one of the more effective forces for keeping tech companies in line have been the tech workers themselves who have tried resist corrupt and unethical behaviour. It hasn’t always worked. There wasn’t enough of it. But what little there was it added meaningful friction at several important moments in tech history.
The less influence and power the labour force has, the more power goes to the executive class and the bigger their share of the pie.
If they lay people off that means there are few people working in any given organization doing the same amount of work. That means more stress and more pressure.
As bad as the added workload and burnout is, that isn’t the only change to the job.
There’s also less novelty.
I’ve noticed that even though there’s an increasing number of apps or software being deployed, they’re all either the same “AI” products or “AI” features released everywhere. Or, they are agent-generated clones of existing software.
None of it is novel. It’s almost exclusively old ideas poured into new containers.
One of the problem we have as software developers, and I definitely count myself as somebody with this problem, is that we are a novelty-seeking group.
These changes mean we’re doing more with less time, probably less reward as well, and the job itself is less interesting.
It’s also the nature of the agent automation to feel more like delegation to a clueless junior developer than automation, and that changes your relationship to the work.
In woodworking, the power tools didn’t reduce the woodworker’s emotional investment in what they made. The worker still felt like they’d made that thing out of wood themselves.
Some woodworkers might still use hand tools for a variety of reasons – some of them romantic, some of them, you know, just because it’s more fun.
But having a table saw doesn’t make you feel like you’ve asked somebody else to make the shelf, or table, or chair.
The nature of automation with Large Language Model tools feels substantially different from prior attempts at automation in coding. It’s an abstraction that feels like a delegation, not an automation, and separates you from the output.
That has an effect of making you feel less invested in the output and that takes away much the joy of coding. The flow state is also gone, the sort of focused state of mind that you get into when you’re really deep into solving a problem.
On top of the changes to the job itself, the industry itself has changed as well.
There is more uncertainty. People know there are layoffs every quarter. They know they’re probably next on the block. Or, if they aren’t, they’ll be on the block next quarter instead. They know the reasons for dismissal are going to be arbitrary. They know they generally aren’t going to be able to do anything to improve their odds.
And if they talk about how bad all of this makes them feel, the “AI” true believers will turn around and say that it’s a skill issue. It’s your fault.
There’s a lot of stress, and worry, and pressure in the industry that wasn’t there before.
I suspect that everybody’s suffering. Some are just better at pretending.
This makes people want to get ahead of the lay-offs and quit. Especially those who have some savings.
But even that lines up with the incentives of the executives who are running these companies. People who quit don’t get a nice severance package. People who quit might not be fully vested.
And even if you have savings, you can’t time your quitting based on your finances if you have to get out to preserve your mental health.
Once you’re gone, the company can hire somebody to replace you out of the vast pool of now unemployed software developers at what will probably be lower pay.
Executive incentives are aligned to perpetuate the situation and even to make it worse. From the perspective of the executives who are running a software company, this perpetually precarious situation for their employees makes up for their inability to replace software developers completely with “AI”.
Instead of being able to outsource all your software development to an “AI”, they can effectively semi-outsource it to “AI” and have the automated shitstorm managed by a group of people in a perpetually vulnerable state, who are unable to ask for a higher pay, unable to negotiate for a better position, and unable to say no to unreasonable demands.
It lets management lower costs and strengthens their bargaining position versus the labour class that they’re dominating.
This isn’t a situation that will fix itself when the “AI” financial bubble collapses.
The bubble will pop. It’s much too big to be sustainable.
But that won’t fix software industry employment because the Large Language Models don’t really need to work that well for this process. They only need to sort of work.
As long as you can find more overworked software developers, and push them hard until they burn out trying to maintain a minimum standard of quality, you can keep the “AI”-driven labour alienation going. They can use cheaper Chinese models. Or, if they switch to local models, they can make owning a high-end laptop a job requirement.
“You need to spend $5000 USD on a laptop before you apply for this job. That filters out the ‘AI’-generated resumes, you see.”
Why take on the massive costs of RAM when you can use that cost to lower developer pay by proxy?
The overall dynamic doesn’t change once the costs increase for the hosted models. Those models will just get more stagnant; they might not get updated as often; the training data will be stale; the code it generates will be perpetually on the verge of outdated; and they’ll be run with more errors.
Management won’t care.
The work of making up for that is going to be pushed on the developer adding to their load and pressure.
My worry, based on the research I’ve been doing and looking at the systemic incentives of the companies in the software industry, is that this is probably about as good as it’s going to get for a software developer at a large company in the tech industry.
This misery, this pressure, the overwork, the underpayment, the detachment from the work – this is probably what making mainstream software is going to feel like from now on.
It doesn’t surprise me that the people within the software industry who seem to be spotting the issue and leaving preemptively seem to be user experience designers.
Designers have slightly more training in looking at the system within which the user operates, and it doesn’t take that much of a shift in framing for them to understand where they stand themselves.
But software developers are coming to this conclusion as well, especially students. There was a new story recently here in Iceland that year-on-year, sign-ups for computer science at the universities here in Iceland had declined on the order of 30% decreases every year.
And Iceland doesn’t seem to be alone in this.
People are realising that this is a bad career to be in and an industry that’s hostile to its workers.
I suspect many more people over the next few years are going to be having identity crises on top of the career issues.
“Who am I if I’m not in tech? If I’m not a software developer?”
Because if you’ve built your entire identity around software and tech, around being a programmer, around being a tech person who works to enable “progress” and drive the world to the future, you’re going to have a bad time after being pushed out of that role and seeing your former industry studiously dismantle everything you thought you were building.
I don’t know what we can do, except understand and accept that this is happening.
It’s a horrible thought, but you can’t deal with change if you stay in denial.
I think it might be time – if you’re working in software – to think about a possible future outside the tech industry proper.