What does it mean to develop software?
I stumbled onto this post on BlueSky the other day [[1]], mainly because people were trash talking the sillyness of both pointing at limitations and also calling this a revolution. I get that, but it is not the topic of this post.
Two things are true: 1. Claims about AI capabilities are overblown 2. AI has completely revolutionized software development, and will do the same to many other fields I'm a software developer by trade. I'm going to attempt to explain what this revolution feels like "from the inside".
— Extropian bodily autonomy insurer (@rekmarks.com) 2026-10-30T19:26:03.532Z
What I think the thread spells out in excellent detail is how software developers think about software development and about their role in the world at large. In trying to explain what the job is, a whole lot gets revealed.
It is nice to comment on this, as the account is anonymous (I will call the author Ebai). This makes it easier to I am comment on not one person, but on our field as a whole.
What is coding?
What is it like to do coding, you ask? Ebai explains:
You, the human developer, sit at a computer in an app called an editor, and type out code by hand. You are extremely attentive to detail, down to the individual bracket and parenthesis.
Nothing about this is even about making software, it might as well be about knitting or sewing, typing patterns into a machine and caring about detail. Nothing that starts with: as a developer I make software that people ultimately will have to use, for which I use code.
Care about the code, not the product
This I think is incredibly revealing:
Most of your professional pride is invested in your facility with code.
For me this is distrinctly untrue. In the work I have done—for example with Hedy—my pride came from seeing people use the software. Getting emails from people all around the world using Hedy, telling that their teaching got so much easier using Hedy. Of course, it was a very exciting challenge to build an Arabic compiler, I loved the puzzles that that work gave me, but that was not my professional pride.
However, I have come to see that in this regard our field is more Ebai than it is Felienne, most people do get their pride from their coding prowess.
Thinking about humans = thinking about developers
In so far as developers are encouraged to think about how other humans will interact with their software, these other humans are other developers:
You think about how to make sure that other humans—including your future self—can read, understand, and thereby maintain your code. Mostly, you're reading code that other people wrote and trying to figure out if it's good or, if not, how to make it better.
How the user of the product will interact with the software is distinctly out of view, not in the minds of the programmers, but are abstracted away:
Let's say you're fixing a bug in your software reported by a user, a huge part of the job. First, you spin up the app or whatever in a dev environment to reproduce the bug.
I know that for programmers this sounds 100% normal and logical, but this is such a weird way to deal with people that use your stuff. The first step is not to talk to the user, to observe them, send them an email, give them a ring and ask in more detail what is up. If that is anyone's job, it is not yours (it might be a product owner or a QA person). And now I get it that Meta-engineering can't call their users if they can't login to Facebook, but even for small companies where it would be, it is simpy our culture to not do this. Users aren't to be interacted with, their needs are translated into Jira tickets and that is what we work on.
Now this was all the old way of working, what are software developers doing now?
Ebai explains:
A bug that would've taken hours or days now takes maybe 30 minutes to an hour of my attention at most. I can and do have several of these workstreams going at a time.
Bugs don't require attention anymore, by which they means attention to the code, not to the user or the product as that was never attention that was spent anyway.
What is software even for?
Now here comes a quote for the ages. Sofware is just a thing that solves a problem, in the view of developers:
All commercial software exists to solve a problem of its users; to "do the thing", whatever it is. If the software does the thing, it works, and that's the end of it.
This shows the extremely limited view of a class of people that is deciding what the world looks like. Now here are some questions that my field nevers asks. What problem does Facebook solve? Chatting with your friends? Does it solve that? And how about the problem of 'there is too much democracy in the world'? Does it solve that problem to our satisfaction? Does YouTube solve the problem that men were getting a bit too exited about gender equality? The view of software are simply solving problem and "that's the end of it" is exactly what got us into the AI-mess. Ah we were simply trying to make a machine that can mimic text really well and look: is mimcs text really well.
A distinct lack of class consciousness
All fun? Ah no, Ebai is not feeling happy about this progress after all:
But LLMs, critically, take care of not just the drudgery, but also the most interesting, prestigious, and mentally demanding parts of the job. Fancy algorithms. Architecture and systems design. They know the literature better than you do, and they can put it into practice.
Our jobs will be less fun!
To repeat myself, you still need people who can talk to the LLMs about the actual software to make them do things. But to the extent that remains to be true, it's hard to imagine a future where there aren't fewer and worse software engineering / other white collar jobs.
Well look at that. Some people (not you) decided that you have to use technology to do you job, and it is making the job worse. Less pay, more drudgery. This has never happened to anyone else in the world has it? Well of course it has, it has even happened to direct colleagues of developers now feeling the pain or automation rather than the enjoyment of building it. But in those days, they never cared or worried as they were getting a nice paycheck out of it.
One would think that this situation would lead to a spectre haunting the software development world: Hey, wait a minute! This is no fun! This is exploitation! However the world view of programmers is such that they probably not even know what I am alluding to here. They have simply never learned to think about the world in a larger way that the code base. Not even the customer is in full view, let alone the whole world!
Who is in power?
Something else entirely out of view is the power dynamics. Editors, compilers, they were ours, software that we owned, in many cases collectively as open source projects. Language models on the other hand as we have them now are centralized, and using them takes knowledge from developers and moves it to large companies. So using an LLM to program has side effects (har har har) that are not about the code that it produces, but about the world that it will. But this too remains entirely out of scope (har har har).
What a disapointment for a field that said to care about openness, that claimed to hate proprietary code and making money selling software not even that long ago...
In case this post was fun and interesting... This is exactly the type of thing I talk about in my new book Vierkante Ogen. Sadly the book only exists in Dutch for now..., but maybe one day it will be translated!
[[1]]: Read it as one post with a nifty service that aloows you to turn a BlueSky thread into a post by simply a t in front of the URL. Smart!
Member discussion