Software Engineers are Obsolete

In my first interview for a developer position, I shared a link to my personal project with the interviewer. It was a website for learning how to program. I created it from the ground up. I built the PHP app, designed the database schema, made a nice design to tie it all together. I wrote down my process, and it became the first tutorial on the site. Then I collected tutorials from all over the web and displayed them on my website, which acted as a portal.

There was a section for PHP tutorials, for Ruby on Rails, for .NET, etc. Each one individually curated by me. My interviewer was so impressed. I got the job.

Later, I added a section where anyone could submit their own tutorials. It was fascinating how quickly people found my website and started submitting links. The tutorials were coming in so fast that I removed the verification system and let people upload links directly. But then my mind wandered. What if I start a blog? Yes, I had another blog before this one. I built an entire blog engine from scratch.

A colleague found my blog. He was so excited that he shared his own with me. At lunch, we would discuss ideas, and that same evening after work, we would buy a domain name and start a new project. We shared tips and tricks on how to rank on Google. We had a skill, being web developers, and we took full advantage. When we had an idea, we would fire up our computers that same night and build it.

Friends and family would come to us for validation. We were the ultimate deciders of what was a good idea and what was a bad one. We were the gatekeepers. We knew how to program, and nobody outside our circle could say otherwise.

Now, friends and family don't come to us anymore. They go straight to ChatGPT, and it tells them their idea is brilliant.

They launch their favorite AI agent, which builds their entire product from a single prompt. Some of them even manage to host it on the web, accessible to the world, and they are seeing their first customers.

People who used to confuse Java with JavaScript now tell me they have a platform. People who don't even know what programming is are standing at the forefront of software innovation, advocating, evangelizing, and making money. This skill I spent years honing has been made obsolete by everyday people.

We, the developers, are no longer the gatekeepers. In fact, now we need to keep up or risk being left behind. Some commenters online tell me I'm just jealous, that I need to embrace progress.

I don't want to be obsolete. I'm on openclaw, moltenclaw. I have accounts on all the video generation websites. I have accounts on ChatGPT, Claude, Gemini, and Mistral. Just as I'm getting a hang of one tool, my friend who works in a warehouse tells me, "just use Perplexity for that." But Perplexity isn't enough, because another friend says GenSpark is better.

For some reason I can't sign into my Manus account anymore. And apparently, to get the most out of it, I need to get Meta Ray-Bans. Everyone is empowered, no one needs me, and that's that. The developer is now obsolete.

But then, I opened LinkedIn. My peers, fellow developers who for some reason all have the word "AI" in their job title, are saying the opposite. "Developers are not losing their jobs to AI," they say. "Developers are losing their jobs to other developers who use AI." They are vibe-coping to the max.

The history of technology has always been a story of nearly missing out. I remember another job I applied for and totally didn't get. The company had moved all their client-facing apps to Silverlight. If you're wondering what Silverlight is, you might understand why I chuckled when the interviewer described their plight: they were struggling to find developers to help them migrate to HTML and JavaScript. I'm fairly sure that chuckle is why they never called me back.

It's one thing to embrace new technology. It's another thing entirely to put all your eggs in one basket. Companies are betting everything on Silverlight. Sorry, I mean AI. Without thinking through what happens if things don't pan out.

AI has lowered the barrier to entry. That's a good thing. More people can now bring a fresh pair of eyes to the software engineering field. But there's a problem. Those new entrants won't become better engineers over time. Why? Because they are not writing code, not reading code, not debugging code. Their growth path, with time and experience, is to become better prompters.

What this means is that, amid all the noise, my role as a software engineer may seem obsolete. But in the long run, we will be back to square one, where engineers writing code with their own meatware will hold all the cards. These are the people who learned the hard way: by reading documentation, by debugging broken apps, by having their seemingly perfect Stack Overflow question closed as a duplicate. These are the engineers who will hold the keys to software. Not because they're guarding secrets, there are no secrets. It's simply that the new developer is not, and will never be, interested in learning.

---

While we pride ourselves on producing more software than ever, it doesn't take long to realize that software is never truly finished at delivery. It has to be maintained. It's strange, computers whose entire purpose is to repeat the same process over and over, perfectly, somehow manage to degrade over time.

My tutorial website, seemingly working fine, returned an error when I visited it after months of neglect. I restarted all the services and brought it back up. It was now full of spam and NSFW URLs.

An application that worked perfectly yesterday is broken today. It could be a memory leak, unexpected input, or just users with fat fingers. Your completed application is suddenly incomplete, and you have to fix it.

In an ideal world, we wouldn't keep producing more software. We would have working software, and less of it to maintain. AI thrives on quantity. If you need me, I'll be in the back, patiently waiting for you to realize you can't prompt your way out of a Silverlight migration. My rates just doubled.

---