I want to talk a bit about where we came from in software and where we are now. Not the industry in general, although you can read it that way too. More what happened in my own life. It has been 20 years since I started writing software professionally, and a lot has changed in my job and in how I think about it. This is mostly me talking to myself.
At university we wrote our exam answers on paper. Code, on paper. My C professor gave us problems in the midterms and finals, and we solved them by writing the code by hand. You can imagine. We tried to be careful with indentation, pushing the lines a bit to the right. Semicolons, brackets, curly braces, all on a sheet of paper, and then you handed it in. The professor and the assistants read those papers with their eyes and graded them. You had to get the syntax right, and you had to write it hoping it would compile and run. Nobody was going to run it. Hello to him from here, if he ever reads this.
Before that I was already making websites. I started with HTML, using Dreamweaver and FrontPage. Not really code, websites. I made some money from it. Later I built web applications in Flash. I also bought and sold domain names, a couple of them for a good price. That was the world I came from.
When I started working, we learned by watching the screens of the older guys. You pull a stool next to them, sit and watch. Which shortcuts they use, which windows, how they move so fluently between them. You pick things up like that, and after a while you can write code on your own.
"Did you type it letter by letter?"
It is September 2026. A few weeks ago a young guy asked me something in the middle of a conversation.
"Did you use to write code character by character?"
At first I didn't understand the question. I thought he meant something technical and I tried to figure out what. Character by character, what does that mean? Is this about encoding? I couldn't work it out, so I asked him.
"I mean, did you type it letter by letter?"
When he said that, I had a strange moment of clarity. Yes. Of course the answer is yes. But I had never thought about it like that. Not once did I think "I am pressing keys and writing code letter by letter." Lines of code, sure, that was a unit we were used to. Keywords, naming, all of that. But of course that is how we wrote it. When you think about it now, how else would you write code?
Nobody writes it anymore
This month I watched DHH's opening keynote at Rails World on the YouTube livestream, and it said almost exactly what I had been thinking.
At 37signals they have gone "pencils down" on writing code by hand. It is now, in his words, an exceptional state, like seeing a bug in Sentry. Something went wrong, why couldn't the agent produce what we wanted? Then he asked the room how many people are still writing material amounts of code by hand every week. About five hands went up. In a room full of Rails developers.
The talk was called "Look at all the things I'm not doing". That line is actually 21 years old. He said it in 2005, showing how much code Rails let you skip. Now he says it again, about all of it.
So the machines write the code now. Agents, LLMs, whatever we call them. We just tell them to write. That will probably change as well. We still use the old programming languages, maybe those will change too. Right now we are trying to fit the old structures to this new way of working. We use version control. We try to keep code review. There is still this thing called a diff. Will we need it? There are a lot of questions like this on the table.
This is how I work too. I think I haven't written code by hand for about a year. And compared to the year before, I didn't measure it, but I probably produced 10 or 20 times more code. We say "produce" now, because we don't write code anymore. We have it written.
Let me talk only for myself here, I'm not trying to lecture anyone. Code is losing its importance, in quotes. Of course it matters how code is written, which standards it follows, how it performs. Yes. But even when the agent doesn't write it that way, the cost of getting it into that shape is not what it used to be. A refactor, a performance improvement, applying some pattern you want, these take minutes now.
And when you start seeing it this way, that we are producing the final result and not the code, the colour of the job changes. The definition changes.
What is my job now
Before, like every developer, I wrote code for days, nights, hours. Like that young guy said, character by character. It worked, it didn't work, you debugged, you focused, you got into what we call the zone. Long hours, no breaks, no interruptions, and at the end the goal was pieces of code that did what we wanted.
Of course that was never really the goal. The goal is the same today. Build a product, solve a problem, build a system. Improve something, fix something. That hasn't changed. But before, we had to focus much more on the code. Now we focus on the problem instead. Actually we don't even need to focus on the problem that much. You say what you want and it happens.
Is this vibe coding or not, I won't go there. Maybe someone who knows code and someone who doesn't work the same way today, maybe differently. I don't know if it matters much. I'm trying to think and talk as someone who could write code before there were LLMs.
I wanted to build products then and I want to build products now. The difference is that back then I had to write the code myself, because I didn't have the money, the resources or the network to get someone else to build it. Or a company was paying me to write code. Not for any other skill of mine. They paid me a salary and I pressed keys. We were learning and writing at the same time, very close to the technology.
What I have wanted for a very long time is to build my own products. If I rewind 20 years, these things took forever. I struggled to find the time, I never had enough resources. Today I can tell an idea to the agents and have a working version in a few hours. For me this is an incredible change.
What I can do, what I can't do, what I will be able to do, all of it changed. A small example. I used to not start a project just because it didn't have a logo. That was a mistake in my thinking, you obviously don't need a logo to start a software project, but I know I'm not the only one who thinks like that. I need a designer, I need a logo, and then I'm trying to convince a designer. Today this problem doesn't exist. And if you still want to start with the logo, you can get one that does the job in a few minutes with a design tool. Almost for free.
Everyone can build now
This opens a lot of doors. The amount of code written and the number of apps built in the last year or two has gone up a lot. For the App Store there are numbers now: app submissions roughly doubled in the first half of 2026, to around 560,000, while downloads grew about 2%. Twice the apps, the same number of people downloading them. Everyone is building their own app, and instead of paying for something, people say "I can build this myself".
When I put myself into this picture, I see it like this. I could already build whatever I wanted. It took weeks, maybe months, but I could do it. Now this skill has been given to everyone. There is a tool in everyone's hands that has much more than the experience I spent years collecting, and it writes better code than most developers on the market.
DHH is very happy about this. In the same keynote he says agents will let everyone become a programmer, and then: "Who's afraid of a little competition?" I agree with part of it. I don't think we should be afraid. But I have a question, and I know this part may annoy some people, so let me say it anyway.
We could already do this and we were doing it. We chose this path. But when people from completely different professions come and say "I'm going to make money by building software", it feels a bit strange to me.
Why strange? Because this wasn't your job. Building products or software was not your job. I don't want to name professions, but you were doing something else in life. Did you give that up once you could produce code? That is what I'm curious about. Did you not want to do the job you were doing? Or, because you used to need people like me, you couldn't build the ideas in your head, and now you can, and you jumped into the "get rich quickly" dream that developers have been living inside for years? I really wonder. Why is everyone leaving their work to build software right now? I don't know.
What do I expect? I have no idea what will happen, but I think there is a hype here. Yes, everyone can write code now and everyone is writing code. But after a while these people will also see that it is not that easy, that writing code was never the real problem. We developers have known this for many years. You wrote the code, fine, then what? You shipped the app. Who is going to use it? How will you sell it? How will you keep it updated? And more than how, is this what you actually want to do? We wanted to do this, that's why we did it. I'm not sure the new people want it. They had other jobs, and looking from above, they are changing careers.
Maybe they are doing it on the side for now. They will probably see that it can't be done as a side thing either. So I think this crowd will get smaller. "Let's do the job we know. Software isn't for me after all. I don't want to deal with all this." Okay, you don't see the code, you have nothing to do with the code, but in the end you still have to look at the final product to put it in front of users. If users are going to use it. It is an interesting topic for me.
There is a small story in the keynote that fits here. In the spring, 37signals had their designers vibe code the final features for Basecamp 5. Each pull request looked reasonable on its own. Together, 20 or 30 of them left the architecture "like a Swiss cheese". To be fair, DHH says the models got better a few months later and that it would probably work now. Maybe. But someone still has to look at the whole thing and notice the holes.
The edge I lost
By the way, what happened to me? Maybe my biggest strength was that I could build software fast. MVPs, products at MVP scale, with Rails, I could build in days what took other people weeks. People found that impressive.
I have lost that edge now. I still have it, but everyone has the same weapon in their hands. Everyone can use it.
DHH told a story about this that I keep thinking about. His great-great-grandfather was a portrait painter who spent years on a single painting of the Danish royal family. Then the Kodak Brownie came out and photography became cheap. Painting reality as exactly as possible stopped being a skill people would pay for, and painters like him had to find something else to do. DHH calls Opus 4.5, released in November 2025, "the Kodak Brownie of our era".
Of course I'm sad that I lost it. But new doors are opening for me too. Because I have done this for 20 years, I think I see the opportunities in front of me differently. I know better what to do and what not to do.
Recently the CTO of one of the biggest e-commerce companies in Turkey said something like: "I wish all this had happened 20 or 30 years ago. We would have explored it more comfortably, played with it more. Having this come into our lives after 50 is tiring." I agree. I also wish AI had arrived earlier in my 20 years of writing code character by character, and I had written less of it. But these things happen when their time comes, and it has given all of us a strange adventure. I'm sure in a few months something completely different will come out.
Moving away from code
I think we will move away from code completely. The best example is again from the keynote. 37signals have an email product called HEY, which is a web app. They are now rebuilding it as six native apps on the front end, and moving the backend, which is really a mail server at its core, to Rust. DHH hates Rust. He said looking at it is like pouring acid in his eyes. But it is very fast, and their rough calculation is that the new backend needs 99% less CPU and 95% less memory, and HEY's peak traffic could probably be served from a single Raspberry Pi. The agents write the Rust. His words: "I don't know any Rust at all. I consider that a feature." This is someone who gave his life to writing beautiful code, and he has no relationship with this code at all.
He also talked about software principles. His point is that abstractions, and the idea of not repeating yourself, were a trade-off. When thousands of agents are changing a code base, abstractions become choke points, and the price of repetition and of keeping things in sync has dropped to almost zero. I would push it a bit further. I think many of the concepts we learned, abstraction, polymorphism, coherence, exist because humans need to understand code together. They are practices for several people building one thing. Are they useful for agents? Instead of don't repeat yourself, maybe some things should just be repeated. Is it better for the agent to write it again? Do we need the abstraction, or is it better to continue with concrete classes? I think we are at the start of redesigning the methodologies we know from the ground up.
At the beginning I mentioned version control. I think the systems we use now will have to change as well. The way we write code probably needs to change. Maybe it already has, we just don't know, because we don't read what the agents write.
Maybe the languages too. I'm not an expert here, but the languages we use were designed for humans. Okay, LLMs learned them, they write them well, and we can read them. But if we are not going to read the code anyway, maybe the models will write in languages that suit them better. I don't know. Right now we write code in English, or in Turkish. DHH said it nicely: "English is a better programming language than Ruby." And this from the Ruby guy.
Would I do it again?
Where we are now is unbelievable to me. I worked really hard to become a developer. I knocked on a lot of doors, I read a lot, wrote a lot, thought a lot. I was very unhappy sometimes, and I had very happy moments. It was a real fight. I'm proud of it. But when I look at that fight and then at where we are today, it is hard to believe.
DHH said on stage that he has "retired from being a professional programmer". He isn't sure of the exact date, somewhere around March. And he says we should look back at the years of chiselling code by hand "not with regret, but with joy of what was."
I like this, and I mostly agree. But I can't say it is only joy for me. There is some sadness in it too.
If someone had told me 20 years ago that in 20 years I wouldn't be writing code at all, would I still have chosen this job? Yes, because I enjoyed it. But I would probably have looked at it differently. As something more temporary. Instead we gave our whole lives to it. What is good code, what is clean code, how do you write it, how do you read it, how do you not write it. And now we are in a very strange place.
Of course our output went up a lot. The other day I shared a screenshot. On the day Opus 5.5 came out and the day after, I think I did something like six months of work in those two days. Had it done. Had the code written.
DHH put a slide on the screen that just said "What a time to be alive". He meant it with full excitement. I mean it too, maybe a bit more quietly.
Interesting times, let's say.
I recorded this as voice notes in Turkish while walking around Istanbul. I transcribed them with Whisper (large-v3-turbo), then translated and edited the text into English with Claude. The thoughts and the story are mine. It is the first of a few.
