GoKawiilThe other month I finally ran an experiment we had been postponing for over a year at .txt .
The goal was to test our structured-generation algorithms and their open-source counterparts, replacing the naive “does it accept this string?” with something closer to the real problem: “does it produce the right token distribution?”
The experiment kept coming up in conversation, then returning to the roadmap. Last month, I spent half an hour explaining the method to Codex. A few hours later, it had produced a working first version. That’s all it took.
Coding agents are transforming the way individuals write code. In a way, that has already happened. And yet I remain skeptical of the story people usually tell about what this means for software as an industry: that individual productivity gains will translate into the industry moving substantially faster. I have been stuck on that tension for months.
Time to re-read the classics.
Impactful software tends to be written by many humans that need to collaborate.
Discussions on coding agents almost exclusively focus on the individual productivity gains. But collaboration is the interesting unit of analysis.
This is definitely not a new idea. Fred Brooks wrote about it in 1975 in The Mythical Man Month, and it was not new then; Gerald Weinberg introduced the idea in 1971 in The Psychology of Computer Programming. Software is what’s left over after a group of humans finishes negotiating with each other about what the system should do. The code matters, but it is the residue of the harder work, not the work itself.
For fifty years the residue was expensive enough to keep our attention on it. Typing speed, language design, framework choice, IDE plugins, code review tooling were all about reducing the cost of the residue. With coding agents the cost has fallen far enough that we can see what’s underneath: people trying to agree.
Negotiating, agreeing, communicating the shared picture of what we are building has become the work. And it’s just as hard as it was.
... continue reading