When The Readme Is No Longer For Humans

Programming is taking a turn most people aren’t ready for. I opened the README of fastrender (wilsonzlin/fastrender), a vibe-coded browser built from scratch (that works), expecting orientation, context, and a clear map. Instead, I found something else entirely. The README isn’t written for humans, it’s written for models. To navigate the project you need an LLM, and running it or understanding it without AI feels almost impossible.

This is a sharp break from what READMEs have traditionally been. They explain what the project is, where it stands, how to install and run it, and often show screenshots to help form a mental model. Here, that layer is gone. There’s no overview designed for human consumption, no gentle entry point, no narrative guidance.

Education takes a hit as a result. There’s no guided path into the system, no explanation of how the pieces fit together. You either read raw code or rely on an LLM to translate and contextualize it for you. Human-to-human knowledge transfer is no longer the primary design goal.

That shift isn’t accidental, it’s a signal. Code is no longer mainly written to be read by other engineers. It’s written to be navigated, extended, and regenerated by machines. Humans are moving into a supervisory role, approving and steering rather than deeply reading.

Still, there is a bright side. People are building competitors to existing engines. This breaks the monopoly that Google for example, has on tech.