The Lisp curse, now coming to your ecosystem

In a March 2026 essay now circulating on Hacker News, Tim Zöller argues that AI coding tools could weaken open-source ecosystems the way Lisp's flexibility is said to have weakened its own. The "Lisp curse" is the idea that Lisp makes custom solutions so easy that programmers build their own instead of using shared libraries, so few common frameworks ever form, and the poorly documented code they write alone pushes the next developer to write their own as well.

Zöller takes the claim he keeps hearing, that generating code with a large language model is now easier and cheaper than learning an existing library, and asks what follows if people act on it. He expects two effects. Fewer developers would use open-source libraries, so maintainers would get less feedback. Fewer developers would publish new libraries, because the hard part of open source is maintenance, such as versioning, backward compatibility and hosting, and someone who can generate a module tailored to their own stack has little reason to take that on. He argues the two effects could reinforce each other: with fewer libraries for new technologies, people would be even more inclined to generate their own.

He does not accept the premise. He would rather rely on proven, well-tested libraries than generate hundreds of lines that solve one local problem, and calls "not generated here" a new version of "not invented here." Still, he expects some version of the scenario to arrive, perhaps less drastically than some thought leaders proclaim and more by accident than design, with open-source collaboration declining to some degree. The essay offers no measurements of library use or contributions; it is a reasoned prediction, and its value is in naming a mechanism that could be checked against such data.