The fate of “small” open source

LLMs can now generate small JavaScript utilities like the ones in blob-util, so the author thinks fewer developers will search npm, install a package, or need a library for simple blob handling. That reduces demand for small libraries and also reduces their value as a way to teach JavaScript and problem-solving.

He argues that open source is not over, but its center of gravity is moving. He sees more value in larger projects, novel research, inventive tools, and niche topics that an LLM cannot just produce on demand, and he points to work like fuite, memory-leak posts, and new frameworks such as Ripple.js as examples.

Reading notes

#

• blob-util is a small JavaScript utility library for Blob-related tasks and still gets millions of weekly downloads.
• The author says an LLM can generate similar Blob utilities on demand, which makes a package like blob-util easier to replace.
• He thinks using an LLM can reduce extra dependencies, but it also changes the role of small open source packages.
• He says blob-util was written with a teacher’s mentality, with a whimsical README meant to help people learn JavaScript.
• The point of the library was not only to solve one problem, but also to teach people how to solve future problems.
• He thinks the AI approach prizes instant answers over teaching and understanding.
• He says llms.txt fits the trend of pointing an agent at documentation instead of asking people to read prose.
• He concludes that small, low-value libraries are over, even if open source itself is not.
• He suggests the remaining value of open source is in bigger projects, more inventive projects, and niche topics outside an LLM’s training data.
• He gives fuite, memory-leak research, and Ripple.js as examples of work he thinks still has room for human novelty.