A coder's biggest hit timing looks like a lottery, and ChatGPT changed nothing.

In science and the arts, a person's biggest hit usually happens at a random point in their career, like a lottery ticket drawn from the things they make. At the same time, some people simply tend to make bigger hits overall. Researchers wanted to see if this same rule applies to open-source software creators, and whether generative AI tools changed it. They tracked millions of coders on GitHub and measured a project's impact by counting how many other software packages depend on it. They found that open-source careers follow the exact same pattern. Peak success arrives mostly at random, though with a slight lean toward earlier work that grows over longer careers. Lasting personal skill explains only about a fifth of repeated impact, while temporary momentummomentumA temporary effect in which one success makes future successes more likely, then fades. explains the rest. Finally, generative AI did not change this career structure at all. Individual success patterns stayed the same.

Why it matters

Open-source platforms, employers, and funders often judge coders by their past hits. This paper suggests those hits are mostly draws, not reliable signals, and that generative AI hasn't changed the underlying career structure. So reading a contributor's track record as evidence of lasting skill needs care.

Questions answered