I feel like 1/10 comment I make on HN are about this.
So merged PR were until LLMs a good proxy for the ability to code and contribute to a software project. Consequently they were used to estimate if a candidate was potentially good for a position. Merged PR on popular project were thus precious credentials one could "trade" for potential work. Since then the desire to provide PR changed from contributing to a project for its own sake, to make the actual project progress, to signalling.
A new proxy must be found to establish the ability to contribute to a project.
There is, it’s being able to tell bullshit from actual cosplayers.
Ladybird’s blog post is arguing that it used to be tolerable to spend the amount of time evaluating this for every PR, now it’s just unmaintainable for their effort-time.
I'm not sure. I didn't think about this seriously in an adversarial context.
What I can imagine though is that
- the value of merged PRs might drop (as it's not a good metric anymore)
- while the cost of submitting them goes up (as price per token is radically changing, e.g. Github announcement just this week)
so maybe it's also only temporary.
Also if all this is correct, including the value of PR on popular repository being more important, then the long tail of projects might not have to worry about this.
I feel like 1/10 comment I make on HN are about this.
So merged PR were until LLMs a good proxy for the ability to code and contribute to a software project. Consequently they were used to estimate if a candidate was potentially good for a position. Merged PR on popular project were thus precious credentials one could "trade" for potential work. Since then the desire to provide PR changed from contributing to a project for its own sake, to make the actual project progress, to signalling.
A new proxy must be found to establish the ability to contribute to a project.