Fuck AI

So... Algorithmic profiling?

Not exactly loving the sound of that.

Stay your pitchforks kids, this is just a logical response to a problem we all have. It may not be perfect (and lacks transparency), but so far it's the best idea I've heard to solving the problem of what is effectively spam in the FOSS community.

It doesn't seem unreasonable to me that a project should be able to vet would-be contributors based on a variety of metrics. We do this with spam blockers in email all the time and while imperfect, is better than leaving email unusable.

The risk of false-positives though is definitely something to think about. Someone shouldn't be blocked from submitting a PR just because their account is new, but a system which combines a series of metrics:

• account age
• size of PR
• number of similar PRs in the wild and the size of those PRs
• accept/reject ratio
• any other sort of reputational flags that might make sense

...into a calculable score (as this one appears to do, again: better docs, more transparency please) could be quite valuable.

My main complaint is that it's on GitHub. Can we all just get the fuck off that dumpsterfire already?

Not a single mention in the README of how the profiling works and what signals it responds to.

This is likely to just exclude new contributors and damage the pipeline for beginner programmers to get mentored and become contributors.

If what you really want is to shut out AI contributions then there must be much more inclusive and human ways to do this.