>>>> "RWMJ" == Richard W M Jones
RWMJ> I think we'd want to consider the security implications of
RWMJ> accepting packages after only automated review.
New packagers would need to be sponsored and so a human would still have
to be involved at some point. (At least I certainly hope that nobody is
thinking of removing that step.) Existing packagers could just push a
malicious change to one of the packages they already maintain; that's
not a change from the current system.
Now, it happens that I strongly disagree with the idea that we should
let anything more than a narrowly defined set of new packages into the
distribution without human review. I just don't think the security
argument is one I'd use personally to argue my position.