On the other hand, if this site was seeded initially with thoughtful participants (as I believe it was before I got here, and as I believe it continues to be by visitors who come here from pg's personal website), being popular (upvoted) rather than unpopular (downvoted) may be a fairly reliable mark of a comment worth reading. Everyone here reads a lot of the best of what is on the Web, and observes the behavior of other users here, and I think there are many examples here of highly upvoted comments (or, from another point of view, participants with high karma) being signals of good quality. It may be that there are some equally clear examples here of comments with many downvotes, all the way down to the system-imposed limit of a score of -4, or users with very little karma, being examples of poor quality.

After edit: so my assertion, in friendly disagreement with you, is that popularity DOES correlate with validity to a useful degree here, so even if not every popular comment is factually valid, and not every factually valid comment is popular, noting popularity can still be a guide to busy readers here about what to devote time to reading.

Unfortunately, this doesn't scale. It is true now and that makes HN valuable. As HN grows there may be more people who express opinions that are not thoughtful and well reasoned, and then popularity no longer corresponds with useful. I think that's the problem that pg is trying to solve.