I'm surprised to see so much focus on the comments. I think the front page is the primary problem. I wonder how much of the comment problem would fix itself if the front page had more signal and less noise. Maybe that's naive.
Personalization is most certainly the wrong answer for HN, but when I thought last week of my ideal solution to the problem this is what I came up with:
1. Personalized weighted point calculations. Each vote is multiplied by a number which ranges from 0 to 2 in .1 increments. Everyone starts at 1. Everyone who up-votes the same story as me gets .1 added to their weight. Everyone who up-votes a story I down-vote gets .1 removed from their weight.
2. New users can't submit stories during an initial probationary period. They also start out at .1 and get .1 added to their weight each week they're active on the site. After 10 weeks of activity they can submit stories.
3. Weights are applied to comment rankings but not derived from them. Comment rankings also need to be much harsher. We want fewer comments and higher quality comments. Maybe ((weight*2) -.5) for calculating comment points. But the floor is always 0.
I don't know if ideas along these lines could be successfully de-personalized.
These ideas I think are mistaken:
1. Any notion of explicit control. Such as: hard coded karma values, comment size, non-bayesian content filters, etc. (New user probation being the 1 exception)
2. Anything based on unweighted karma values.
These ideas I'm suspicious of:
1. Economic solutions. They strike me as having the same problem as micro-payments. I don't want to have to think about how I'm spending my alloted money each time I up-vote or down-vote. Also the purpose of money is trading, not creating artificial scarcity. And even assuming the goal of artificial scarcity is worthwhile (I don't) then it implies some kind of hard-coded explicit control to determine purse size, which will always be a mistake.
One way to counteract this is to have the new user match, say, pg's profile.