I think that personalized ways or algorithms to sort comments could be a really good thing, if only for the fact that experimentation in this area could possibly lead to something new and interesting and a competitive advantage for hubski.

In order to compete with already existing websites (i.e., reddit), hubski has to differentiate itself in some significant way. No one's going to switch from a really popular site (because simply being the largest community gives it network advantages and more diversity, more ideas, more everything that's a draw) without a significant reason to.

What the comment sorting system needs to do is not just highlight comments that the user agrees with, but, rather, ones that interest the user. I think a successful implementation has to recognize that users follow other users for different reasons: some for the links they share, the comments they make, sometimes because their interests completely align, and sometimes because they make controversial or unusual, but high quality, comments.

A system that only bumps up the comments of people you follow a bit is probably going to be too simplistic- it might not work because it doesn't differentiate these motivations finely enough. Development of a significantly more robust recommender system would be required. It would be significantly more complicated, and harder to do, but would be the real competitive advantage hubski had towards making it a superior place to discover content, and effectively, the only real incentive to draw users away from established sites.

--

I mean, hubski was built with this idea in mind anyway, otherwise why do we even have a "following users" feature? It's founding idea on some level had to have been that content aggregation can be made better by better personalizing it, by better utilizing a user's evaluation of other users when we choose to implicitly vouch for the content they bring by following them.