if visual scores create problems, why not just obscure them, and use an algorithm to find a better relative popularity between the comments; and show a simple 3 level scale (spam/normal/popular).
If you still need more depth, add a fourth level to the scale.
I'm the same. It is nifty and a cool trick... but didn't deserve to enter the mainstream (in it's current form). But there are underlying motivations that warranted the change before it was ready for mass deployment.
One of the best ways I've found is to volunteer your time. This makes sense for a few reasons.
A. The people you are with are also likely volunteers (opensource, event/meeting organizers, etc), which means you will meet people with a favorable attitude/mindset/values. The people with the most to give are the ones with the most value.
B. You have the same interests as them.
C. Volunteers are commonly a very diverse spread of people (experience/jobs/projects/goals/etc).
D. Meeting people like this makes it easy to keep things casual, or move into a partnership type relationship.
I generally try and batch read HN (via rss) daily, and sometimes I feel that I am missing out on new comments (made after I finished reading and closed the tab; most commonly the new/recent posts)...would be nice to avoid bookmarking and searching for the new comments