My interpretation on utility of upvotes
The way I use HN is this - I browse the Front Page and the "new" section as often as I can. I go through a couple of pages (as permitted by time) for both sections and upvote those that are relevant to whatever I may be learning at the time. Anything I upvote (threads and comments) are saved, providing me with easy access to all of the information within vicinity.
This is the most useful thing about HN, this ability to collect knowledge in an organized manner with possibly most minimal effort. And I am afraid that most of us do not realize this.
If it was so (that we use upvotes similar to how we use bookmarks) - then I personally would expect to see a broader distribution of votes among topics that would parallel the necessarily broader range of interests in any large collection of human beings.
Instead, we observe a centralized tendency where predictable topics garner hundreds of votes, while most languish with 1 or 2. This is a result of muscle memory from using Facebook's like button.
I'll try to make time and add some visual evidence of this in a few months using HN's API.