I would:
1. Look at the top 50 HN contributors, looking at their profile (and possibly checking up on some of their posts) to figure out if they are already committed to some project or company that would interfere with working with you.
2. Auto-select any names you recognize as people you "like". You know they're smart, and not merely prolific at posting to HN. Also look into any names you know you like who aren't among the top 50.
3. Figure out your own strengths, and, based on your lists comments, narrow down to a handful of folks with contrasting strengths and interests. A team of hardcore developer that churns out code like a machine, and a generalist that doesn't mind closing vim/emacs when other tasks need doing who also happens to have either design skills or at least an eye for great design, would be my ideal combination (but this may be reflective of my own limited experience rather than any blinding insight).
4. Contact the short list directly, one or two at a time, with a "Hey, would you be interested in trying an experiment with me, based on RFS #X?"
5. Profit!
An HN thread sounds just as reasonable an idea as any other.
That said, I don't think this is a particularly good idea. The best way to find out if you'd make a compatible team is to actually go and work with someone. When I've turned down startups, it's often not because I thought the idea or other team members were bad (often times, they're very good), it's because I don't feel any particular passion for working on that particular problem with those particular people.
No, but you've gotta filter somehow. There are thousands of HN users. At least the ones with a lot of posts are a known quantity. It's the same reason I like to hire developers that work on Open Source software. They may not be the best (or they may be), but I know with much greater accuracy what I'm getting.
I personally wouldn't start a startup with someone I met on Hacker News, and didn't know in any other way. But OP wanted to know, and so I told him how I'd do it if I had such an inclination.
It seems like something that could happen, but making it a community (as in HN) effort seems like it has too many issues.
The community could then put it to all poll.
The other option is to simply have a database showcasing every interested applicant along with contact info. Then it would be up to them to find a team that works well together.
