I think I do the opposite: I think I value candidate potential more than you, because you propose screening based on how cordial or fluent a single interview with their prospective team is, and I don't buy for a second that you get a realistic picture of how well someone will communicate or gel with that team in the course of a single adversarial interview.
How many tech interviews are one session long anymore (especially for the good ones)? Like I said in my post above this one, I'm arguing for the point after candidate potential has been acknowledged. If you know a candidate can do his/her potential job well - then I think good team fit is the obvious next step. If nobody wants to work with a potential candidate - then it will clearly be a detriment to a team (short term and most likely long term as well).
I think the problem is that everyone on HN thinks things are binary, when practically every scenario in this universe is not. You can't just eschew potential team fit because you want a rigorous hiring process that depends only on candidate qualification. The opposite holds true as well. Here are the two things you need for an awesome team:
