Rhetoric works to win debates because debates are time-limited (and the audience's attention is both time- and attention-limited), and so you can win just by e.g. preventing your opponent from ever getting a word in edge-wise, or forcing them to answer for trivial side-issues and therefore never have time to say what they really want to say; etc.
What I'm railing against here is the https://en.wikipedia.org/wiki/False_balance: the rhetorical rearrangement of non-equal evidence into a debate presented to the audience as if each side had equal support behind it.
And, make no mistake, that sort of rhetorical rearrangement is what discussion forums with comment voting are doing when the community rearranges the comments with upvotes/downvotes: they're curating which comments appear first (and thus which comments appear at all, to those without interest in reading the whole page.) They're changing the impression of aggregate weight-of-evidence someone gets by skimming the debate.
But my claim is that this is not a bad thing, when it is done in the goal of presenting the debate accurately. If one side of a debate has little evidence, but more voices, those voices should be rhetorically damped down to compensate, until the force of their combined voice matches the volume of the evidence on their side. That is the proper goal of editorial curation in e.g. news journalism: to let authoritative primary sources speak loudly, secondary sources quietly, and kooks and rumormongers not-at-all. And it should, IMHO, be the proper goal of a community's editorial curation of a comments section, as well.
I think Carl Schmitt, for instance, should have been put in prison. His ideas are horrible. I still think they are worth reading.
IMHO, on an ideal platform for debate, the audience's experience of the debate should itself resemble the statistical distribution of opinions held by the debaters.
I.e., if 80% of people think X while 20% of people think Y, then comments saying Y should not make up 90% of the debate by any metric (number, word-count, above-the-fold representation, etc.) If Y did take a majority share by any metric, that'd be a misrepresentation of the debate, that is likely to cause the audience to come away with a misunderstanding of the issue.
I don't know how close we are to an "ideal platform for debate" on HN (probably not very)—but I feel like a certain degree of people reflexively downvoting contrarian opinions (no matter the subject) probably bring us closer, not further, from that point. Contrarian opinions should not be presented as if they were orthodox opinions. They should be given some representation—I'm not advocating for censorship—but they should only be highlighted in any algorithmic way, to the exact degree that they're recognized as something some appreciable group of people really believes.