How would you propose a method of limiting the downvoting of comments that are salient and relevant but otherwise elicit a knee-jerk emotionally disagreeable response? Or do you consider that ability a feature? It seems to me to erode curious debate.

Not sure I understood your question, but you can't really control what others will do. E.g. people can downvote or flag without any reason whatsoever.

The question relates to how can you create a system that helps align peoples incentives with the goal of the forum.

Imagine if you could allow users to delete comments they don’t agree with. That would create the antithesis of a curious discussion, so that feature isn’t available. So, yes, a good system can constrain what users are allowed to do with intelligent guardrails.