The other axis remains and still has a traditional censorious downvote – but my hypothesis is that it will then be used less, avoiding the negativity created by its current overuse for multiple purposes.
That 'spite' you mention isn't a constant across all communities, across all time: it grows with cycles of negative, stinging punishment-feedback. (The study which started this thread was suggestive of that, regardless of its other limitations.)
I strongly suspect each of the following is true:
• threads on certain contentious subjects have lots of competitive and retaliatory downvotes, driven from the impulses "I can't let only these favored comments sink/fade" and "if those partisans are downvoting me, I'll downvote them"
• that habit/mood, once learned, spills over into other threads, to the detriment of discourse and positive/productive feelings
A separate axis for disagreement might catch some of that and express it where it does less spite-generation.
With "social problem" I meant something else, namely the fact that there are trolls and low quality comments etc. I don't think those can be eradicated by a technical solution, because once you have a high quality community there is an incentive for people to take part in it even at the cost of having to defeat countermeasures.
What is this "technical fix" you allude to? Eliminating downvotes sounds like a non-starter to me. You could limit the people who can cast them, but that lowers their effectiveness because the probability that a bad comment is downvoted would plummet.