Taking a deeper dive, I agree that the spam situation is not bad, aside from a few concentrated pockets. I do wonder if part of it is decreased usage. Free downvotes will certainly help, especially if there are crowd-sourced and crowd-driven downvote programs. As far as I can tell, efforts to neutralize spam/blatant abuse are led by a small number of people.
It would be great if simply sorting comment threads by pending payout would produce the most useful result. But in practice, (a) many decent comments get no votes, (b) a few low-value comments get some votes, and (c) some insightful comments are downvoted far below 0.
Increased curation rewards (and interest in the platform) could lead to more intelligent automated voting which would help smooth out the difference between (a) and (b). There's a history of trying to solve issues with more complexity, where a simpler solution would suffice.. at scale.
Curated mute-lists could be part of a solution, though I'd imagine UIs picking a default mute-list for guests would be a further source of contention.
They may be (though some of this is probably the larger efforts being more visible) but by far the biggest improvement was simply the switch to RC which cut spam by some very significant percentage overnight (my guess would be 80%-ish).
This could be an example of solving a problem with an arguably more complex system, but one which was well thought out and had broad application (unlike many of the tweaks which were more reactive and impulsively made in the past, but had the effect of layering on unstructured complexity in a more harmful way).
Anyway, I do think we should assess the effects of HF21 which at this point we can only make rough estimates ranging from very limited to profound across the platform.