It worked quite well with a mostly-academia audience. For the general public, of course, moderation is indispensable. The spam in Usenet began when businesses and scammers discovered Usenet. My point in the root comment is that Usenet clients contained all the ingredients for efficient, structured, long-running, deep discussion, and those are orthogonal to moderation, which you nowadays need in any case.
The problem with Usenet style interfaces isn't just spam, it's low value comments. Comment voting somewhat solves this but most people vote based upon how much they agree rather than the comment's discussion value.
This should be solved by moderation when necessary, like in traditional web forums. I agree that voting is a bad mechanism, because people then start posting for likes and reaction stats, and also involuntarily judge posts by their like count instead of only by their own judgement. Flagging/reporting to moderators is fine.
I'm not sure moderation is that much better. Some moderators strive to be fair but the majority will still vote and moderate based on preferences. There's also the quantity issue, most low-value comments don't break any rules but also aren't very insightful.
It really depends on the community. You can certainly have rules against low-value or one-sentence comments. Here on HN you can get banned if you post too many low-value comments, it doesn’t need voting for that.
I haven’t seen the lack of voting to be a major problem in newsgroups or traditional web forums. But I’ve come to despise the prevalent dynamics of liking/voting in newer platforms.
Slashdot solved this (arguably) by having you classify comments as "interesting", "funny", and several other categories, rather than plain up/down. Although Slashdot failed here by mapping these categories to a single score.
