I agree that hubski has a large advantage over reddit in terms of preventing nonsense, but it is still susceptible to it. A lot of the tags on hubski are very general and would be susceptible to overload by nonsense. Imagine if #space were overloaded by people commenting/posting about us never going to the moon, or #politics overloaded by a specific political movement that you disagree with and find annoying.

The hubski solution would be to exclusively follow users at that point, but then what tags do those users post under? If it's politics related, they'll post it to #politics. Yes you'll be filtered from the nonsense content posts, but those nonsense people will end up seeing the posts that you are following since they are tagged #politics. So, the comments will still be overrun by nonsense comments.

The twitter/hubski tagging solution to the comment filtering would be to make up a new tag. The adoption rate of the new tag would take a long time, though, and a new dead tag will take awhile to gain traction, so you'll still be subscribed to the old one for practical reasons. If you manage to develop a really obscure but quality tag, then you'll suffer attrition as people lose interest, people lose time to post, etc. You'll have to go back to the original tag that all the nonsense users are posting on or organize a movement to generate a new tag again, which your attrition will hinder. The transparency involved with advertising this new tag would end up with nonsense users joining up on the new tag as well, or intentionally sabotaging it by trolls.

I just did a test and it is possible to post without any tags being added. After generating a good user base of followers, convincing people to stop tagging their stuff would be another potential solution. The problem with this is that untagged posts would make it harder to find new and quality users to subscribe to. When going out searching for new people, the only way to find the users in the first place is to scan common tags and see who is posting good content or not, so this is a bad idea.

Another problem with the last two solutions (new tags, no tags) is the community tag feature. The community tag feature would get your #supersecretpolitics post added to #politics as well, therefore drawing in all the nonsense commenters.

There are two solutions that hubski provides as well that I haven't mentioned yet, which are muting and ignoring people. This is a useful feature, but when you get a large userbase of annoying, trolling, negative, or meme-oriented users this is like fighting email spam with a blacklist. It doesn't work when there are too many sources of bad content. You can blacklist all day, ignore/mute 200+ users every day, and if you are dealing with 1+ million users, you're not even going to make a dent. That's why email uses bayes filtering for spam, which is a more complicated algorithm than simple blacklisting. That being said, having run my own email server, bayes filtering isn't good enough with one user (myself), and I had to switch over to a more public spam aware mail server with multiple users. For spam filtering, bayes would work just fine with multiple user inputs since spam is a pretty obvious thing that is defined in pretty much one way. Comment quality is subjective though, and the giant meme community would call a cat video quality while I and others might not. Bayes wouldn't apply to this situation at all as a result.

So my main point here is that yes, hubski is more resilient to these problems than reddit. You can easily filter content by avoiding subscribing to tags and subscribing to users by monitoring tags for quality users. Hubski is still vulnerable to comment quality issues, though, based on the popularity of comments that are meme-like by a large percentage of the userbase.

I may be underestimating the mute feature, but I'm not sure. It might help against nonsense users, but it can't defend against trolls who generate new accounts to bypass your muting. I have to think about the large scale implications of muting, but I think you'd still run into issues with it being a blacklist and it not being quite enough.

I'm curious if the inverse of muting, whitelisting users, might be an interesting solution. That might be too drastic, but it's worth some thought.