This all presumes that the "goal" of aggregators is organizing the library of articles that each person submits. Topic based organization is helpful to users, and conversations certainly develop around topics, but I would argue that simply organizing content is not the goal of an aggregator. If we think about a library, sure I find all of the information that I want, but I don't have conversations in the stacks about what I am looking for. Rather, the goal of an aggregator is to filter good content from shit content, and to foster/moderate discussion.

All that hubski is is a different approach to that problem. If we just follow hashtags, then it is nothing more than a decentralized reddit model (if hashtags are thought of as individual reddits). Your still going to see the same degradation of content around each hashtag. The biggest problem is discovery (the reason that you noted further down) that new users can't make heads or tails of the follow a person model, and hashtags are the only way to keep your feed populated. That doesn't mean that the follow model is wrong, it just means that it hasn't been developed enough.

I don't come here to talk about one thing, Reddit works just fine for that, I come here because I've found a few people to follow who submit consistently high quality content, and I can look through it without having to worry about being inundated with cat pictures. Is it perfect, no, but it works pretty well for what it is.