Search functionality was a separate issue from the rest. I'm not sure how to deal with the issues I brought up. kleinbl00 has talked about the power user effect before -- if a new user submits something and kb or such shares it, it gets seen; if not, it doesn't. That's a bit of a problem. I think the best way to get around this is to encourage people to browse tag backlogs for cool content rather than sticking to their feeds. I love to do this with #goodlongread and #music when I have a few moments to myself. Mostly, the amount of submissions to the individual tags isn't prohibitively large, so a submission can sit idle for a week and then get noticed pretty often. It's just a matter of hubski users actually doing this with their favorite tags. (And also getting them to use very specific tags when tagging their posts so that people can find content they want reliably. And then making sure the list of tags is mostly comprehensive and easy to find. After a while, tags start to look like subreddits in many senses. Again, not sure what the answer there is.)
Thanks. Definitely some good food for thought. I am sure that tags can be leveraged to better improve discovery, and that's the main reason why I updated them to so we can store data with each tag. Still, I want to steer clear of the subreddit effect, so I think we'll have to focus on tags as discovery mechanisms rather than tags as destinations. I also think we aren't using user relationships to their full potential. On a related note, one of the bottlenecks here is presentation. How do you present options for lateral wandering, and when? IMO the navigation heiarchy on Hubski is a bit unecessarily complex. I think there are probably ways to make the interface for similar elements more uniform.