great points, @lukestokes.
The idea is that using the largest tags doesn't actually give you the best opportunities. In v1 we will average by the # of posts and can definitely have an exclude list of tags to never show (like witness related tags). I wonder how we could capture some of the spamming of tags by users. I'm sure we could pull that data and display it eventually. Ideally people will know the best, most relevant tags to use and where their tribe actually hangs out instead of just posting in "steem" each time (for example).
In v2 we plan to add "fish scales" which will give in depth analysis of who is posting, voting, commenting, etc (not calling out names, but %'s) so that people know what each tag/community consists of. This should show if the resteems/votes/comments are by low-weight bots or actual dolphins or whales, etc.