It still has vulnerability to Sybil type attacks. If all it took to make it to the top of the trending page was a high volume of comments, you would start to see massive groups of people just generate a lot of relevant although not very helpful comments. Enough to not be considered spam, but still up the count.
One of the biggest differences between medium and Steemit is the money part. When you introduce the money part, you introduce all kinds of skewed incentives which corrupt a lot of the algorithms that work on a purely content driven site where all the users care about is content.
Good points @timcliff there is almost always a downside.
It wouldn't work just on comments alone, no. Also when something reaches the height of the trending page, more people will notice if people exploited functions to get there. If the rep of the user needs to be above 50 or 60 for comments to count could that work? theres the reputational risk if people see a user spamming basically. And if it makes the trending page, it will be seen, and maybe flagged
Reputation is a great feature, but not everyone who participates is going to have a high reputation score. Curators and investors who may not post and comment will have no say on what makes it on the trending page.
I see (and ty for explanation) At a certain point then trending will actually mean 'promoted'. As any curator who has a hefty weight is going to be someone who earned the SP from posting, but more likely a whale, or investor. The latter would basically pay to influece the trending page, sure, and that is great. But it woud be misleading to have that as 'trending' tab
I noticed that there's the active and hot sections which don't seem to reflect post activity very well. Further developing the tags in more specific and effective ways would help. Not necessary to shut-out other means of influencing content noticeability.
i can barely guess at how the 'hot' and 'active' tabs filter.
Replying to your lower comment
I don't know what active does but Hot is a pretty simple algorithm. The posts start in the same order as New (newest first), but each time a post gets an upvote, it gets a boost back toward the top of the list. How much of a boost is determined but the SP and weight of the upvote. Posts that don't get upvotes will slide down fast, while posts that continue to get upvotes will stay near the top, then eventually slide down as well.
I'm pretty sure active puts posts that were either just created, edited or had a comment added put at the top of the list, which then just continues to scroll until the next edit or comment, again just bumping to the top of the list.
Simply hitting edit and update post with no changes will put you to the top of active too.
could there be a cheetah bot that focuses on content under trending? it would catch copy pasta in comments that might otherwise look like a human response
It would become extremely complicated and political to police. It would also be very subjective. At what point is a comment a legitimate comment vs an attempt to game the system? Should 20 minnows who all say something like "I really like what you did. Great post!" be flagged?
I was musing about if comments with minimum word count and upvotes of comments could be factored in as well. not simply number of comments. So maybe it wouldnt be necessary to flag things like that. Rather, for example, if someone resonded to a post about philosophy from something they copypasta from a philosophy blog they didnt write, just to make comments. (but I realize it's still complicating and much work)
Any solutions dont make sense without steemit having a direction as either blogging or short-form (or split up as both with different guidlines). Posting guidines and expectations can mean implementing other features like minimum word count replies, which right now, would be limiting.