"a lot of the voting power is aimed at posts from top authors"

There's that economic and game-theory imbalance we were just talking about.

"upvoting posts from popular authors early with an autovoter consistently brings upvoters 3 to 5 times more than upvoting authors not consistently earning $100 a post"

Thank you for including the data tables, it's extremely helpful and I realize one of the more laborious and underappreciated additions to a post.

The simplest solution here would be some sort of curve where curation rewards are substantially reduced for popular authors. "Popular" could be defined a number of ways - payouts, followers, reputation.

However, one thing we can all agree on is that these users do not need curation to get their posts seen. Therefore, the curation reward which is supposed to encourage unearthing and promoting high-quality posts is failing, not on a technical, but on a fundamental level.

The curation formula must be changed, lest Steem become (more of) a bot vote-trading playground. I have actually written a series of curation articles proposing various changes to curation, everything from vote-refunding for highly effective curation to changes in the reward curve.

"When competition from automatic voters gets too high on an author relative to the manual upvotes coming later, the rewards tend to drop a bit "

This makes sense. The auto-voters are probably all getting in at what they deem the optimal moment. My guess is that this is slightly before 30 minutes, in an endless game of "wanting to trigger just before all the other auto-voters trigger, while balancing the 30 min curation curve".

What this means for us manual voters is two things:

1. When we see your post during normal, human browsing (ie, later that evening or in a day or two), your rewards are already very high. In many cases I look at a post and think "I'd have voted for this if it was making far less, but it now has rewards sufficient to match its quality."

2. Users hunting for curation rewards will pass over articles that have already been mostly upvoted, which for human browsers, is probably most of the articles they see. This is because they expect to make fewer curation rewards. In some cases, they would even be better off voting on 0-reward comments of the author for curation value than the original post itself.

Breaking off this wall of text again into another self-reply.