These numbers make me sad, but they are the logical consequence of a flawed system that rewards these actions. What we need is to move towards a system where delegating to bid-bots is not the best way to earn an interest on a Steem investment with little effort.
While I was and still am sceptical about 50/50 author/curation rewards, this would certainly incentivise curation over self-voting and vote-selling.
Then, we would need pre-curation trails instead of curation trails, for example if @utopian-io would vote with a low-SP account - let's call it @utopian-voter - and follow this vote with @utopian-io 5 minutes after, curators could trail the @utopian-voter account and reward good content while earning good curation rewards at the same time.
Thirdly, we need more projects raising funds through delegations. Even though ICOs are called "dead" by some, on Steem we have the unique possibility to raise funds through delegating SP without loosing the initial investments. When delegations are used for auto-voting their delegators like @dstors does I am very sceptical about that, but if delegations are used for manually curating content authored through these emerging dApps like on @steemhunt and @travelfeed, this benefits Steem, these projects and delegators who will receive SMTs for their investment which could be worth more than the STEEM they are getting from bid-bots.
Ultimately, SMTs will bring the opportunity to reinstall proof-of-brain not for STEEM, but for a range of new tokens that each have their separate rewards pool: If the project account behind the SMT holds e.g. 25% of the total funds and uses it for actively curating the best content once a day, this brings huge curation rewards for votes that come in before. SMTs could also choose a 50/50 split in author and curation rewards, effectively making curating more profitable than self-voting.