I don't think an analysis of the reward design can be done without including an analysis of the Curation Rewards.
For example,
A 'deviant whales does not maximise profits by voting on their own post', if their own post is subpar. They would be better off voting (at an optimal time) on a superior post. Assuming they spend an hour knocking up a plausible post, they would be better spending that hour voting on 5 or 6 superior posts. The Curation Rewards would outstrip the Author Rewards. The only advantage of the author rewards is being able to obtain liquid SBD rewards.
It is also worth noting that a deviant whale cannot just post anything and reap rewards. A sham post can easily be spotted and downvoted to zero by other whale/ voters, as that behaviour is against the long term interest of SP holders. The deviant whale will lose any rewards and also lose some of their reputation. In this sense the design is self-policing assuming the majority of participants are rational actors.
RE: Blog rewards CAN’T be widely distributed