Thank you for your feedback, @biophil. You found a mistake and explained the issue much better than me. So, corrections are in order.
Don't you mean in USD terms?
Yes, I mean in fiat terms. I'm going to correct this.
I think you're making an assumption somewhere that the market cap (in USD terms) of STEEM is constant?
Yes, I make that assumption to keep the calculations simple. I've mentioned it in the post: "If the amount of fiat investment in Steem remains constant" I will update this also.
If 100% of the total stake were voting actively, those conclusions might make sense. One issue is that much less than 100% of the total stake is voting actively -- probably less than 60% (though I haven't checked the number myself). Thus, 60% (or possibly much less) of the stake is allocating the entire reward pool, so depending on exactly how full the reward pool is and what these numbers actually are, you can easily beat dilution by self-voting.
Yes, the assumption that 100% of the total stake voting actively was implicit in the calculations. Do we have a reference handy where this percentage data can be verified? If not, I can google this later.
I believe the solution to this problem is to get everybody to use their voting rights every day. If they can't do that, they should delegate it to someone they trust who would work as an "editor" on Steemit.
Or have I misunderstood you somewhere?
No, you have understood me correctly.
I need to update the post, because the thesis here won't hold if less than 100% of the stake is voting actively.