Nice of you to highlight this post and provide some of your thoughts!
First, a bit of a correction -- Greymass is an organization with @jesta, but also myself and a few others. This article was actually written by me. :)
It's cool that you've linked back into the Worker proposals newly created here on Steem. Indeed, it's something I was a big proponent of, partially due to this reason.
Interestingly, Steem doesn't really hold the same problem that EOS (and other DPoS systems) has, due to the rewards pool. Since the inflation has a much broader distribution, and the witnesses only gain a portion of it, the convergence is not nearly as much of a problem (i.e., they are unable to achieve collusion / byzantine fault without performing active market participation / buying tokens from others).
The worker proposal system just makes things even better in my mind. I'm less concerned about even having witnesses dip from this pool -- so long as their goal is to sell the tokens to achieve financing of the project. Selling the entirety of the rewards to the market alleviates the problem, as I somewhat point this out in the update at the bottom of the post. Though, it should be in entirety.
EOS doesn't have any other source of inflation, and their governance has already seen accelerated centralization. I believe one of the reasons of this is a form of the effect I wrote about.
Oh ok - haha.. Thanks for the correction. I appreciated the clarity and focus you put forward in it. I will update the post.
With steem, the overall size of the userbase influences the distribution of the rewards to posters and curators, so as the userbase decreases, the witnesses have more ability to access more inflation. In any case, I understand what you are saying and it makes sense - I did think about that too but figured that if the top witnesses are the main engineers building projects from the SPS then their ability to draw steem to themselves multiplies not just from the direct payments but from their increased ability to generate profits from other steem holders who are now their customers. This could, in my untested thought process, lead to the point where domination might become a more genuine concern. The same did hold for bid bots and their operators but maybe won't in the future. The other issue is psychology and public perception - even if there weren't a mathematical risk of domination, many investors will not invest if they think there is such a risk.
I'm not sure that true decentralisation is possible in a profit generating system that is itself run by members within it who are competing. I think a distributed network where everyone has equal power to compute and co-run the network is the only true solution to the equation. :)