Good point, but even if selling of verified accounts is not policed, it wouldn't any worse than the power the whales have now. And those sub-communities that want to be functional and rewarded, need real followers, so it would still benefit them regardless. So I'd advocate not inferring with the free market, and do not try to stop the selling of the verified accounts.
These still need to be real people to achieve verification, so the value imparted to the voting power (over some reasonable time horizon) by a unique follower needs to be less than threshold of what they earn to become verified and this threshold has to be small enough that it isn't encouraging professional systems of hiring people to become verified who do nothing else but that, yet that threshold also needs to be significant enough that it isn't too easy to become verified such that is trivial to game it by having a bloc that upvotes #verification blogs (and thus not really be verified). In other words, the threshold can't be allowed to be so small that many can sneak under the radar of the community.
Another problem is that as the usership grows, people could reverify under numerous aliases and the community would probably forget these faces. But even so there is probably a limit to how frequently someone could fool the community. Typically the community rewards a comment post that exposes a fraud.
The game theory quagmire of distributing a pooled debasement of the money supply via voting won't be solvable without some community manual effort. Algorithmically it can't be solved purely automatically, because it is a Tragedy of the Commons self-referential given the externality of the shared pool everyone is trying to game. We have to introduce some external reference point.
RE: How To Fix Steemit For Communities & Viral Engagement