So the problem is not strictly in the design of systems, but in the psychological dependencies of users of systems . . . Exactly.

Frank Herbert would've agreed.

Among other things I study networks and correspond with many other academics on the subject. Most of them believe at the moment also that no automatic design can get around the issue of psychology. And the issue of the specific preferences of users.

Experiments like Steem will make it clearer whether that is the case. Such systems allow users to cash out. Therefore they arguably supply sufficient incentives for users to behave in a way that reveals their preferences.

Then observed user behavior can test theories in a valid way, in the sense of experimental economics in the tradition of Vernon Smith.