Conjecture

Reputation-weighted voting does concentrate influence relative to one person one vote, but the concentration is not by itself the failure. Against the realistic alternatives, weight bought with tradable tokens and the informal elite that skewed participation produces anyway, earned non-transferable reputation measured as the least concentrated of three DAO weighting models compared, though still highly unequal (median voting-power Gini 0.46 against 0.75 and 0.98, in an exploratory study without statistical tests). Non-transferability is the clearest design difference between those models, not a demonstrated cause, and it reduces concentration rather than preventing it. What decides whether the concentration harms is what the reputation is earned on: volume of activity reproduces the participation skew and entrenches it, while scored accuracy on the kind of decision being voted on is the one basis on which concentrating weight has been shown to improve collective judgement. Two defections bound that result. Accuracy has to be scored relative to the crowd on the same questions, or participants gain weight by choosing easy questions to be scored on. And weight that cannot be transferred can still be rented, through vote-buying or bribery contracts, so non-transferability needs a check on paid or delegated voting. So the design question for a merit-weighted franchise is the scoring rule, the decay, the scope of each score and who defines what gets scored, not whether to weight.

Institutional design