If we start with the assumption that all seats must be filled, we inevitably end up with the final seats being won too cheaply (e.g. on remainders), the price (as in RRV) or representativeness (as in Greedy Monroe) of a seat gradually decreasing, or the rate of load-unbalancing (in Phragmen) gradually increasing, due to exhausted ballots. This is unfair to exhausted surplus voters, who may not only outnumber unexhausted voters, but in fact may collectively have quotas, and would presumably prefer their quotas (or remainders, if the largest) to go to null candidates (i.e. keep the seats vacant), thereby increasing the power of the winners they supported. Asset voting is, of course, a better solution; however, because of transaction costs, it’s only a partial solution. I propose a vacancy solution, preferably in conjunction with asset voting (i.e. should seats exceed quotas after the trading window expires).
To be clear, I am not suggesting that seats should never be won on remainders. Rather, the remainder should have to be greater than the exhausted surplus vote divided by the number of remaining seats (I call this the Divided Exhausted Surplus, or DES quota). Or, at the very least, vacancy, treated as the candidate all exhausted ballots vote for, should be able to win seats like any other candidate. This version would be ideal for systems without quotas, like RRV or Phragmen. In STV, a voter should even be allowed to give “vacancy” an intermediate rank.