Brian-- My goal has been to just try to make everyone's representation as equal as possible (or, for true divisor methods, everyone's representation _expectation_). I do that by making each cycle's overall s/q = 1, as nearly as possible.
But you've suggested something less traditional, but something that makes perfect sense: Directly tinkering the result of each census apportionment, so as to meet one of the standards that you listed. It would be a more complicated method, with a more complicated rule. And it would be farther from tradition than the methods we've discussed so far. But maybe better, if one of the standards is considered more desirable than just trying to make the s/q as equal as possible. Just when I thought that CW, AR, WW and W were the only ones to propose, someone suggests an approach to _directly_ enforce one of those standards. So now there's an additional kind of method for proposing Of the standards that you listed, I prefer the one that minimizes the greatest under-representation. Mike Ossipoff ---- election-methods mailing list - see http://electorama.com/em for list info
