Chris, as I remember, after MCA was invented various attempts at generalizing it eventually resulted in ER Bucklin (whole) for ranked ballots, even though strictly speaking (as you point out)that method is not a generalization of MCA.
The nice thing about ER Bucklin (whole) was that it satisfied the FBC. I pointed out that generalizing MCA to Cardinal Ratings ballots (now known as range ballots) with any number of slots also satisfied the FBC, but this information was ignored because range ballots were not in style back then. I agree with you that after three slots we get diminishing returns, at least for Bucklin like methods, since the candidates that people really care about will get ranked in the top two slots. However, if we use Range Bucklin to get the initial candidate c0 in the GSA path, then the lower slots will have more value in guiding the course of the path. Forest ---- election-methods mailing list - see http://electorama.com/em for list info
