It probably does that with any method.
Suppose it's Nader, some Democrats, and some Republicans.
You equal-top-rank Nader and all the Democrats.
Nader wins. If you and others like you had left Nader out, a Democrat would
have won.
The Republicans have ranked all the Republicans and Democrats, and protected
them all with AERLO, in order to
protect against the election of anyone they like less than the Democrats.
Nader's win, therefore, triggers the Republican AERLO, bringing lots of
Republicans and Democrats to top-ranking
among the Republican ballots.
That gives the win to a Republican.
If you had only top-ranked the Democrats, and left Nader out of your ranking, a
Democrat would have won.
That violates FBC.
Sure, a verbally-described non-numerical scenario like that isn't as conclusive
as a numerical example, but it sounds
like something that could happen, and it's convincing. AERLO probably spoils
FBC compliance.
So I don't propose AERLO for public elections. I want it to always be possible
to assure voters that there is absolutely no
way that they could regret voting their favorite at top rank or rating.
I feel that majority working-together, to enforce majority rule, would be more
difficult with more than 3 ranking slots. A
3-slot method makes it easier to assume what your compromise's supporters will
do. AOCBucklin would involve more
strategic guesswork and risk than would MTAOC.
So now I don't know if I'd propose any method with more than 3 rating slots.
Well, in my conversations with people new to voting systems, I've been told
that ABucklin is too complicated anyway.
Mike Ossipoff
----
Election-Methods mailing list - see http://electorama.com/em for list info
