Michael Poole writes: > Forest W Simmons writes: > >> Has anybody explored this idea? >> >> Make an electrical circuit with a terminal for each candidate. >> >> For each pair of terminals attach a diode that has a different >> resistance in each direction: the resistance in the direction from >> candidate i to candidate j is proportional to the number of ballots on >> which candidate j is ranked above candidate i. >> >> After all of the diodes are attached measure the circuit resistance >> between terminals i and j. >> >> The candidate X against whom the maximal resistance is minimal is the >> winner. > > I do not understand the analogy to electrical circuits. > > 1) Diodes do not have an impedance that is easily characterized by a > single number. Roughly, they conduct a lot of forward current if the > forward voltage is above a certain value, and have a (negative) > breakdown voltage beyond which they allow current to flow easily in > the reverse direction. > > 2) If there are independent paths between point X and Y in a circuit > (that is, the endpoints are the only places that any path intersects > with any other path), the impedance from X to Y is the geometric mean > of the impedances of each path. I do not understand why this would be > desirable for defeat paths.
Sorry, not the geometric mean; the total impedance is the reciprocal of the sum of the reciprocals of each path's impedances. Michael Poole ---- election-methods mailing list - see http://electorama.com/em for list info
