As in all my postings, my main objective is to encourage mental limbering up, so that we don't get bogged down in a lack of imagination and thereby overlook the wide variety of possibilities whose surface we have barely scratched. All elections are influenced by biased random bits of information and disinformation. To ignore this is to be extremely naive. All of this biased random information amounts to a front end lottery that influences the outcome of the election. Why not fight uncontrolled biased fire with unbiased controlled fire? Use an unbiased random sample of the voters at the front end of the election as a "voter jury" style panel for the purpose of getting true polls of the right type, etc. This official voter panel (where every voter has an equal chance of being chosen) would offset some of the damage done by the biased corporate polls, where only voters that are near their telephones during the polling hours get polled, the questions are leading, the anwers are misrepresented, the answers have no bite, so theycan be attempts at manipulation, etc. In the presidential election this panel should have somewhere between a thousand and ten thousand voters. There are many possible uses for this panel in setting up the election for its final stage. Here's just one of an infinite variety of possibilities (and I hope that you will exercise your imaginations in finding many others, totally different from this one): Step 1. A randomly chosen member of the panel proposes a "target candidate" T_0 . Step 2. The panel compares (by voting head-to-head) the other candidates with T_0. If no candidate beats T_0, then T_0 is presented to the voters as the "Target Candidate." Else the candidate that beats T_0 by the greatest number of votes is designated T_1. Step 3. A beat chain is thus generated T_0 < T_1 < ... until some candidate T_k is either unbeaten or appears for the second time in the chain. The previous candidate T_(k-1) is presented to the voters as the "Target Candidate." Step 4. The panel is dissolved, and the general election is held by Approval with this proviso: if no candidate receives more than fifty percent approval, then the designated "target candidate" is elected. Note that if the panel is truly representative, then the outcome for the general public will probably be the same as the outcome for the panel, which means that T_k will probably be elected. If not, the surprise winner will be someone with greater approval than T_k relative to T_(k-1). Though step 1 is stochastic, the panel process is much more controlled than standard pre-election chaos, and so the final deterministic approval step is much better informed than are most votes in actual elections up 'til now. Note that T_0 is not necessarily the favorite of the randomly chosen panelist. He's proposing a "target" not picking a winner. It seems to me that this procedure could be adapted to choose the beatpath winner most of the time, without requiring the voters to fill out ordinal ballots. The panelists would do the ground work, and the rest of the voters would provide the final deterministic confirmation (or rare surprise upset) using approval style ballots. If you don't like this method, see if you can figure out a better use for the front end panel. There's gotta be a better way! In small groups the entire electorate could be the front end panel, since repeated ballotings are no problem in that context. Forest
<<winmail.dat>>
---- Election-methods mailing list - see http://electorama.com/em for list info
