Craig et al.,
I am not telepathic, and I can't read too many long posts (other than my own!). I still don't get what worries Craig about Approval (as versus other methods) nor what is the point of the display of the 64-candidate ballot - for use with Approval or any other election method. No election method, intended for universal application to 'all elections large and small', can save a situation where 64 or more candidates are unleashed on a mass electorate who on average know few (and have prior reason to know few) of the candidates, or anyhow much about them. Sorry, I don't groove on Hadamard matrices. I also don't know what anyone means by 'Hadamard theory', let alone what questions this theory focuses on and investigates. (Any relationship to Hadamard's approach to the Prime Number Theorem??) As a mathematician I'm overspecialized, or anyhow specialized in a different direction (or two). Craig writes: 'Testing is not realistic if it is not done on paper or a computer since too few points are involved. Surely at least 1,000 points are needed. It could be tricky to blindly test the so called IRV method (with 3 candidates) and find a grave defect using <=1,000 sample points. Why should the results from the test be "local" (of a region in USA) while there was no theoretical testing?' I agree that it sounds silly to claim a need to test locally what should already be understood - or if not, then be tested - universally. However, please bear two things in mind. First, Craig and I are talking partly about testing different things: theoretical inherent properties of a universal abstract method vs. practical logistical properties of a local concrete implementation. Methods which can be implemented so as to work straightforwardly with most voters and election officials may fail in Long Beach as implemented there. Such logistical failure will likely betray itself with many fewer than 1000 or even 20 distinct 'points' (which I take synonymous with 'runs', or 'instances' of use). Second, many people will pay attention only when a method (such as IRV!) is seen to fail a 'practical' test, even though the failure is readily predictable and inherent on grounds long-understood by theorists. Much of Craig's response is a critique of a message in which I suggested points to make in a rebuttal letter vs. pro-IRV statements. Some of these points ARE perhaps a bit incautious, as Craig urges. In particular, general scholarly approval of Approval might indeed be questioned, so a more cautious rebuttal letter might note particulars such as EMAC or Brams or ... Joe Weinstein Long Beach CA USA _________________________________________________________________ MSN Photos is the easiest way to share and print your photos: http://photos.msn.com/support/worldwide.aspx ---- For more information about this list (subscribe, unsubscribe, FAQ, etc), please see http://www.eskimo.com/~robla/em
