I replied privately to Mr Jan Kok who was being misled by the electionmethods,org website. Subscribers that could criticise the site are likely to drop out. I assume that Mr Forest Simmons (who has to wait for me to unsubscribe and then a delay or week or so (unless he has an arrangement with the owner to find out if I had unsubscribed ), won't be dropping out. Simmons did not seem to want the site removed.
At 2003\03\02 13:22 +0100 Sunday, Markus Schulze wrote: >you wrote (2 March 2003): ... >> > ---------------------------- >> > A 19999 80004 >> > B 1 5 >> > BA 19997 19997 >> > CB 40002 40002 >> > DBA 20001 20001 >> > ---------------------------- >> > Total: 100000 160009 >> > AV Winner: A B >> >> That looks like a "classical participation axiom" problem (but >> it is not since B got 4 votes). > >In your example, the fact that the winner is changed from candidate A >to candidate B is caused only by the addition of the 4 B voters. Your >conclusion that "the large support rise for A causes A to lose" is >false. >
Where is the "participation" word ?.
Also it is unsafe to say that A started to lose because there were extra 4 votes against A. The Alternative Vote negates votes (etc.) so pure support rises can harm the candidate supported.
I recall that the B votes are rising because that was a consequence of requiring that all of the altering be done with FPTP-style papers.
The fraction, 60,005/16,009 gets big fast:
Num Candidates : Fraction that arrived late to harm their candidate 4 3.0000812454 / 8 5 7/16 = 0.4375 % the next are extrapolations only 6 15/32 = 0.46875 7 31/64 = 0.484375 8 63/128 = 0.4921875 9 127/256 = 0.49609375 10 255/512 = 0.498046875 11 511/1024 = 0.4990234375 12 1023/2048 = 0.4995117188 = 1/2 - 10**(-3.3113299523) 13 2047/4096 = 0.4997558594 = 1/2 - 10**(-3.612359948) 14 4095/8192 = 0.4998779297 = 1/2 - 10**(-3.9133899436) etc.
Mr Shulze can say if he had an IRV passing rule that stops exactly that type of problem.
--
This next text mismatches with text of the text you in reply to a message of Mr Rouillon. Here are the two.
_________________________________________________________________ Statement S2 :
>Actually, when IRV is being used and candidate A is the unique winner, >then a set of additional voters who strictly prefer candidate A to >every other candidate can never make candidate A lose. _________________________________________________________________
_________________________________________________________________ At 03\03\02 00:19 +0100 Sunday, Markus Schulze wrote: >Dear Steph, > >you wrote (1 March 2003): >> I suppose you meant:
Statement S1 :
>> Suppose that candidate A is the winner. Suppose that >> a set of voters, where each voter strictly prefers >> candidate A to candidate B, is added to the original >> profile. Then candidate B must not become the new winner. > >Yes. You're right. > _________________________________________________________________
I comment on the two texts.
(1) Statement S2 talks about the win-lose state of candidate B while S1 talks about the win-lose status of candidate A. But both have candidate A be the preference of the two (if two) that is nearer the first preference. So the two are totally different. They word participation is switch and Mr Schulze doesn't want admit that he can see that problem, I assume.
(2) Also the bad wording "[caged kangaroos] who prefer candidate A to every other candidate..." in S2 presumably is referring to all candidates, rather than only the candidates named on the added papers. So it must mean something like 'A is the 1st preference'.
In S1 (underneath S2), symbol A only has to be before symbol B in the list. So the rule S1 is garbage when there are many candidates
(3) S1 has a consideration of 2 candidates and S2 (above S1) hasn't.
(4)
The words "strictly prefer ... to every other candidate" are very unclear. Here is the text:
>then a set of additional voters who strictly prefer candidate A to >every other candidate can never make candidate A lose.
What if the ballot paper is this ?: (CA). Also candidate B's preference is off the RHS edge of the ballot paper
No one knows whether A is strictly preferred by the breathing Australian kangaroos of the Mr Schulze's mathematical, over candidate B or not.
That is same mistake again because it means that with 3 lines Mr Schulze struggled to get out, has plopped him back into the dogs breakfast kitchen floor mess of brandishing the stick for many. I.e. my P2 rule, a rule good enough for me to use, since great stuff and not something from Mr Schulze, and its is also a rule that passes the Alternative Vote, has candidate A hold a greater influence than B in this paper (CA), than happens with this paper: (CAB).
E.g. if there are 4 candidates: then (CA) has maybe around 1/2 the power of (CAB), at promoting B. I.e. (CA) = (1/2)( (CAB) + (CAD)) by rule P2. Candidate B is out of the back of the paper, and the voters may jump to conclusions but not have sharp and strict views on what they like.
____
Apologies on the nuclear waste disposal operation. I was going to run an experiment proving that American members would copy the "IRV" term, but not speak up for Mr Ritchie. So the sensitive amongst us would know that missed Mr Ritchies view of 'Why IRV' and so on. The latter is hidden.
Another wording mistake was the argument on calling "rules" "criteria". Mr Schulze was calling it a criterion. I don't need to correct my bad sentence wording if the name was abandonded.
I might quit soon if Mr Schulze doesn't transfer a fraction of the emphasis on materiality of descriptions out to the idea with an reality that something to describe arises. It is all done with English and I am really doubt that extending a plank of common understanding that English is material, is going to be helpful to the exposition when the others are in the Commonwealth (I am thinking of New Zealand and African countries. Persons in Africa can't remain confused forever, though it seems possible for Nth Americans).
Craig Carey
----
For more information about this list (subscribe, unsubscribe, FAQ, etc), please see http://www.eskimo.com/~robla/em
