[EM] Droop fails the Markus Schulze Rule????? [2]

[Bart Ingles]

I received this one Friday, but not the one that was re-sent today.
-Bart



Tom Round wrote:
> 
> In other words, Donald seems to be saying (a) The Hare quota allows to to
> produce a result where no ballot is  wasted (but only if we transfer the
> last runner-up's ballots to the next-preferred candidate remaining and deem
> them to be real preferences for the latter), but (b)  Under the Droop quota,
> almost a full quota of votes may be wasted (because it would be some kind of
> "fraud" to transfer the last runner-up's ballots to the next-preferred
> candidate remaining and deem them to be real preferences for the latter).
> 
> Sorry, but Rule No #1 of comparing electoral methods must be that we judge
> them both by the same yardstick. "Sauce forthe goose ..." and all that. So
> if it's legitimate to perform this operation under the Hare quota,
> re-allocating all losers' votes among the winners (assuming, as Donald
> suggests, that no votes exhaust), then it's also legitimate to do it under
> the Droop quota -- and under _both_ methods it will also "prove" that there
> are no more wasted votes. Saying "we can't make sure every vote elects
> someone under Droop because Droop is a fraud from the very beginning" is a
> circular logic if one is arguing that "Droop is a fraud from the very
> beginning because under Droop we can't make sure every vote elects someone".
> 
> (PS: I personally think the best measure of wasted votes is to continue
> elimination until only one more candidate remains than the number of
> positions. We can then meaningfully compare support for the five winners
> with support for the last runner-up. Translated  to a single-seat system,
> this becomes the "two-party-preferred" count so often cited in Australia,
> although the term is inaccurate in cases when an Independent or minor-party
> candidate wins or comes second, and is harder to  apply to a multi-member
> result. "Final-count preferred" is  the term I think they use in Ireland.)
> 
> Rule No #2, specifically in relation to proportional systems, is to
> recognise that votes can be "wasted" as much on elected as on defeated
> candidates. Otherwise, if only losers' votes counted as "wasted", it'd make
> PR systems inferior to those like the former Japanese Single
> NON-Transferable Vote, where if six candidates divide the votes 40%, 30%,
> 20%, 5%, 4% and 1%, we elect the five highest and congratulate ourselves
> that a mere 1% of the votes were wasted. Of course a lot more than that were
> wasted, ifthe first one elected has 40% when s/he needs only 16.67% (under
> Droop) or 20% (under Hare) to be elected and the extra 23.33% or 20% should
> be transferred (in which case they may well help Mr/Ms 1% defeat Mr/Ms 4% or
> 5%).
> 
> >Date: Fri, 15 Oct 1999 20:07:00 -0400
> >To: "[EM]" <election-methods-list at eskimo.com>
> >From: donald at mich.com (Donald E Davison)
> >Subject: [EM] Droop fails the Markus Schulze Rule
> 
> [... snip ...]
> 
> >     It is best to use Ideal Conditions when we compare the two. Ideal
> >Conditions are when there are enough informed choices made by the voters so
> >that no ballot becomes exhausted.
> >     This may never happen in a real election, but it is mathematically
> >possible.
> >
> >     Anyway, the Hare STV election will not have any wasted votes. Every
> >ballot will end up on one or another of the winning candidates. Each
> >elected member will have received the same number of votes - an equal part
> >of the total votes.
> >     The Droop STV election will not fare as well. The Droop will produce
> >one quota of wasted ballots.
> >
> >        <B> Droop fails the Markus Schulze Rule. </B>
> >
> >     Also, while the elected members will all have the same number of
> >votes, the amount will not be an equal part of the total votes.
> >     Now, I am not suggesting that the last Droop quota be transferred - no
> >way am I saying that.
> >     I am saying that the Droop is a fraud and should be discarded before
> >we even start the count. [... snip ...]