<div dir="ltr"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">"Strong monotonicity" is my term for the stronger mono-raise where<br>raising A can't change whether the outcome ranks some other B ahead of<br>C. Range passes it, for instance, but it's very rare in ranked methods.</blockquote><div>Am I misunderstanding this, or is it just independence of irrelevant alternatives?</div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, Mar 26, 2024 at 5:15 PM Kristofer Munsterhjelm <<a href="mailto:km_elmet@t-online.de">km_elmet@t-online.de</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">On 2024-03-26 00:42, Closed Limelike Curves wrote:<br>
> On Mon, Mar 25, 2024 at 12:39 PM Richard Lung <<a href="mailto:voting@ukscientists.com" target="_blank">voting@ukscientists.com</a> <br>
> <mailto:<a href="mailto:voting@ukscientists.com" target="_blank">voting@ukscientists.com</a>>> wrote:<br>
>> The non-monotonicity of STV just owes to it trailing a remnant of<br>
>> plurality voting, in its "last past the post" exclusion count.<br>
> <br>
> <br>
> No, the issue is the multi-stage method—the more elimination rounds you <br>
> have, the more likely you are to have a negative voting weight event. <br>
> Pretty much every sequential-loser method has the same problem.<br>
<br>
Yes, though I think three candidates suffice for weighted positional. <br>
Say you have a Condorcet cycle, A>B>C>A, and the base method passes <br>
majority and monotonicity but fails either LIIA or strong monotonicity <br>
for three candidates. Then it's possible that raising A can lead the <br>
loser of the first round to go from C to B, e.g. the base method's <br>
ordering going from something like<br>
<br>
B>A>C<br>
<br>
to<br>
<br>
A>C>B<br>
<br>
which, in the context of an elimination method, means that in the <br>
original election, C is eliminated and then A beats B pairwise; but <br>
after A is raised, B is eliminated and C beats A pairwise.<br>
<br>
"Strong monotonicity" is my term for the stronger mono-raise where <br>
raising A can't change whether the outcome ranks some other B ahead of <br>
C. Range passes it, for instance, but it's very rare in ranked methods.<br>
<br>
I'm prettty sure you can prove such a strong monotonicity failure for <br>
every weighted positional method by using linear programming. And <br>
probably for a bunch of other methods as well.<br>
<br>
>> The Surplus transfer election count, especially Meek method, is<br>
>> monotonic.<br>
> <br>
> That's also not correct; see Pukelsheim's book on apportionment for why <br>
> every quota rule method is nonmonotonic. I do think /most/ of STV's <br>
> nonmonotonicity comes down to the IRV elimination, though.<br>
<br>
To my knowledge, his 2014 book only mentions house size monotonicity and <br>
vote ratio monotonicity. House monotonicity is compatible with quota. <br>
(Imagine a method that eliminates the one winner that won't cause a <br>
quota violation later on, and then continues doing so until the right <br>
number of seats has been reached.)<br>
<br>
Vote-ratio monotonicity seems to be closer to mono-add-top, since <br>
Pukelsheim talks about a party's weight increasing while another party's <br>
weight is fixed at unity. I haven't looked into the proof in detail, but <br>
I suspect the reason that this works is because adding weight changes <br>
the value of the quota. Plain old mono-raise wouldn't do that, because <br>
the number of voters doesn't change.<br>
<br>
(Here's an interesting thought: Set the elimination order to some <br>
predetermined order before running STV. Use the same logic: while <br>
someone has more than a quota, elect him and redistribute surpluses. <br>
Then eliminate the remaining candidate who's ranked last on the <br>
predetermined order, and repeat. Is this method monotone? Doing it like <br>
this would remove the IRV/strong monotonicity failure-related problems <br>
mentioned above.)<br>
<br>
>> I have invented an STV exclusion count which is an iteration of the<br>
>> election count; both therefore monotonic; a "scientific" one-truth<br>
>> voting method, unlike other voting methods the world uses. I posted<br>
>> programmers links, none of which, the list manager has so far redeemed.<br>
<br>
I just downloaded the EM archives from <br>
<a href="http://lists.electorama.com/pipermail/election-methods-electorama.com/" rel="noreferrer" target="_blank">http://lists.electorama.com/pipermail/election-methods-electorama.com/</a> <br>
to run a grep through the links and find any software links.<br>
<br>
And I think this is it:<br>
<br>
<a href="https://github.com/Esrot-Clients/STV_CSV/tree/master" rel="noreferrer" target="_blank">https://github.com/Esrot-Clients/STV_CSV/tree/master</a><br>
<br>
referenced in this post:<br>
<br>
<a href="http://lists.electorama.com/pipermail/election-methods-electorama.com/2023-October/004988.html" rel="noreferrer" target="_blank">http://lists.electorama.com/pipermail/election-methods-electorama.com/2023-October/004988.html</a><br>
<br>
So it has already been posted :-)<br>
<br>
-km<br>
</blockquote></div>