The Ranked Choice Voting (Surgery on the Flesh) Bill

[Miestră Schivă, UrN]

BE IT ENACTED yadda yadda yadda that, in conformity with international best practice, El Lexhatx B.14.7 shall be amended to read as follows:

14.7. If, after any iteration, there are two or more candidates with the fewest ballots assigned to them, the candidate with the fewest first preferences assigned to him shall be eliminated.

14.7.1. If these candidates all have the same number of first preferences assigned to them, then B.14.7 and this paragraph shall be applied to each preceding count until one candidate is excluded.

    14.7.2. If no such distinction can be made between these candidates under the above 2 paragraphs, the remaining iterations shall be conducted under multiple scenarios. Each scenario shall eliminate one of the tied candidates.

        14.7.2.1 If the different scenarios described by B.14.7.2 result in the same winner of the election overall, the winning candidate shall become the Senator.
        14.7.2.2 If the different scenarios described by B.14.7.2. result in different winners of the election overall, the result will be considered a tie between the winners of the different scenarios and will be resolved in accordance with the Organic Law.

[Baron Alexandreu Davinescu]

The change only needs to be a narrow one, since the disputed language is in 14.7.  My suggestion:

If, after any iteration, there are two or more candidates with the fewest ballots assigned to them, the candidate with the fewest first preferences assigned to him shall be eliminated. If these candidates all have the same number of first preferences assigned to them, the candidate with the fewest second preferences assigned to him shall be eliminated, and so forth.

should be amended to

If, after any iteration, there are two or more candidates with the fewest ballots assigned to them, the candidate with the fewest first preferences assigned to him shall be eliminated. If these candidates all have the same number of first preferences assigned to them, the candidate with the fewest second preferences on ballots that are neither exhausted nor otherwise assigned assigned to him shall be eliminated, and so forth.

[Baron Alexandreu Davinescu]

There's a case pending on this, but I'm not sure it will actually get heard since it's clearly a "live controversy" and those aren't supposed to be subject to advisory opinions.  So we should definitely fix this, either way -- good initiative, Dama Miestra.

[Miestră Schivă, UrN]

The change only needs to be a narrow one, since the disputed language is in 14.7.  My suggestion:

If, after any iteration, there are two or more candidates with the fewest ballots assigned to them, the candidate with the fewest first preferences assigned to him shall be eliminated. If these candidates all have the same number of first preferences assigned to them, the candidate with the fewest second preferences assigned to him shall be eliminated, and so forth.

should be amended to

If, after any iteration, there are two or more candidates with the fewest ballots assigned to them, the candidate with the fewest first preferences assigned to him shall be eliminated. If these candidates all have the same number of first preferences assigned to them, the candidate with the fewest second preferences on ballots that are neither exhausted nor otherwise assigned assigned to him shall be eliminated, and so forth.

That's not how RCV works. Reversion to the last round is how you resolve ties.

[Baron Alexandreu Davinescu]

The change only needs to be a narrow one, since the disputed language is in 14.7.  My suggestion:

If, after any iteration, there are two or more candidates with the fewest ballots assigned to them, the candidate with the fewest first preferences assigned to him shall be eliminated. If these candidates all have the same number of first preferences assigned to them, the candidate with the fewest second preferences assigned to him shall be eliminated, and so forth.

should be amended to

If, after any iteration, there are two or more candidates with the fewest ballots assigned to them, the candidate with the fewest first preferences assigned to him shall be eliminated. If these candidates all have the same number of first preferences assigned to them, the candidate with the fewest second preferences on ballots that are neither exhausted nor otherwise assigned assigned to him shall be eliminated, and so forth.

That's not how RCV works. Reversion to the last round is how you resolve ties.

This is the existing language crafted by @Ian Plätschisch and @Glüc da Dhi S.H. , with only sufficient changes to resolve the current ambiguity to match their intentions.

[Miestră Schivă, UrN]

Yeah, and we should change it to fit in with international best practice!

[Baron Alexandreu Davinescu]

Yeah, and we should change it to fit in with international best practice!

I'll look to the original authors to chime in about that, since they know more than I.

[Ian Plätschisch]

I agree that this method of tiebreaking would be less confusing than our current one. However, I would rephrase as follows:

14.7. If, after any count, there are two or more candidates with the fewest ballots assigned to them, the candidate with the fewest ballots assigned to them after the previous count shall be eliminated. If the candidates were also tied after the previous count, the candidate with the fewest ballots assigned to them after the count before that shall be eliminated, and so on until one candidate is eliminated.

14.7.1. If no such distinction can be made between the tied candidates because all have the same number of ballots assigned to them after each count, the remaining counts shall be conducted under multiple scenarios. Each scenario shall eliminate one of the tied candidates.

        14.7.1.1 If the different scenarios described by B.14.7.1 result in the same winner of the election overall, that candidate shall be declared the winner.
        14.7.1.2 If the different scenarios described by B.14.7.1 result in different winners of the election overall, the result shall be considered a tie between the winners of the different scenarios and shall be resolved in accordance with the Organic Law.

[Glüc da Dhi S.H.]

The change only needs to be a narrow one, since the disputed language is in 14.7.  My suggestion:

If, after any iteration, there are two or more candidates with the fewest ballots assigned to them, the candidate with the fewest first preferences assigned to him shall be eliminated. If these candidates all have the same number of first preferences assigned to them, the candidate with the fewest second preferences assigned to him shall be eliminated, and so forth.

should be amended to

If, after any iteration, there are two or more candidates with the fewest ballots assigned to them, the candidate with the fewest first preferences assigned to him shall be eliminated. If these candidates all have the same number of first preferences assigned to them, the candidate with the fewest second preferences on ballots that are neither exhausted nor otherwise assigned assigned to him shall be eliminated, and so forth.

That's not how RCV works. Reversion to the last round is how you resolve ties.

This is the existing language crafted by @Ian Plätschisch and @Glüc da Dhi S.H. , with only sufficient changes to resolve the current ambiguity to match their intentions.

I should note that this is not how I ever intepreted that clause. This has affected earlier election results as well. See for example Florencia dec '19 where according to Ians interpretation it should have been a four-way tie, but my interpretation, as written out here: https://talossa.proboards.com/thread/13773/chancery-dec-florencia-senate-results?page=1&scrollTo=168434 lead to Acafat del Val being elected.

From what I can tell Ian did write the bit about the tiebreak, so perhaps my interpretation is wrong, but then at least this earlier result would also have been wrong.

[Glüc da Dhi S.H.]

Anyway I agree that Miestrăs proposal would be a better solution and I also like Ians rephrasing.

[Baron Alexandreu Davinescu]

Anyway I agree that Miestrăs proposal would be a better solution and I also like Ians rephrasing.

Great.  Seems fine to me, then, if both of you think it's the way to go.

[Glüc da Dhi S.H.]

I should note that this is not how I ever intepreted that clause. This has affected earlier election results as well. See for example Florencia dec '19 where according to Ians interpretation it should have been a four-way tie, but my interpretation, as written out here: https://talossa.proboards.com/thread/13773/chancery-dec-florencia-senate-results?page=1&scrollTo=168434 lead to Acafat del Val being elected.

From what I can tell Ian did write the bit about the tiebreak, so perhaps my interpretation is wrong, but then at least this earlier result would also have been wrong.

Correction, according to Ians interpretation it would have been a two-way tie between Mximo and Acafat because of the multiple scenarios clause (no matter who you eliminate first, it never ends in Breneir or IV actually winning)

[Glüc da Dhi S.H.]

Something still bothers me about this (other than contradictory interpretations of the law having decided two elections in different ways).

In the Maricopa case there is a strong argument that a tie is the just outcome. After all without the second preference of someone who already preferred Carlüs it would be a tie which includes Carlüs.

But that is not true in the Florencia case. Any two-way matchup between Acafat and any of the other candidates would have resulted in Acafat winning. So Id argue Acafat winning was also a just outcome. But this proposal (or the new consensus interpretation of the current law) would result in a tie.

Looking at previous rounds wouldnt catch it, because there were no previous rounds to look at.

The multiple scenarios clause only catches part of it. When all remaining candidates are tied and theres more than two candidates you're always gonna have a tie because you always have at least one scenario where each candidate is eliminated.

This is why Mximo stays in the race even though 2 out of 3 voters with a preference preferred Acafat.

That doesn't feel correct to me.

[Ian Plätschisch]

I guess we could change it so that whoever wins a majority of the scenarios is declared the winner, but that also doesn't feel right for some reason I can't put my finger on.

[Glüc da Dhi S.H.]

I guess we could change it so that whoever wins a majority of the scenarios is declared the winner, but that also doesn't feel right for some reason I can't put my finger on.

Yeah, I had the same thought and I understand what you mean. That doesn't feel right to me either.

I'm currently more thinking along the lines of

-if the scenario thing shows that some candidate can still win and some can't then wouldn't the next logical step be to continue counting with only those candidates that can still win before we declare an "organic" tie-

-in the Florencia case that means doing the whole scenario thing, coming to the conclusion that Acafat and Mximo can still win and then redoing the count without Breneir and IV in which case Acafat wins.-

,but how to put that into law in a way that is simple and unambiguous and also doesn't create an infinite loop of multiple scenarios?