Talk:Ranked voting/Archive 2

Are the last few sections useful?

Pros and cons of RCV

Is this section helping anyone? I believe it started out as an anti-FPTP polemic and got reined in. There’s nothing much wrong with it, but it’s a bit essay-like and not over-interesting. Also RV systems differ so much between themselves that arguing for or against them as a bloc is bound to concentrate on secondary concerns.

Uniqueness of votes

Does this help anyone? It’s all true and relevant, but looks like mathematical detail for its own sake.

Use by politics

A lot of work has gone into this, but the articles on individual voting methods provide similar information. I suspect no one really benefits from it and either it’s a maintenance burden or it goes out of date.

Proposal

If no one objects I will delete the section on Pros and Cons (with apologies to its authors), moving a mention of the need for machines to the lead para of the article. I will leave the sections on Uniqueness and Use as they are. I would need very little encouragement to delete the first of them, but in view of the amount of work that’s gone into it, I don’t want to do more than record my doubts about the value of the second.

I added tags to the sections in question. Tags are even more obtrusive than unwanted material. I hope I will remember to simply delete the Pros and Cons section if there is no dissent, and to remove the tag from the Uniqueness section if there is no support for removing it. (If I forget, please do it for me.) But I would leave the tag in place for the Use section until some sort of consensus emerges. Colin.champion (talk) 08:14, 1 August 2021 (UTC)[reply]

Second thoughts: There is no list of IRV elections in the IRV page (and the page is already rather long). On the other hand, there is such a list in the STV page, with the consequence that every change in the status of an STV system needs to be recorded twice on Wikipedia. So I suggest creating a new page, ‘List of elections using instant-runoff voting’, and copying all the IRV entries in the table to it. It would be worth checking that all the niche methods (such as supplementary vote) are covered by their own pages. Then the change would be non-destructive and I would be willing to carry it out in the absence of dissent. Colin.champion (talk) 13:12, 1 August 2021 (UTC)[reply]

Just as long as it avoids duplication that sounds reasonable, thanks. Reywas92^Talk 20:24, 4 August 2021 (UTC)[reply]

Thanks reywas. It turns out that there was a page History and use of instant-runoff voting which was a natural destination for the IRV entries in the table. There is some overlap with the prose account there, but I didn’t feel I was the right person to deduplicate further. Colin.champion (talk) 09:24, 6 August 2021 (UTC)[reply]

Accuracy of ranked voting methods

Here I would like to add a section, but I’m afraid someone will complain that it’s ‘original research’. I would describe it as a routine calculation (Wikipedia:No_original_research#Routine_calculations). At any rate, the appended source code makes it verifiable.

Accuracy of ranked voting methods

m method	3	6	10	15	25	40
FPTP	70.6	35.5	21.1	14.5	9.3	6.4
AV/IRV	85.2	50.1	31.5	21.6	12.9	7.9
Borda	87.6	82.1	74.2	67.0	58.3	50.1
Condorcet	100.0	100.0	100.0	100.0	100.0	100.0

It is possible to measure the accuracy of different voting methods under a given model of electoral behaviour (see comparison of electoral systems). This is particularly easy if we assume that voter opinions come from a unidimensional Gaussian distribution, and that the rightful winner of an election is the candidate closest to the mean/median. The table shows the percentage accuracy for four different voting methods, assuming an infinite number of voters and m candidates who are sampled randomly from the voter distribution.

All Condorcet methods give 100% accuracy, as does Coombs’; a single representative example (minimax) was included in the trials.

The software to produce these numbers was 89 lines of C (to which a few general purpose library functions and definitions need to be added). It follows (without the additional code). The greatest difficulty was in implementing IRV; the evaluation itself is less tricky. The evaluation works through the voter distribution from L to R; every time it crosses the mid-point between two candidates, their positions in the ballot are interchanged. The proportion of votes for each ballot is the difference between two gaussian tail areas. Once we have a list of ballots with frequencies, we invoke the voting methods in turn. The (optional) arguments to the program are the number of candidates and the offset of the candidate distribution from the voter distribution in standard deviations (always zero, but other values can be used to illustrate a weakness in the Borda count). A million simulated elections are generated.

C program

 #include <math.h>
 double rtlnorm(double),gaussv() ;
 
 int fptp(int **bal,double *qwt,int nbal,int m)
 { double q,*score=vector(m) ; 
   int t,winner ; 
   for(t=0;t<nbal;t++) score[bal[t][0]] += qwt[t] ;
   for(q=winner=t=0;t<m;t++) if(t==0||score[t]>q) { winner = t ; q = score[t] ; }
   free(score) ; 
   return winner ;   
 }
 int borda(int **bal,double *qwt,int nbal,int m)
 { double q,*score=vector(m) ; 
   int balno,t,winner,*b ; 
   for(balno=0;balno<nbal;balno++) 
     for(q=qwt[balno],b=bal[balno],t=0;t<m-1;t++) score[b[t]] += q * ((m-1)-t) ;
   for(q=winner=t=0;t<m;t++) if(score[t]>q) { winner = t ; q = score[t] ; }
   free(score) ; 
   return winner ;   
 }
 int minimax(int **bal,double *qwt,int nbal,int m)
 { double q,*qmin=vector(m),**r=matrix(m,m),qmax ; 
   int i,j,balno,t,winner,*b ; 
   for(balno=0;balno<nbal;balno++) for(q=qwt[balno],b=bal[balno],i=0;i<m-1;i++) 
     for(j=i+1;j<m;j++) r[b[i]][b[j]] += q ; 
   for(i=0;i<m;i++) for(qmin[i]=1,j=0;j<m;j++) 
     if(i!=j&&r[i][j]<qmin[i]) qmin[i] = r[i][j] ; 
   for(qmax=winner=t=0;t<m;t++) if(qmin[t]>qmax) 
   { winner = t ; qmax = qmin[t] ; }
   free(qmin) ; 
   freematrix(r) ; 
   return winner ;   
 }
 int av(int **bal,double *qwt,int nbal,int m)
 { int t,tau,balno,winner,loser,mleft,**b=imatrix(nbal,m) ;
   double qmin,*score=vector(m) ; 
   for(balno=0;balno<nbal;balno++) for(t=0;t<m;t++) b[balno][t] = bal[balno][t] ;
   for(mleft=m;mleft>0;mleft--) // mleft is the number of remaining candidates
   { for(balno=0;balno<m;balno++) score[balno] = 0 ; 
     for(balno=0;balno<nbal;balno++) score[b[balno][0]] += qwt[balno] ;
     for(qmin=loser=t=0;t<mleft;t++) if(t==0||score[b[0][t]]<qmin) 
     { loser = b[0][t] ; qmin = score[loser] ; }
     for(balno=0;balno<nbal;balno++) for(tau=t=0;t<mleft;t++) 
       if(b[balno][t]!=loser) b[balno][tau++] = b[balno][t] ;
   }
   winner = b[0][0] ;
   free(score) ; 
   freeimatrix(b) ; 
   return winner ;
 }
 
 int main(int argc,char **argv)
 { int (*alg[])(int **,double *,int,int) = { borda , fptp , minimax , av , 0 } ;
   char *algname[] = { "borda" , "fptp" , "condorcet" , "av" , 0 } ;
   int testno,i,j,t,l,r,ntests=1000000,nalg,rightful ;
   for(nalg=0;alg[nalg];nalg++) ;
   int m = (argc>=2?atoi(argv[1]):3) , m2 = (m*(m-1)) / 2 ;
   double offs=(argc>=3?atof(argv[2]):0) ,ql,qr,q ;
   int *win = ivector(nalg) , *ballot = ivector(m) , **bal = imatrix(m2+1,m) ; 
   double *c = vector(m) , *qwt = vector(m2+1) ; 
   xi *cm = xivector(m2) ; 
 
   for(testno=0;testno<ntests;testno++)
   { for(i=0;i<m;i++) c[i] = gaussv() + offs ;
     realsort(c,m) ; 
     for(t=i=0;i<m-1;i++) for(j=i+1;j<m;j++,t++)
       cm[t] = xi((c[i]+c[j])/2,(i<<8)|j) ; 
     xisort(cm,m2) ; 
     for(i=0;i<m;i++) ballot[i] = i ; 
     for(ql=1,i=0;i<=m2;i++,ql=qr)
     { if(i==m2) qr = 0 ; else qr = rtlnorm(cm[i].x) ; 
       qwt[i] = ql - qr ;
       for(t=0;t<m;t++) bal[i][t] = ballot[t] ; 
       if(i==m2) break ; 
       l = cm[i].i >> 8 ;
       r = cm[i].i & 255 ; 
       for(t=0;t<m-1&&ballot[t]!=l;t++) ; 
       swap(ballot[t],ballot[t+1]) ; 
     }
     for(q=rightful=t=0;t<m;t++) if(t==0||fabs(c[t])<q)
     { rightful = t ; q = fabs(c[t]) ; }
     for(t=0;t<nalg;t++) if(rightful==alg[t](bal,qwt,m2+1,m)) win[t] += 1 ; 
   }
   printf("m=%-2d, x=%.2f",m,offs) ; 
   for(t=0;t<nalg;t++) printf(" | %s: %4.1f",algname[t],win[t]*100.0/ntests) ; 
   printf("\n") ; 
 }

Library functions and definitions

I have a header file which includes stdio and stdlib; it defines a xi as a double/int pair and provides vector/ivector/xivector functions which allocate freeable space for doubles/ints/xis; it provides matrix and imatrix as matrix allocators for doubles/ints with associated functions to release memory; it defines a swap operator for ints; and it provides realsort/xisort which sort ascending on real values. All memory allocators zeroise the space allocated.

rtlnorm and gaussv return respectively gaussian right tail area and standard gaussian deviates.

If no one responds, I’ll probably try my luck by adding this section to the article.

It wasn't easy for me to find in this article a definition of "Condorcet" and a link to an article that discusses that. There is a section on "Condorcet completions", but I can't just jump there and learn what it means. There's a link to Condorcet criterion in the second paragraph of the first main section on "Principal ranked voting systems", but I had to search carefully to find it.

Might someone else modify the "Condorcet completions" section, so it defines the term and cites the separate article on ""Condorcet criterion"? I don't know enough to easily do that.

Your table is interesting and potentially useful. However, you should state clearly that you think that that you assume that a Condorcet winner should win when there is one. Then you should note that your simulations produced a Condorcet winner in every one of the million simulations you did to produce that table.

Am I correct that Arrow's impossibility theorem says that there may not be a Concorcet winner, and your simulation says that such events are extremely rare, to the extent that your simulation methodology matches actual voting behavior?

Have you done a literature search for a publication that includes something like this table? If no, I suggest you do that. If someone else has already done something like this, you can cite that work. If you have done such a lit search, it should identify a journal that might like to publish your simulation and other authors who have published similar work, whom you could ask to review this simulation and article and comment. (You probably know that there are Wikipedia articles on Condorcet winner criterion and Condorcet method. Recent references in those and the "Arrow's impossibility theorem" articles should give you people and / or journals to contact.)

What you've done seems quite valuable to me, but I'm not an expert in voting systems.

I'd also like to see your same analysis applied to data from real elections where they've used something like ranked choice voting, e.g., the recent Democratic primary if the 2021 New York City mayoral election: Was there a Concorcet winner? If yes, did that person actually win? If that person is eliminated, is there still a Concorcet winner, and would that person have won?

It would be good to do similar analyses of other elections where ranked choice voting was used, especially ones where there was controversy.

It is my understanding that Gayle McLaughlin was elected as Mayor of Richmond, CA, using some form of ranked choice voting. However, I don't see that documented in the Wikipedia article on her. If you can get the raw data on that election, it could be interesting and help sell your analysis and result to a wider audience.

Thanks for this. DavidMCEddy (talk) 14:45, 1 August 2021 (UTC)[reply]

David – thanks for your comments, which are very helpful.

Your first point is that the expressions ‘Condorcet this’ and ‘Condorcet that’ are introduced without definition and with no prominent wikilink. The article assumes (and I think always has assumed) that the reader is familiar with these concepts; but now that you raise the matter, I agree that this is indefensible. ‘Ranked voting’ might well be the first article a reader sees on the topic, and should not assume prior knowledge of other articles which might be quite involved. I have sought to address this by adding a short section on the theoretical concepts needed. It is quite condensed, but I hope it covers the necessary ground.

When we come to the table, I have more to say in my own defence. I do not assume that ‘a Condorcet winner should win when there is one’ – in my opinion, this would be as bad as assuming that a Borda winner should win whenever one existed. What I assume is exactly what I state, which is that ‘the rightful winner of an election is the candidate closest to the mean/median’ of the voter distribution And given the symmetry of the gaussian distribution, can you think of any reasonable alternative assumption? I suspect that the results in the table look too good to be true to you and that you suspiciously attribute a circular argument to me, when in fact the results are indeed true with no circular assumptions thanks to the under-appreciated median voter theorem.

Also my evaluation does not show that the absence of a Condorcet winner is ‘extremely rare’ under the assumed circumstances: this absence is known a priori to have probability exactly 0, and the table merely confirms numerically something that has been mathematically proven through the median voter theorem.

I have not performed a specific search for evaluations like my own, but I am reasonably familiar with the literature of empirical evaluations, having come to this article from writing about such evaluations in comparison of electoral systems. Even 40 years ago the evaluations which people published were much more complicated than my own simple table; and while the complexity of the evaluations serves a purpose, the results are quite hard for the lay reader to interpret. (Follow the links to papers with Tideman, Green-Armytage or Darlington as first author from comparison of electoral systems if you doubt me.) I was wanting to offer an ‘evaluation for dummies’ in place of the evaluations for voting experts which can be found in journals. I believe that my results are consistent with those in recent papers (especially Darlington’s latest), but you have to abstract out a lot of complexity to see what Darlington was actually showing.

Can I politely decline to get involved in measurements of real elections? I don’t want to get sucked in too far. Colin.champion (talk) 08:16, 2 August 2021 (UTC)[reply]

That's great. More thoughts:

The text accompanying your table should say that it's a simplified version of other published research and include at least one note giving citations to some of the best and most recent of the relevant papers in that regard.
If it's a simplified version of what's been published in refereed academic journals, that strengthens the case that it's encyclopedic, I think, because it suggests that it might not be publishable by itself but still relevant.
Of course it's OK for you to "politely decline": I'm not in a position to tell you what to do.

Thanks again. DavidMCEddy (talk) 15:14, 2 August 2021 (UTC)[reply]

@Colin.champion: Might it be convenient for you to find a reference documenting the circumstances under which the probability is zero of an outcome violating the Condorcet Criterion? If yes, might it also be convenient for you to add something about that to the articles on median voter theorem and Arrow's impossibility theorem?

For me at least, it greatly strengthen's the case for using a Condorcet criterion with ranked voting. And it also increases the value of the median voter theorem while explaining why Arrow's impossibility theorem can be ignored for all practical purposes. Thanks again DavidMCEddy (talk) 18:24, 2 August 2021 (UTC)[reply]

David – thanks for your comments, I’ll wait to see if anyone else adds anything. I’m afraid I don’t know of any references of the type you ask for; I doubt there’s much to say, and if there’s anything at all, it’s quite likely that no one would have said it. Until a few days ago I thought it was only Condorcet systems which possessed the median voter property, and I was surprised to find while researching for this article that Coombs' method also possesses it. I’ve never seen a general statement. Colin.champion (talk) 07:10, 3 August 2021 (UTC)[reply]

I know about voting systems and Accuracy of ranked voting methods seems completely obscure, so if I can't easily see what the heck this it, I'd call it nonsense, 100% OR, not simple calculations, that deserves no attention on wikipedia. Tom Ruen (talk) 07:54, 3 August 2021 (UTC)[reply]

Let x₀, x₁, x₂ be real numbers understood as the positions of candidates along a continuum. Let B(x₀,x₁,x₂) be an indicator function whose value is 1 if, given infinitely many voters whose positions are governed by a standard gaussian distribution, the Borda winner is the candidate closest to the mean of the voter distribution, and whose value is 0 otherwise. Then the entry for m=3 against the Borda count in the table is the integral over x₀ x₁ x₂ of B(x₀,x₁,x₂)×g(x₀)×g(x₁)×g(x₂) where g() is the standard gaussian density function.

The integral needs to be evaluated numerically. This is what I would call a routine calculation. I don’t think you question the accuracy of my figures, and there’s no difficulty in principle to verifying them. I think what you regard as ‘obscure’ and ‘nonsense’ is the interpretation of this integral as the accuracy of a voting system. You’re at perfect liberty to think so, but my interpretation is by no means original research, being well established in the literature. See for instance T. N. Tideman and F. Plassmann, ‘Modeling the Outcomes of Vote-Casting in Actual Elections’ (2012).

Obviously if this (and other) references are needed, there’s no difficulty in providing them. The papers in question are not an easy read. Colin.champion (talk) 09:11, 3 August 2021 (UTC)[reply]

Specifically, if you download the Tideman and Plassman paper from researchgate, look at the para running from the page numbered 10 to the page numbered 11. Compare it with the Introduction (which is easy to read) of T. N. Tideman and F. Plassmann, ‘Which voting rule is most likely to choose the “best” candidate?’ (2012). Together they describe an evaluation which is the 2-dimensional generalisation of the one in my table, but with many added complexities. Colin.champion (talk)

Further thoughts: anyone who questions the accuracy of my figures may compare them with Table 5 in a 1978 paper.^[1] The simulated elections there were in 4 dimensions with 4 candidates, a finite number of voters, and a small number of trials. The ‘Condorcet efficiencies’ were: Coombs:99%, Borda 86%, IRV 60%, and FPTP 33%. 1000 voters in this paper is ‘a poor person’s approximation of an infinite election’, and said poor person is permitted to identify Condorcet efficiency with accuracy.

At present I’m hanging fire on my table, not because I doubt its truth, relevance or verifiability, but for fear that it shows IRV in a bad light, and that without relevant context it might exaggerate IRV’s failings. There are at least two reasons why IRV might be more acceptable in practice than the figures suggest. The first lies in the fact that some of the advantage of electoral reform comes from removing the fear of wasted votes, which leads to candidates being imposed by party machines rather than truly reflecting voters’ preferences. A voting method might effectively combat this without being particularly fair in any other sense.

The other fact lies in the limitations of a spatial model. Consider a US primary election in which there are 5 candidates, numbered 0–4 from left to right. (Caveat: I know nothing about American politics.) 1 is the incumbent; 3 is a celebrity challenger; 0, 2 and 4 are local politicians. Voters see 1 and 3 as the viable winners, and choose between them according to their positions on the spectrum. If they vote sincerely, they may vote according to patterns like 1-3-0-2-4, whereas if they vote tactically they may vote 1-0-2-4-3: neither pattern conforms to a spatial model.

There are good reasons to expect Condorcet methods and the Borda count to work well given sincere voting. FPTP and IRV might work better than spatial evaluations suggest. Tactical voting might have different effects from in purely spatial elections, and not much is known about it even there. Coombs’ method might work disastrously, eliminating the two strongest candidates in the two first rounds.

So the trouble, as I see it, is not that my figures are original research, but that original research is needed to put them in a proper context. Colin.champion (talk) 13:57, 6 August 2021 (UTC)[reply]

Doesn't the literature in general and your analysis specifically favor Condorcet when there is a Condorcet winner?

Am I correct that Condorcet says that a candidate who wins all pairwise contests should win the election?

Am I also correct that the relevance of your simulation hangs on two things:

How often are the assumptions of the median voter theorem (or a multivariate generalization you outlined) violated?
When those assumptions are violated, what percent of those cases do not have a Condorcet winner?

This analysis suggests that it would help to analyze data from real cases where ranked voting has been used, and compute the following from those cases:

(a) How often was there a Condorcet winner?

(b) When there was a Condorcet winner, how often did that person actually win under the rules used? (And maybe how often would that person have won under other rules that are discussed in the literature?)

(c) When there was not a Condorcet winner, what can we say about who would have won under alternative voting rules, and how does that compare with what actually happened and what people think is fair?

Thanks. This is interesting. DavidMCEddy (talk) 16:36, 6 August 2021 (UTC)[reply]

David – many questions, some quite involved... I’m not an expert, and can only answer in part.

* Doesn’t the literature favour a Condorcet winner when he or she exists? Many people would favour a Condorcet winner in such cases, but I personally would be more hesitant. In a famous example I would say that neither the Condorcet nor the Borda winner was a compelling choice from the ballots cast.

* Your summary of the Condorcet criterion is correct.

* All significant evaluations draw voters from a symmetric spatial model. The conditions of the median voter theorem may be violated by small sample effects, but as the sample size increases the optimality of Condorcet methods is increasingly ensured by the experimental design. To my mind this is unsatisfactory, and I wish people had tested the robustness of their assumptions by using intentionally asymmetric distributions (eg. Gaussian mixtures or hybrid models). At the moment we can say that under a normal approximation Condorcet methods are streets ahead of the alternatives, and that there’s no reason to suppose that this conclusion would be materially affected by a different model, but that the matter hasn’t been fully explored.

* I think people have looked quite hard at how often there is a Condorcet winner in practice, and generally concluded that the number of cycles seen in empirical counts is no more than would be expected from small sample effects. Tideman and Plassmann’s ‘Modeling the Outcomes of Vote-Casting in Actual Elections’ is the only paper I’ve read which says this, but I believe there are others.

* Tideman and Plassman specifically conclude that a 2-dimensional Gaussian model fits their observations quite well. I would treat this conclusion with caution (they didn’t prove that equally plausible models didn’t fit equally well, and they reduced all elections to 3 candidates), but I suspect that the pathological examples sometimes constructed to illustrate weaknesses in voting methods have little connection with reality.

* And I have no knowledge at all relevant to your final questions. But I’m amused by para 8.5 of a paper by Darlington in which he states that IRV was rejected in Burlington when its failure to elect a Condorcet winner was judged ‘bizarre’.^[2]

And finally... my current intention is to put my table in the article on comparison of electoral systems and summarise the topic in the current article. It’s silly to discuss simulated evaluations without giving any indication of the results which arise from them and silly to leave the reader dependent on the over-complicated presentations found in reasearch papers; but it’s dangerous to present such stark results except in surroundings which allow the reader to recognise the caveats they need. Colin.champion (talk) 08:52, 7 August 2021 (UTC)[reply]

Thank you for doing this. It seems important.

I know people pushing ranked voting in California and Missouri. I should ask them about their understanding of the issues raised here and in Comparison of electoral systems. DavidMCEddy (talk) 10:55, 7 August 2021 (UTC)[reply]

Thanks David. I’ve added the table to Comparison of electoral sytems, and will add something here later depending on how it fares. Colin.champion (talk) 11:36, 9 August 2021 (UTC)[reply]

References

^ J. R. Chamberlin and M. D. Cohen, ‘Toward Applicable Social Choice Theory...’, (1978).
^ R. B. Darlington, ‘Are Condorcet and Minimax Voting Systems the Best?’ (v8, 2021).

re: "and the counting procedure – depending on the nature of the voting method – is more complicated and slower"

If it "depends" then we shouldn't say "is" we should say "could be."

Shouldn't we change it to something like: "and the counting procedure, if done manually, would be more complicated and slower." myclob (talk) 00:02, 25 December 2021 (UTC)[reply]

I concur. Wikipedia:Be bold. Make the change. If others don't like it, they can revert or change it further. DavidMCEddy (talk) 00:46, 25 December 2021 (UTC)[reply]

Done. Thanks! myclob (talk) 02:54, 26 December 2021 (UTC)[reply]

re: "often requiring mechanical support"

What is mechanical support? Shouldn't we just say "computer counting"? In developed countries we typically use computer to scan forms, and all the math is done and "mechanical support" doesn't really make sense. Shouldn't we say "if done manually" or something? myclob (talk) 00:04, 25 December 2021 (UTC)[reply]

This is similar to the question above, so I'll respond here. I think the section should be substantially expanded, and there's a lot to say about the election administration difficulties in a ranked choice system that don't exist or are very different in a single-vote system. I don't feel strongly about wording changes, I think WP:BB is in order. - Astrophobe (talk) 00:44, 25 December 2021 (UTC)[reply]

I used "and the counting procedure, if done manually, would be more complicated and slower" and deleted the part that said "mechanical support" because I don't know what mechanical support means. Thanks again! myclob (talk) 02:55, 26 December 2021 (UTC)[reply]

The image and the table under "Spatial models"

The image and the table are supposed to go together. However, I don't know how to do it. How they are now, it looks like the table is with the other section. I would like them to be side-by-side, and the image should have a border... Can anyone help?

This is the image and the table:

Ballot	Count
A–B–C	36
B–A–C	15
B–C–A	15
C–B–A	34

Spatial Model

Myclob (talk) 21:07, 29 December 2021 (UTC)[reply]

Better Image

Should we use this image?

Myclob (talk) 23:12, 29 December 2021 (UTC)[reply]

Merge in Group voting ticket

Could I propose we merge in the Group voting ticket to here, or strip it out from here? The article on its own is very short, and mostly duplicated here and there. Felix the Cassowary 03:49, 12 July 2005 (UTC)[reply]

That sounds like a lot of work. I don't know how to do it. However, I did add Group voting ticket under "See Also". I hope that helps! Myclob (talk) 23:35, 29 December 2021 (UTC)[reply]

Propose move

I propose we move this page to ranked ballot, as that seems to be the most unambiguous term we can use to describe the content of the article. As of now, the lead section looks more like a disambiguation page than it does the start of an article, and I don't see how we can overcome that unless we actually disambiguate preferential voting properly. Scott Ritchie 10:56, 17 December 2005 (UTC)[reply]

So are you proposing that "preferential voting" should become a disambiguation page? I agree that this page is valuable for that purpose. KVenzke 04:22, 9 January 2006 (UTC)[reply]

I think people who live in countries where this system is actually used should have some say in this. In Australia we call it preferential voting. I'm not sure what they call it in Ireland, but I'm pretty sure it is not "ranked ballot." Adam 05:14, 9 January 2006 (UTC)[reply]

I would oppose this move; these systems are called preferential voting in English. I would have no idea what ranked ballot means until I read the article. Adding an article preferential voting (disambiguation) may help. Septentrionalis 05:58, 9 January 2006 (UTC)[reply]

Reading the article more thoroughly, I agree it's a mess. The disambiguation section is good. The next section is good if we are discussing ranked ballots, which is what this article apparently was for originally. The rest of the article contains information specific to Instant-runoff voting and Australian electoral system. KVenzke 06:01, 9 January 2006 (UTC)[reply]

Not moved. And yes, please diambiguate. —Nightstallion (?) 08:31, 10 January 2006 (UTC)[reply]

[Ranked ballot] now re-directs to [ranked voting]. Is that what you wanted? If so, can we archive this? If not, tell me. I was trying to figure out disambiguation, but I am not awesome. — Preceding unsigned comment added by Myclob (talk • contribs) 23:49, 29 December 2021 (UTC)[reply]

Category:Voting systems vs. Category:Preferential electoral systems

Category:Preferential electoral systems is itself a category within Category:Voting systems. — Robert Greer (talk) 20:07, 7 May 2009 (UTC)[reply]

Category:Voting systems Doesn't exist any more. Myclob (talk) 23:55, 29 December 2021 (UTC)[reply]

"See Also" at the top, and at the bottom

I don't think we should have a "See also" for two things at the top, and then more at the bottom.

I am going to delete "See also: Single transferable vote and Instant-runoff voting" from the top.

Myclob (talk) 00:02, 30 December 2021 (UTC)[reply]

Criticism of preferential voting?

Oughtn't there be a section about criticism of preferential voting. In California, where a number of municipalities have adopted it, it has come under fire. Should some of the criticism be in this article? 173.213.247.42 (talk) 18:56, 19 August 2011 (UTC)[reply]

The article says that RCV is more complex. What other critisisms were discussed in California? You should add them Myclob (talk) 00:11, 30 December 2021 (UTC)[reply]

re: "In the following years, the English mathematician Charles Lutwidge Dodgson (better known as Lewis Carroll) and the Anglo-Australian Edward Nanson published new voting methods."

If these other voting methods are not ranked-voting, than I don't see why we would include them. Can we just delete this sentence? I mean, this shouldn't be a history of every voting system. This should just be a history of ranked-choice voting. There are other articles for more general voting histories. myclob (talk) 23:42, 25 December 2021 (UTC)[reply]

The sentence doesn't have a reference: It should have two, one for what Dodgson / Carroll did and another for Nanson's proposal(s). If someone thinks this sentence belongs, they can provide credible reference(s). Thanks for raising this question. DavidMCEddy (talk) 23:54, 25 December 2021 (UTC)[reply]

Thanks. I'll remove it. myclob (talk) 01:35, 26 December 2021 (UTC)[reply]

It’s trivial to verify that Nanson's method is a form of ranked voting; so is Dodgson’s. Otherwise, as you say, there would be no point in mentioning them here. There is no need to reference every statement, only every statement that can reasonably be challenged, which does not apply in this case. Myclob has made a mass of other changes (over Christmas) which I don’t think are justified, but fuller consideration from me will have to await decent access to WiFi. Colin.champion (talk) 15:53, 26 December 2021 (UTC)[reply]

It may be "trivial to verify that Nanson's method is a form of ranked voting; so is Dodgson’s" PROVIDED you know what they are.

Sadly, it was NOT trivial for me to get that information yesterday. If it had been, I would have done so before supporting the suggestion to delete that recommendation.

That's why we need references.

I wish you "decent access to WiFi." The world needs your input. The best Wikipedia articles tend to have had editors with very different perspectives, who fought over critical questions of substance, according to Feng Shi; Misha Teplitskiy; Eamon Duede; James A. Evans (4 March 2019). "The wisdom of polarized crowds" (PDF). Nature Human Behaviour. 3 (4): 329–336. arXiv:1712.06414. doi:10.1038/S41562-019-0541-6. ISSN 2397-3374. PMID 30971793. Wikidata Q47248083., cited in the section on "Articles on contentious issues" in the Wikipedia article on "Reliability of Wikipedia". DavidMCEddy (talk) 17:23, 26 December 2021 (UTC)[reply]

Well the wording as it was didn't really add anything to the article. It just said: "In the following years, the English mathematician Charles Lutwidge Dodgson (better known as Lewis Carroll) and the Anglo-Australian Edward Nanson published new voting methods." It should read "new ranked-voting method" if we want to keep it in the article, but really if they added to our understanding, shouldn't we say what those methods were? Also, if we want to be informative, what years? Theory and writing might not really qualify as worthy of mention, unless they were able to get anyone to try their counting methods. Right? Those might be years worth mentioning. myclob (talk) 18:21, 26 December 2021 (UTC)[reply]

@DavidMCEddy: I’m sorry the information wasn’t easy to find. When I left the article it mentioned Nanson’s and Baldwin’s methods in the text (see https://en.wikipedia.org/w/index.php?title=Ranked_voting&oldid=1038395158#Alternative_vote_/_instant-runoff_voting_(IRV) ) with a supporting link. This sentence got deleted without good reason by whoever expanded the discussion of STV. Colin.champion (talk) 09:01, 27 December 2021 (UTC)[reply]

Is this the wording you want back in? "Baldwin’s and Nanson’s methods use more complicated elimination rules based on the Borda count. They are Condorcet-consistent."? myclob (talk) 15:29, 29 December 2021 (UTC)[reply]

Should we say that the counting method of ranked voting is or could be slower and more complicated than that of FPTP? What I wrote originally was correct: it is slower and more complicated. The only reasons for the caveat ‘depending on the voting method’ are (a) to protect against the pedantic quibble that FPTP could itself be implemented as a trivial form of ranked voting (in which case it would not be slower), and (b) that the degree of extra slowness depends on the method. To say that it “could be” slower presents a categorical fact as a hypothesis.

I can’t see why you object to the word ‘mechanical’: the article on vote counting talks about counting being done ‘manually or by machines’. It is not general to use mechanical support for counting even in developed countries: the process is purely manual in the UK and the same Wikipedia article says that FPTP ballots are ‘often counted manually’. The machines may indeed incorporate software, but this fact does not need to be mentioned.

There was no mystery about Nanson’s and Dodgson’s contibutions. The section on history begins with a {{main}} header:

The article Electoral system contains a discussion of the ‘single winner revival’ which mentions these two contributions and supplies links and references. The article on ranked voting is largely a summary of matters discussed in greater detail elsewhere, with {{main}} headers pointing to fuller treatments. No one should complain that information is missing until they have followed the links. Moreover Dodgson’s method gets a link further down the article.

I see that you made a huge number of changes, mostly minor rewordings. It is extremely discourteous to try to impose your own stylistic preferences in this way. Your written English is not better than mine, and – so far as I have been able to determine – you have appreciably reduced the accuracy of the article. No one can be expected to vet a huge number of unnecessary changes to see how many errors have been introduced. I shall revert to the state of a few days ago. I suggest you limit your changes to those which you believe to be genuine improvements, and argue them here if you are in doubt. Colin.champion (talk) 14:59, 30 December 2021 (UTC)[reply]

Ranked Voting is a type of instant runoff

Doesn't it make sense that this is the larger category: https://en.wikipedia.org/wiki/Category:Instant-runoff_voting. Why else is it called a "category?"

There are multiple types of Instant-Runoffs, and Ranked Voting is one of them.

Instant-Runoffs makes sense as a broader category. It is any system that you do a run-off, without having to do a second vote. It happens in an instance.

Ranking the candidates is one way of doing an instant runoff. Ranking is more specific.
You could also grade them Score voting, or
mark all the ones that are acceptable Approval voting.

There are many ways of doing an instant runoff.

I know this page says that instant runoff is a type of ranking, but that doesn't make any sense: https://en.wikipedia.org/wiki/Instant-runoff_voting.

As you can see, I came up with 3 types of instant runoff. One of them is Ranked Voting, so Ranked Voting MUST be a type of Instant-runoff voting.

Myclob (talk) 00:46, 30 December 2021 (UTC)[reply]

That's not true, see for example Borda count, which is a type of ranked voting in which instant runoffs are not necessary. A ranked voting system is any system in which voters can rank more than one candidate on their ballots. Instant runoff voting is a very special example of such a system. But the logic isn't what matters really, the important question is verifiability. One complication is that in the U.S., "ranked choice voting" is often used interchangeably with "instant runoff voting", but the topic of this article is clearly not just that American linguistic quirk but rather all types of ranked voting systems, since it includes as subsections the Borda count and even instant runoff voting itself as an example of the broader topic. And there are plenty of examples of reliable sources clearly treating instant runoff voting as an example of ranked choice voting. In this paper, Instant runoff voting is listed as the fourth variety in the section "Types of Ranked Voting". If ranked voting were a type of instant runoff voting, then it would not be possible to have examples of ranked voting that are not also examples of instant runoff voting, but this article shows that Preferential block voting as such an example. And just for variety, here's a computer science paper proposing a type of ranked voting that is not a type of instant runoff voting. Definitions of ranked voting also make it clear that it is broader than the specific instant runoff voting idea. On this point I'll be brief and just cite this article, where the first sentence of the abstract is "Ranked voting data arise when voters select and rank more than one candidate with order of preference". Clearly that's a much broader idea than instant runoff voting, and you would find the same pattern if you click on any of the links here. Even this American county clerk says "Ranked Choice Voting (RCV) is a type of ranked preferential voting" — remember that American election officials use "ranked choice voting" to mean "instant runoff voting" (a fact that is itself attested in lots of sources, see e.g. here and examples of the phenomenon here and here), so in the lexicon that currently (and in my opinion correctly) prevails on Wikipedia, the claim she's making is that instant runoff voting is a type of ranked voting. Here's an article that gives a nice accounting of the situation: "we use the term instant-runoff voting [...] Before the system was implemented in San Francisco the Department of Elections changed the name to ranked-choice voting [...] In academic comparative studies IRV is one of several types of preferential systems and is called the alternative vote." Similarly this advocacy group uses that lexical difference, defining "ranked choice voting" to mean "instant runoff voting" and then use "preference voting" to mean what Wikipedia currently calls "Ranked voting". So if you're bothered by the use of language here you might propose a page move from this page to something like "preference voting", which could help eliminate any confusion that the current page title causes. - Astrophobe (talk) 02:24, 30 December 2021 (UTC)[reply]

User:Astrophobe Thanks for explaining that. I am freaking out now because user:Colin.champion undid all of my changes, and so I guess I have to create a discussion justifying each one below. I will read your explanation that verifies the categories used by experts. I am just trying to use normal English and logic to understand the meaning, but you are right that verifiability is what matters. The ranking of voting, allows instant runoff, because you go to the alternate votes. And so I understand the American linguistic complication of the two, I think. Anyways, thanks again for explaining all that. I do like your idea, because ultimately dumb people like me who are reading about the subject for the first time but have advanced degrees in other subjects, should be able to to understand the logic of the groupings, or at least have smart people explain it so the groupings make sense. Thanks! Myclob (talk) 16:46, 30 December 2021 (UTC)[reply]

We need to re-write this whole article. The Wikipedia manual of Style says: "editors should avoid ambiguity, jargon, and vague or unnecessarily complex wording."

See the manual of Style (https://en.wikipedia.org/wiki/Wikipedia:Manual_of_Style).

It says: "editors should avoid ambiguity, jargon, and vague or unnecessarily complex wording."

I think this whole article needs to be re-written.

Can anyone tell me what this means?

"Logical voting criteria can be thought of as extrapolating the salient features of examples into infinite spaces of elections. The consequences are often hard to predict: initially, reasonable measures contradict and reject otherwise satisfactory voting methods."

I say we delete it or replace it with this:

"Logical voting criteria can be thought of as gathering the crucial features of (examples into, LOL) all the possible combinations of elections. The results are often hard to predict. Initially, reasonable measures can have unintended consequences."

But even then, who is gathering the crucial features? A computer? Who thinks what is reasonable? What measures are we talking about? Are we talking about voting requirements? If we are, can we just say "voting requirements"? WHO IS REJECTING "otherwise satisfactory voting methods?" Myclob (talk) 18:55, 30 December 2021 (UTC)[reply]

Probably! A binary search in the article history might help us identify the person who wrote these bad paragraphs. Tom Ruen (talk) 19:10, 30 December 2021 (UTC)[reply]

The author is @Colin.champion: from Aug 6, 2021. [1]

References

[chamberlin-1] J. R. Chamberlin and M. D. Cohen, ‘Toward Applicable Social Choice Theory...’, (1978).

[darlington-2] R. B. Darlington, ‘Are Condorcet and Minimax Voting Systems the Best?’ (v8, 2021).

[1]

[2]