The thing about Cumulative Voting is it gives each voter multiple votes, say three when three seats are to be filled.
This can be good - it means one less change if people are already casting multiple votes - say in city elections where wards are not used. But in all provincial and federal elections -- now anyway -- voters cast just one vote. CV would require change to multiple voting in those cases. And another thing about CV is it uses multiple-member districts. That is great - MMDs or some pooling of voters is required for any form of PR. Cumulative Voting ironically would give each voter multiple votes then allow them to cast all of them just on one candidate - In fact voters of a small minority group would likely have to lump all their votes on one candidate to get him or her elected. So under CV, we give each voter multiple votes, then push each of the minority group voters to vote just on one candidate with all their votes. (thus doing the equivalent of what single voting in a multiple-member district would do) For prov and fed elections, it is easier to leave each voter casting just one vote and switch to MMDs, and immediately you have SNTV, a crude form of PR - but one that reaps same results as STV in many cases. Add transferable votes (ranked ballots) and you have STV. STV ensures mixed representation at the district level and seat allocation in due proportion to party votes at (MMD) district level (although derived by direct vote for specific candidates).
We can look at the benefits said to be produced by Cumulative Voting - more bipartisanship, less polarization, more voter choice, better representation. And we see that the first two are also aspects of STV where they are deeper than that produced by CV. I believe that where votes are transferred across party lines, such as STV, you have more "blending"/"bleeding" AKA overlap between parties.
Such flow does not happen under CV -- there are no transfers under CV but a voter can vote for candidates of different parties - just there would be no way of knowing that, due to secret voting. Under CV, if a candidate gets 99 votes, you don't know if 33 voters gave him 3 votes each or if 30 voters gave him 90 and nine came, one each from a voter who also voted for another candidate - you just don't know. But I think any voting system will not reconcile the irreconcilable - just as under Malta STV you have two-party system just as under NZ's MMP you have antagonism between Labour/Greens and National Party. Even under CV, you are not going to have much sharing between Republicans and Democrats. You may have more of a tolerant relationship though, such as perhaps shown by western Allies to USSR during WWII. But CV in Illinois, or any form of PR today, alone means that a party that makes the reform and any party that supports it was/is showing they recognize the right of the other to have representation - a step in the right direction. CV produces voter choice just as any system using MMD does. More candidates... Multiple candidates even within the party - although not for minority groups that are working hard to concentrate their group's votes to get just one elected. There would be no choice, or no real choice. There might be two candidates say of the Latino community, but Latino voters would be pushed just to concentrate on one.
A candidate asking for a voter to vote for him or her always happens - but in STV you can at least give back-up preference to another of the same party in case it is needed.
But in CV where minority voting bloc is small, it can afford just one candidate if it is not to split the group's vote.
===========
Under FPTP a candidate says "vote for me."
Under Block Voting, a candidate says, "give one of your votes to each of the members of the slate I am on."
Under STV, a candidate says "mark your number one for me and mark secondary preferences for the other candidates of my party."
Under CV, candidate perhaps says "give all your votes to me."
Under MMP, district candidate says "vote for me and mark your top-up vote for my party."
It seems that to a large degree only under STV can a candidate loosen up and be to a degree free from the zero-sum games that is competitive electioneering.
======================
"Better representation" - true if a minority group does in fact manage to get representation.
I would say under any other form of PR, mixed representation is more certain than CV.
Like CV, SNTV is also plurality but mixed representation is assured under SNTV -- no party can take all the seats in a district unless it has like 80 percent of the votes and they carefully distribute their votes. And fair, scientifically-balanced mathematically-derive representation is almost certainly more dependably produced under STV and MMD than it is under CV. CV uses plurality method, as flawed as it is under FPTP, CV uses multiple voting - as obscuring as it is under Block Voting CV uses lump-able voting - something that Canadian government elections have not used since 1910 (or so) in Toronto school board elections. On the other hand CV uses MMDs, which is a good thing - less gerrymandering, more voter choice. CV uses multiple voting, which is already done in city elections. Under CV, a voter's votes can be concentrated behind just one (or sometimes two) candidates to try to ensure that some representation is elected by the voting bloc that the voter belongs to. Such concentration is also done under STV. Concentration under STV is done by votes being transferred, as back-up preferences come into play as the less-popular candidates of a party (or of the more general left or the right) are eliminated.
Under CV, a voter can concentrate his or her votes on a specific candidate; but under STV a voter only has one vote so the group as a whole (but not the individual voter) can form up behind a lesser and lesser number of candidates to have best chance of getting representation - until the last seat is filled. Still under STV, a party or voting bloc will not get representation if it does not have quota (or if it does not have plurality when the field of candidates is thinned to one more than the number of remaining open seats, if that in fact happens). In CV, concentration is done by voter himself or herself lumping votes on one candidate Whether that candidate is elected or not elected. The votes are locked in, and concentration strategy may work or not, with no back-up mechanism. Under CV, even with the possibilities allowed by lump-able multiple voting, as even under STV, not every minority group will get representation. STV has built-in adjustment ability that CV does not have.
If voter guesses wrong under CV, because the votes are locked in, a minority group, or a majority group, may not get the representation its support in fact means it is due.
(And sometimes under CV the voter relies on the expected behaviour of the masses and adopts the strategy of giving one vote to the leading candidate and two to the lesser candidate so that both are elected. But if all - or if a great number of voters - do it, such a strategy would actually elect the less-likely candidate instead of the leading candidate. So you must not only know whom you want elected but also know how others will vote if you are to plan your votes right.
This can almost happen even under STV - on past occasions people tried to help up the less-likely candidate by voting for him over the leading candidate, then were surprised when the leading candidate barely squeaked in. If people don't vote for him, the leading candidate is no longer the leading candidate.
CV's multiple voting means that knowing the amount of voter support is I think impossible to actually measure.
if a CV system allows three votes per voter, say with 200,000 voters, altogether casting 600,000 votes, Democrats say would take 240,000 votes
you could say that Democrats altogether got 40 percent of the vote, but how many voters voted for one or more Democrat candidates?
Of the 240,000 votes, are these one or two from each voter so 100 percent support? Not likely.
Or is it two votes from each of 120,000 voters - 60 percent of the voters? Possibly. Or is it support of just 80,000 voters - 40 percent? Maybe.
Can't know just from vote totals.
If a specific candidate gets 30,000 votes, is it (partial) support of 30,000 voters or support from just 10,000? Can't know just from vote tallies. Single voting as in STV or SNTV, or in both levels of MMP, does give you that knowledge. (And the open process of STV transfers means you know where votes came from in many cases). it makes sense that some would tout the virtues of CV - because it is a form of PR - but unfortunate that they do not see that CV is not the only form of PR and that the good that CV does, is also done by any form of PR - And I would say any other form of PR does what CV does and does it better.
==============================
Further discussion of various forms of PR...
It is said that
"A reform in our voting procedures. to include PR, would go a long way to ensure government decisions at all levels reflect the needs of all Canadians"
And we see this through the high proportion of effective votes under PR compared to FPTP.
PR as I see it means that most votes are used to elect someone, not at all the result in a chamber made up of regional-sweep winners -- roughly balanced overall but each is a dis-proportional winner in a different region -- nor the wonky results under FPTP, when just plurality is all that is required to elect all (the one) seat in the district.
We can see/show this through these tables, where the low proportion of effective votes under FPTP is clear:
FPTP Ottawa Centre 2021 federal election (our national capital, picked as random sampling of FPTP results) Votes as cast Effective votes Naqvi Liberal 33,825 Naqvi Liberal 33,825 elected (46 percent of valid votes) MacEwan NDP 24,552 Clemenghagan Conservative 11,650 Keller-Herzog Greens 2115 [and more ] Thus the vote-to-seat conversion process is wonky under FPTP. less than half the votes are effective. Here's the same kind of thing under STV-PR, showing high proportion of effective votes that that system produces in each district and each time it is used. STV Edmonton Alberta provincial election 1944 five to be elected valid votes: 37,834 Votes as cast Effective votes Manning SC 14,271 Manning SC 6306 17 percent Roper CCF 5253 Roper CCF 6306 17 percent Page Independent 4603 Page Independent 6333 17 percent Williams 3532 Williams (Vet. & Armed Forces) 5535 15 percent Caine Independent 1400 James SC 781 James SC 5397 14 percent [and a few more] TOTAL 80 percent Thus the vote-to-seat conversion process is very strong under STV. a high proportion of effective votes means most votes actually elect someone - there is no way that that would not produce a representative result. (although under STV votes are not necessarily used to elect first preference candidates - some are but others are "merely" candidates preferred over the others still remaining. Some votes are declared exhausted - not enough usable back-up preferences to allow vote to shift to a home - these are counted as being among those recorded as not used to elect anyone. Others were marked more fully but still failed to be used to elect anyone. The 80 percent is a true 80 percent -- they are 80 percent of valid votes originally cast. (not just 80 percent of votes still in play at end, etc.)
Votes that are not effective the 20 percent not used to elect someone includes - those where the voter's first preference was never eliminated but was not successful in the end. in some cases, these votes had back-up preferences that were marked for candidates who were elected but those back-up preferences were never referred to, so there is no way to know if the voter was pleased with the election of others or not, even if the vote itself was known to be not used to elect anyone.) AND - those where the voter's first preference was eliminated and the vote was transferred perhaps repeatedly but all of his or her back-up preferences were eliminated one by one until all the marked preferences were used up and the vote had to be declared exhausted.
and also includes - many where their first preference was elected but the vote was transferred on due to being among surplus votes and never was used to elect anyone. (Although voter himself or herself had the satisfaction of seeing their first choice elected even if their vote was not used to produce that result) and - many who saw their first preference eliminated relatively late in the game and all their marked back-up preferences had already been elected so the vote was never used effectively -- the vote only had the opportunity to be transferred to any other candidate only after that candidate was already elected. (so voter himself or herself had the satisfaction of seeing one of more of their lower choices elected even if their vote was not used to produce that result, and even if their first choice was not elected) and includes - voters who plumped because they felt that if their first preference was not elected, they did not care either way who was elected out of the rest. (in Canada's STV history, plumping was allowed and is now more and more common even in Australia's use of STV.) Under STV systems where plumping is allowed (all Canadian elections in the past), voter did not have to mark more choices just to see their vote used to elect someone. The voter had right to mark only those he or she truly cared about and not carry on to give lesser choice whom they don't really care if they were elected or not. So the vote in those 80 percent of cases was used to elect someone whom the voter had indicated as one of their preferences - with no forcing. (Some arbitrarily say that any choice below the number of seats to be filled -- say the sixth choice marked when five seats are to be filled -- should not be considered enough of a high enough preference to be enough to make the vote an effective vote - I don't know about that But that at least is broader than saying the first preference is the only person the voter really wants elected, as defenders of FPTP seem to say sometimes. Defenders of FPTP (try to) put holes in STV when PR supporter say every vote is used or that most votes are used - by saying that only the first preference is in fact whom voter wants to see elected. But they ignore that under FPTP many voters vote for people that they do not prefer most of all, (due to strategic voting) and they ignore that under STV, the back-up preferences are actually choices made by voters, and perhaps are almost as preferred as the first preference itself. Especially when party lines are strong, the specific candidate chosen may be less important than the identity of the party in government -- or in opposition. Under the seat/preference analysis, the percent of effective votes may be lower than the 80 percent I quote, but I think usually we can't know for sure.
In Canadian election reports, we don't know what level of preference was used for the vote to find a home - after the first or second count of an election process. As soon as surplus votes are taken from among a mixed pool of votes with some being first preferences and some being previously-transferred votes, there are quantities that cannot be defined exactly. Is it the second or third or a lower preference that is being used to set the direction of the transfer? We can't know. Scottish STV elections for local authorities have massively fulsome reporting and the exact preference used can be known but that is exception anywhere any time STV-wise. Even under STV, not all votes will be used to elect someone, but an old book I was reading pointed out that even if vote not used to elect someone and even if no one was elected of the party preferred by the voter, still a voter under STV will usually find someone that they agree with on some things among the range of members elected. This is virtue of any system that both uses MMD and prevents one party from taking all the seats in the district. Block voting allows one party sweeps so I would exclude it from list of systems that share this virtue. FPTP is similarly excluded - no range of members possible when only one member elected. CV would be included on the list of system that have this virtue, at least on those times when it secures mixed representation. And to be fair, we should note that MMP produces asimilarly high proportion of votes used to elect someone (about 80 percent or more) although there, many seats are filled through party support, not through direct election of specific candidates as happens under STV.
People being able to cross party lines with their votes or preferences may be considered a good thing or a bad thing, just depends on your point of view -- how much liberty you want voters to have and how important parties are to you. Under STV a vote can go to one of the specific candidates preferred even at expense of initial party proportions.
Say a voter marks his or her ballot:
Smith (Party A) 1
Macadams (Party B) 2.
If the vote is not used to elect Smith of Party A, the vote might be used to elect MacAdams of Party B.
If the vote is used to elect MacAdams, the final result will appear dis-proportional as a vote thought to be initially cast for Party A goes to elect a member of a different party.
But it was the will of the voter -- likely the voter had opportunity to mark back-up preference for another candidate of Party A (say Jones) but preferred MacAdams over Jones, although it meant crossing party lines.. Under MMP top-up, people's vote is stuck with their choice of party even if that means the member elected is not a specific candidate of choice. Under CV a voter can vote for specific candidates of two different parties, and see their votes used to elect both, one or the other, or neither. (As well/anyway they might see someone else of their preferred party/ies elected, without the help of their votes, or they might not.) Under STV, a voter will see their vote used to elect just one or none at all. and in the MMD that is used in STV, without using the person's vote, other members might be elected that suit the voter, or not. Under MMP, where two votes are used, one for district contest and one for top-up, a voter might see their preference elected in the district contest and someone of their preferred party elected in the top-up, or just one of those things, or neither. Which of these vote "super-powers" is better is a personal choice of the electoral system designer -- and choices made become the burden or a source of satisfaction for the voters who will have to vote under the system.
====================
Here's an article on Illinois use of Cumulative Voting from 1870 to 1980.
But I am not sure the writer - Steven Hill - understands CV.
He wrote "For 110 years until 1980, Illinois used a method called cumulative voting. Instead of single-seat "winner take all" districts, legislators were elected by three-seat districts and a candidate needed 25% of the popular vote to win one of the three seats."
But there is no magical quota under CV as there is under STV. and with each voter casting three votes, 25 percent of the vote does not mean the same as it does under any system where the voter just has one vote. (In the 1966 Illinois House of Representatives election, in the 1st district, Harold Katz won a seat with 19.88 percent of the vote. In 1956 a member was elected in that district with just 21 percent of the vote.) (https://en.wikipedia.org/wiki/Illinois%27s_1st_House_of_Representatives_district)
Cumulative Voting gives multiple votes to each voter and allows him or her to lump all the votes on just one candidate. This lumping is easier to organize if a party can nominate just one candidate. Then seats are allocated just on a plurality basis, with the most popular being elected even if they do not have majority support.
Illinois moved to go back to Cumulative Voting in the 2001-2010 period.
But somewhat oddly in 2005, the Illinois Legislature debated the change back to CV under strange rules:
the proposed legislation would have changed the House of Representatives from a 118-member body elected from 118 districts to a 117-member body elected from 39 districts of 3 members each. and would have prohibited a political party from limiting its nominations to less than 3 candidates in a Representative District, so a party would not be able to force a concentration behind a single candidate, the virtue of CV in the first place.
This rule was done consciously, in effect to prevent sweetheart deals between parties where a party would agree to run only one candidate in a three-seat district. But does blunt the fairness of CV elections, in my opinion.
And the record of Illinois's Cumulative Voting is not so great -
in 1976 only 14 black members were elected to the Illinois state legislature, with perhaps a hundred members in total being elected in 1976. The 14 does not appear to be proportional to the black portion of the population. As well, perhaps only 14 districts out of 59 districts elected a black representative.
Steven Hill does says that CV, as good as some think it is, is surpassed by what he called Proportional Ranked Choice Voting - what I and others call STV.
He says
"P-RCV [STV] has all the benefits of cumulative voting but it also has other desirable features that prevent spoiler candidates, split votes and wasted votes that can sometimes happen with cumulative voting."
I think due to the wasted votes, it is possible to win a seat with less than the 25 percent that some seem to think is the set rule of three-seat-district Cumulative Voting.
But I have not seen any district election results to confirm my guess.
====================
On the use of Cumulative Voting in Canada
Cumulative Voting was used to elect city boards in Toronto starting in 1903.
The Proportional Representation Review (September 1903) described it like this: "Cumulative voting as applied to the Board of Control, means that each elector will have four votes but that he need not give each of them to a different candidate. He may do so if he wishes; but he has also the power to give all his four votes to one candidate. This makes "plumping" four times as powerful as it was by the old "block" vote system, when if you "plumped" for one candidate, you threw away three out of your four votes. Now you have the benefit of your full voting power, whether you plump or not.
And plumping is the correct thing; in fact proportional representation is simply effective representation with the addition in the best systems of a provision for transfer of votes, so as to prevent wasting too many on one candidate... [
Besides permitting an elector to give all four votes to one candidate, the cumulative plan enables him to give two of his votes to one candidate and two to another, or he may give three votes to one candidate and his fourth to another candidate. In fact he may distribute or cumulate his four votes as he pleases....
If one-fourth of the voters give all their votes to one candidate, they can elect him, no matter what the other three-fourths choose to do...."
Thus Cumulative Voting if used carefully produces both majority and minority representation. But STV is more able to produce them dependably without the risk of foul-ups.
The confusion about the portion of the votes that secures elections demonstrates some vague-ness. In the theory of Illinois's use of CV, 25 percent of the votes was enough to win one of three seats in a district, but in Toronto, 25 percent of the voters was said to be enough to win one of four seats. (Perhaps the one is based on the Droop quota and the other on the Hare quota, when actually no quota at all is used in Cumulative Voting.)
====================
"Alternative Electoral Systems as Voting Rights Remedies"
by Edward Still and Robert Richie
Journal of Electoral Administration (No. 18, 1996)
much on Cumulative Voting (esp. p. 19 and on)
(the issue also has article on the use of CV in Illinois)
I see the "Alternative Electoral Systems as Voting Rights Remedies" article explains that there are two types of Cumulative Voting
both share basis of CV:
CV has each voter casting multiple votes (same as for Block Voting)
CV allows voter to lump all his/her votes on one candidate or to spread them over two or more.
But
one variant has voters allocating each of his/her votes on a candidate. lumping is allowed.
the other variant has votes marking their choices same as for block voting, with the votes then mathematically divided evenly among the marked choices.
in five-seat district,
if voter marks one choice, that candidate is given five votes
if a voter marks two choices, each of them are given 2.5 votes
if a voter marks five candidates, each of them are given one vote.
and so on.
I can see the second variant working better than the first variant.
The first variant depends on voters being taught to lump their votes - something never done in our elections.
In the second variant lumping is inherent (AI), not performed by the voter him/her self
plus the first variant is mathematically/psychically challenging for the voter
do I like this person two-thirds of this other one?
if I give 1 1/3 votes to this one and 5/6ths of a vote to that one, how much can I give to this other one?
But the idea of lumping, taken to its more efficient extent, simply means single voting - same as is used in SNTV - so that is simplest of all.
plus with single voting (SNTV or STV) you can see how many voters support each candidate.
with multiple voting you don't know if a party slate of five candidates that gets 40 percent of the vote (where 10,000 voters cast 50,000 votes) got five votes from 4000 voters or two votes from each of the voters.
or you have the instance where a party slate of five takes more votes than there are voters voting. (Edmonton in the 1921 Alberta election)
If you plug that into a party share statistic or top-up allocation, you have a problem.
========================
Comments