[EM] Two simple alternative voting methods that are fairer than IRV/STV and lack most IRV/STV flaws

Abd ul-Rahman Lomax abd at lomaxdesign.com
Thu Jan 14 12:42:14 PST 2010


At 08:06 PM 1/13/2010, Kathy Dopp wrote:
>For those who need a system for substituting for a top-two runoff
>election, I devised two fair methods to suggest to her that do not
>have all the flaws of IRV/STV. (They both may've been devised by
>others before me. My goal was to create a fair method without
>IRV/STV's flaws which solve the problem of one person/one vote which
>is necessary to get a voting method approved by US courts.

Unfortunately, Kathy, you are entering a field where a great deal of 
work has been done, and there is some precedent, and you aren't 
terribly likely to come up with something new (though it's possible). 
I'll comment specifically.

>...
>
>Both methods below solve the problem of every voter having a vote of
>value one and, unlike IRV, treat all voters alike by counting all
>their choices
>
>So, here are two possible methods that are fairer than IRV/STV and
>which are monotonic (unlike IRV/STV):
>
>1. A rank choice ballot method:
>
>Any number of candidates may be running for office and any number
>allowed to be ranked on the ballot.
>
>Voter ranks one candidate vote =1
>
>Voter ranks two candidates, denominator is 1+2 = 3
>votes are worth 2/3 and 1/3 for first and second ranked candidates
>
>Voter ranks three candidates, denominator is 1+2+3=6
>votes are worth 3/6 and 2/6 and 1/6 for 1st, 2nd, and 3rd choice respectively
>
>Voter ranks four candidates, denominator is 1+2+3+4=10
>votes are worth 4/10, 3/10, 2/10, and 1/10 for 1st, 2nd, and 3rd and
>4th choice respectively
>
>ETC. Just follow the same pattern

This is a variation on Bucklin, something similar was tried in 
Oklahoma. It's similar, also, to Borda Count. Bucklin is instant 
runoff approval. It counts each rank as additional votes in an 
Approval election with a sliding scale down as to approval cutoff 
with each rank's counting.

You seem to be under the impression that counting more than one vote 
simultaneously is a problem with Approval Voting. That isn't a 
position that has been sustained by the U.S. courts in general. 
Minnesota is an exception, and they knew that they were contradicting 
the general position of the courts, and no other court confirmed 
Brown v. Smallwood, the relevant case; further, that case has been 
misunderstood: the objection was to *any kind of multiple voting*, 
it's not clear that the vote splitting scheme decided would have 
avoided the opinion of the MN Supreme Court.

With the scheme you proposed, and without any 'runoff' ranking, 
voting for additional candidates would seriously weaken the voter's 
power. Consider it this way. Suppose the voter wants to vote for a 
frontrunner. But the voter prefers someone else and wants to express 
that. If they vote for an additional candidate who is not a 
frontrunner, they have wasted part of their vote, it will be moot, it 
could be eliminated and have no effect on the outcome (that is, the 
fraction that isn't for a *winner* could be struck!) So if they 
prefer a non-frontrunner, and dare to express that as their first 
preference, they only get one-third of a vote to use in the real 
contest. That's quite a penalty to pay for a sincere vote!

Oklahoma Bucklin was a little more sensible. It's a Bucklin method, 
counted in rounds until a majority is found. The first rank vote was 
a full vote, the second rank vote was 1/2 vote, the third rank vote 
was 1/3 vote. The method was never actually used in an election, 
because it was found unconstitutional, not because of the fractional 
voting, per se, though that might have been involved, but because 
adding additional ranked votes, under some conditions (more than two 
candidates?) was mandatory or the ballot was spoiled. I agree with 
the Courts decision, though they could have left the method in place 
and only struck the mandatory part. Instead, they tossed the whole thing out.

Basically, Kathy, sorry, but the method is a bad idea. There have 
been equivalent suggestions with Approval Voting, i.e., if you vote 
for two, each vote counts as a half-vote. This essentially means that 
if the voter votes for two, they end up having only a half-vote of 
real voting power. Bad idea. I'm not at all sure that it's better 
than Plurality. If a majority is required or there is a runoff, 
maybe. You can do a lot of things with a majority requirement.

The method is a variation on Warren Smith's Asset Voting, only with 
fixed ratios rather than purely voter-determined quantities (the vote 
must sum to one in Smith's Asset). It's legitimate there because no 
votes are wasted in Asset, and both half-votes, say, could end up 
being assigned to the same candidate, or to different ones, and this 
is a multiwinner method, as proposed, and so both votes can even go 
to different candidates and still be effective votes. A way of 
looking at it is that if the voter votes for one, the voter is 
creating a voting proxy, and if the voter votes for more than one, 
it's a virtual proxy committee that can amalagamate the voter or 
split it up, whatever those who get the votes decide to do. Two 
proxies instead of one, with two votes.

>2. A point system where a total number of points per voter per contest
>may be allocated by the voter to any of the candidates running for
>office:
>
>Two candidates running for office, give all voters 2+1=3 votes to
>cast.  They may cast all three votes for one candidate or split the
>votes any way between the two.

Cumulative voting. In a single winner election, it's generally 
foolish to vote for more than one. Multiwinner, maybe, it depends. 
The general opinion seems to be that unless the voter can coordinate 
with others and make trustworthy voting decisions collectively, the 
best strategy is still to cast all votes for the favorite.

You seem to think that there is a legal problem with approval voting. 
The best opinion is that there is not. In the end, only one of 
multiple votes cast is effective for election, or none. There is no 
situation where both votes have an influence on the outcome. The 
voting power is one person, one vote. Another way to think of this is 
that two votes in approval is like mutual alternatives, If not this, 
then that, and if not that, then this.

If you strike all multiple votes not for the apparent winner, and 
look again at the vote totals, you will find that the vote percentage 
for the winner doesn't change at all. If it was a majority supporting 
the winner, it's still a majority. It's still the same percentage, 
the percentage of voters who were willing to support the winner.

Approval voting is in current use, but not called that, of course. 
When there are multiple conflicting ballot questions, and both gain a 
majority, the one that wins is the one with the most Yes votes. If 
only one passes, nobody complains that some voters also voted for the 
other, or against the other. Two votes were cast, only one, in the 
end, influences the outcome.

A single winner election is quite equivalent to multiple conflicting 
ballot questions. Completely equivalent if a majority is required to win!

>[...]
>
>The advantage of these two methods over IRV/STV include:
>
>1. easy to count, precinct-summable (unlike IRV)
>
>2. fair, treats all voters' votes equally by counting all choices of
>each voter (unlike IRV)
>
>3. gives each voter a total of one vote total over the entire vote
>counting process satisfying the US courts (unlike IRV)
>
>4. is monotonic -- preserves the right to cast a vote that has a
>positive affect on a candidate's chances of winning (unlike IRV.)
>
>5. Allows all voters to participate in all the rounds since these
>methods require only one (1) round (unlike IRV)
>
>6. can begin the counting immediately without waiting for all the
>late-counted provisional and absentee ballots to be ready to count
>(without fear of having to restart the entire process again from the
>beginning unlike with IRV/STV)
>

Bucklin. You want Bucklin, Kathy.

Easy to count, precinct summable.
Fair, counts all votes and uses them, from each voter, from and up to 
the last rank counted.
Only one vote is actually effective, all others are alternative votes 
that turn out to be of no effect.
Monotonic.
Counting is immediate, and if the number of other votes is known to 
be limited, the contest may be decided without counting them under 
some conditions. My opinion is that all the votes should be counted 
anyway, even if they are moot. It's useful to know how many people 
were willing to express a lower preference.

Another advantage to Bucklin: it's been tried, there is a history, it 
worked, it was popular, it was called American Preferential Voting by 
the political scientists.

Another advantage is that it's quite easy to understand. It's really 
phased Approval and converts to pure Approval if all the ranks are 
counted, and a great deal of work has been done on Approval, it's an 
astonishingly good system for it's terminal simplicity. Simple 
ballot, same as Plurality. Just Count All the Votes, don't toss an 
overvoted ballot. Candidate with the most votes wins. Don't vote for 
another candidate if you'd be displeased if your second choice turns 
out to get the most votes.

Take Approval and allow a first choice vote that is counted first, 
and if a majority is found, done. That's Bucklin.

And Bucklin is slightly improved if multiple voting is allowed at any 
rank, not just the last one. That allows voters who have little 
preference between two candidates, both better than all the others, 
to equal-rank them. And will reduce spoiled ballots. Just Count All the Votes.

Now, isn't that a slogan for an election integrity expert to get behind?

Beyond this, want some improvement, used a Range or Score voting 
ballot, which amounts to the ability to cast fractional votes at 
will. But the limit is one full vote maximum *per candidate.* Note 
what happens if the voter voted maximum for all candidates.

NOTHING in terms of result, except that the voter has shown equal 
approval of them all. It is a vote with no effect on the result, 
unless a majority is required to win, in which case it supports all of them.

If there were ten candidates, the voter didn't get ten votes to help 
decide the outcome! One one of the ten votes would have any actual 
effect, the one cast for the winner. 




More information about the Election-Methods mailing list