On this page:
– The proposals
– Q & A
– Current elections and their problems
We like Condorcet methods for accuracy, but let’s also use a very simple point system for practicality.
Here are a two-ballot and a single-ballot proposal, named for the state in which nonpartisanship is a long-standing tradition. While I do want to use these election methods in Nebraska, please feel free to use them anywhere.
Nebraska Rank/Rate Primary
(1st Ballot)
Voters may rank two candidates that they like.
A write-in line will be provided.
One 1st choice per ballot is counted as 10 points.
One 2nd choice is counted as 6 points.
The top four scorers qualify.
(Using two rating levels, with each limited to one candidate, provides a balance between allowing a voter to choose more than one, and not allowing the largest party to dominate all of the spots on the next ballot.)
Nebraska Rank/Rate General
(2nd Ballot)
There will be a maximum of four candidates.
Voters may rank 3 candidates that they like, one per tier.
There will be 4 tiers listed, so those who want to forgo the 6-point 2nd rating may still rank 3.
If one candidate is ranked 1st on over 50% of ballots cast, that majority winner is elected.
Otherwise, proceed as follows.
Two finalists will be the ones having the highest scores, when 1st is 10 points, and 2nd is 6 points.
(Eliminating two who have the lowest scores. This shortcut is simpler than instant runoffs, and works better, because it considers the 2nd ranks of all ballots instead of only a few.)
Head-to-head ranking comparison, the finalist preferred by more voters wins.
(That’s a pairwise comparison, or one proper instant runoff of the final two.
3rd ranks can come into play here.)
Nebraska Rank/Rate Single-Ballot
Voters may rank up to 5 candidates, one per tier.
A write-in line will be provided.
The top 3 in score will qualify.
(1st = 10 points, 2nd = 6 points).
Double-win final. One who is preferred head-to-head, on more ballots, over both of the others, will be elected.
(This Condorcet check of only the top three scorers will confirm a winner in the vast majority of instances.)
If one candidate has one win, one tie, and no defeats, and the other two have one or more defeats each, the undefeated candidate will be elected.
(A “defeat” refers to losing a head-to-head comparison, by being preferred on fewer ballots than one opponent.)
If two candidates tie head-to-head, and they each defeat the 3rd, the tied candidate having the higher score of the two will be elected.
If three candidates all defeat one another, or if there are two or more ties between the three, the one having the highest score of the three will be elected.
Illustrations: https://americarepair.home.blog/2024/07/18/nebraska-rank-rate-method-quick-guide/
Q & A
Q – Why points? Why choose more than one?
A – The election is supposed to measure the people’s will. Adding points and allowing multiple rankings makes a better measurement than the old choose-one, which treats our 2nd-favorite the same as our worst enemy.
When we speak of a point value, that’s a rating. A ranking, on the other hand, is technically independent of points, and indicates preference level compared to other candidates.
I have merged ratings and rankings in a way I hope will be intuitive. For example, the one with your 1st ranking also has your 1st rating, and 10 points.
The points make a slick shortcut, to approximate what would happen if we used a full ranking tabulation. (Despite its accuracy, I’d rather avoid a full Condorcet ranking method at this time, because that makes a lot more work for vote counters.) The 1st rating has the most points, representing how a 1st rank is the rank most likely to help a candidate win. The 2nd rating carries fewer points, because the average 2nd rank would be roughly 40% less important than 1st rank in the average ranking comparison. So while 1st is 10 points, 2nd is 6 points. Of course, the true importance of 2nd ranks compared to 1st will vary. So in the end, the total scores really will be an approximation, but one based on the two most important rating levels, and no more than two picks per voter.
Q – Why single-ballot elections?
A – While the presidential election draws a huge amount of public interest, elections such as city council, legislature, and state auditor do not. These elections frequently attract few candidates as well, so there’s little need to have two rounds.
A single-ballot election costs less than a primary and general.
Candidates who lose in a primary could run for a second office in the same year, because we need more candidates.
And, perhaps most importantly, many busy people would much rather just vote once instead of twice.
For a single ballot that collects more of each voter’s opinions, we can maximize accuracy with a top-3 scoring round, and then 3 pairwise comparisons.
I would replace Nebraska’s current nonpartisan primary elections with this accurate single-ballot election, adding state officers, sheriff, and perhaps others that are currently partisan but need not be.
The new nonpartisan primary, on the other hand, would be used for at least governor, congress, and mayor of large cities. And maybe in the future, president.
Q – Why use pairwise comparisons, when you could just use point values, and the highest score wins?
A – Sometimes a survey of rankings on all ballots can reveal that the 2nd-place scorer is actually preferred, by more voters, over the high scorer.
One way this happens is that these point values are arbitrary, vs a pairwise ranking comparison, which is a precise measurement.
Another way is vote splitting, illustrated as follows: Two popular conservatives, dividing the 1st ranks of conservative voters, will sometimes both have lower scores than a lone popular progressive. But in a head-to-head pairwise comparison, the preference of a conservative-dominant electorate would be to elect a conservative.
Q – You called the point values an approximation, then you called them arbitrary. What the heck?
A – Every election is an attempt to approximate the people’s will. The result is our best guess. No election will ever be perfect. But generally speaking, choose-one is inferior to choose-two, and score, and ranking methods.
Every point system is arbitrary, because only God could know what the correct point values should be.
A score election with point values of 3, 2, and 1 would eventually have a different outcome than the same election but with values of 4, 3, and 2.
I have chosen 10 and 6 as a good ratio. I would also accept 3 and 2, or 2 and 1, but 10:6 is between those two, and it works well on my scratchpad.
10 and 6 are even numbers, so any odd total will flag an error. And it’s easy to multiply by 10.
Some may not like that I’ve left a 3rd rank with zero points. I considered 4 points, but when the value is less than half of a first rank, it is weak, and often won’t matter. At the same time, the more points each voter can use, the more opportunity there is for the largest party to dominate or try shenanigans, such as Republicans putting an incompetent Democrat into the final instead of two top Republicans. Or an incompetent instead of a qualified Democrat… the incompetent one might win and do serious damage.
What matters more than the specific point values is that the voters have the opportunity to understand the election method, and the available point values, and they will adjust their choices accordingly.
And everyone’s 2nd choice will count for something.
Q – Why would I want to help my 2nd favorite, if it might cause my favorite to lose?
A – Because you might feel it’s important to help both of your two favorites get into the final.
One might rephrase the question as follows.
In these unlikely circumstances:
1. Your favorite’s total score was less than 6 points lower than the total of your 2nd choice, AND
2. That tiny difference in score might cause your very unlucky favorite to be eliminated in 3rd or 4th place, AND
3. Your favorite would have won the final had they not been eliminated,
Is trying to predict and prevent that extremely unlikely situation worth changing your vote?
Probably not.
My advice is don’t worry about trying to manipulate anything, just vote honestly.
However, you will have options.
You can be normal and do a normal ranking, 1st, 2nd, 3rd, and leave it at that.
Or you can bump your 2nd rank down to 3rd, so they won’t get any points, but you’ll still prefer them in the final if your favorite doesn’t qualify.
Or you could just choose one, if it makes you feel better to limit your own influence.
Q – Isn’t this a solution looking for a problem?
A – Choose-one ballots are inaccurate.
Your 2nd-favorite gets the same rating as your worst enemy.
That’s a big problem.
That’s really all you need to know.
We have settled for a bad system because some people wish for everything to have minimal complexity, regardless of the side effects.
We’re all carrying supercomputers in our pockets. Complexity is good.
And we can certainly do better than choose-one.
Q – You endorsed Ranked Choice before, and now you want something else instead?
A – First, a ranked choice proposal and this hybrid proposal share one important trait: a greatly improved ballot. Getting that more accurate ballot is a high priority to me, so I would support either method.
After we implement a ranking ballot, then we can debate about which tabulation method to use.
Second, Australia-style ranked choice is more likely to eliminate the most popular candidate before the final two. And it is more likely to put the least popular candidate into the final two.
So even though that ranked choice method will designate the right winner around 94% of the time (as seen in Australia), we can do better by using a scoring elimination round.
Q – Gasp! Talking against Ranked Choice? Are you a bad guy?
A – “Ranked Choice” refers to Instant Runoff Voting (IRV), also called Hare method because Thomas Hare invented it for a great multi-winner method. Then other people decided to use it for single-winner elections, for which it isn’t so great.
A “Condorcet winner” is a candidate who is preferred by more voters over every single opponent. The Condorcet winner wins every possible head-to-head matchup.
In Alaska’s very first ranking election, whoops, the clear Condorcet winner lost. IRV gave him last place (3rd), because in the first round, it only looks at 1st ranks. This is a BIG problem.
If we ask the voters, “Do you prefer Nick or Mary?” They pick Nick.
Then we ask, “Do you prefer Nick or Sarah?” And they pick Nick again.
Should we eliminate Nick first?.. it’s the opposite of what any sensible person would do… and that’s what happened.
Even in subsequent rounds, IRV can still exhibit the same problems as a choose-one method. Voters whose 1st choice makes the top two are stuck with their 1st choice, while other people will use their 4th, 5th, or 6th… A weirdo’s 10th choice whim might pick the winner, while mainstream voters will never have their 2nd choice considered at all.
With IRV, a Condorcet winner is certain to be eliminated in 3rd place from time to time. But they’ll never be 2nd place, because when IRV randomly allows a Condorcet winner into the top 2, they always win!
The only rock-solid ranking comparison is to compare two, and only two candidates at a time. That’s why a method that acknowledges the Condorcet criterion is a more accurate way to run a ranking election.
Compared to Aussie Ranked Choice, Condorcet is better, score is better, and my methods that use simplified scoring plus a simple ranking final (à la STAR) will be better, with more accurate results, and an easier tabulation.
My 2nd ballot might also eliminate a Condorcet winner in 3rd place (out of four), but only when two others score higher, which will be rare.
And remember, the scores come from 1st and 2nd ranks, so someone who is a Condorcet winner for being a lot of people’s 3rd choice out of 4, well, maybe they shouldn’t win against 2 higher scorers.
That would be way better than eliminating a Condorcet winner based only on 1st ranks.
And my method doesn’t count fringe weirdos’ 10th choices, but it does count everyone’s 2nd choice, which makes a lot more sense than the reverse. Seriously, think about it, your 2nd choice should matter.
Q – Your ideas are unnecessary because we have STAR voting and professor-published Condorcet methods we could use.
A – STAR is good, but it can get weird. For example, when you give 1 star to a candidate who’s your 2nd-last choice, the point attached to that star might help him win.
My methods won’t cause your ballot to help candidates you don’t like.
STAR has no limit on scores, a radical change which will offend people in various ways. Maybe Oregon can handle it, maybe we’ll use it someday, but STAR isn’t right for the here and now.
Condorcet-consistent methods could be wonderfully accurate, but most election commissioners, and their allies, will fight hard against drastically increasing the complexity of a hand count. (10 candidates make 45 possible pairwise comparisons, 11 make 55, 12 make 66…)
My methods use compromise: limited points for only high ratings, and no more than 3 pairwise comparisons, still quite simple. A good step in the right direction, but not a step-too-far.
Q – Why did you add on a majority winner rule? Doesn’t the majority rule enough?
A – The two-ballot general election does have a majority winner rule tacked on, because it is possible (though unlikely) for a majority winner to have a 3rd-place score, and be left out of the 2-way final.
The single-ballot election (and the primary) will elect a majority winner either way, because it’s mathematically impossible for three opponents to all outscore a majority winner.
First, my hope is that these improved methods will give us three or four viable candidates to choose from, making a 1st-rank majority winner a rare event. So it might not matter much.
Second, I crafted these procedures to emulate a Condorcet-consistent election, but with a simplified process.
Every 1st-rank majority winner is also a Condorcet winner, because they never lose a head-to-head matchup.
So one way for my method to elect the same candidate as Condorcet is to use majority winner as an easy first check.
(Some Condorcet winners are not majority winners, but there’s only so much I can do with the scoring shortcut.)
Third, that general ballot will be used for very important offices, including governor and congress. If my method would ever thwart a majority winner, the majority would be mad enough to write new rules that I wouldn’t like.
So it seems necessary that a majority winner will always win.
Side note: Some people who advocate alternative election methods want to minimize the importance of a 1st-rank majority winner, because the rule can influence voter behavior to cause less accurate results. Voters might betray their real favorite candidate, in order to join a near-majority to help their party (which could thwart their real favorite). Others might decide to not even vote, if they believe their opposition has a majority (which can be a self-fulfilling prophecy).
With this in mind, and realizing that my method, with its strict limits on points, treats 2nd ranks as a sort of alternate 1st rank, it would not be a great injustice to remove the majority rule from that general election, and allow points to set the stage, for just a few, rare, high-consensus winners, instead of a few partisan strict-majority winners.
I can only guess how much that rule change would affect voter strategy, or election outcomes, and my guess is not much.
End of Q & A
For your reference, the following tells how our elections currently work, and why change is needed.
Elections in Nebraska
– Choose-One Nonpartisan Primary
Voters may select only one candidate.
This coerces some voters to select one they believe can win, rather than accurately marking their actual favorite.
A voter’s 2nd-favorite (or their actual favorite whom they betrayed for another) is placed at the same preference level as the voter’s least favorite.
So the ballots are inaccurate, so the tabulation of those ballots must also be inaccurate.
The top 2 favorites will qualify for the 2nd ballot.
We have seen recent elections that were a virtual tie between 2nd and 3rd place, or a virtual 3-way tie.
Considering how voters feel coerced, and the ballots and tabulation are inaccurate, it is probable that the truly most popular candidate will periodically be eliminated in 3rd place.
– Choose-One Nonpartisan General
Voters may select one.
This indicates their preference between the final two.
This general works perfectly.
It elects the more preferred of the final two.
What causes the problems in the primary is that “Choose-one” only works right when there are two candidates, and no more.
– Choose-One Partisan Primary
The most popular candidate will sometimes be the 2nd-favorite Republican, or the 2nd-favorite Democrat.
There is an exclusive group of voters in each primary, and they don’t get a 2nd choice, so the overall best candidate is sometimes eliminated, or chooses to not run at all.
– Choose-One Partisan General
If there are exactly two candidates, it will work fine.
But we need more than two.
Back to blog index page:
https://americarepair.home.blog/2020/09/05/index-of-blog-thus-far/
America Repair 