Sunday, September 13, 2009

The Instant Runoff Virus hits play elections!

IRV can't even ensure a majority winner in a play election!

Was taking a look at some IRV articles on line, and found yet another shining example of how IRV doesn't ensure a majority winner in a single election - and not even a real one!

The Lyndale Neighborhood Association hosted a demonstration RCV election presumably organized by FairVoteMN.

Lyndale Neighborhood Association hosted a demonstration Ranked Choice Voting election at its annual meeting on June 22nd to educate residents about the way they’ll vote in the upcoming November elections in Minneapolis.

Contestants were desserts brought by neighborhood residents. The ballot line- up included:

Salted Nut Bars
Sweet Potato Pie
Caramel Pecan bars
Scones & Cream
Bundt cake

After sampling the candidates, guests filled out a ranked ballot, marking their first choice, and second and third choices if they wished. By ranking candidates in preference order, voters know their vote will continue to count if there is a runoff and their favorite candidate is eliminated in round one. A candidate needs 50% + 1 vote to win in a single-winner election. If no candidate receives a majority of votes outright, a runoff is triggered. The candidate with the fewest votes is eliminated and votes for that candidate are redistributed to the remaining candidates based on the second preferences on those voters’ ballots. This process is repeated until one candidate has a majority of votes.

Results of the Lyndale dessert election

A total of 29 ballots were cast. One voter marked two candidates for first choice, invalidating the ballot. As in any election, you can only vote for one candidate at a time. With Ranked Choice, you may vote for one candidate for 1st choice, one for 2nd choice and one for 3rd choice.

28 votes cast in the first round, and one guy turns in a ballot with no choices made. But with 28 votes cast, the threshold for winning on the first round is half of 28 plus one vote - or 15 votes. No one had 15 votes at the end of the 1st round, so it went to IRV.

It was a competitive election and no candidate received the majority (50 % + 1) of votes needed to win, triggering an “instant” runoff. Bundt Cake was the most popular, receiving 10 votes.

The candidate in last place, Sweet Potato Pie, was eliminated and that voter’s ballot was reallocated to the voter's second choice on the ballot, Caramel Pecan Bars. No candidate still had a majority of a votes.

In the next round, Scones & Cream was eliminated and votes for that candidate were redistributed to those voters' next preferences indicated on each ballot.

Another round was needed to determine the winner. In round 4, Caramel Pecan Bars was eliminated and those votes were redistributed to next preferences on each ballot. Five of those ballots did not have additional preferences marked and were exhausted.

Bundt Cake, the front runner in round 1, won a majority of votes cast in Round 4!

Demonstration elections are an excellent way to teach voters about Ranked Choice Voting and to get them ready to vote 1-2-3 in November.

Contact us if your neigbhorhood or organization would like to conduct a mock election. You provide the candidates and we can provide and count the ballots! We can also provide a speaker to explain how Ranked Choice Voting works. Contact

Now let's do a real analysis of the election. Bundt cake had 10 votes at the end of the 1st round. There were still 28 votes at the end of the 2nd round, but no one had reached the threshold of 15, so they went to the third round. Bundt cake picked up no votes in this round.

After the 3rd round, still no majority winner of 15 votes, so they went to the 4th round. Bundt cake picked up 3 votes for a total of 13 votes at the end of round 3 - but still not enough votes to reach the threshold of 15 and win the election.

Funny thing happened in the 4th round - the turnout dropped! 5 people just didn't give a damn anymore - they must have thought this was a silly game (I agree with them). So instead of 28 votes, they had 23 to deal with. Half of 23 is 11.5, and you round up to the next highest number which is 12. So bundt cake still had 13 votes which wasn't enough votes to win the contest at the end of round 3 - but 13 votes was enough to win the race at the end of round 4 without picking up a single extra vote - because the threshold changed simply because 5 people dropped out of the election!

In other words, the bizarre and complicated rules of IRV allow someone to win by changing the threshold at the end of the election - dropping the threshold from 15 to 12 to avoid having a real runoff election.

Bundt cake won the race because of a manufactured majority made possible by Instant Runoff Voodoo! Strangely enough, there is no indication if the good people of the Lyndale Neighbohood Association questioned the results of the race!

I wonder if they understood they were being conned by Instant Runoff Voodoo?

But perhaps the real question is why would anyone thinking that voting for food is a good example for use in Instant Runoff Voting? It's not like anyone is really voting for the dessert that will be served at all functions from now on. This vote has no real-world consequences other than to make people feel comfortable with IRV.

The mere fact that this community association accepted the results of the election and didn't object to the threshold lowering shows that people don't understand it well enough to use it on Election Day. I only hope that they pay attention while the votes are counted and ask questions when they don't understand something and don't accept whatever FairyTaleVote tells them as the Gospel.

Would food allergies and other physical health issues effect just the voting, or would it also effect the selection of deserts to be tasted? That is another way that such silly examples of desert and entre sampling and beer tasting are very lame.

At my local Flying Saucer, they have real runoffs for beer tasting. They determine a winner like you do with sports teams - head to head contests between two beers, with the winner advancing to the next round. Finally you have two beers going head to head (no pun intended) to get a winner. Having to drink several beers and rank them makes no sense. Will your order of ranking be effected by the order in which you taste the beers - will it change if you go from dark to lite (or strong to weak) vs going the other way?

1 comment:

Voter said...

I vote for "Bad Penny" microbrew, and I don't want any other.

As for IRV style tasting parties for beer, Joe Soucheray at the Pioneer Press says "But if it worked like IRV, you might end up drinking some swill you never voted for."