Tuesday, July 18, 2017

Should Guelph Have Ranked Ballot Voting?

In my last post I went into exhaustive detail about why I believe the City Council made the right decision about going back to using paper ballots for elections. In this post I'd like to outline a potential change to our voting system that might make our municipal elections even safer and potentially better. The province recently passed legislation that allows our city to use this new system, but Council decided to avoid discussing the change. But I think Guelph voters should at least be aware of this new system because if they decide that they want it, the time to raise the issue should be during the 2018 election.

&&&&

Many of the problems I raised in the previous post arise from the fact that "data mining" allows modern political parties to carefully parse out the voting public's reaction to policy. This allows the leadership to create "boutique policy" aimed at the relatively small fractions of the voting public that are highly motivated over issues that the vast majority of people are indifferent about. This means that a party can create a winning platform by ignoring the opinions of most citizens by adding bits and pieces of the electorate to their core constituency. One example I referred to was the decision by the Mike Harris Conservatives in Ontario to come out against photo-radar. Data-mining also allows parties to figure out which voters support other parties, which is how it is able to direct voter suppression campaigns against these citizens, as in the "robo calls" incident. But underlying all the above is one simple factor, the fact that politicians do not have to get majority support to gain complete control over the apparatus of government. What I am talking about is the "first-past-the-post" system.

A sadly large number of people are under the impression that the people we elect to run our country are elected by a majority of voters. But in actual fact, the vast majority of democratic elections in Canada are settled by a plurality of votes, not a majority. A plurality vote system doesn't care about whether a majority of people vote for a candidate, just who got more than anyone else. In the last three Federal Elections the winning Liberal candidate for Guelph got 49% (2015), 43% (2011), and, 32% (2008). Provincially, the situation is much the same, Liz Sandals won with 42% (2014), 42% (2011), and, 41% (2007). Across the entire country, the Liberals won a majority with only 39% in 2015, and the Conservatives had a majority with 40% in 2011. Provincially, the Liberals won with 39% in 2014, and, 38% in 2011.

&&&&

Our Prime Minister, Justin Trudeau, made a promise in the last election that he would change the voting system of Canada to stop elections being decided by plurality voting. Many people thought that this meant that he would bring in a proportional system. (That is, where seats are assigned in Parliament based on the percentage of votes cast for each party.) But instead, it appears that what he really had in mind was introducing a system of ranked ballot voting, which is somewhat different in result. Since he couldn't get any of the other parties interested in pursuing this change, the Liberals have dropped this idea---at least until the next election. Since all three levels of government are toying around with the idea; either to use in federal elections---or, as only allowed recently allowed by the province---to use municipally; I thought that I'd put some effort into explaining the system theoretically and then discuss how it would operate in Guelph.

&&&&

The key concept of a ranked ballot is the idea that in an election voters aren't generally so polarized that they love one candidate and hate all the others. People are torn. They like the policies of one person, but fear that she simply won't be popular enough to win, so they choose someone else that they don't completely support because they are afraid that someone else that scares them will win if they don't. This problem is usually described as "wasting your vote". For example, someone might want to vote for Jill Stein (Green Party), but they are afraid that if they don't vote for Hilary Clinton, Donald Trump will be elected. A ranked ballot gets rid of that problem by giving people a second choice.


If the US used a ranked system, someone could vote the following way:  Jill Stein (Green Party) #1, Hillary Clinton (Democrat) #2, and, Donald Trump (Republican) #3 (or blank.) When it comes time to count the votes, if it turns out that Jill Stein received fewer votes than either Hillary Clinton or Donald Trump, her ballots are then recounted and people's second choice would be added to the pile of that candidate. So the person who was forced to vote for Clinton before, still ends up supporting her, but she has also been able to vote for Stein without running the risk of helping Trump. With a ranked ballot, there are no more wasted votes. 

Getting rid of wasted votes has a wide range of implications for our electoral system---most of which are really valuable. For one thing, it gets rid of the self-fulfilling prophecy that since someone is not going to win, there's no sense voting for him---which of course means that he is not going to win. (And we end up with president Kang forcing us to build a space cannon.) 

Secondly, it imposes a price on candidates who throw mud during an election. In order to win an election with a ranked ballot, candidates have to be the second and even third choice of voters who support other candidates. If the first candidate acts in a vile way towards the other candidates, their supporters will refuse to make him their second or third choice. This forces politicians to show respect for each other simply out of self-interest. As a result, a ranked ballot reintroduces civility into election campaigns where it has been in decline.

As well, with a ranked ballot people don't win elections with a plurality anymore, they win it with a majority. This means that it isn't possible to carefully parse out a plurality by "throwing red meat to your core constituency" and then adding in enough "boutique policy planks" to push your vote count slightly above someone else's. You have to attract over 50% of the votes, and you can't do that without putting forward policy that is appealing to the majority of citizens. For example, it doesn't help you win if you get 5% more votes for promising to get rid of photo-radar if this means you lose twice as many second choice votes from people who like it. This dramatically changes the debate that happens during elections. This again, will improve the quality of what voters hear during campaigns.

Finally, when politicians always have to win with a majority instead of a plurality, it gets harder to manipulate elections through vote rigging and voter suppression. Remember that robocalls in Guelph failed not because Elections Canada caught the Conservatives doing it, declared the election invalid, jailed everyone responsible, and, then did another vote. Instead, the fact of the matter was that Frank Valeriote got so many more votes than the Conservative candidate that the number of votes diverted from the Liberals simply didn't matter. It might be possible to suppress or steal 1,000 votes without anyone noticing, but it is going to be a lot harder to do this with 10,000. In a system that forces winners to get at least 50% of the vote, it is a lot harder to steal or suppress votes without getting caught.

&&&&

This man had a devious plot---
majority rule through compromise and consensus!
Photo by A.K. Fung, c/o Wiki Commons
If this is such an improvement over the status quo, why did the other parties besides the Liberals reject it?

The NDP and Greens refused to endorse this idea and held out for a proportional system. The obvious inference is that they did this because they felt that in a ranked ballot system they would probably end up losing seats in Parliament. The theory is that the Liberals would end up becoming the second and third choice for the overwhelming majority of voters---simply because of fear of the Conservatives. The Conservatives, on the other hand, opposed a proportional system but rejected ranked ballots too. They did this for probably the same reason as the others. Many of their voters would put down the Liberals as their second choice because they would be afraid of the Greens and NDP. The result would be that no party could win by having the vote split amongst the parties and then getting a bare plurality. Recently, this has been the way that the Conservatives gained power because of a Liberal/NDP/Green vote split. But before that, Jean Chretien followed the same formula to gain Liberal majorities when the right was split between the Progressive Conservatives and Reform Party.

In other words, the unspoken reason for the opposition was ultimately the fear that the Liberals would never again lose an election if a ranked ballot was introduced simply because they would end up being the first or second choice for most voters. This is probably not true, however. That's because changing the way we count ballots would have a tremendous impact on the way parties behave. If the other parties changed their policies so they would be more appealing to a majority of voters, (or, at least put more energy into educating voters about why they should support hitherto unpopular policies), they could win with a transferable vote too. After all, isn't the core idea of a democracy that parties that represent the majority opinion should win? (Damn those perfidious Liberals---what the heck are they thinking!)

I suppose on a deeper level the issue is whether or not you believe that the ideal of compromise should be at the heart of democracy. Putting up your second and third choice in an election shows that you really are willing to compromise your ideals in order to create a government consensus. "True believers" and polarized voters hate to compromise because they don't see the other point of view as having any merit at all. Trudeau himself suggested that he supports a ranked ballot because he wants to avoid the polarization that has recently damaged democracies all around the world. In the end, a ranked ballot system is probably not going to ever be as attractive to the people who devote their lives to a minority political party as it might be to the elected members of "the natural governing party". So it's hardly surprising that the non-Liberal MPs on the committee looking into electoral reform simply dismissed it out-of-hand. To paraphrase Stalin, how you count the votes is often more important than who votes. And professional politicians will always be tempted to conflate the good of their party with the idea of "fairness".

&&&&

I don't want to over-sell the ranked ballot. There are some mathematical complexities in it that are very hard to understand, which is why people often reject it. (Later on I will try to explain this issue using the specific system that the provincial government has imposed upon municipalities.) But before I get to that, I'd like to outline a system called "range voting" that mathematicians say is the absolute best system of vote counting that has ever been devised. I won't get into the mechanics of why it is the absolute best, as I couldn't quite figure out what they were saying in the few times I've listened to someone explain it. But it does have a big value in being plug simple to understand and do, which is profoundly important for anything political.

Anyone who has ever watched competitive gymnastics or figure skating has seen this system in action. You give everyone who votes the option of giving a number between zero and ten for each candidate. If someone wants to give ten to everyone, that's fine, but they can give any other combination of numbers too. So you could give Jill Stein 10, Hillary Clinton 8, Ross Perot 3, Kang 1, Kodo 1, and, Donald Trump 0. When you count the ballots you simply add all these numbers together and declare the person with the highest score as being elected. It is a form of ranked ballot simply because the score you give each candidate gets a different rank in the form of the score you give them. It also had the advantage of giving you a more nuanced rank than simply putting an order of preference. For example, my fictional voter likes Stein a little better than Clinton---but she really doesn't like Trump at all. This is important, because the intensity of a voter's feelings about a candidate are just as important as their relative order of preference.

Range voting happens a lot in sports like figure skating.
I have no idea at all where this image comes from,
If it's yours and you want it removed---contact me.

Please note, that everyone has the option of putting whatever number they want (between zero and ten) besides the name on the ballot. In the last election many people would have put a "10" next to Trump and a "0" next to Clinton. If everyone has the option, then no one is being unfairly disadvantaged.

As I mentioned, range voting is also a form of ranked ballot. As a result, all the benefits I've mentioned above in ranked voting also accrue to range voting. There is one last point, however. That is, it is tremendously easy to make it proportional too. All you have to do is expand the size of your electoral districts and elect more than one MP. The easiest way to do this would be in large cities where the ridings are small. For example, you could stick four together, have a vote, and the top four candidates would be on their way to Ottawa or Queen's Park.

Know who this guy is? No?
Then stop having strong opinions
about knowing your local candidate.
Photo from his website
Unfortunately, in our large, rural Electoral Districts, putting four together would create enormous areas. People will argue that in such large electoral districts it would be impossible to know your local MP or MPP. To this I'd ask the average voter---"do you know the name of your current MP? Would you recognize them if you saw them?" I know most wouldn't be able. So is it really all that important? Every voter would know about the different parties, so they could easily vote for "Joe Blow of the Whatever Party"---because she'd have heard of the "Whatever Party", even if she didn't know who "Joe Blow" is. (If people decide to be honest, they'll admit that that's how most folks vote now.)  I don't know if the Election Reform Committee had this option explained to them, but I suspect if they did, someone probably didn't like it.

That's just the way it is, isn't it? The main thing to realize that there are a lot of very smart people who have thought long and hard about how to make our voting systems better.  The transferrable vote is one suggestion. Range voting in a multi-member district is another. These systems work fine in other countries, but I suspect that lots of wailing and gnashing of teeth can be expected before any real change takes place.

&&&&

Let's move from Ottawa to Guelph. How does its elections compare to national and federal ones?

In the 2014 municipal election the Mayor received 50.75% of the vote. The rest of Council got their seats by receiving between 33.46% and 19.97%. The average for the entire Council comes in at 25.59% of the tally.

The average support for each of these people was 25.59%---which is pretty darn good!
Photo of current Council from Guelph City website, some manipulation of image---I removed the Mayor.
This might sound awful, but remember that we have a system where voters cast two votes and elect two people.  And when we vote, we can only vote for one particular candidate once---the other has to go to someone else. That means that if every single voter in an election voted for one of the candidates and split their other vote among several other candidates, the absolute maximum number of votes that that candidate would receive is 50%. (That is, unless they can convince voters to only vote once. If everyone single voter in a ward did this, the candidate could theoretically receive 100%.  This is called "plumping", and it used to be a factor in the Guelph's old "at large" system where the city was treated as one large constituency, and people voted for up to 12 council members---lots of people only voted for a few of the positions.)

In the system Guelph uses, a majority win is 25%, so an average vote count of 25.59% for winners is quite good because it translates to 51.18% in terms of popular support. Looking back at the eight elections that Guelph has held since it switched from an at-large system to a ward system, these results are the norm instead of the exception. This means that Guelph Council is one of the rare places in Canada where most politicians get elected with a majority instead of a plurality.

&&&&

Before people get too comfortable, however, we should remember that the voter turnout in 2014 was far from praiseworthy---a miserable 43%. Even this was a bit of an improvement over the 2010 election, which had an even worse score of 34%. This was one of the arguments in favour of Internet voting. Personally, I suspect it had more to do with a very competitive race for the Mayor, which tends to raise voter turnout. The novelty factor of a new way to vote also might have had something to do with it.

Unfortunately, with the death of the Daily Mercury, I suspect that voter turnout will be less in the next poll. First, lots of people in Guelph simply won't even know that there is a municipal election. They won't be seeing anything about it in what news coverage they follow (probably mostly international on-line, with perhaps national news on tv.) They commute or work with commuters, so there won't be any "common wisdom" to pick up in the lunch room because people live in different communities from each other. Even if they do know that there is an election, they probably won't have a clue about the personalities or issues. And, of course, it is totally reasonable to not vote if you don't know enough to have an informed opinion.

And most people only have a certain amount of energy to devote to the public sphere. Many are happy to read about the election in a newspaper delivered to their door. But most don't have the energy to do the poking around necessary to inform themselves---hence my attempt to fill part of the gap with the Guelph Back-Grounder. (Just to give an example, it is darn hard to find statistics about voter turnout in municipal elections. It just isn't something that has been traditionally recorded or reported in election results statistics. Why? Perhaps because it is somewhat embarrassing.)

&&&&

Now let's look at the specific transferable vote system that the province now allows municipalities to use. That's right, I wrote "allows municipalities to use" because in Canada cities are specifically under the tight jurisdiction of the province. And in Ontario, the province is extremely jealous of its authority. That means that there is zero opportunity for the city to "play around" with the voting system we use. We either use what we have now, or, the specific form of transferable voting described in the Municipal Act---no fooling around with stuff like range voting!

So here's the system that the province says municipalities have to follow.

First, the city can design the ballot as:

A List

Columns

Or a Grid
Images from Govt of Ontario Website

&&&&

Next people have to wrap their heads around a little math. If we switched over to a ranked ballot, electing the Mayor would be very simple because it is just one person for one slot. What gets complicated is when you elect more than one---like we do in our Wards. That's because we have to decide what to do when we get a situation where one person gets elected easily---and there is a real struggle for the other slot.

First of all, we have to calculate the "threshold" of the vote. The ministry provides a formula for calculating this number:   [(A – B) ÷ (C + 1)] + 1, where "A" is total number of ballots cast, "B" is the number of rejected ballots, and, "C" is the number of spots to be filled. Let's look at an actual example from Guelph to see how it would work. In Ward One during the 2014 election the following vote count arose:  Bob Bell 2,984, Dan Gibson 3,419, Terry O'Connor 1,705, Maria Pessano 1,386, and, Karolyne Pickette 2,705. So, in the case of Ward One in the last Guelph election, the numbers would work out as: 12,199 votes were cast, (let's forget the rejected ballots), and there were 2 spots to be filled. So the number is [(12,199 - 0) ÷ (2+1)]+1= 4,067. So to win a seat on Council, someone would have to get 4,067 votes--both initial and transferred.

Let's assume that the numbers we have above represent the first choice of all the voters.

Still from the 1960s Irwin Allen tv show "Lost in Space" via Google Images
Used under the "fair use" copyright exemption
(Readers must remember that the assumption that the vote each of these candidates got in the last election was the voter's first choice pretty much negates the whole idea of transferable votes. People vote strategically in first-past-the-post, plurality systems because they don't want to waste their vote. So if---totally for the sake of understanding this point---we assume that some people who really like Karyolyne Pickette and Terry O'Connor voted for Bob Bell because they don't like Maria Pessano, we can see a totally different outcome if we switched to a transferable vote. Under a single transferable vote, these voters would vote for their real choice first, then mark Bob as their second choice in order to prevent the dastardly Pessano from seizing control of a seat on Council. So please remember that in what follows I'm just making this assumption to explain the math---there is a real difference in how people vote under a plurality system versus transferable votes!)

OK, so let's assume that the votes mentioned above are the real first choice of voters. With a threshold of 4,067, neither of the two front-runners---Dan Gibson (3,419)  and Bob Bell (2,984)---have enough votes to be declared a winner. At this point, it's time to start transferring ballots. The Municipal Act describes two ways of doing this: the "single elimination method" and the "batch elimination method". In the single elimination method, the people counting votes look at this first count and say "Maria Pessano has the lowest number of first choices, 1,386, so it's time to take her out of the race and put her voters' second choices into play".

Let's also say that 700 of the people who voted for Maria Pessano put down Dan Gibson as their second votes. This gives Dan Gibson 4,119 votes, which puts him over that threshold of 4,067 votes and makes him a successful candidate. Oddly enough, not a single person who voted for Maria Pessano put down Bob Bell or Terry O'Connor as their second choice, and instead put all their votes (686) towards Karolyne Pickette, which then gives her 3,391 votes.

At this point, the Municipal Act says that the city should go on to a third round of vote counting. You might think that at this point we just eliminate the lowest candidate an redistribute their votes. If you did, you'd be wrong. The Act says that at that point we need to recalculate the transfer ratio. We do this by following this formula:  (F – E) ÷ F . "F" is the number of votes cast for the successful candidate (Dan Gibson's 4,119) and "E" is the threshold for the office (4,067), which gives us "52". Divide that number by 4,119, and we get the ratio number of  0.013 (I rounded off using the rules of scientific accuracy.)

What is the transfer ratio used for?

Well, now that we have a winning candidate, we need to have a mechanism for fairly transferring the extra votes that people cast for him (Dan Gibson) to the other candidates that are still in the running. The point is, we don't want to get into a situation where too many people vote for a successful candidate and by that waste their votes that could be used to elect someone else. To understand this, consider a hypothetical situation where 90% of the people in the ward voted for Dan. If that happened, then if we didn't have some way of taking into account these people's second choices, the next candidate could get elected with only at most 10% of the voter's intentions being followed. That would hardly be fair, would it?

So with that transfer ratio in hand, we can now count the second choices of the people who voted for Dan Gibson, and then multiply that count by 0.013, and then add that to the second choices of the people who voted for Terry O'Connor (because he had the second lowest vote count.) Terry got 1,705 votes, and let's say that his voters gave 700 to Karolyne Pickette, 400 to Bob Bell, and, 605 to Dan Gibson. The votes transferred from Dan Gibson are multiplied by the transfer ration of 0.013, which gives us a total of 8 votes.  The 700 votes that Karolyne got from Terry's supporters are added to the votes she got as the first choice (2,706) plus the second choice votes she got from Maria's supporters (686), plus the 8 votes that are transferred from Dan's supports which puts her at 4,091 votes. This gives her 4,100 votes, which takes her over the threshold (4,067) and gives her the second seat on Council.

This is how I feel after trying to explain how
a transferable vote works, Wiki Commons
Sorry to say, I've significantly simplified the process involved in selecting a candidate in order to make things seem at least somewhat comprehensible in this example. For example, I've ignored the issue of "exhausted ballots". These are the votes where someone voted less than than they could have, for example, if they only voted for Dan Gibson and left every other option open. In order to avoid the distortions caused by "plumping" (as I mentioned above), the clerk is supposed to add up the number of exhausted votes after every round, and recalculate the threshold. There's also a complex mathematical system for eliminating all the losing candidates all at once---which I haven't even tried to understand. If anyone finds a mistake in my calculations, I wouldn't be surprised. The point of the whole above exercise is to show how complicated a transferable vote system looks when you have two or more people elected at once with one vote cast.  
&&&&

I went into somewhat tedious detail outlining the transferable voting system recommended by the province because I'd like readers to understand some of the problems that the city Clerk and Council would have trying to explain this new system to voters. I don't think that any city would be foolish enough to try and sneak in a transferable vote system, but the province would declare that illegal if they tried. That's because the amendment to the Municipal Act that allows transferable votes specifies a process of public consultation and education before it can pass a bylaw changing the system. Moreover, staff will also have to figure out how much the change will cost and identify the vote counting specifics. Given the current context, that would probably involve leasing some vote counting machines that are designed to deal with transferable ballots. (And that opens yet another can of worms---as outlined in my previous article.)

Faced with all these issues and the fact that the province only passed the legislation allowing transferable votes in 2016, staff recommended that Council not change the voting system for the next election campaign. Since they would have had to rush the whole thing, and the system we currently have seems to work relatively well, this was probably a wise decision. But that doesn't mean that the change doesn't have merit. Let's see what the future Council thinks.

&&&&

If you really like this article and can afford it, why not toss something in my tip jar?  (It's on the top right side---just click on the "donate" button.) If you really feel generous, why not subscribe using Patreon (just above the "donate" button)? Even a dollar a month would make a big difference and help me find the time to do more research and writing. Also, don't forget to share my URL on FaceBook, Twitter, and other social media. If my readers won't share, the word doesn't get out!


&&&&

It is REALLY HARD to get most people to support any type of change that they have a hard time understanding. So if Guelph Council ever decided to switch from the present system to a transferable vote where we still elect two people per ward, I wish them luck. It needn't be that way. Many are the things in our world who's operations are totally beyond our comprehension (the cell phone, the automatic transmission, etc.) We simply accept that doctors, systems analysts, mechanics, etc, know more about their specialty than we do, and, accept what they say. But it rankles voters when you ask them to embrace something that is difficult to wrap their heads around. That's why we don't have proportional representation of any form in Canada. It always comes down to a referendum, and in these people ignore the experts, throw out some sort of bogus argument, and, vote for the status quo. (This is why I support range voting---it is so easy to understand I suspect it would scare a lot less people.) 

Having said that, it is really important to avoid throwing out the baby with the bath water. Transferable voting has some advantages---really big advantages that more than outweigh any discomfort that comes from math anxiety. 

No more need to vote strategically!
Clint Eastwood from "Dirty Harry"
Fair Use copyright exemption
First of all, it ends strategic voting. No matter how much you dislike Donald Trump, you will no longer have to vote for the slightly less annoying Hillary Clinton. Instead, you can vote for the Green Party Candidate first and put Clinton down as your second choice. Clinton may end up winning anyway, but at least you don't feel that the system is putting a gun to the side of your head and threatening you with the Trump bullet in the brains if you make a mistake and refuse to vote strategically.
Secondly, the transferable vote can make a difference. In the hypothetical example I gave, Karolyne Pickette got elected instead of Bob Bell. In a transferable vote system this often happens when a really polarizing figure seems to be one of the front runners but gets almost no support for second or third votes---which means that they start out in the lead but end up losing. In the recent French election (which uses run-offs instead of transferable votes---same principle, different mechanism), Marie Le Pen (a very strident anti-immigrant politician) came in second with 21.3% in the first round of voting against Emmanuel Macron who got 24%. When it came time for a second round of voting, the people who supported candidates besides these two then swung decidedly for Macron, who won the next round decisively with 66.1% versus Le Pen's 33.9%. 

Another example comes from the recent leadership race for the federal Conservatives, who use a transferable vote. (Funny that a party that has built it's election strategy around exploiting pluralities created by first-past-the-post and fights tooth-and-nail against voting reform for us plebes would opt for a transferable vote in their internal elections---.)  In the first round of vote counting, Andrew Scheer only received 21.82% of the vote, which put him well behind Maxime Bernier who had 28.89%. But after  an exhausting 13 rounds of recounts, the vote transfers finished off with Scheer at 50.95% and Bernier at 49.05%---giving Scheer the leadership.


Andrew Scheer likes
transferrable voting.
Photo by Marcos Oliveira, AgĂȘncia Senado
c/o Wiki Commons
Maxime Bernier maybe not.
Photo by Marcello Casal Jr.
AgĂȘncia Brasil, c/o Wiki Commons 





One last point that voters and Councillors should think long and hard about. Just because Guelph currently elects most of its Council through majorities doesn't mean that it always will.  Could there come a time when we end up with crazy splits being the norm, and extreme people routinely get elected to office? And if it does, will any of those people elected in that system want to switch to a system where they won't get elected unless they can appeal to the majority of voters? I have yet to meet a politician who would willingly bring in a system of voting that would guarantee that they would lose the next election. This time of stable government might be exactly the right time to change our voting system.  

No comments:

Post a Comment