On Canada’s Voting Reform Survey: Part 2

In Part 1, I critiqued the first half of Canada’s #ERRE survey on electoral reform, introduced the various single-winner voting systems that exist, and then told you to forget the whole thing because single-winner systems don’t make sense in multi-winner scenarios. One of the reasons for this is well-known: it’s called Gerrymandering.

Gerrymandering Explained

Christopher Ingraham explained this problem in “How to steal an election: a visual guide”:

Suppose we have a very tiny state of fifty people. […] And just our luck, they all live in a nice even grid with the Blues on one side of the state and the Reds on the other.

Now, let’s say we need to divide this state into five districts. Each district will send one representative to the House to represent the people.

Fortunately, because our citizens live in a neatly ordered grid, it’s easy to draw five lengthy districts — two for the Reds , and three for the Blues. Voila! Perfectly proportional representation, just as the Founders intended. [grid 1 above]

Now, let’s say instead that the Blue Party controls the state government, and they get to decide how the lines are drawn. Rather than draw districts vertically they draw them horizontally, so that in each district there are six Blues and four Reds. […] The Blues win 5 seats and the Reds don’t get a single one. Oh well! All’s fair in love and politics. [grid 2 above]

Finally, what if the Red Party controls the state government? The Reds know they’re at a numeric disadvantage. But with some creative boundary drawing — [grid 3] — they can slice the Blue population up such that they only get a majority in two districts. So despite making up 40 percent of the population, the Reds win 60 percent of the seats. Not bad!

Gerrymandering (achieving various outcomes by manipulating district lines) is possible whenever people that favor different parties live in different locations. It exploits divisions such as rich vs poor, industrial vs agricultural, black vs white, or Anglophone vs Francophone.

The unfairness of district boundaries is easiest to see and exploit under FPTP, but in fact all the single-winner systems suffer from the same problem to some extent.

Even if the district lines are drawn by an independent commission, like in Canada, it doesn’t really solve the fundamental problem that single-winner systems produce non-proportional results. Even when there is an independent commission, using single-winner districts tends to have the following effects:

1. Small parties are punished. For example, in the last election the Green party got 3.4% of the votes, but only 0.3% of the seats (i.e. one seat). Whereas gerrymandering exploits differences between people in different areas, small parties are punished by similarities. For example, if a party has 20% of popular support, but that support is spread perfectly evenly throughout the country, the party cannot win a majority in any one location (except in case of a five-way split) and gets zero seats. Small parties require regional variation to win anything at all.
2. Regional parties are rewarded. For example, in the 1993, the separatist Bloc Quebecois became the official opposition (19.1% of the seats with 13% of the vote) and in 1997 the Reform party of Western Canada became the official opposition (23.2% of the seats with 19.3% of the vote).
3. Individual candidates need highly concentrated local support. Candidates whose popularity is too spread out geographically cannot win.
4. Citizens get a distorted view of the popularity and electability of each party. First of all, if we’re talking about FPTP, no one can see how many voters voted “strategically” for a big party (the “lesser of two evils”) when they really wanted more power to go to a smaller party. And regardless of which single-winner system is used, the media will probably focus only on seats won and ignore the popular vote. I mean, just look what happens when you Google “Canada Election Results 2015”:

News coverage focuses on districts and seats, not votes

Notice how only one of the search results shows the popular vote for each party, rather than the number of seats won or the geographic district winners. If you were watching TV during the election, you would have seen that same bias. The media tends to ignore how voters voted, choosing instead to focus on who won and the big map and the “horse race”.

What, I would ask, does any of this have to do with democracy? How is it “democratic” to make the election results depend on where people live and whether or not they are geographically distributed in a heterogeneous manner? How is it fair to say one candidate can win because her fans are geographically concentrated while another cannot win because her fans are geographically dispersed? The answer, of course, is nothing. It’s not democratic at all.

These flaws (and more which I haven’t listed) are fundamental results of districting. The only way to avoid them is to use systems that were designed for multiple winners. Every systems I’ve seen that was designed for multiple winners has been, in one sense or another, proportional, so we could call them “Proportional Representation” systems, although some multi-winner systems like STV and DR are not usually called by that name. Proportionality is not required, either: any given multi-winner system could be modified to make it non-proportional. For instance, it’s certainly possible to bias a PR system to make majority governments more likely, even though, as far as I know, it has never been done.

Let’s get back to the survey, which has started asking about Proportional Representation:

Proportional representation questions

This section introduces two types of proportional representation, “open list” and “closed list”. In doing so it forgets to mention that there are many more ways to do proportional representation than they’re telling you. For example, the Wikipedia page for open-list PR mentions several approaches with names like “Relatively closed list”, “More open list”, “Most open list”, “Free list or panachage”, and “Arbitrary list”. So, after the survey’s inadequate description of the options we have available, it asks some confusing questions:

• Voters should vote for political parties (not specific candidates) and the seats in the House of Commons should be allocated based on the percentage of votes obtained by each political party. This describes Closed-List PR, and my answer depends on the alternative. If the alternative is FPTP, you already know I STRONGLY AGREE, because everything is better than FPTP and multi-winner systems are better than single-winner systems. But among the various PR systems, I’m not a fan: closed-list wastes the opportunity to listen to voters, since it only asks what party you support and nothing else. Because of this tension, my final answer is NEUTRAL.
• Political parties should determine which of their candidates get elected from their list. What does this question mean? Is it describing Arbitrary List? Arbitrary List adds some flexibility in the system and provides a way for parties to differentiate themselves from each other — but I don’t think that’s what the question means; maybe this question is asking about Closed List a second time, so my answer is NOT SURE.
• Voters should determine which candidates get elected from a party’s list and the seats in the House of Commons should be allocated based on the percentage of votes obtained by each political party. AGREE. It’s not my favorite, but for the sake of voter choice, Open List is better than Closed List.

Single Transferrable Vote (STV)

Canada’s electoral system should produce a proportional Parliament (where seats roughly match the parties’ vote share) through the direct election of local representatives in multi-member electoral districts.

WEAKLY AGREE: STV is a reasonably good system. In part 1 I expressed my disdain for the idea that voters should be forced to vote for someone “local”, but STV is a reasonable compromise between the “localists” and those like me that prefer total geographic freedom. STV uses local districts that are larger (typically 3 to 5 times larger) than the districts we have now, and they often provide voters with a choice among multiple candidates from the same party, typically on a preferential ballot.

The disadvantages of STV are that the math that picks the winners is slightly complicated, that the results aren’t completely proportional (only approximately), and that the reward to each winner is binary (win/loss), not proportional to votes received. For example, if the most popular candidate gets three times as many first-choice votes as the second-most popular, both of them win one seat of equal value in the legislature.

Note: if STV is used for a single-winner election, it degenerates into IRV, but STV was developed by different people, for a different purpose. As a roughly-proportional system, I currently think STV is better in multi-winner scenarios than all single-winner systems, even though I think IRV is a bad system for single winners.

Note: Stephen Dion’s P3 is similar to STV. Of the two, I think I prefer STV.

Mixed Electoral Systems

Personally, I think if we’re going to “mix” electoral systems, it should be voter-driven on a per-election basis. These are several ways to do this, but I’ll just explain one. Imagine if voters could pick which system they preferred, party-list PR or Simple Direct Representation (SDR), and vote on the corresponding ballot for that system. Suppose 68% of voters vote on a PL-PR ballot and 32% vote on an SDR ballot. Then 68% of the power in Parliament would go to the candidates elected through PL-PR, and the remaining 32% of power would go to candidates elected through SDR. In this way, voters could choose not only their candidate or party during each election, but also the electoral system itself.

However, I doubt the committee will consider “crazy” ideas like that. Probably they are only considering Mixed-Member Proportional (MMP). This system mixes party-list proportional representation with a heavy dose of nostalgia, using a single-winner system (most likely FPTP, but maybe not) as a centerpiece.

As you know, single-winner systems produce non-proportional results. MMP solves this problem after-the-fact by granting extra seats to parties that didn’t earn their fair share of seats during the election. Typically, some fixed percentage of seats (say 20%) would be reserved for this compensation process. It doesn’t directly solve the problem of “wasted votes”, but it gives voters that lost in their own riding a consolation prize in the form of party-list seats. Oddly, the system lets you vote for the Liberal party but vote for an individual who is a Conservative, or vice versa, but this doesn’t seem to break the system as a whole.

• Voters should cast two votes on their ballot: one to directly elect a member to serve as their representative, and a second for a party or parties to fill seats in the House of Commons based on the vote share they receive. WEAKLY AGREE. MMP is a decent system, I’d happily accept it. It seems better than party-list PR because it gathers more information from voters.
• Seats in the House of Commons should be allocated in proportion to the percentage of votes received by each political party. AGREE, but isn’t this a question about PR in general rather than MMP in particular? And what if I want a system based mainly on the election of individuals rather than parties?

Mandatory or Compulsory Voting

• Canadians should be required to cast a ballot in a federal election. (This could include spoiling a ballot.) STRONGLY DISAGREE: citizens should only vote after giving the matter sincere thought, and not make a superficial decision in the voting booth. Forced voting encourages people to vote who haven’t given it much thought. If you’re reading this, you’re working harder than 99% of the people out there. After spending a lot of time looking at platforms and watching debates, do you really want your vote to compete with people whose only exposure to politics was TV commercials and yard signs? Without mandatory voting, many of the couch potatoes will stay home, as they should.

Online Voting

• Canadians should be able to vote online in a federal election. WEAKLY DISAGREE: As a computer geek and Slashdot reader I have heard a lot about the security risks of electronic voting and online voting. Many places in the U.S. are using insecure voting machines already that are vulnerable to hacking and tampering by technicians. Large-scale vote-rigging is much easier with electronic voting machines than with paper. Protecting the integrity of online voting is even more difficult. Ensuring that each eligible voter only casts one vote and verifying voter identity while ensuring the process is verifiably unhacked and unhackable — but also providing a “secret ballot” — seems like a nearly impossible “holy grail”.
• There is a public good or value associated with voting in person.
  Good question! For once I’m legitimately NOT SURE.
• I am concerned about the security and reliability of online voting. AGREE.
• Online voting should only be considered as an alternative for people unable to vote in person on election day. AGREE. But I see a flaw in the design of this question. Two groups with opposite opinions will both say “disagree”! People who think everyone should vote online will answer this question the same way as people that think no one should vote online.

Moving Forward on Electoral System Reform

• Any plans for a future Canadian electoral system should be determined by a majority of members of Parliament. AGREE, but isn’t it a given? How could the system change without the support of MPs? Do I misunderstand the question?
• Any plans for a future Canadian electoral system should require broad public support, in addition to parliamentary approval. WEAKLY AGREE. Some positive reforms in the past, such as giving women the right to own property and vote, were driven from the top down, while other reforms came from the grassroots. This is a positive reform being driven from the top down, and there is no reason for anyone who believes in democracy to oppose it, so why should we risk letting voters turn down the new system before they have even had the chance to try it? On the other hand, the public does seem to generally support electoral reform, so there’s little reason to disagree here.

Broad public support should be gauged through…

• In-person and online consultation with Canadians representative of Canadian society (demographically and geographically). AGREE. That’s sort of what they’re already doing anyway. This is the best way to do it, because it gathers opinions from people who actually care about the issue (for or against) rather than confusing the matter by gathering opinions from people that are indifferent.
• The creation of a citizens’ assembly. WEAKLY AGREE. This was done earlier in B.C. and seemed to work fine. I think creating such an assembly is redundant, but as long as it doesn’t prevent electoral reform from happening before the next election, I’m on board.
• A direct vote by Canadians on an option or various options for a future Canadian electoral system (through a plebiscite or referendum).
  DISAGREE. You know the saying from Winston Churchill? “Democracy is the worst form of government, except for all those other forms that have been tried from time to time.” It’s true. Random people on the street aren’t in a position to make good decisions about how government should run. We don’t have democracy because it’s the best system, we have it because it’s much safer than the alternatives. Some would argue that a “benevolent dictatorship” is better, but there is no way to ensure that a dictator will always be benevolent or even intelligent enough to do the most important job in the country. We should get the best electoral system we can, regardless of whether the people are smart enough to want it, so that at least the voting system measures our opinions accurately, and makes fair and mathematically sound decisions about whom to put in parliament. At least that way, when we choose the wrong people to lead us, we can no longer blame it on the voting system.

Here is a paradox that represents a core problem with the referendum approach. Let’s say you choose 5 possible electoral systems. WHICH VOTING SYSTEM DO YOU USE IN THE REFERENDUM ITSELF? Condorcet? IRV? Range Voting? First Past The Post? Officials must pick a voting system for the referendum, but the voting system used, and the wording of the question(s), will bias the results one way or another. I think the best system for the referendum is Range Voting, even though I know enough to reject Range voting as a way to pick Parliament as a whole. We need an intelligent decision by people who understand why that’s true. Average voters won’t understand this, though.

Asking voters which voting system they want is like asking home buyers what kind of shingles and insulation they want for their new roof. It’s an important question, and the answer directly affects them, yet they really don’t know. They just want something that works well. An expert can educate them if they’re prepared to listen and learn, but then again, why not have the expert make the decision in the first place?

I see the irony of saying we shouldn’t have an election to choose the election system. Yet for all our lives we’ve suffered under FPTP and until now no one ever offered us a choice; none of the big parties (with the 55% seats they got from 40% votes) were willing to consider changing the system that so clearly benefitted them.

It makes a lot more sense to ask voters about systems they have experienced than about hypothetical future voting systems. So I have imagined a “meta-system” where we have two completely separate electoral systems and voters choose, each election, which of the two they prefer. The less popular system would lose up to 10% of the seats allocated to it in the next election, and if one of the two systems drops to 0% seats, some kind of process would be automatically initiated to replace the unpopular system with something new and different, and the seats allocated to the new system would jump to 50% so that voters get a chance to experience the new system over a period of time. (Notice that under this meta-system, replacing a voting system should take more than 5 election cycles or 20 years, so voters have plenty of time to evaluate it.)

I’m sure the ERRE is not looking for solutions that are so sophisticated. But they are listening, and I’m not looking this gift horse in the mouth.

Final thought

If you can only take away one lesson from all this, let it be this: Multi-winner systems like “open-list PR” and “MMP” are good, whereas First-Past-the-Post is the worst. The survey has one page at the end where you can have your say in 1000 characters or less. Make sure they know that Parliament should be elected with a multi-winner system, not a single-winner system with districting. As long as they do that, the system should work reasonably well.

So what’s my favorite voting system? Direct Representation (DR), which the ERRE is not looking at. I can’t blame them, either: to start with, it should probably be used experimentally at a smaller scale. But I thought so much of DR that I designed Simple Direct Representation (SDR).