The folly of finding a new flag

Dr Eric Crampton
The Dominion Post
6 November, 2015

Will the coming flag referendum pick the right flag? It looks like we might never know.
It is hard to imagine agreement on what the right flag is. But I expect that we would all agree that if one of the proposed flags beat each and every other option in a head-to-head race, that flag should be considered the winner.

It would be easy to tell if any option were really a "champion of all" flags – or at least champion among the ones chosen by the panel. Voters are already being asked to rank as many flags as they like, with unranked options being treated as last-equal. And so the Electoral Commission will know, for each voter, which flag is preferred in any two-flag race. Tallying it up across all voters would let the Electoral Commission tell whether there were any overall champion – what voting theorists call a Condorcet Winner.

Unfortunately, the voting process Parliament chose not only means that we might fail to choose a Champion Flag, if it exists, it also means that we may never find out whether any champion-of-all flags existed.

Instead of asking the Electoral Commission to check first for a Condorcet Winner, Parliament told the Commission to follow a simple elimination procedure. It works as follows. If any flag is the first choice of a majority of voters in the first round, things are simple: that flag would by definition be the champion of all. But things get trickier if there is no first-round winner. In that case, the option that was the first choice of the fewest voters is dropped.

Let's take a simple example. Suppose that a million people vote in the flag referendum. 350,000 voters prefer the Black, White and Blue silver fern design but rank the Koru as second and all other options below that. 340,000 prefer Red Peak, rank the Koru as second, and all other options rank below that.

Finally, 310,000 prefer Koru as their first choice, rank the Black, White and Blue silver fern second, and all other options lower. Under the elimination procedure, the Koru is eliminated first because it is the first choice of the fewest voters.

A majority then prefers the Silver Fern to Red Peak. But, had Silver Fern faced Koru head-to-head, the Koru would have won – and it also would have beaten Red Peak in a head-to-head race.

Whenever a flag is the second choice of many voters but the first choice of few, the elimination procedure that the Electoral Commission will use risks choosing a flag that would have been beaten in a head-to-head race by another option. It might not matter, as there are only some configurations of voter preferences that would yield that kind of result. But it is disappointing that the New Zealand Flag Referendum Act did not ask the Electoral Commission to check.

Even worse, the Electoral Commission is forbidden from publishing the data that might let us know whether the voting procedure has resulted in the wrong flag being chosen. I asked the Electoral Commission whether they could provide the data that would let other researchers test which flag might have been chosen under different voting rules. Their Senior Legal Advisor replied that Section 50 of the New Zealand Flag Referendums Act 2015 prohibits them from disclosing that information and that they would decline any Official Information Act request on that basis.

If their interpretation of the law is correct, we will never know whether the flag announced as the winner on referendum night was really the champion.

I could suggest that the government rush through legislation under urgency telling the Electoral Commission to check whether any flag would beat all other contenders before running the run-off.

But, this year's flag referendum comedy has had enough of that kind of thing. And among the options available, the kind of run-off election being used in the flag referendum is far from worst.

It just could have been a little bit better.

And perhaps we could consider it to be a feature, rather than a bug, that we might never really know which flag was really best. If your preferred flag does not win and if there is no overall winner in the first round, why not pretend that the flag you most liked really was the winner and would have been chosen under a better voting system? You could even be right.

Let a half-dozen different flags bloom.

For what it's worth, the flag that looks like the back end of a monkey with a curly tail has already won my heart.

Stay in the loop: Subscribe to updates