Anyone for a Bullet in the Foot? Instant Runoff!

by Devin Ray Freeman

Posted August 11, 2006

“A vote not from the heart is a vote wasted” quoth Michael Badnarik in 2004. If only most voters would think in this way, and not in dismal thoughts of “electability,” drearily determined to vote against the greater evil.

No wonder, though, that so many are locked into that mindset. The current voting method (called “Plurality Voting” in voting theory talk) effectively locks us into the lesser-of-two-evils dilemma. This is well-seen, but there are some who say they hold the key. They call it “Instant Runoff Voting.” “IRV, easy as one, two, three” they tout. It'll “guarantee a majority,” “open the field to minor candidates,” “improve democracy!”

They're gravely mistaken. Instant Runoff is in no way suitable for multi-party democracy.

What is Instant Runoff?

So, you may ask, just what is Instant Runoff? It's complex and hard to put into words, but the upshot is that Instant Runoff is:

• really hard to recount and it's really hard to count in the first place.

• illusory in its ranking and in the majority it delivers.

• sure to maintain duopoly as well as Plurality does.

Instant Runoff is a tally and elimination scheme, retallying without revoting (reallocating), that repeats until a majority of votes reshuffles into one pile.

To run through this step by step, examples may be helpful.

So let's say my man Larry, Hal, that fascist Katherine, that complete bozo Sleazworthy and that one guy whose name I keep forgetting all run for mayor. I go vote.

It's Instant Runoff, so the directions are:

Rank all candidates in order of preference.

Easy enough, though I'm not real sure where to place that one guy, whoever he is. OK, then...

...1^st - Larry, 2^nd - Hal, 3^rd - that one guy, and 4^th, 5^th - Katherine, Sleazworthy in whichever order.

The Algorithm

Now for the hard part - the many steps to the counting process:

Round One

1. First choices are tallied up.
2. If there's a majority you've got a winner.
3. If not, the candidate who was the first choice of the fewest is eliminated.

(Deviate that I am, Larry's first to get booted out.)

Round Two

4. Reallocation - second choices of those whose first choice got eliminated are tallied up.

(With Larry eliminated, my vote now sits on Hal's stack.)

5. If there's a majority candidate now, we've a winner.
6. If not, the second choice of the fewest gets eliminated.

Round Three

7. Reallocation - third choices of those whose second choice got eliminated are tallied up.

(My vote for Larry that got switched over to Hal is now a vote for that one guy.)

8. If there's a majority holder, then it's a win.
9. If not, the third choice of the fewest gets eliminated.

Round Four

10. Reallocation - fourth choices of those whose third choice got eliminated are tallied up.
11. There is a majority. We have a winner.

The third round eliminates a third candidate, which leaves only two, so final reallocations in the last round assure a majority. Depending on how the ballots tumble, my vote either stays with that one guy, or else he gets eliminated and my vote goes to either Kathy the fascist or Sleazworthy, whoever it was I put 4^th.

Had I put only Larry and Hal on my ballot and not bothered to rank candidates I don't like or don't even know about, my ballot would have been eliminated for lack of someone to reallocate to, as is essential to guarantee a majority outcome. Voila! Democracy improved! (There are variations on the theme. I've presented the version in use in most of Australia.)

Instant Runoff's Big Drawbacks

If the example above hasn't adequately illustrated the illusory nature of the ranking and majority yielded in this IRV reshuffling scheme, mathematicians have. They've shown that in ordinary scenarios, if a group of major party supporters conspires to rank their favorite candidate lower they can push up that candidates chances of a win! And vice-versa. Ranking lacks meaning in IRV.

To illustrate how this illusory ranking can be exploited through tactical voting to alter the outcome, let's suppose Sleazworthy supporters take a poll in town before the election. They ask for first and second choices for mayor from people on the street. They find to their delight that Sleazworthy is the first choice of almost 40%, which is several % better than Kathy. Hal and Larry are last with 4% and 6% respectively. As for second choices, though responses are mixed and some give no second choice when asked, they find to their horror that nearly half say Larry! This means Larry could win in the second round! Oh, no! What to do?... Easy! Tell all your Sleazworthy-loving friends to vote Hal first, Sleaz second! Then Larry gets eliminated in the first round and none of those second choices for Larry ever go to Larry. Nobody need ever know what a crowd-pleaser Larry is and the win for Sleaz is in the bag! Just as IRV can make a come-from-behind victory possible for the underdog, IRV makes strategic voting workable for major candidates to insure their victories.

Furthermore, mathematicians agree that IRV, or any such elimination and reallocation scheme, makes the count very labor intensive because the data can not be summed up for transfer. I'll explain what's meant by “not summable.” Because IRV entails eliminating and reallocating, all the data has to be gathered up from all polling places into one central location before the elimination and reallocation process can begin. This means a given polling place can add up nothing but ballots of like-orderings. Otherwise, the path of reallocation would be lost and nobody would know where to restack any ballot. The number of possible orderings, then, equals the number of candidates factorial! In the above example it's 5!, or 120 possible orderings! On six candidates it'd be 720 orderings! Not all possible orderings are likely to occur, but most probably will. So, each polling place has to divvy into that many stacks and provide all those subtotals, and can provide no summed results for its precinct.

Of all voting methods going, Only IRV and IRV-like variants are not summable. This means that counting up the votes is super complex if there are more than four candidates, and recounting every bit as complex. This means that IRV data is so unwieldly it's problematic in any election held over more than one polling place. This unsummability that makes a recount a huge to do and makes it hard to count up the votes in the first place, makes it more susceptible to tampering.

I say IRV is clearly well-designed to keep the same two parties in control, but if you're not convinced that IRV is wholly unsuitable for multi-party democracy, don't take my word for it. Simply look to Australia, where Instant Runoff (called “Preferential Voting” there) has been in place for a century, and two barely distinguishable major parties bicker but reign together with impunity. Just think. Where's it gotten Australia? Click and see what the Australian Electoral Comission itself says about Preferential Voting (IRV).

A few U.S. cities now use IRV, or an IRV-like reallocation short-form under the IRV banner! *gasp* It's beginning to catch on in the States!

The pro-IRV people are deluded. Instant Runoff does nothing that its proponents say it does. IRV can be trusted to maintain a two-party system. When lawmakers on the hill realize this, what's to stop them from building whole-hearted bipartisan suport for IRV? Some in Congress already say they like it! If nothing is said against IRV, one day we'll all be worse off than we were with Plurality Voting! IRV lovers don't know where they're pointing their pistol!

A Vote from the Heart

Still, one need not lose heart. There are many methods of voting that are used, or have been used, or have been proposed, and even ones that have yet to be proposed. Among the many, there are some that have the potential to truly improve democracy, if only they are allowed to replace the likes of Plurality and Instant Runoff which fail so miserably.

Of those voting methods that are suitable for multi-party democracy, none are widely touted, nor widely known. The most widely heard of good one (good in my pompous opinion) is the one that you've been using on this site for voting on platform planks - namely, Approval Voting.

19 Comments

Printer Friendly Version
Top of Page