Alternative voting method discussions should note the “bug” in RCV’s instant runoff voting

How can we build a bet­ter bal­lot and make vot­ing a more sat­is­fy­ing expe­ri­ence?

This is a ques­tion that pro­gres­sive activists and orga­ni­za­tions have been ask­ing for a long time, but with height­ened inter­est fol­low­ing the Supreme Court’s deci­sion in Bush v. Gore, the 2010 vic­to­ry of mil­i­tant Repub­li­can guber­na­to­r­i­al can­di­date Paul LeP­age in Maine, and Don­ald Trump’s 2016 Elec­toral Col­lege win.

Each of those elec­tions demon­strat­ed the weak­ness of the plu­ral­i­ty vot­ing sys­tem we have tra­di­tion­al­ly used here in the Unit­ed States, which is some­times called first past the post (FPTP for short, an allu­sion to rac­ing) or win­ner take all. This sys­tem requires vot­ers to mark a bal­lot for just one candidate.

The can­di­date who receives the most votes for each posi­tion being vot­ed on then becomes the win­ner — even if they haven’t received major­i­ty support.

In a first past the post elec­tion where there are more than two can­di­dates com­pet­ing for votes, it is pos­si­ble for two good can­di­dates to siphon votes from each oth­er, allow­ing a bad third can­di­date to swoop in and emerge victorious.

This is called the spoil­er effect, and, as men­tioned, this egre­gious defect has reared its ugly head in numer­ous high pro­file elec­tions in the past few years.

In part to address the spoil­er effect, many elec­tion reform advo­cates are call­ing for the adop­tion of an a dif­fer­ent vot­ing sys­tem known as ranked choice.

Tech­ni­cal­ly speak­ing, there are mul­ti­ple ways to imple­ment ranked choice vot­ing, so that term can refer to more than one alter­na­tive vot­ing sys­tem. How­ev­er, most dis­cus­sions of ranked choice vot­ing equate that term with instant runoff vot­ing, or IRV, so IRV is the imple­men­ta­tion that we will pro­ceed to dis­cuss in this post.

As implied by the name, all ranked choice sys­tems afford vot­ers an oppor­tu­ni­ty to rank the can­di­dates who appear on the bal­lot in a giv­en order.

With IRV, if no can­di­date receives a major­i­ty, then the top vote get­ting can­di­dates each instant­ly advance while can­di­dates receiv­ing the fewest votes are elim­i­nat­ed and their votes redis­trib­uted in order to deter­mine a winner.

Pro­po­nents of IRV, like the nation­al orga­ni­za­tion Fair­Vote, claim that this sys­tem “helps elect a can­di­date that bet­ter reflects the sup­port of a major­i­ty of vot­ers” and “helps to more fair­ly rep­re­sent the full spec­trum of voters.”

But what instant runoff mar­ket­ing mate­ri­als won’t tell you is that IRV hap­pens to have a major glitch of its own: it’s actu­al­ly pos­si­ble, a frac­tion of the time, for a can­di­date to lose an elec­tion by becom­ing more popular.

The eas­i­est way to explain this defect is to show it to you.

Below is a sim­u­la­tion cre­at­ed by Nicky Case which shows an imag­i­nary can­di­date going from being on the verge of win­ning to instead los­ing… by becom­ing more pop­u­lar with vot­ers. In the sim­u­la­tion, can­di­date Tra­cy Tri­an­gle is ini­tial­ly win­ning as vot­ers shift towards Tra­cy and away from fel­low can­di­date Steven Square.

But as the video shows, as vot­ers con­tin­ue to move clos­er to Tra­cy,  there comes a point where Tra­cy ends up in a runoff with Hen­ry Hexa­gon… and los­es.

“How often does this actu­al­ly hap­pen in real life?” Case asks rhetor­i­cal­ly, pro­vid­ing the fol­low­ing answer: “There’s a cou­ple con­firmed exam­ples, and math­e­mati­cians esti­mate this glitch would hap­pen about 14.5% of the time. But sad­ly, we can’t know for sure, because gov­ern­ments usu­al­ly don’t release enough info about the bal­lots to recon­struct an IRV elec­tion & dou­ble-check the results.”

“So, not only is Instant Runof­f’s glitch as unde­mo­c­ra­t­ic as First Past The Post’s glitch, it’s pos­si­bly worse – because while FPT­P’s count­ing method is sim­ple and trans­par­ent, Instant Runoff is any­thing but. And a lack of trans­paren­cy is an even dead­lier sin nowa­days, when our trust in gov­ern­ment is already so low.”

In Wash­ing­ton, the Leg­is­la­ture is con­sid­er­ing a pair of bills that would explic­it­ly autho­rize local gov­ern­ments to adopt instant runoff voting.

One of these is Mia Gregerson­’s HB 1722. There is a Sen­ate com­pan­ion, SB 5708, spon­sored by Guy Palum­bo. HB 1722 was heard today in the House State Gov­ern­ment Com­mit­tee and is sched­uled for exec­u­tive ses­sion on Feb­ru­ary 22nd.

It may not reach Gov­er­nor Inslee’s desk, but it is cer­tain­ly spark­ing interest.

There is a risk of vot­er con­fu­sion with these bills. If this leg­is­la­tion were to be imple­ment­ed, then we’d like­ly soon see a very com­pli­cat­ed bal­lot with plu­ral­i­ty vot­ing being used in some races and instant runoff being used in oth­er races.

That’s because these bills do not man­date a tran­si­tion to instant runoff vot­ing; they only make it an option for cities, coun­ties, and oth­er local governments.

NPI whole­heart­ed­ly agrees that we should work on aban­don­ing win­ner take all as our vot­ing sys­tem. But if we’re going to say good­bye to first past the post, then the alter­na­tive we adopt should be bet­ter than what we have now.

If we adopt ranked choice vot­ing with instant runoff, we should do clear-eyed, rec­og­niz­ing what the down­sides are. We’d have to live with the glitch described above, but we can cer­tain­ly insist on requir­ing ample resources for vot­er edu­ca­tion to deal with the poten­tial for con­fu­sion in juris­dic­tions that adopt RCV.

Nobel win­ning econ­o­mist and math­e­mati­cian Ken­neth Arrow has the­o­rized that all vot­ing sys­tems where can­di­dates are ranked will ulti­mate­ly be unfair in some way.

It’s also impor­tant to know that instant runoff and ranked choice vot­ing are not the only alter­na­tive vot­ing sys­tems out there. There are alter­na­tives avail­able to us that don’t involve rank­ing can­di­dates at all. Like approval vot­ing.

With approval vot­ing, you check the box (or fill in the oval) for every can­di­date that you approve of. So, if you like, say, three of the Demo­c­ra­t­ic can­di­dates cur­rent­ly run­ning for Pres­i­dent of the Unit­ed States, then you could vote for those three. Here’s a hypo­thet­i­cal bal­lot which con­sists sim­ply of the declared can­di­dates for Pres­i­dent so far for 2020 on the Demo­c­ra­t­ic side:

• Cory Book­er
• Eliz­a­beth Warren
• Kamala Har­ris
• Kris­ten Gillibrand
• Julian Cas­tro
• Tul­si Gabbard
• Bernie Sanders
• Amy Klobuchar
• Pete Buttigieg
• John Delaney

Wow. Ten con­tenders for the Demo­c­ra­t­ic nom­i­na­tion already! And the field is expect­ed to get even big­ger. Would­n’t it be nice if Demo­c­ra­t­ic vot­ers had the option to express their sup­port for more than one can­di­date? There can only be one nom­i­nee, but plen­ty of Demo­c­ra­t­ic vot­ers are like­ly to be fond of more than one of the can­di­dates, espe­cial­ly with so many good options to choose from.

“Wait, pick­ing more than one can­di­date? Does­n’t that vio­late the one-vote-per-per­son rule?” Nicky Case asks in ref­er­ence to approval vot­ing, writ­ing in response: “Well, your vote was nev­er a sin­gle check mark, your vote was always the whole bal­lot. And on this bal­lot, you get to hon­est­ly express all the can­di­dates you approve of, not just your favorite or strate­gic second-favorite.”

If our goal is to improve the vot­er expe­ri­ence to bol­ster par­tic­i­pa­tion and safe­guard the future of our democ­ra­cy, then we should design a bal­lot that encour­ages peo­ple to vote hon­est­ly. And it’s essen­tial for peo­ple inter­est­ed in vot­ing jus­tice to know that we aren’t lim­it­ed to exper­i­ment­ing with IRV and RCV (which Pierce Coun­ty already tried and reject­ed and which is being used by sev­er­al locales around the coun­try). We also have approval vot­ing and star vot­ing open to us… and if we’re going to pass a local options bill for alter­na­tive vot­ing meth­ods, we could allow those to be used in addi­tion to RCV.