Instant Runoff Voting “won” a big battle this past week in court, and it looks as if Minneapolis is going to take a whack at the voting system.
In the wake of the recount, I think it’s a terrible idea; it is to the best of my knowledge impossible to conduct an actual recount of IRV elections, and if you don’t think you’ll eventually need to, you are naive enough to be a Minneapolis liberal.
But I’m not going to talk about that (yet). I’m going to go after the voter-facing side of IRV (adapting a piece I wrote a little over a year ago in a Saint Paul online forum).
Background: I analyze systems – software, hardware, processes, print publications, what have you – to empirically determine how usable they are.
And speaking not as a partisan, but as a professional whose entire line of work involves figuring how to make things easier for real people to use, there’s a truism at work whenever people design systems; the designer always thinks he/she has designed something so intuitive that someone’d have to be an idiot not to be able to figure out how to make it work as intended and designed. It’s true for programmers writing websites, for executives designing processes for other people, for engineers building freeway ramps, for architects designing public spaces; everyone designs things to be blazingly intuitive – to people like them, other programmers, executives, engineers or architects.
And when those programmers, managers and engineers watch real people in controlled usability tests actually trying to do real-world things with those websites, processes, ramps and spaces, and making mistakes and doing things they were not intended to to, they tend to have one of the following reactions:
- “Nobody’s that stupid!” But it’s not usually a matter of stupidity. It’s human nature – especially if that human is not a programmer, executive, engineer or architect.
- “It’ll never happen in real life!” But it just did!
- “Wow. Who knew? We gotta redesign this!” These are the good programmers, executives, engineers and architects.
Although those who stump for IRV – the idea’s “programmers, executives” and so on – express it via rose-colored glasses (that, too, is human nature); they don’t say people who might hypothetically make mistkes voting in an instant-runoff election are “idiots”.
But I can see several places where confusion is potentially built into the system.
Allow me to walk through a fairly simple conundrum that faces usability people and, by the way, real people using real systems, drawn not from political ideology (of ANY sort!), but from the experience of someone who has had to ask these questions of programmers, executives and engineers for a living for the past decade.
Proponents explain the core of IRV pretty simply:
“you simply rank the candidates in the order in which you prefer them”
So when “simply” ranking, say, five candidates from top to bottom, do you number them 1-5, or 5-1?
Remember – in many Asian cultures, 1 is “better” than 5, while many people think bigger numbers are “better” than smaller numbers (like a hockey score).
And if you answer “that’ll be explained in the instructions”, please bear in mind that people – REAL people – tend not to read instructional writing, and retain even less for any amount of time. That’s not the cynicism of a former tech writer talking (although it’s there!); the research on much explanatory writing, on forms or website, that people read and retain is comically small.
So – how do you make sure everyone gets the directions the same way? Verbal instructions from poll staff? Mightn’t those be potentially legally-problematic?
Will people be able to cast “Tie” votes if they have no preference? Rank everyone “1” (or “5”), or rank five candidates “1,2,2,2,3” or “1,1,3,3,5?”, or “5,5,5,5,5”? (If you don’t think people will try, think again!) What’ll happen to the ballots if people try to do that? More importantly, how will people KNOW the consequences of trying that, whatever they are, and whether it’s OK or (emphatically) not?
All of you who chant “count every vote”: how many potential disqualifiers do you see in the above paragraph?
Let’s move past “process”, to mechanism. On what medium do people cast their vote in an IRV system – a paper ballot? Marked with what? Pencil? If they change their mind before submitting the ballot, how are changes made? Erasing numbers? How does one know, for audit purposes, WHO erased the number, then? What if they do a poor job of erasing (with older people with arthritic hands, this is not uncommon); how are ambiguities caused by poor erasing and faint handwriting resolved?
How about people who don’t erase, but scribble or overwrite?
And don’t bother replying “tell them to get a new ballot”. That’s a not-insignificant part of the current voting instructions – and we all know how many “spoiled” ballots turned up in the Coleman/Smalley race, don’t we?
And let’s not forget that immigrants frequently write numbers differently than Americans do; I run into this myself, since I usually use German numbering, and sometimes people read my “1”s as “7”s, and my “7s” as “4”s (I cross my 7s, European-style); how are these ambiguities to be resolved? And if the answer is “by telling immigrants to make sure they use American numbers”, do you realize the problems you’ll run into?
Indeed, how are the votes of the handicapped to be tallied? How would someone with, say, arthritic hands vote? (I won’t even ask the obvious question about voting for the blind; I’ll have to assume SOMEONE’s on top of that one).
And none of this even touches on the issue of “how the ballots are designed”. And that is a huge issue. Remember – whomever designed the infamous Broward County Butterfly Ballot thought they had a perfectly workable, usable design!
——
Bear in mind that NONE of the issues I raised above is, in my decade’s experience as a usability geek, outlandish, or even especially far-fetched; certainly none of them are remotely political. These are the sorts of issues someone in my field expects to see when any new system intersects with new users. Smart system owners run usability tests before their system “goes live”, and fix the issues they encounter. Dumb ones…well, thank goodness for them, since usability disasters keep me employed.
I’d be very interested in seeing a real, live, end-to-end, empirical test of an IRV system and all of its components – the ranking system, the ballot and media, the counting process, the system of explaining the process to new voters in various languages – and seeing how it really works in a reasonably-complex, contested polling.
I say “contested” for a reason, by the way; IRV seems to have only been tried in locales with relatively monobloc politics, from what I’ve seen. Without trying to judge the politics themselves, professionally speaking, that’s not necessarily a thorough workout.
Answer those questions, IRV proponents (preferably never using the phrase “nobody’s that dumb” in the process; it’s not “dumb”, but it’s human nature).
Then we can move on to the other questions:
- How do you do recounts?
- Why do all of you lefties who spent from 2000-2006 whinging about how Diebold and its electronic voting machines were in the bag for the GOP square that with the fact that IRV tallies are entirely, 100%, irrevocably computer-based?
Sound off!