Building a Better Table (/Voting System)

In my last post, I explained why trying to make a third party work in the current American electoral system is like trying to make a functional table that stands on one spindly leg -- it's not improbable, it's effectively futile.

In this post, I provide a catalogue of the current contenders for best Vastly Superior Table Design.

*****

Under Ranked-Choice Voting (RCV), aka Instant-Runoff Voting (IRV), aka Alternative Voting (AV), instead of picking a candidate, you pick a little list of your favorites. You keep going until you don't know who these people are or you don't care. If your first-choice candidate gets the fewest first-place votes, they get magically lopped off your list, and everybody else moves up a spot. Now your #2 choice is your #1 choice. Repeat. But, thanks to technology™, we can do this all at once now -- we take everybody's ballots, and we tally up the first-choice votes, we eliminate a candidate, and we check to see who the new first-choice "loser" is. Eventually there are two people left (or somebody gets >50% of the vote), and the winner of that standoff wins. This person tends to be the candidate most people are broadly okay with.

Why does this work better? You don't have to lie. You can vote for whatever nutjob or savior you want, and when they lose, your vote isn't wasted -- it still counts. That's why RCV, unlike FPTP, is a "transferable-vote" system.

Is it perfect? No.

• It'd be tough to implement on the current generation of voting machines, so we'd need new ones (though we'll need some eventually anyway, so what the heck).
• You can't really do regional tallies, because you have to have everybody's votes in before you start the runoff. Otherwise, relatively small differences in votes can cause snowball effects.
• Ties are pretty common early on while there are still a lot of candidates -- you have to eliminate somebody every round, and when all the randos are "tied" at close to zero, it can be kind of a headache to sort through (especially since we already know that none of them is likely to win). Although, using a % cutoff in the first round can quickly sort the wheat from the chaff. See e.g. this awesome visualization of the recent Minneapolis mayoral contest!
• It's hard to see what's happening from the outside -- there are ballots, then Math occurs, and then it spits out a winner. Though as the link above demonstrates, it's pretty digestible after the fact.

BUT. Can it sustain third parties? Can it offer people real choice?

YES! Well, in the long run there will still tend to be two dominant parties, but which ones they are may change, the other parties can genuinely compete for top billing, and nobody has to hold their nose at the ballot box EVER AGAIN.

Oh, also: if the GOP Primary were RCV, Trump would not be the nominee. But hey, NBD...

Is this some hippy-dippy political science pipe dream?

NO. There is an organization called FairVote.org that is making huge strides. A number of American cities already use it, there are were several (successful!) ballot initiatives LAST NOVEMBER to use it at the state level to elect legislators and governors, and Australia has already been using it for a while now. We only haven't heard about it because we have our heads up our asses trying really hard to make a super tall table with crappy parts.

*****

Under Range Voting, aka Score Voting, you say how much you like each candidate. It's Yelp, it's Amazon, it's Olympic gymnastics. That's literally it, that's the whole thing. We could use zero to five "stars," or a scale of one to ten, it doesn't really matter. This means it's not "first-past-the-post" either -- you don't win once you reach a large enough number of votes, you win by pleasing enough people.

Can this sustain third parties? Obviously. If everybody in the country agrees the "top two" candidates are not actually that great, they lose. It's the ultimate level playing field. Because you've used Yelp and Amazon, you get how this works. Sometime the big name brands make a crappy product, some scrappy mid-tier contender puts together a winner, and voila.

But ALSO because you've used Yelp & Amazon, you also know the slight pitfall here. People with agendas can really influence the vote -- all those crappy one-star reviews by the idiots who didn't pay attention to what they're ordering, or worse, sponsor competing products? Their votes still get averaged in. RCV avoids this problem because you can't spike the ball on people you hate (or even mildly dislike), you can only rank them lower.

Does that mean the system is broken? No. It's still better than just asking "meh, what's popular?" It's just that on its worst day, it ends up performing similarly to what we have now.

Moreover, unlike RCV, it'd be a snap to implement on current voting machines; you CAN do regional tallies; ties don't matter unless they're near the top; and it's really obvious what's happening. You can easily sort through ballots by hand if you want, and just take the averages.

So, pluses and minuses relative to RCV. But still better than our current system pretty much any day of the week. Is *this* a political science pipe dream? Well, kind of. There are very, very smart people who would take this over RCV -- see rangevoting.org -- but not really a committed ground game at the moment. Still, spread the word, and we'll see what people think.


One footnote: how do we deal with people you don't have an opinion on? Leave it blank. Only take the average of actual scores for a candidate. Wait, wouldn't that mean some random UFO enthusiast with 1000 fans could just come in at the end and win?!? Yeah, you'd probably have to restrict the final vote to like the top five or ten candidates. But right up until election day, leave 'em in -- as soon as the UFO guy randomly wins a poll, everybody will be like "who's that? OH." And, if the people decide he's crazy, he will never harass the polls again.

*****

Under Approval Voting, you say "I do/do not like this person" for each candidate. It's basically just Range voting, but with a scale of zero or one, instead of one to ten. It's the Rotten Tomatoes to Range Voting's IMDB. (Alternatively, you can think of it as: you can vote for as many people as you want, and the person with the most votes by all voters wins.)

That means you can't express as much information about your preferences, but it's also less sensitive to extreme opinions. If you're like me, and you hate to pretend things are black and white, this makes you very mad, but if you worry a lot about the system getting hijacked -- if, for example, it turned out one political faction tended to vote ten for their candidate and one for everybody else, while another used the full scale -- then Approval is a little less risky.

Even easier to implement than Range; still levels the playing field tremendously; means races will be closer, and voters won't get to say as much about their preferences. But a solid choice, and again, still vastly superior to the current system.

*****

There you have it, folks. If you're truly a Gary Johnson or Jill Stein fan, or think further outside the box; or if you hate Hillary and Donald, whether or not you can pick one; or if you just think we can maybe stop pretending the whole US is two perfectly, fundamentally opposed groups of people for one goddamn minute so we can think straight and accomplish things, then get on board the screw-this-table train and get involved.


Probably the shortest path to this is signing up to volunteer with FairVote, but Range is great too; in any case share, discuss, do what your heart tells you. That's the whole point!

*****

PS - if you want the ULTIMATE POWER MEGAZORD METHOD that has the fewest concessions to bad math, and don't mind if it's a little complicated, say hello to your friend Condorcet Voting! This is actually a criterion rather than a method per se, but there are two most common versions, I think: one that's like RCV but asks you to rate both your top and bottom candidates, and one that requires the voter to rate each head-to-head combination of candidates running. You can see why these are... less likely to catch on, despite being pretty bomb-proof mathematically. (Probably don't muddy the waters by lobbying for this right now, thanks.)

PPS - ...Did I mention with RCV primaries, Trump wouldn't be the nominee, to the relief of most Republicans and everyone else?

(Also, did I mention there was a candidate trying to get into the Dem debates last fall who had FairVote's system on his platform [cf. footnote]? Did I call this jazz coming? Not that I'm bitter or anything... Oh Lessig. <3)