With a better voting system (e.g. Approval or Condorcet) and nothing preventing a dozen candidates from running, that would be much more feasible.
These go hand in hand: our current voting system breaks down if multiple candidates draw votes from each other, but a better voting system would be immune to that failure mode and could give people a free choice of several reasonable candidates without having some of them withdraw and throw their support to others.
why is it that there are not the best 5, 50 or 500 people on the ballot and I as a voter can mark any and all that I think I know will do the job the way I want them to
Only (relative) nerds would be able to figure this out in practice. Using a voting system that “the common man” doesn’t fully understand is unfortunately kind of elitist/classist in its effects even though i understand that’s not the intention of the many who wish for it.
> Only (relative) nerds would be able to figure this out in practice.
"Check the box next to anyone you approve of" is not hard to figure out.
"Write down the candidates you find acceptable, in order of preference" is not substantially harder, either, and provides additional benefit (the ability to express a preference among the candidates you approve of).
This is just the thing you “voting system optimizers” miss!
Yes, “check each one you approve of” is easy if you the voter aren’t meant to understand how it works. “If you just shut up and do what you’re told by the political scientists, it’ll all work out.”
It’s the ‘how it works’ part that loses people. Strategies employed by those who understand it fully can even advantage more savvy voters over the ones just strictly following instructions. For instance, manipulating ranked choice by strategically not choosing any second or third choices. People know that they don’t know fully how systems besides FPTP work, and it makes them feel like these are a way to cheat them.
The instructions are simple, the mechanism and the possible ‘optimization strategies’ are more complex.
I’m not saying people are stupid, just not experts in this flavor of nerdery, and suspicious.
> manipulating ranked choice by strategically not choosing any second or third choices
This is not a useful strategy for good preference-ordering systems. "strategically" not making a second or third choice is on balance much more likely to cause you to fail to get your preferences, because the scenario you thought you were hedging against didn't happen.
> Yes, “check each one you approve of” is easy if you the voter aren’t meant to understand how it works.
It's easy to understand how approval voting works, as well. That's the primary advantage of approval over Condorcet: ease of understanding. "Vote for anyone you approve of and would be satisfied to see in office. This way, you can vote for candidates you like more, as well as candidates you think have a better chance of winning, and you don't have to choose whether to vote strategically to keep someone you don't like out of office or to vote for your favorite candidate. We tally all the votes, and the person with the most votes wins, which means they're the person approved by the most people."
While Condorcet is a bit more challenging to understand (the easiest explanation being based on "who would beat everyone else in a head-to-head race", leaving aside the cycle-breaking mechanisms), it's much more resilient against the need for strategic voting. In Condorcet, there isn't an incentive to vote strategically; just write down your preferences.
Approval does have scenarios in which strategic voting may be desirable, notably when a less mainstream candidate is close to overtaking the mainstream candidates and enough people are willing to drop their approval of the closest mainstream candidate. That's a dangerous strategy to try, because it also gives an opportunity for the most popular opposing candidate to try to win. The existence of that strategy is one of the biggest arguments for Condorcet over Approval.
So, the massive tradeoff here is:
Do you want a system that's trivial to explain to people with no additional complexity under the surface, trivial to adapt existing voting procedures and ballot designs for, and is better than FPTP along most axes, but still retains one of the notable flaws of FPTP, namely having particular scenarios that still encourage strategic voting? That's Approval.
Or do you want a system that's better than FPTP along pretty much every axis, encourages and benefits from honest voting with no "strategy" required or desired, but is a bit harder to explain to people and has some additional complexity under the surface, and is harder to adapt existing voting procedures and ballot designs for? That's Condorcet.
Also, frankly, if you're going to talk about systems that present voters with something easy to understand on the surface but have complexity under the surface that most people don't understand and that does cheat people, we already have that system in place: FPTP with the Electoral College, gerrymandering, myriad forms of voter suppression... Compared to that, both Approval and Condorcet are simple.
Yeah I do find the commenter you responded to is, shall we say, overly concerned about the mental capacities of the average voter. "Mark each one you like" is not that hard and not substantially harder than "mark the one you like most".
I am concerned tho that preference voting (is that the right word?) gives people too many opportunities to screw up—there are too many things that can go wrong on the ballot (ones I've seen have three columns for 1st, 2nd, 3rd preference, and there should be at most? exactly? one mark in each column). Also vote counting has too many ifs and whens and seemingly arbitrary rules how my ballot will be counted in case my 1st choice doesn't clearly make it on first count.
Writing down candidates names will make for some fun days when it comes to counting. Plus, what if multiple candidates have similar names? And, a ballot with a scribbled "Miller!!1!" on it, do you mean Ms Annie Miller from Portland Oregon? That one? Lastly, when was the last time another person was able to read your scratches?
>I am concerned tho that preference voting (is that the right word?) gives people too many opportunities to screw up—there are too many things that can go wrong on the ballot
Here in Australia you number the box next to the candidates name from 1 to N this is not normally an issue for the local seat in the lower chamber (house of representatives) where there are a manageable number of candidates, but in the upper chamber where senators are elected on a state wide basis you end up with sometimes farcical situations like the infamous "tablecloth ballot paper" which had 264 candidates.
There has been some attempt to reform this such as forcing people/party's to pay a registration fee in order to appear on the ballot (this is done to cut down on so called nuisance candidates but it is arguably undemocratic, although I believe you are eligible for a refund of the fee if you poll above some threshold % of votes which is a fair compromise in my opinion).
The deposit system in the UK seems to work fine for us.
It's not enough (£500 currently) to be a meaningful expense to any candidate actually trying to win, but it's enough friction for 'fun' campaigns that we get enough to be funny but not enough to really interfere with anything.
And, yeah, ours is called a deposit because if you get 5% of the vote you get it back - and the phrase "lost their deposit" to reference a candidate who thought they were real utterly cratering on election day is pretty widely known.
Of course the UK's system of government is largely a patchwork of oddities that, put together, mostly work for us, so how well any particular oddity will work in another system is invariably debatable.
Fond though I am of ranked choice conceptually, there are a lot of people I would not enjoy (nor be particularly successful at) talking through voting under such a system.
Given especially US multi-race ballots, I find it a lot harder to see the incremental complexity of approval voting having a noticeable effect (though note for calibration that there's plenty of criticism out there for how confusing many multi-race ballots are as is).
Your assessment may be swayed when you discover that the current de-facto bipartisan state of US politics is a logical and near-inevitable outcome of a first-past-the-post voting modality.
Yeah, ballot design for preference voting is an interesting UX problem. Any design needs to be obvious for voters, reasonable to count unambiguously, and support mail-in paper ballots (tangent: which ideally 100% of people would have the option to use).
You could list the candidates in random order with a number next to them, and then have people write down those numbers in order. That, like anything else, would have failure modes, though.
An NxN matrix for N candidates allows using "marked or not" detection, but it creates a huge ballot even with just a handful of candidates.
Voting is already a pricey endeavor. Simply do the UX studies and find what works. Than publish the studies, and after a couple of cycles run a meta-analysis, and we should have close to optimal strategy.
Dowdall's method assigns 1, 1⁄2, 1⁄3... points to the 1st, 2nd, 3rd... candidates on each ballot, then elects the candidate with the most points. Ranked voting systems vary dramatically in how preferences are tabulated and counted, which gives each one very different properties.
Ranked voting systems are usually contrasted with rated voting methods, which allow voters to indicate how strongly they support different candidates (e.g. on a scale from 0-10).[1] Rated voting systems use more information than ordinal ballots; as a result, they are not subject to many of the problems with ranked voting (including results like Arrow's theorem).
That's a bucket full of problems. I remember that I had a similar problems years ago when I wanted to consolidate several lists about the "daily usefulness" of Chinese characters. How much does it weigh when you put A, B, C in that order? is A=1/1, B=1/2, C=1/3? or maybe A=0, B=-1, C=-2? To quote the above, "Ranked voting systems vary dramatically in how preferences are tabulated and counted". This should be a red flag! I'm happy to see that alternatives to first-past-post are tried out in some locations such as Alaska, Maine and Australia, but why ranked voting?
Ranked voting is only good when you think of yourself as a sole, lonely voter. You put the candidates into this order: A, B, C. But there will be other voters with other preferences. Let's say you're three people and the rankings are ABCD, EBCD and FBCD. Now you have three favorites A, E, F that are preferred choices, and there's that also-ran B who only was second best. But B got three second-best votes whereas each of A, E, F only got one vote.
So answer my question: how many second, third, fourth votes will it take to surpass a given number of first votes? If there's another ballot in the same election with GCBD on it, now you have B with 3x2nd and 1x3rd place. You cannot add, subtract, multiply or divide these numbers. There's no clear answer (although practical procedures can and have been proposed and used) and that bugs me. The entire thing looks like a game of Bridge, Skat or Poker to me, full of justifiable but ultimately very arbitrary rules, hard to explain and tricky to get right. https://en.wikipedia.org/wiki/Condorcet_method says that much to me. One also has to account for unranked candidates and decide whether it should be possible to give one rank to several candidates, and then whether (as done in sports) the next rank must be left empty.
It just goes on and on, the list of options to do the ballot and to do the counting is truly mind boggling. This is the reason I've come to much prefer preference voting over ranked voting. Ranked voting is much too complicated for general use in a general election whereas preference voting is simple and clear and the number of "design choices" as it were is drastically reduced when compared to ranked voting.
"preference voting" and "ranked choice" are largely used as synonyms, as both involve an ordered list of preferences. When I mentioned writing out a list of numbers in order, I was talking about ballot UX design here.
The ordinary case of preference voting is extremely explainable: if more people prefer A to B than prefer B to A, A beats B; pick the candidate that beats everyone else. "It's like running a bunch of head-to-head elections all at once."
They're also really easy to tally into a table: "A ranked above B" is a point for A over B; "B ranked above A" is a point for B over A; A and B ranked equal relative to each other doesn't add anything to either of those two cells.
The corner cases like cycles almost never come up. And even then, it's relatively easy to explain a high-level understanding of the most common methods: "if there's a loop of people who each beat the next one in the loop, ignore the one with the smallest margin of victory". If A beats B in a landslide, B beats C in a landslide, and C very narrowly beats A, A wins.
Also:
> Let's say you're three people and the rankings are ABCD, EBCD and FBCD.
Every ballot implicitly ranks every option, and the usual assumption is that if someone doesn't even care enough to list a candidate, they prefer every candidate they did list over any candidate they didn't list. (There are UX design ideas that could let someone say "this candidate is last" without ordering every candidate, but that adds complexity.) So these ballots are:
A > B > C > D > E = F
E > B > C > D > A = F
F > B > C > D > A = E
Writing it out that way shows that B beats A by 2:1, B beats E by 2:1, B beats F by 2:1, B beats C by 3:0, and B beats D by 3:0. No corner cases or cycles here; a majority of people prefer B to every other candidate.
Pleas do not google preferential voting, no country has ever implemented it. Definitely not Ireland.
The current system is the best thing in the world.
On serious note.
This will never be allowed in US as the whole 3rd candidate is wasted vote argument goes away. And suddenly there is a crack in the door for a other candidate to win.
Sure or some simplified version of ranked choice voting. Perhaps where only the top two choices get counted instead of having unlimited ranked choices (which can get messy and be harder to audit).
Once you've already had people rank their choices, why ignore meaningful information they've given you? In particular, models like IRV encourage people to continue ranking a compromise candidate on the top, to avoid the scenario where the compromise candidate is eliminated.
In models that ignore some of people's preferences, voting "1 Preferred, 2 Lesser Evil, 3 Greater Evil" is dangerous, because your preference for Lesser Evil over Greater Evil is ignored. This creates an incentive for people to keep voting the way they do now, with the compromise candidate on the top.
Approval makes sense because it's easy to explain. Condorcet makes sense because it's relatively close to ideal at the cost of making it slightly more complex to vote. Any model of the form "Ranked choice, but ignore some of the preferences" is the worst of both worlds: more complex to vote, but doesn't produce as good of an outcome.
I think the poster's point above was a form of ranked choice voting where you get exactly 2 choices, instead of 1 choice as in the current system. So you can vote "I like X best, I like Y second best" and that's it. This doesn't ignore anyone's choices, and is easy enough to understand. Details can be quibbled over (maybe 3 would be better than 2, for example), but it's quite clear in practice that any ranked choice voting system has to have a limit on how many choices you can express.
All this is moot though - the states can't agree on creating a popular vote for the president, even though the system to do it is already being used, and there is overwhelming support. They're not going to overhaul the voting system when they can't agree on the national vote even mattering.
It ignores preferences below the top 2, just like the current system ignores preferences below the top 1. It's probably an improvement, but the minor advantage in ballot design doesn't seem worth rolling out a half-measure.
> but it's quite clear in practice that any ranked choice voting system has to have a limit on how many choices you can express
In practice, 99% of ballots will only include the candidates listed on the ballot. Perhaps set a limit of N write-in candidates for some reasonable N, to avoid someone playing edge-case games like "you are legally obligated to count my ballot containing 200 million names!!!1!". But there's no obvious reason not to allow as many rankings as candidates on the ballot plus number of permitted write-ins.
> All this is moot though - the states can't agree on creating a popular vote for the president, even though the system to do it is already being used, and there is overwhelming support.
There's widespread opposition from the party that regularly loses the popular vote. I do hope it passes someday, though; it'll be a first step towards serious voting reform.
These go hand in hand: our current voting system breaks down if multiple candidates draw votes from each other, but a better voting system would be immune to that failure mode and could give people a free choice of several reasonable candidates without having some of them withdraw and throw their support to others.