How our voting system (and IRV) betrays your favourite candidate
ฝัง
- เผยแพร่เมื่อ 28 ก.ย. 2024
- Most people know that you can't always vote for your favorite candidate in plurality voting. Your favorite might be a spoiler and your vote might cause your least favorite candidate to win. No matter how much you like him, you may have to betray your favorite candidate and vote for a candidate that actually has a chance of winning.
Does Instant Runoff Voting (aka Ranked Choice Voting) fix this problem? No!
This video clearly shows why sometimes, even in instant runoff voting, supporting your favorite candidate can cause your least favorite to win. You may need to betray your favorite and put for your first choice the candidate that has a better chance of beating your least favorite.
Nader image by Don LaVange (cropped) (CC-BY 2.0): www.flickr.com/...
Isn’t it fine if the subjectively “bad” candidate won if they earned the majority vote after elimination?
The "Good" candidates was preferred to the "Bad" candidate by a majority of voters. Either way the issue is basically that this is the spoiler effect and encourages artificial duopoly.
@@MustSeto true. RCV does still have the problem of polarizing candidates winning. This is why I support voting methods that encourage consensus candidates to win like in Approval
"Good" wasthe true preffered party.
You can still get a worse result by expressing support for your second favorite candidate.
Yes but that's rare and esoteric. The broader point is that you can get a worse result by ranking your favorite in first place.
True, but it is likely to harm you as you cannot show support for some candidates over others. Also, if you word it that way (“YOU get hurt” instead of “first choice gets hurt”) IRV fails that.
Approval voting is, I am convinced, a major improvement over Plurality and superior to IRV, but not perfect; it is "indeterminate". To predict the outcome in hypothetical elections, you have to assume the voters use some particular "rule of thumb" to decide how many candidates to approve. Different rules of thumb lead to different weaknesses. The simplest example is when voters all "vote for only your favorite", in which case it acts exactly like Plurality. I've looks at other "rules of thumb" and found examples that manifest the Spoiler Effect under Approval.
Voters who favor "third" parties and independents would have reason not to bullet vote. Bullet voting would work against their interests. An interesting question, however, is under what circumstances, if any, voters tend to acquire a correct understanding of what strategy works best in a given voting system for promoting their goals. Approval (and more generally Score) compared to ranking systems has a lower Kolmogorov complexity, so maybe it is easier to learn and remember the strategies that work for them.
Note that Approval is equivalent to finer-grained Score if voters know to vote probabilistically and if the election is big enough to smooth out the effect of such behavior. So for instance if my strategy would tell me in a Score 101 election to give say Romney a 100 and Ron Paul a 99, but I am faced with a Score 2 election i. e. Approval, I should obtain a random number with a 99% probability of being 1 and otherwise being 0, and if it is 1, vote for Paul.
William Waugh I made this comment a year ago and I'm not as weary anymore about range or approval voting. As of now, I am completely against ranking candidates (for public elections) and feel even more strongly that non-partisan primaries with approval voting is the way to go. One really good benefit of such a system is that I think it removes some of the tribal behavior among parties in which they are treated like sports teams.
The issue I had with straight approval and range voting is that it could be a problem if there is a statistical tie between your favorite, your compromise, and your hated candidate. If you support your compromise candidate, it could hurt your favorite's chances. I am not as worried about this anymore.
For Presidential elections, I think that Mark and Clays rated-IRV system should be used. The winner of the state should get to decide whom the states electoral votes gets to go to.
Mutex50, Yes, I suspect that in the long run, the tribal behavior would diminish with Approval (or more generally any range). That's because when candidates agree on some of the issues, they won't have to think of one another as competitors. They can cooperate to promote those issues. A vote for candidate A is not necessarily a vote against candidate B.
I also agree with you that Approval is preferable to any ranking system.
When you refer to "rated-IRV", that's the same thing that is discussed at groups.google.com/d/msg/electionscience/JK82EFn7nrs/Lble3V2CW4UJ as "Instant Score-off", right? Note that the discussion doesn't wind up with participants expressing much support for the scheme.
John Hodges I have been advocating under a misconception that Approval and more generally Range wholly eliminate the spoiler effect. I was mistaken. Discussion at groups.google.com/forum/?fromgroups=#!topic/electionscience/uIzIoBSK-wc
Using approval with a nonpartisan top-2 primary for every seat would fulfill both requirements. An approval vote top 2 primary would put two broadly appealing candidates into the general election and a plurality vote during the general would allow a veto of tactical voters from first round.
Thanks for making this video.
STAR Voting also passes the Favorite Betrayal Criterion and eliminates the spoiler effect.
Actually, it doesn't. It has significantly less violations of FBC then IRV, but it can still fail the criterion.
There are very few voting methods that pass FBC. This table lists most of them: en.wikipedia.org/wiki/Comparison_of_electoral_systems#Compliance_of_selected_single-winner_methods (FBC is listed as no favorite betrayal on that table).
lol, good luck getting people to understand it. complicated voting methods lead to fear of corruption and voter mistrust
STAR is an attempt to balance FBC and LNH.
no it doesn't. it fails FBC and IIA. score voting (and by extension, approval voting) on the other hand satisfy both.
@@alanivar2752 star voting is radically simpler than the instant runoff voting method we used when i lived in san francisco and berkeley. there's tons of data on this.
This does not explain a problem with IRV, it explains a problematic SCENARIO for IRV. This scenario (1) is extremely unlikely, and (2) would be a problematic scenario for STAR voting as well. In fact, I can't think of any voting system that would do well in the scenario depicted in this video. It's such a specifically bad-case scenario that it's just a perfect storm of election problems.
No, this is not a problem for STAR voting.
medium.com/@ClayShentrup/strategy-with-star-voting-and-irv-1ec22028399d
This is absolutely a problem for IRV. It doesn't matter how rare it is, it just an issue of _relative_ probability. For instance, in the current system, Greens who tactically vote Democrat don't *know* that there's going to be a spoiler effect (the Green is going to get more votes than the Dem minus the Republican). They just know that it's _more likely_ than the outcome where the Green wins, thus they might as well vote Democrat _just in case_. Just like my aunt in Iowa voted for Biden even though she really preferred Warren, because she thought Biden would have better odds against Trump. This is _exactly_ what we're dealing with under the IRV method. Again, STAR voting and approval voting don't have this problem.
'is extremely unlikely' except that time Aspen, CO adopted it, had it happen in the very first election, and immediately abandoned it. or that time Burlington, VT adopted it, had it happen in the very first election, and immediately abandoned it. it happens all the time in Australia and Ireland
This is literally the only claimed benefit, and it's spoiler effect is actually worse than FPTP. RCV is somehow even worse than FPTP
Approval Voting doesn't have this problem
This makes no sense to me. You start the video saying that the majority like the Good candidate, but then state that when the Ideal candidate actually wins, those that had Good first, chose Bad second. Surely that's simply Democracy in action? Not everyone has the same notion of Ideal, Good, and Bad, so what might seem a spoiler vote for me, is actually the optimal for someone else. The greatest good for the greatest number is surely the most fair, no? Maybe I'm just missing the point this video is trying to illustrate, but as of now, I don't see how runoff voting has created an overall spoiler effect here...
A majority of voters preferred Good to Bad; even those who put Bad 2nd still put Good 1st.
A voter switching from Good>Ideal>Bad to Ideal>Good>Bad caused Bad to win. Since Ideal could easily be politically farther from Bad, why should a voter moving farther from Bad cause Bad to win?
Those voters, by being honest about their favorite (Ideal), cause their least favorite to win (Bad), instead of their compromise (Good) who could have won if they were put first. The Ideal voters' 2nd choice was never considered. If Ideal didn't run, Good would have won.
These scenarios are assuming that Good and Ideal are political allies. The point is that if Ideal dropped out of the race, then Good would win, so Ideal is a spoiler for Good. But if Bad dropped out, then Ideal would win (in the second scenario), which means that in the Good-Ideal voting block, most people actually prefer Ideal. Thus, voters should ignore their first preference (Ideal) and strategically rank their second preference (Good) higher in order to prevent this spoiler effect.
That's because it hasn't produced a "spoiler" effect. That every bloc in question here ("Ideal>Good>Bad", "Good>Ideal>Bad", "Good>Bad>Ideal") has placed "Good" over "Bad" is immaterial. The question at hand was not "Do you prefer Good over Bad?" The question at hand was "What are your preferences given these three candidates?" Some voters preferred "Ideal>Good>Bad", and some preferred "Good>Ideal>Bad", but some preferred "Good>Bad>Ideal". IRV has accurately reflected the preferences of the electorate *as a whole* given this question, and revealed that a substantial portion of the "Good" party actually aren't members of the "Good>Ideal" bloc. They'd sooner have "Bad" than "Ideal". And if enough of them feel this way, then the "Bad" candidate *should* win, because the "Bad" candidate is the most preferred option out of all of the available options.
That's not a bug. It's a feature. It wasn't the folks voting "Ideal>Good>Bad" that caused this outcome. It was the folks voting "Good>Bad>Ideal". They indicated that "Bad" was their second preference, and since a substantial piece of the electorate also said "Bad" was their first preference, "Bad" won. Whining about this state of affairs is undemocratic, because the "Good" candidate only wins with this electorate when this electorate is denied "dangerous" choices. The "Good" candidate only wins when you're not allowed to ask this electorate how they feel about the "Ideal" candidate at all (because that is a "dangerous" question).
Approval voting doesn't solve this problem. It *embraces* it by obligating voters to send the message "I really dislike 'Bad'" rather than "I really like 'Ideal'" (because if voters attempt to strategically bullet vote under Approval voting by only selecting "Ideal" and leaving off "Bad" *and* "Good", then they run the risk of "Bad" winning, so they select "Good" too and end up ultimately hurting what was their first choice). Approval voting doesn't let the electorate fully express its preferences and thus the electorate cannot evaluate those preferences and adjust them over time. Approval voting *stymies* the electorate by forcing it to only ever meaningfully consider muddled, milquetoast candidates.
Ultimately, the schism between RCV/IRV/STV-like systems, and Approval-like systems, is a question of priorities. Is it better to find the most preferred candidate, or the least disliked candidate? And, depending on your priorities, you'll fall on one side of the schism or another. Perhaps somewhat obviously, I fall on the RCV/IRV/STV side of things. I've had more than enough of systems that force me to pick the "less evil" option rather than the "not evil" option.
@@aspensmonster A candidate is a "spoiler" when they "spoil" (i.e. change) the outcome of the election by running. In this example, the majority of people prefer the Good candidate over the Bad candidate. That's not immaterial, that is the whole point of whether or not this election was spoiled. By being a part of the election, the Ideal candidate caused the will of the majority of the population to be usurped. The majority of people would have been happier with the Good candidate than they would have been with the Bad candidate, thus they have objectively received a worse result.
Let me try to demonstrate why this is a spoiler effect. If the election had been between the Bad candidate and the Ideal candidate, then you are right that the majority of people preferred the Bad candidate. That is simply the will of the people and democracy in action. If we then had the Good candidate enter the race, there are three possible outcomes. Option 1, the Bad candidate still wins. Thus the good candidate had no appreciable effect, the majority still prefer the Bad candidate, democracy still works. Option 2, the Good candidate wins. This is also fine, it just means a better option was provided that the majority preferred. But it still means democracy is working. Option 3, the Ideal candidate wins. What the hell??? How did this happen? THIS is the spoiler effect. The majority of people did NOT want the Ideal candidate, that candidate's level of support has NOT changed, and there is no logical reason they should have won.
It is the equivalent of if someone asked you if you would prefer chocolate or vanilla ice cream. You give an answer, let's assume you ask for chocolate. Then they tell you "oh! we also have the option of strawberry." And you say "well, in that case I'll have vanilla." It makes absolutely no sense. If you prefer chocolate over vanilla, then adding another option shouldn't change that. You'd logically either stick with chocolate or switch to strawberry, but there is NO reason you should switch to Vanilla. Yet in some voting systems, that's what happens, and when it does it is known as the Spoiler Effect. IRV does not solve the spoiler effect. It just makes it less common and less obvious.
You're absolutely right that Approval isn't perfect either. No system is. And I actually do prefer IRV over Approval voting. My favorite type of voting though is STAR voting. It is kind of a combination of IRV and Score Voting (it literally stands for Score Then Automatic Runoff). I really like that it allows you to say how much you prefer one candidate over another, while removing the temptation to bullet vote.
@@periculum69 That a majority placed good over bad (at all) *is* immaterial. The *order* of preferences matters. "good>bad>ideal" is *not* equivalent to "good>ideal>bad". And it's not the voting system's job to police the electorate and tell them "your order of preferences is silly and makes no sense." Its only job is to *measure* the preferences of the electorate and identify the candidate with the *strongest* support. And in this "spoiler" scenario, the "good" candidate absolutely does *not* have the strongest support. They have the *least* support, and are eliminated, as expected. And since enough people said not only that they prefer good over bad at all, but that they *also* prefer bad over ideal, bad *should win*.
Ideal is only a "spoiler" here if you think that the most *tolerable* candidate should take precedence over the most *preferred* candidate. That's what cardinal systems like approval voting (and STAR) do. But ultimately, whether a candidate is a "spoiler" or not is going to boil down to what you think the election system should be optimizing for.
Candidate List is a method that combines the virtues of both Party list and STV.
Candidate List requires each individual candidate to publish a list of all the candidates on the ballot, in order of THAT candidate's preference. This is essentially a "ranked choice ballot" issued by the candidate, instead of by the party or the voter.
In a single-seat election, each voter votes for ONE candidate. In a multi-seat election, they could have one vote for each open seat, with all candidates eligible for all seats (so, for example, if a voter desired, they could vote for the same candidate for every open seat, or distribute their votes between several favored candidates, in the style of "cumulative voting"). The votes are tallied for each candidate. The candidates' ordered lists, with their vote totals, are then used as the input data for an STV election (for a multi-seat election) or some variety of Condorcet (for a single-seat election).
Selling this idea will probably require explaining STV and Condorcet to the public, but it may be that a lot of people will be satisfied just to know that STV has been in use in several countries for decades, and Condorcet has been vetted by scholars as being the best and fairest method for single-seat elections; "round-robin tournament instant runoff", not that hard to understand, the only complications being in tie-breaking.
So, Candidate List gives us Condorcet for single-seat elections, and STV-proportional representation for multi-seat elections, with the voters using just an ordinary ballot on which they pick their single favorite, or their favorite for each open seat. They don't have to deal with ranking 20 or 30 candidates. The administrators of the election don't have to deal with thousands or millions of ranked ballots. The STV or Condorcet calculations will be simple enough (with only a limited number of different kinds of ranked ballots as input) to be published, and inspected by the public, after the election, so that the public will be satisfied by the logic of the selection.
A commenter has pointed out that we could allow other parties, other than candidates, to put lists on the ballot; for example, the Sierra Club or the National Rifle Association could mount whatever was needed (petition drive, filing fee, whatever) to get on the ballot, and post their own list of all the candidates in their organization's order of preference, even if they were not running a candidate themselves.
John Hodges Why not use proxy assignment? Each voter would decide who represents them, and legislators would wield the the proxies assigned to them when they vote on legislation and motions.
Help! I downloaded the transcript, and am confused by this statement regarding IRV:
00:48: "...but what happens when your ideal candidate does even better...what if he actually beats your second choice, and your second choice gets eliminated? Unfortunately, that puts us completely at the mercy of these voters that we have no control over...those who voted for the good candidate...and who did they put for their second choice?...it would be nice if they had all put our ideal candidate second, but they didn't...it takes only one-fifth of them putting the bad candidate second for him to win..."
Can someone try to explain to me what the above means? I am trying to figure it out, using Trump, Clinton, and Sanders as the three candidates, and am having trouble understanding how the spoiler effect would come into play.
If Sanders beat Clinton, and then Trump won, wouldn't that just mean that the first-round Clinton voters preferred Trump over Sanders (chose Trump over Sanders as their 2nd choice)?
Likewise, If Clinton beat Sanders, and then Trump won, wouldn't that just mean that the first-round Sanders voters preferred Trump over Clinton (chose Trump over Clinton as their 2nd choice)?
"If Sanders beat Clinton, and then Trump won, wouldn't that just mean
that the first-round Clinton voters preferred Trump over Sanders?"
It means some of them did. But the majority of all voters may still have preferred Clinton over Trump. Once Clinton is eliminated, IRV longer cares that the Sanders voters preferred Clinton over Trump.
@@MustSeto Thanks...I get it now...if voting for your 1st choice ends up killing off your 2nd choice, you'd have been better off voting for your 2nd choice in the first round...which means that strategic voting is useful with IRV, which is a failure of IRV, since a perfect system would NEVER require strategic voting. Thanks!
If you had a negative vote, that is a vote that says under no circumstances elect this person along with IRV it would eliminate the problem
Approval Voting already has that baked into the math
Does the Center for Election Science have anything to say about Proportional Representation?
I said, "I believe they are in favor of it." On edit: I was wrong; see the link below.
Link doesn't work.
OK, scorevoting.net/PropRep.html
We're supportive of it. Looking at our blog, you'll find it mentioned quite frequently. www.electology.org/blog
lol, Guyana has a lot to say about Proportional Representation
but the mistake youre making is that with instant runoff, there would be more candidates in the race than just one, on one side of the isle. And "bad" party voters might swift to the "ideal" candidate!
As this video states, this doesn't happen if the third party takes from both candidates equally. But it's very common that one of the third parties will be more similar to one candidate than the other, and to take more from that candidate. For example, if one party wanted to run multiple candidates. The only reason they do not right now now is because it would split their voters. In fact if a third candidate takes from all others equally, I don't think it causes a spoiler event even in normal plurality. But vote splitting is common in plurality with more than 2 candidates, unless there is a lot of strategy. Either way, two-party domination is ensured. IRV doesn't fix it.
IRV is successful in what it proposes to do. IRV creates the possibility of an election in two rounds in only one election. The criticism lies not only with IRV, but over any major election in two rounds. The voter who prefers the most voted candidate can vote in third place because he knows that if the third win the second place, their candidate will have a less competitive election in the second round.
One way to avoid this is to analyze each combination of the first and other possible candidate and see which one offers the greatest risk for first place. If any candidate gets more votes than the first, this candidate is elected, if not, the first candidate wins.
Clay Shentrup You should not be so on the offensive. You accuse me of not knowing Condorcet systems ... I researched on various voting systems and have visited your site several times. I am not against Scoring Voting or Approval Voting, I have my criticisms, just as I have with IRV ... People should not fight over this. We should sit together and think. Maybe someone has a better idea. What I proposed is that instead of IRV eliminate the least voted candidate and redistribute their votes should analyze combinations of two candidates at a time. The first and second, first and third, first and fourth ... If another candidate in the race to win the one on one with the first, he could be elected. I'm offering possibilities ... I think this is an idea that can be developed.
This video assumes all candidates are males, smh
No it doesn't. I just picked pronouns to make an example. This is more important than the point being made? It seems you are looking for things to be upset about here.
@@amtiv Lol chill out. Someone's getting triggered
@@ClayShentrup There's "they" which is the standard way to do it in English
@@amtiv Don't know what you're on about. This is a "comment" on the video. It doesn't have to be an academic response to the central thesis. I'm commenting on an aspect of the video...
Chill out
@@ClayShentrup he and she are both grammatically equivalent in English and acceptable academically, but he is more commonly used.
Later-no-harm is actually bad in other ways. If you are *willing* to compromise with a consensus candidate, and saying so honestly hurts your first choice, well that is the entire premise of consensus and compromise, not a bad thing. In other words, the emergence of a consensus candidate *should* alter the chance of a polarizing candidate, even if the consensus candidate isn't anyone's favorite. So it's not just that the principle is peculiar and has a problem. We don't even want later-no-harm in principle.
Aaron Wolf shouldn’t voters be able to vote for their favorite candidate without worrying about who the consensus/compromise candidate is?
@@YamadaDesigns YES they should! Instant-runoff and later-no-harm do NOT let them do that!
In instant-runoff, you can't safely ignore who the consensus/compromise candidate is! If you do, they might get too-few 1st-choice votes, then get eliminated, and then your favorite STILL gets removed in a later round, and the winner is NEITHER your favorite NOR the consensus, it could be your *least* favorite.
So, the point is that we want a system that DOES let you vote for your favorite without worrying about who the consensus/compromise candidate is. Exactly as you say.
STAR Voting lets you do that because all the scores are counted no matter what. If your favorite doesn't actually have enough support, your vote will *still* go for the consensus candidate over your least-favorite. They don't get awkwardly eliminated before your preferences get counted.
But to reiterate my original point: what you want is to be able to vote your favorite without worrying about (A) who the consensus turns out to be or (B) whether you are risking your least favorite or (C) whether you are risking your favorite winning on a plurality but living in a society that falls into civil war.
The goal is to be able to HONESTLY vote for your favorite while still getting a result that is an effective enough representation of the overall voters such that democratic governance is feasible.
So Later-No-Harm is not needed. What's needed is not having favorite-betrayal. We want to vote honestly for favorite. Later-no-harm isn't about that. LNH is about voting for later-preferences without *any* reduction in chances for your favorite. But I don't prioritize having the maximum-chances for my favorite. I prioritize being ABLE to express my favorite without worrying about screwing up the outcome.
If my favorite has enough support, they can win. If they don't have enough support, I want as good an outcome as I can get otherwise. There's no system where I can both maximize the chances of my favorite winning *and* maximize the chances of an acceptable outcome when my favorite doesn't win. Those are tied together too much. And I don't want all-or-nothing, my-way-or-screw-it, I want a functional society. But I do not want the consensus candidate to falsely think they have a strong mandate or that my favorite who lost actually had no support. I want to express my support for my favorite honestly. That has political significance whether they win or not.
In short: I want to vote honestly, not think about any of the strategy otherwise, and I want the best reliable outcomes that satisfy the most voters overall.
Sorry the "bad" candidate was liked slightly more than the "ideal" candidate. Under FPTP, the bad candidate would've won even more
This video isn't promoting FPTP. But it does suggest that IRV doesn't prevent duopoly, and virtually every method behaves the same as FPTP when there are only two strong candidates.
This is the second time I've seen this spurious argument.
Democracy says we live with the "Bad" candidate if they don't get enough votes.
In NY we have a runoff if no candidate gets 40%+.
Its hard to see how RCV is a worse system, given 3rd party candidates are so unpopular.
This video fails to consider the ""Good" in 1 strong party and 2 not so popular parties.
Your candidate may lose if they are not popular enough. Thats not "Bad"
Should they lose if they are the most popular, and preferred by a majority vs every other candidate, as Good was?
Being scared of voting third party because of a manipulative voting system is vastly different from simply being unpopular.
The LP has run too many candidates to not notice these spoiler effects.
Democracy says each voter should have equal power to each other voter. IRV fails that test. www.equal.vote/theequalvote
You're assuming that all the bad votes are still for the bad instead of ideal.
Theres no argument for the bad party wouldn't vote for the ideal too.
Actually I think they did mention this, at 1:52
Ideal still loses. Their second choice vote do not count. This is an argument against IRV.
1:20 what happened? what happened is that the majority voted for the "bad party" can someone explain the problem
+Juan De La Cruz The problem is explained later at 1:43.
Intuitively, if people vote with the ranking: (1: Ideal, 2: Good, 3: Bad), that should increase the chances of the Ideal candidate winning, and decrease the chances of the Bad candidate winning.
However, the only difference between the 1:20 situation and the 1:43 situation is that more of those voters participated in the first scenario, yet that actually ended up helping the Bad candidate and hurting the Good candidate, even though those voters ranked the Good candidate above the Bad candidate.
Everyone who started as Good or Ideal preferred Good to Bad. They were a majority.
Bad is only preferred by a majority vs. Ideal (not "Good"), which is what's being shown at 1:20.
The problem is that Ideal's 2nd choice was never considered, because their 2nd choice was "Good" but they were already eliminated. Elimination didn't count 2nd choices, only 1st.
The pair-wise preference is bad > ideal, ideal > good, good > bad, which is a cycle and means that there is no Condorcet Winner. In that sort of scenario, the people who preferred ideal the most would rather have good win than bad, but since good was eliminated in the first round, bad ended up winning. In other words, if the ideal candidate had dropped out of the race, then good would have won, which means people who sincerely prefer ideal would be best served to strategically put good above ideal on the ballot. It's pretty much the main weakness of IRV.
This assumes there are few candidates running but there's no need for that. Say you have 5 candidates running, or 7 or 9, al of the sudden you don't really have any safe choices any more.
This spoiler or (unintended outcome) is hidden in IRV/RCV regardless of number of candidates.
It is possible to have many candidates and know that most of them have no chance at winning. There could be a big gap between second and third, or between third and fourth.
There is a flaw in your explanation, you made it seem that only the good party would vote for a third party as their first choice, people who usually vote for the "bad" party are also casting votes against the "good" party" (obviously the good party is not good for them). Not only ranking choice drastically mitigate the spoiler problem, but also creates a political enviroment with much more competition between the candidates beacause the major candidates don't take the election for granted.
I don't think that changes anything. 2nd choices can still be eliminated before they are considered.
Both could agree on a third party and it still could be eliminated.
This is saying that there is a "good and bad" candidate and that candidate definition is based on bias. What is good for one, may not be good for another. What if they are both "bad candidates" to another voter?
You can of course consider this from the perspective of an impartial observer. They will still be able to see that a spoiler-like effect occurs.
This appears to assume that both parties wouldn't fracture. Republicans could be divided into Trump vs Lincoln Project parties and Dems could be divided into Progressive vs Moderate. Ranked Choice seems to be able to handle a high number of candidates. Granted, I think you'd end up with a popularity contest á la Schwarzenegger in Cali.
PS: I'm not well read on this, just a "more educated than typical" passer-by. Feel free to entertain me with education on the matter. My brain isn't keyed for game theory.
Why would fracturing change any of this? If anything I think it would make it worse, since more parties/candidate means more rounds, and it can happen in any round
Sorry, but isn't your point at 1:15 the whole point? The majority of people in the 3 way split voted for the "bad" candidate, and the candidate with the majority of votes won in the end.
Late reply, but the problem is that by voting, you have caused your least favorite candidate to be elected. That is to say: your views would have been better expressed had you not voted at all. If you notice, at 1:45 they state this explicitly.
That's fundamentally flawed, if we're in agreement with the Voter Betrayal Criterion.
The majority of people preferred "Good" to "Bad" (everyone who put "Good" first", plus everyone in "Ideal" who put "Good" before Bad, which in this example was all of them).
Principle isn't a matter of scale, it's a matter of yes, the candidate adheres to my principles or no, the candidate doesn't. I would argue that only voting for your ideal candidates in IRV and not tactical voting for mediocre candidates that only sort-of disagree with some of your oppositions values will actually give you the candidate you want sooner. If you let people see how bad things can be under the power of your opposition then they'll be more likely to swing further the other way the next time.
polling has shown that many, if not most, people negatively vote. The goal is to express oneself and not be bound to how others are voting.
@@amtiv I agree. Voters are not like a jury that needs a unanimous verdict to avoid a mistrial, or employees who may not cooperate if they get to vote on a policy and people in the minority try to make the majority's plan not work and by doing so hurt their employer. It's not a voter's job to please other voters or know or guess how other people will vote.
But who exactly determines which candidate is electable or not in the scenario you set up? I get it that we live in a nation where Dems and Reps are the usual top two vote getters. But I suspect if we had IRV during the 1992 presidential elections, Ross Perot would have almost certainly won. As it was, Bill Clinton won with only 43% of the national popular vote in that 3-way race.
To go back to my first sentence ... with IRV, Bernie Sanders would have likely destroyed Hillary Clinton and later Trump in the general using IRV. But the establishment will forever and always claim Bernie would have lost in either or despite knowing full well they would be lying to themselves thinking that ... and that is why IRV, as a great a system as it is despite potential flaws as you have attempted to speak about here would be a much better way to elect our leaders.
How would IRV change the effects of an (almost) 2-person election?
Well, I suspect it wouldn't. IRV is rather useless in a two way race. But even if the 3rd choice is a weak candidate, in the end, the one most likely to have won had there not been a 3rd party to begin with probably wins anyways? To be sure no system is perfect, and I personally believe IRV would give more candidates a chance to run that otherwise wouldn't waste their time. Can a bad candidate win in such a setup? Sure, but the benefits of expanding choice among the people outweighs the negatives. At least, I am of the opinion it would.
How do you know Gary Johnson wouldnt have won in 2016 with IRV? More voters likely would have been ok with him than Trump or Clinton.
Also as the 2016 Dem primary was basically between 2 candidates, IRV wouldnt have changed much there. But it likely would have resulted in a different GOP nominee than Trump since the other 16 splintered the vote in primaries.
Wow, those are impressively bad examples of the kind of spoilers you're trying to highlight. The 1912 election is at least constructed the right way for the kind of scenario you're setting up, but Taft, Roosevelt, and Debs were all more similar to each other than any one was to Wilson, so you'd need almost half of Taft's voters ranking Wilson second... why, out of spite? for instant runoff to get the result you're describing. And 1992, 1996, and 2000 are even worse examples -- exit polls suggest supporters of Perot and Nader, who both finished distant third and are therefore not the kind of candidates you're talking about, were split pretty evenly between each major party and "I wouldn't have voted" as their second choices.
Ok maybe ranked choice voting isn't perfect but can we at least admit that it is vastly superior to plurality.
The scenario you described of a spoiler election under IRV is much more complicated and unlikely to happen than under the current system where it can basically happen any time a third party candidate votes are more than the margin between the top 2.
It absolutely is better than Plurality Voting. But then again....most voting systems are better than Plurality. Plurality just plain sucks. I don't think "better than garbage" is really the ideal benchmark we should set.
Don't get me wrong. I 100% would support a change over to IRV because again, it absolutely is better than our current system. But it does still suffer from the spoiler effect, it's just less common and less obvious when it does, and it also has a problem with monotonicity (sometimes you can cause your favorite candidate to LOSE by rating them #1, when they would have won if you had rated them #2, which is a paradox that I'm very not okay with).
My preferred voting method currently is STAR voting. It's also not perfect (no system is) but it avoids the above problems and promotes honest voting as always being the best strategy you can take, to give you the most optimal outcome you can hope for.
'is much more complicated and unlikely to happen' except that time Aspen, CO adopted it, had it happen in the very first election, and immediately abandoned it. or that time Burlington, VT adopted it, had it happen in the very first election, and immediately abandoned it. it happens all the time in Australia and Ireland
This is literally the only claimed benefit, and it's spoiler effect is actually worse than FPTP. RCV is somehow even worse than FPTP
@@alanivar2752 In plurality voting, an extra candidate is often a spoiler just by existing. Even if they only garner 2% of the vote, they can spoil it.
In RCV, an extra candidate won't spoil the election unless they have both more support than the candidate who otherwise would have won, and less support than the candidate who would have lost. Which is objectively a rarer and more complicated situation.
Which doesn't really make it all that rare. You're right that it definitely still happens a lot. But is it worse than FPTP? I have trouble with that claim. In both systems, the outcome of the election would have been the same (the extra candidate would have spoiled it), and at least it happens less often in RCV.
But I agree we can do better.
@@alanivar2752 I knew about Burlington, but you do have sources for Aspen, and for how often it happens in Australia and Ireland?
I think this video could uses some hard numbers:
We're in Plurality
0:10 "Good has a 10 point advantage over Bad"
Bad: 45%
Good: 55%
0:27 "If [Ideal] takes away too many votes away from the good candidate, that would cause the worst candidate to win"
Bad: 45%
Good: 44%
Ideal: 11%
Now we're in IRV
0:40 "With IRV when your ideal candidate gets eliminated your vote moves over to help the good candidate win"
ROUND 1:
Bad: 45%
Good: 44%
Ideal: 11% Ideal's voters all vote Ideal>Good>Bad
in round 2 Ideal gets eliminated and the assumption is that all their votes would transfer to Good.
ROUND 2:
Bad: 45%
Good: 55%
0:45 "But what happens when your ideal candidate does even better, what if he beats your second choice and your second choice gets eliminated"
ROUND 1:
Bad: 45%
Good: 27%
Ideal: 28% Ideal > Good > Bad
(Note 27%+28%=55%)
However good gets eliminated in round 2.
0:56 "Unfortunately that puts us completely at the mercy of these voters ... It takes only 1/5th of them putting the bad candidate second for him to win the election"
If good's votes end up being:
6%: Good > Bad > Ideal
21% Good > Ideal > Bad
Then round 2 looks like this:
ROUND 2:
Bad: 45% + 6% = 51%
Ideal: 28% + 29% = 49%
So Bad just won because Ideal was more successful than Good, even though Good would beat Bad in a head to head election.
Therefore Ideal voters who understand this would not vote for Ideal > Good, and would not want to give Ideal their core support.
This is probably the best explanation I've seen on TH-cam about how IRV can come up short. The problem I have with approval and range voting is that it is hard give more support to your favorite candidate than your compromise candidate without it potentially coming back to bite you. Using approval voting with a top two runoff is probably the best way to solve this and is the most pragmatic way to have voting reform.
I also like using a Condorcet method, but we would need to do some sociology experiments to see how likely it is for regular voters to vote strategically. If a very high percentage of voters are going to vote strategically, the worst candidate could get elected with a Condorcet method.
If we have voters who are educated about voting systems the best voting method may be to rank and rate each candidate and allow the range voting winner runoff against the Condorcet winner.
If you want a system where voters can afford to give middle votes, you should look at Bucklin (median) systems such as Majority Approval Voting (or Majority Judgment, or others).
Your other suggestions (approval with runoff, runoff between different system winners) are good but require two rounds of voting.
Seems like a lot of effort to have the voters rate AND rank, when you could get implied ranking from the ratings. (If I rate a:10 b:8 c:1 d:0, that implies a>b>c>d.)
And it's important to note that (based on extensive simulation work) score (AKA range) and approval are more-likely to pick the (true) Condorcet winner than any actual Condorcet method when faced with strategically-minded voters.
But sure, if the implied-rankings-Condorcet winner is different from the score winner, have a runoff. A top-two runoff always helps, a score-v-Condorcet runoff almost has to be better.
Well you can give different levels of support with ranged. AV is just a simplified version of yes/no.
I hear your concern about wanting to rank candidates. This was a major area of study when our organisation was deciding which method to support. What we found was that it is the very act of comparing and contrasting candidates that CREATES strategic voting. When you are allowed to evaluate each candidate individually, on their own merits, that's what creates sincere voting.
Approval voting considerably reduces the Spoiler Effect but does nothing about Gerrymandering.
Excellent Point! There's a mathematical solution to that; algorithmic division of electoral districts (look it up.)
What is the treatment for gerrymandering?
No, it does nothing about campaign finance either. Your statement is a non sequitur. Not all things solve all problems. That is a separate issue. Proportional representation is one possible solution to that.
Mathematically, Approval Voting is not guaranteed to prevent gerrymandering. In all real-life examples, however, it has been shown to make gerrymandering incredibly difficult and even backfire, thus incentivising away from it.
That could be said for any system given this is rather divorced from any voting system.
So, no matter what, you don't always get the candidate you want to win.
This video is terrible and misleading. All voters have a choice between 'bad', 'good' and 'ideal'. If 'good' is eliminated, those voters then have a choice between 'ideal' and 'bad' i.e. their 2nd preference. 'Ideal' voters don't take away from 'good' voters unless 'good' voters prefer 'bad' over 'ideal'. So, the solution is simple.. if you have to choose between your 'good' and 'ideal' choice, make sure they're both in the top 2 of your preferences.
The video makes contradictory assumptions; it assumes that 'good' voters would prefer 'ideal' but are scared and end up voting for 'good' .. but then 'good' voters might put 'bad' as their 2nd preference?? Not a chance.. obviously these reluctant 'good' voters will put 'ideal' as their 2nd choice. A strength of ranked choice voting is that you can put your 'ideal' candidate and still support another good candidate on the 2nd preference. If 'good' voters support 'bad' as 2nd preference, 'bad' was always going to win no matter what system you use.
>"it assumes that 'good' voters would prefer 'ideal' but are scared and end up voting for 'good'"
What do you mean? I didn't get that from the video at all. ALL Good voters prefer Good to Ideal. But _some_ prefer Ideal to Bad, and _others_ prefer Bad to Ideal, which is normal
>"If 'good' voters support 'bad' as 2nd preference, 'bad' was always going to win no matter what system you use"
Not really, systems less vulnerable to the spoiler effect (which already exist) would have continued to pick Good as the winner, as was the point of this video (they weren't saying Ideal should _win,_ they were just saying Ideal shouldn't be a spoiler)
@@MustSeto
'Spoiler' is a value judgement. As long as over 50% of voters put 'ideal' and 'good' as either 1st or 2nd preference, 'ideal' or 'good' wins. If 'bad' ends up winning, it's because more than 50% thought 'bad' was worth a 1st or 2nd preference.. and if more than 50% prefer 'bad' in that way, 'bad' by definition is not a spoiler.
Instant runoff means you get to always vote your ideal preferences knowing your vote will still count if your ideal candidate does not do well. If your ideal candidate does do well, you rely on other voters to also prefer your ideal candidate as 2nd best. If not, your ideal loses because they lacked preferences not because of assumed 'spoiler' effect.
@@aliasadsj >"'Spoiler' is a value judgement
I don't think so, I think it's normally defined as a non-winning candidate who, by running, changes who wins (generally to a candidate considered even "farther" away, to at least some of the voters), which definitely applies here---Ideal was a spoiler for Good
>"As long as over 50% of voters put 'ideal' and 'good' as either 1st or 2nd preference, 'ideal' or 'good' wins."
Unfortunately this is clearly not the case, even in this example. Over 50% of voters put Good as either 1st or 2nd, yet Good lost (in fact, Good was the Condorcet winner, preferred by a majority versus _every_ other candidate 1-on-1)
>"Instant runoff means you get to always vote your ideal preferences knowing your vote will still count if your ideal candidate does not do well."
Again, this has unfortunately been proven to be mathematically impossible. Arrow's Impossibility Theorem proves, basically, that no reasonable rank-based method can be immune to the spoiler effect (or rather Independence of Irrelevant Alternatives failure, which is even more formal). Rating-based methods can _sort of_ get around this, but absolutely nothing can get around Gibbard's theorem, which says no reasonable method of any type can be immune to strategic voting
@@MustSeto
> Unfortunately this is clearly not the case, even in this example. Over 50% of voters put Good as either 1st or 2nd, yet Good lost
Yes, in that situation, ideal wins. I said that ""As long as over 50% of voters put 'ideal' and 'good' as either 1st or 2nd preference, 'ideal' OR 'good' wins." In the videos first few examples, all of 'ideal' transfers to 'good' and good wins.. and then all of 'good' transfers to 'ideal' and ideal wins. So, as long as:
1st preferences (ideal or good, whichever larger 1st prefs) + 2nd preferences (of the other eliminated one) is > 50%, ideal or good wins.
Instant run off has to work like this or I'm really missing something here.
@@aliasadsj >"Yes, in that situation, ideal wins."
My situation was from the video, but Ideal never wins anything in this video, Bad won in that situation. Over 50% put Good (or Ideal) 1st or 2nd, but Bad wins.
>"So, as long as: 1st preferences (ideal or good, whichever larger 1st prefs) + 2nd preferences (of the other eliminated one) is > 50%, ideal or good wins."
In this video, even though 1st prefs Good plus 2nd prefs of Ideal is > than 50%, "Bad" wins
It sounds like you might be thinking IRV works a bit like Bucklin voting?
To put some numbers to the examples:
26x I>G>B,
20x G>I>B,
5x G>B>I,
49x B>?
% who put Good 1st or 2nd: At least 51%
But Good gets the fewest 1st choice votes and is eliminated.
Next round:
46x I>B
54x B>I
Bad wins
But if some I>G>B voters betray their favorite and switch to G>I>B
25x I>G>B
21x G>I>B
5x G>B>I
49x B>?
Now Ideal has the fewest first choice votes and is eliminated
Next round:
51x G>B
49x B>G
Good wins, which is better for the I>G>B voters, G>I>B voters, and G>B>I voters. This would also happen if Ideal simple did not run.
Ideal, by running/not being strategic, caused someone they and their compromise liked less
So the "bad" (from a biased perspective) fairly won a majority. That is precisely the perk of IRV/RCV; a majority's good or even ideal candidate won. If you believe a candidate should have majority support, then this video is a promotion for RCV but attempts to frame it some other way.
The "Good" candidate was preferred to "Bad" by a majority, but was spoiled by "Ideal" as "Ideal" gained support
@@MustSeto Are we using different definitions of a spoiler? A spoiler is someone who makes it so a contest is won by a candidate with minority support. In that sense, there literally can't be a spoiler as long as voters vote in the runoffs. The winner will always be whoever had majority support. If "bad" wins, it's ONLY because they're actually either good or ideal in the eyes of the majority.
@@dustintrombly The spoiler effect is when a non-winning candidate changes who wins by running (splitting support between compromises). If Ideal did not run (or if their supporters betrayed their favorite and concentrated support on their compromise) then Good would have won, which both Ideal and Good would have preferred to Bad winning. Ideal spoiled the election for the {Ideal + Good} "faction".
Runoffs only guarantee the winner will be preferred by a majority vs one other candidate (the other candidate who makes it to the runoff).
But Good was preferred by a majority vs. Bad --- Bad is the minority candidate in that race. Good was also preferred by a majority vs Ideal. If it weren't for the spoiler effect, the winner would have had majority support vs. each other candidate.
"Does [the spoiler effect] happen in every election?" That's irrelevant, since that sort of thing is still mechanically possible in IRV, however statistically unlikely it is. It's also possible, depending on whether every choice must be ranked, for a voter's ballot to either start counting toward candidates they don't want in office, or worse, just stop counting altogether.
The situation in the video is *guaranteed to happen* if there is overlap between the two dominant factions and the third faction is significant in size. The third faction will inevitably lean one way or another. And we know for a fact such overlaps are the norm and significant. So if IRV did somehow encourage people to vote for third parties, the problem would occur.
IRV is inherently a 1-dimensional two-faction system. The existence of more parties and ideological distributions breaks it.
Thank you for sharing this!
It turns out there is a bit of an ideological conflict between supporters of IRV and Approval Voting. I think it comes down to the purity and mathematical robustness of Approval Voting as opposed to the cynicism and pragmatism of IRV.
I believe that voters are not tactical or perfectly rationale enough for Approval Voting to work out. I think that it will quickly devolve back to Instant Runoff where voters simply vote once for their preferred choice. While I tend to agree that approval is the most correct from a mathematical point of view, I think that Instant Runoff Voting will work better in the real world.
This is clearly false. For instance, with the current choose-one system, "strategic" voters are those who *do not* vote for their favorite. E.g. a Green who votes Democrat. Obviously with approval voting such a voter would also vote for the Green, plus anyone else she prefers to her tactical lesser evil Democrat.
@@ClayShentrup That's the thing though, voting for multiple candidates under Approval Voting only makes sense when you think your first choice doesn't have a chance to win. You then cast other votes so that your vote isn't completely wasted. But if you think your first pick has a chance to actually win, then you want to ONLY vote for them, as casting any additional votes will hurt your preferred candidate's chances.
My personal favorite method is STAR Voting. While there's no such thing as a perfect system, the only real complaint I've heard about STAR voting is that it's so new and that people think it would be difficult to convince places to adopt it. But it's a simple concept that promotes voters giving their honest ratings for candidates. I think a system in which honesty is the best strategy has a lot of advantages.
@@periculum69 One issue with STAR is that it fails the participation criterion: causing your less-popular favorite to get in the runoff instead of your compromise candidate can cause your least favorite candidate to end up wining, instead of the compromise candidate winning if you hadn't voted.
@@SolomonUcko While technically true, that doesn't appear to be something anyone could exploit, and would only happen if all three candidates were very close. In a race that was so close, it was too close to call, and any of the three could win, my best strategy is still to vote honestly. And they're all so close that regardless which one of the three does win, even my least favorite, it'll be roughly equal in terms of satisfying the electorate as a whole.
In short, not a problem that particularly bothers me.
@@periculum69 Technically, there there can be cases where a single vote can decide a not-overall-contested election, such as with IRV (due to repeatedly comparing the smallest vote counts), but beyond that you just assume that either the election happens to be close enough, or that you represent a large enough group of voters. Also, if you knew that not voting or voting dishonestly would improve the election result for you, why would you choose to vote honestly?
AV inferior to RCV: say I'm extreme lib & 4 candidates: extreme lib I love, moderate lib I like, mod conservative I'd tolerate to avoid extreme conserv I hate. In AV, vote 4 mod con = i love them as much as libs. No vote for mod con = I'd be just as upset w/ them as extreme con.
How unlikely, how about we talk about condorcet loser criterion?
'unlikely' except that time Aspen, CO adopted it, had it happen in the very first election, and immediately abandoned it. or that time Burlington, VT adopted it, had it happen in the very first election, and immediately abandoned it. it happens all the time in Australia and Ireland
This is literally the only claimed benefit, and it's spoiler effect is actually worse than FPTP. RCV is somehow even worse than FPTP
Approval Voting has been show to be Condorcet in all real-life scenarios
Let's not let perfect be the enemy of the good. Can we all agree that the plurality system is the worst and that other systems would be preferable?
I am very bothered by the statement at 1 minute: "that puts us completely at the mercy of these voters that we have no control over." Exactly who should have control over the voters? Are you suggesting that because the scenario WANTS the Ideal candidate to win, that it should? Obviously, many people liked the Bad candidate. So, I guess they don't think that candidate is bad. What caused the "least favorite candidate to win" was the fact that more people voted for that person as their 1st or 2nd choice. That's democracy folks.
"Let's not let perfect be the enemy of the good. Can we all agree that the plurality system is the worst and that other systems would be preferable?"
Yes, we can all agree on that. However, the point of this video is not to promote plurality over RCV, but to point out a flaw in RCV that other voting systems (such as Approval Voting and Range/Score Voting do not have).
"What caused the "least favorite candidate to win" was the fact that more people voted for that person as their 1st or 2nd choice. That's democracy folks."
That's not true. Most people preferred Good over Bad, but Good got eliminated too early. This occurred because more people expressed their *sincere* preference for Ideal over Good, but those same people still preferred Good over Bad. If you rewatch the video, it is explained fully.
This video is dumb.
The only way it makes sense is if you feel that your candidate is entitled to another candidate's voters via your personal perception of their ideology. This scenario acts as if 100% of the voters of the "bad party" did not vote for the independent candidate.
On a mathematical level, we can all tell why this is very unlikely. On an anecdotal level, I know some very angry folks that would have voted Green if they thought the Greens had a chance so instead they voted Trump to "defeat the Democrats that rigged the primary."
This logic assumes that Dems will never vote for Libertarians, Republicans will never vote for Greens, and Independents go to the closest ideology that you personally perceive them having. This argument is entirely subjective regarding your idea of which party is "good and bad" and who you personally believe is closest aligned to.
Australia uses IRV in our lower house and yes you are correct, people have varying reasons for their rankings, but they do tend to 'lean' in the direction of ideology, but it's never 100% and usually more than 3 candidates. I don't see why its bad to favour candidates with a higher first choice, if they were almost going to win a majority anyway but it's still not a guarantee.
What do you mean? I don't think it says any of this.
>"This scenario acts as if 100% of the voters of the "bad party" did not vote for the independent candidate"
What does this matter? IRV wouldn't consider that in this scenario either, whether ideal or good is eliminated first
>"This logic assumes that Dems will never vote for Libertarians, Republicans will never vote for Greens, and Independents go to the closest ideology that you personally perceive them having."
I don't think it makes that assumption at all, it even sort of mentions this at 1:52
e.g. even if Ideal voters also had a portion of people who put Bad 2nd, this can still happen, and still tends to happen long before Ideal has a chance of actually winning, in the meantime those Ideal voters who prefer Good would only be helping to spoil their compromise
@@MustSeto do you not realize the concept of a “bad party” is subjective?
@@derrickblanton8328 The video using "Ideal", "Good" and "Bad" essentially as names probably isn't helping with clarity at this point. Should we call them A, B, and C instead? Can you rephrase your issues with this video?
The problem isn't Instant-Runoff
The problem is that the Gold Party is too much of a turn-off for the blackest of the White Party, the White Party is more centrist.
The problem is that instant-runoff ignored that the gold party preferred the white party to the black party. If it didn't ignore that, then the white party would have won.
There is no perfect way of electing a single office. The way to get full representation of all the diverse views in a community is to elect a council by some form of proportional voting, and have that council choose a mayor accountable to them.
Wilf Day and then those mayors choose a governor and so on and so forth?
+mmoore, the people at large could elect a five-person or seven-person governing council by PR.
There are possible advantages to such a system, however it does limit democratic accountability in that the people do not have opportunity to separately determine the quite different roles of executive and legislature. They also lose the ability to directly elect the leader of the government. The Mayor selected by the council, as the most powerful individual office holder, will assume a leadership role, but people will feel less involved if they did not directly vote for her. I live in the UK where directly elected executives are still rare, and people usually see the kind of council/parliamentary democracy that results from the system you describe as distant and detached. There's a big advantage, perhaps especially in local government, in having not just a council of representatives but the actual leader of the government elected directly.
One other option is to have a multi-person executive which is directly elected alongside a separate elected legislative council. Swiss cantons and local government in Switzerland are usually organized this way, but this is almost never seen outside Switzerland. That may be because executive governance does not exactly lend itself to not having a single final decision maker (as the American founding fathers recognized, hence the individual President): if a legislative council deadlocks, no new laws are made but services continue to be delivered because the executive keeps going, but the executive may often be faced with decisions which need answers right away, what happens if the executive council deadlocks over a critical decision which needs an immediate answer? This doesn't seem to happen in Switzerland but that may be because of a uniquely consensual political culture there.
+William Waugh Directly elected executive councils of about 5 people are actually common in the Swiss cantons and Swiss local government, but are rarely seen outside Switzerland. As I wrote in my other comment, what happens if the executive council deadlocks on a crucial decision which needs an immediate answer? A legislature can deadlock because day-to-day government is carried on by the executive, all that a legislative deadlock does is stop new laws from being made, but an executive deadlock could potentially cause important services to be disrupted.
Concerns like these are probably the reason why the Founding Fathers gave the US an individual President. It may also be the reason why almost all executive government, local and national, tends to be led by an individual rather than a committee. It is worth noting that in Switzerland there is a uniquely consensual political culture which may be why they can make multi-person executive leadership work. However even in Switzerland there is a provision last used in World War II to appoint an individual commander-in-chief of the military during times of war or great risk of war.
Approval Voting is proportional the more, and smaller the districts are
The potential problem this video focuses on under IRV can be largely eliminated if voters are instructed to only vote for the candidates they would be OK with; voters must be told that they do not have to rank every candidate that's on the ballot. That way their vote will never count towards a candidate they can't stomach.
+Nancy Lord That wouldn't help with the scenario in the video. The problem in this scenario is that the Bad and Good candidate attract more moderate voters and the ideal candidate is more on the extreme end who attracts the base of the good candidate. If the Good candidate is eliminated the moderate voters who supported him are more likely to go to the bad candidate than the extremist ideal candidate.
I think what you're describing is Approval voting, where the choices aren't ranked, which is a system he advocates at the end of the video.
It's disappointing to see reform advocates fighting over which system is the best alternative to plurality. I would think fans of approval voting would recognize that they are "spoiling" efforts at reform by splitting the community, and thus helping us retain the status quo, when they could spread a positive message about "approving" both.
The prior CES video (with the fruit candidates) was quite a bit better than this one. It's a shame to see CES put their name on a video that wastes time going after IRV.
I don't see this video as being just negative. FBC is an important aspect of voting, and you can't explain FBC without looking at a nontrivial example where it's violated.
Also, the prior video cost a whole bunch more to make. Simpler, cheaper videos like this one are an appropriate way to explore specific aspects.
I do agree that this video is not for everyone. It's a bit technical, and only intended for people who are interested at this level of detail.
Drew Spencer IRV advocates are spoiling the chances of voting-system reform generally. They get municipal officials to spend time trying out IRV, everyone gets disappointed by how poorly it performs, then they vote to restore Plurality. This gets them fatigued with voting-system talk, so their ears become closed to Score Voting for single-winner cases and PR systems for multi-winner cases.
William Waugh A majority of voters in every city using RCV support its use: www.fairvote.org/reforms/instant-runoff-voting/ranked-choice-voting-civility-project/what-do-voters-think-of-rcv-/.
Seventeen jurisdictions have voted to adopt RCV in the modern era and only three have voted to repeal it.
No repeal has switched to plurality - they all switched to two-round runoff elections, which have the exact same "problem" identified in this video, suggesting that "favorite betrayal" had nothing to do with repeal.
Drew Spencer, how is the first round conducted for the two-round runoff elections? It's by vote-for-one-only, isn't it?
Drew Spencer, I re-read your comment more carefully and took note of your suggestion that advocates should "Approve" both rating and ranking systems. But the problem that we utilitarians have with IRV is we need a system that will defang the two-party system and institute a democratic republic.