Agreed, and largely (but not wholly) agreed. I’m only pointing out abstract mathematical flaws in the procedure, as brought to our attention by Kenneth Arrow. There is no system of electing leadership that is free from mathematical flaws!
I do not agree that “Winner Takes All” is anywhere near as damaging as gerrymandering. And low voter turnout is voluntary. It’s part of free choice. It isn’t a problem; it’s a valuable and prized feature of our system. If you don’t want to vote…shrug. So don’t. American society would be more harmed by mandatory voting than it could possibly be by people choosing not to vote at all.
(I do not like low voter turnout. It saddens me. But any possible cure would be worse than the symptom itself.)
No, the current Australian Federal parliament has a majority, albeit by a single seat.
The 2010 Federal election produced a minority government, which survived (indeed was quite productive) when four of the cross-benchers declared their support for Labor on confidence and supply
The previous hung Parliament was seven decades previously, in 1940.
To suggest that preference voting (IRV) causes hung parliaments and minority government is simply wrong.
I looked at the link. I’ve seen the “expected outcome” theory before, I’ve seen rebuttals before.
The forced min-max vote is the problem, or one of them. Requiring every voter to give either full approval or full disapproval to each candidate means that some genuinely second (or lower) preferences get counted equal to all first-choices, others not at all. A variety of strange and dissatisfying outcomes are possible. Bush-Clinton-Perot '92 is a frequent example. By polls of head-to-head preferences, Perot would have been the Condorcet loser, but he might have won under AV! Actually, given the same matrix of actual preference-orders, any of the three could have won! Depending entirely on how many people of which groups chose to “approve” their second choices. Kind of ridiculous, don’t you think?
Maybe it’s not a mathematical failure of independence-of-irrelevant-alternatives, but only because AV doesn’t perceive the real difference between “I really like this best of all” and “I hate this less than the third choice.”
Arrow’s work is classic, but only formally addresses rigid ranked systems, I believe.
This is not correct.
Tony Windsor (IND) won the seat of New England in NSW at both 2004 & 2007 elections as did Bob Katter (IND) for the seat of Kennedy in Queensland.
Prior to the nominated period there was 1 IND in 1998 and 4 IND in 2001,
Following the nominated period there were 5 IND in 2010, 5 IND in 2013 and 5 IND in 2016
Yes, preferential voting favours the major parties but it’s not a whitewash.
Not at all. Perot might well have been the best choice! The Condorcet criterion isn’t all that.
An oracle that polled peoples real preferences and optimized for happiness could certainly select any of the three, depending on their exact breakdown and exactly how the optimization happens. Should it go for the centrist that nobody loves, but no one really hates, either? Should it try to make the maximum number of people happy, without too much regard for their degree of happiness? Should it try to maximize average happiness, allowing those with strong preferences to dominate the rest?
Anyway, I’m not trying to claim range voting is a bad thing, necessarily–just that in practice it’ll be almost identical to approval voting in practice, since anyone remotely strategic will min-max their scores. It’s extra complexity with very little payoff.
I think this is correct, but there are other proofs that such flaws accrue to other systems, such as the “market” voting system (you have ten points to distribute among the candidates: do you spend all ten on your favorite guy, or do you hedge your bet a little by putting some points on your second and third favorites?), ranked voting by points (very similar: your first choice gets three points, your second choice gets two points, your third choice gets one point) and the really grotty pairwise voting (for three candidates, A, B, and C, you have three separate ballots: A vs. B, B vs. C, and A vs. C. You have to win two or more of these to win)…
I wandered into a sentence, and can’t find my way out of it. Anyway, all of these (and every other possible system) can break down in certain circumstances.
It’s of piercing important to USAians that many of these breakdown circumstances are in highly polarized elections (i.e., really close races.)
Not in an instant run off voting system. What tends to happen is that the sub parties cooperate to win in states they are weak, and compete in states they are strong. But having the ‘Centrist Dems’ and the “Left Dems” competing in say California doesn’t matter, since they give their preferences to each other over any of the more right wing parties. While in a red state they’d only run one candidate and pool resources. Instant Run Off voting allows these kinds of coalitions, so they form. Your current first past the post system doesn’t really allow for coalitions (yes there’s a rare independent that caucuses with a party).
You still end up with two parties that are dominant, but their dominance is not as absolute, they share power with some smaller parties and have to work with them, or at least do deals to appeal to their voter base. Now I’m not saying IRV is perfect, certainly we’ve had serious issues in Australia, but absolutely it would decrease the dominance of the two major parties in the US.
There’s bugger all evidence that your happy co-operation of sub parties occurs in the three countries which use IRV. Have you observed any flourishing of third parties in the Australian state parliaments? It would be Bambiesque to think that what hasn’t happened in almost a centuries use of IRV in Australia would become the inevitable trend in other jurisdictions.
While fascinating, discussion about the mathematics of different voting systems is rather beside the point. In terms of switching to a different voting system, approval voting is the low-hanging fruit. To quote myself:
It is an easy first step away from our current system. For those of us who want to change, it is a much easier goal than more complicated systems.
Requires changing the ballot (no longer checked-unchecked for each candidate, but a range value). Remember the problems with “butterfly” ballots. Any change away from “check-this-candidate, do-not-check-that-candidate” is going to cause massive voter confusion and unintended votes.
Is harder to explain (no longer yes-no for each candidate, but a value from a range). A more complicated voting system using a more complicated ballot is begging for trouble.
Is harder to evaluate the results (no longer a simple count of votes, but a sum of values). Yes, we Dopers are comfortable with math, but it’s a huge burden for the public to understand it. And when the public has trouble understanding how the ballots are used to determine the winning candidate, the legitimacy of the system is undermined.
Yes, range voting is better than approval voting, but going to it is a bigger step than switching to approval. When voters are extremely reluctant to make any change at all to the voting system, it is much better to choose the system that minimizes the change. Better in terms of actually getting the change implemented and better in terms of voters using it as intended.
I don’t know that voting for multiple candidates will seem less weird than giving grades to all of them. Speaking of which, what about A/B/C/D/F as someone suggested upthread?
I like it a lot! But it’s several steps harder than approval voting.
Yes, voting for multiple candidates will seem weird. Voting for multiple candidates and giving them a grade is more weird. I want to minimize the weirdness, so that a change in the voting system is more acceptable and less likely to backfire.
Imagine explaining a proposed voting system to an intelligent but uneducated 70-year-old who has voted one way their whole life. Approval voting is a large step for them. And any other new voting system is even more difficult.
In your opinion. But I just don’t agree. I think it will be more common for people of any age to feel weird about giving multiple equally weighted approval votes to different candidates rather than demonstrating some distinction about which is your favorite. Taking away the ability to declare who one’s top most preferred candidate is (that is, unless you just approve one, in which case you have lost the point of even having a different system than we have now) is going to be the most disconcerting change in my view. For this reason, as much as I have criticized IRV, I think it would be more acceptable than approval voting. In fact, I would prefer to just keep the current system rather than go to approval voting. But range voting would be a welcome change.
Heh, I think range voting is my favorite, with approval voting the easiest first step toward it. I’d rather keep our current system than use instant runoff.
The fact that in approval voting one can vote exactly the same way as currently (just voting your single favorite) is a feature of that system. It lowers the barrier for voters to adjust to it.
And it’s not really “in my opinion” that approval voting is simpler. Other systems require more than a binary choice for each candidate. Whether it’s a range value or an ordinal, other systems require more input from a voter.
I’m a little surprised you’d accept range voting but not approval voting, since the latter is a subset of the former (the range being reduced to yes/no).
There was no use of the words “simpler” or “simple” in the post I responded to. What I was (and still am) calling “your opinion”, with which I do not agree, was:
I don’t agree that approval voting, compared to range voting, is:
–more weird
–more acceptable
–less likely to backfire
As I said, *my *opinion is that most people have a strong desire to express which of the candidates is the single person they would most prefer be elected. Once that itch is scratched, I think a lot of people would enjoy being able to express that some of the also-rans are still pretty good or at least semi-acceptable, while also indicating that some of them are moderately unacceptable and still others are heinous. Doing that by giving each candidate a letter grade is something anyone can easily understand, and which a lot of people would find fun. Using a 0-99 scale is not nearly as good. And just saying “Elect Joe or Jim, I really don’t care” is, I think, deeply unsatisfying and counterintuitive to most people.