The Independent on Sunday featured an article by Chris Huhne, the energy secretary, arguing that, while the AV referendum result was clearly a setback for electoral reformers, the pressures that still exist within the system will at some point make change inevitable. Now that such a change has been delayed, Chris argues, when the time for reform does come again, it will be on a much bigger scale than the relatively modest reform that AV would have been.
Here’s a sample:
The problems to which electoral reformers are responding have not gone away and will continue to demand an answer. British society is increasingly pluralist, and the trend to diversity is accelerating. In the Fifties, only 4 per cent of voters rejected Labour and the Tories. Now the figure is a third. Once Labour and the Conservatives dominated our politics. Now Liberal Democrats, Greens,Nationalists and others demand a voice.
The attempt to squeeze diversity into the fraying corset of two-party politics is likely to lead to more and more unfair results. Already, the regional representation of the parties is distorted, with the Tories underrepresented in the north and Scotland despite substantial votes, and Labour similarly anorexic in the south. Both parties speak first to their regional bases, respectively ignoring urban deprivation and aspirational affluence.
Nationally, electoral results are likely to look odder and odder. The last Labour government won a firm overall majority of parliamentary seats with just 36 per cent of the vote, the lowest ever recorded in our parliamentary history. How low will that proportion have to sink before defenders of the status quo confess its illegitimacy?
You can read the piece in full over on the Independent website here.
Follow @LibDemVoice on Twitter
Like us on Facebook
Subscribe to our feed
Sign-up for our daily email digest
Tracing the route of Butterley Tunnel on the Cromford Canal
Number 12: The Innocence Mission - Black Sheep Wall
Lib Dems see off Reform in Barnsley with double their vote
Cleanliness at Harris Academy swimming pool #dundeewestend
Welsh councils still facing financical difficulties
Joe Bourke
Mohammed Amin
Jonathan Brown
theakes
14 Comments
This post – complete with graph – similarly concluded that PR may be inevitable: http://www1.politicalbetting.com/index.php/archives/2011/04/25/be-careful-what-you-wish-for-av-through-the-looking-glass/
Adam Corlett (with respect) –
I’m not totally sure that I accept the premise of that link. FPTP has produced hung councils and continues to do so. Surely the 2010 election result shows a system that can produce Coalitions that represent a large part of the electorate? It is also worth noting that a PR system in Scotland has punished Labour, despite their vote holding reasonably static (and yes, there are some qualifications there).
The stark reality is that a 70:30 referendum is the end of the discussion. What is worrying is the way that some people seemed to have convinced themselves that there is no possibility of any sort ever at all that a significant number of people, fully aware of the pros and cons, are OK with FPTP.
“the pressures that still exist within the system will at some point make change inevitable”
While this is probably true, “at some point” currently means “no sooner than 2021” — and quite possibly years later. Nothing will be done in the lifetime of this Parliament, and nothing will be done in the next Parliament, no matter who wins the 2015 election.
“The last Labour government won a firm overall majority of parliamentary seats with just 36 per cent of the vote, the lowest ever recorded in our parliamentary history.”
They won the majority of the seats, that’s how it works, AV wouldn’t have changed this much, if at all. If PR is what Chris Huhne wants then maybe he should have bloody well argued for it and not accepted the AV referendum.
That one’s easy. There’s been a whole load of polls over the past decade on this question and they all came back with the same result: the public overwhelmingly hates FPTP.
That makes it obvious that the referendum result, while it was influenced by many things, was never about FPTP.
This bit is probably true, however. Electoral reform is now off the table for the next ten years or so.
The LibDems must push for Approval Voting. It is superior to and much simpler than IRV (aka Alternative Vote).
http://www.electology.org/approval-voting-vs-irv
Approval voting is useless for elections from the general population. It only functions when voters are educated and aware of both the full behaviour of the system and the political reality. In reality, the general population has a minimal understanding of their voting system (even FPTP – lot of people don’t understand the “spoiler effect” in detail and how to counter it) and pays very little attention to politics.
In a situation where you have uneducated voters, approval voting fails most significant criteria for analysing voting systems; it does not reliably select the majority candidate if one exists (that’s the candidate which over 50% of the population want as their “first preference”), it fails IIA (hence it has a big spoiler effect), and it can elect the condorcet loser (that’s the least popular candidate – one which the majority would have preferred anybody else over). That makes it pretty much as bad as FPTP.
Approval voting is a good system when you’re dealing with an electorate that is highly educated and aware. It is ideal for a society of mathematicians to elect their leader. It is hopeless for selecting a government.
(Range voting, which that site favours above all others, is rather better – that one actually works. But it’s not proportional. Let’s stick with STV)
Andrew Suffield,
Your comments exhibit several of the common intuition-based misconceptions about election methods.
As my link explains, Bayesian Regret calculations (performed by a Princeton math Ph.D. named Warren D. Smith, who has studied election methods for well over a decade, and is perhaps the world’s foremost authority on the subject) show that Score Voting outperforms all commonly proposed alternative systems, with any mixture of strategic or honest voters, and despite wide variances in “ignorance levels”. (Approval Voting is the simplest form of Score Voting, and performs almost as well as when using a wider range such as 0-10.)
http://ScoreVoting.net/BayRegsFig.html
Approval Voting was also recently favored by the London School of Economics, out of a pool of 18 different systems.
http://www2.lse.ac.uk/newsAndMedia/news/archives/2011/04/VotingSystems.aspx
The most obvious and presumably common tactic with Score/Approval Voting is to vote for the same candidate you would with Plurality Voting, PLUS every candidate you prefer to that one (www.electology.org/threshold). With that general behavior, there are a number of significant elections which would have apparently been different with Approval Voting. The most obvious one for an American such as myself, would be Gore vs. Bush back in 2000. But in Plurality elections with more than two candidates, there are commonly situations in which a different voting method would produce a different outcome, with high probability.
So to say that Approval “only functions when voters are educated and aware of both the full behaviour” is simply not in line with the available evidence. Moreover, there is a great deal of evidence about human behavior when using Approval Voting, both from numerous large exit polls, as well as use in large organizations (many of them having a greater population than many towns). All evidence says that common theoretical/academic expectations of voter behavior are consistent with reality. You would have benefitted from reading my original link, which talks about this in detail.
http://www.electology.org/approval-voting#TOC-Where-has-Approval-Voting-been-used
“In reality, the general population has a minimal understanding of their voting system (even FPTP – lot of people don’t understand the “spoiler effect” in detail and how to counter it) and pays very little attention to politics.”
This is quite correct. I am a software engineer in San Francisco, where we use Instant Runoff Voting (aka “Alternative Vote), and I recently cited an online conversation with a very smart co-worker (who votes) demonstrating that he did not understand the process.
http://www.electology.org/san-leandro-inamdar
This highlights the importance of using a SIMPLE system, like Approval Voting. Approval Voting is empirically the simplest system there is, for voters. It results in the lowest rate of ballot spoilage (meaning, with Approval Voting, voters screw up their ballots less often than with any other system).
http://ScoreVoting.net/SPRates.html
“approval voting fails most significant criteria for analysing voting systems; it does not reliably select the majority candidate if one exists (that’s the candidate which over 50% of the population want as their “first preference”)”
There are numerous severe problems with this statement. For one, the case where a majority preferred candidate fails to be elected with Approval Voting is statistically highly improbable. But Approval Voting SOLVES a VERY COMMON problem that occurs with Pluality Voting (and all ranked systems), in which a “spoiler” candidate causes the wrong outcome.
http://www.electology.org/spoiler
But there is actually a lot more to this issue, so I’ll simply provide this link for further reading.
http://www.electology.org/majority
“it fails IIA (hence it has a big spoiler effect)”
Wrong. Score Voting and Approval Voting are the only systems which SATISFY Independence of Irrelevant Alternatives.
http://www.electology.org/spoiler
“and it can elect the condorcet loser (that’s the least popular candidate – one which the majority would have preferred anybody else over).”
Another extremely unlikely scenario. Actually, there is evidence that Approval Voting is MORE likely to elect a Condorcet winner (when one exists) than ACTUAL Condorcet methods.
http://ScoreVoting.net/AppCW.html
Furthermore, it is mathematically proven that the Condorcet winner is not necessarily the most popular candidate, and that the Condorcet loser is not necessarily the least popular. So it is a fallacy to cite this remote possibility as a “flaw”.
http://www.electology.org/majority-criterion
“That makes it pretty much as bad as FPTP.”
Utterly false.
“Approval voting is a good system when you’re dealing with an electorate that is highly educated and aware. It is ideal for a society of mathematicians to elect their leader. It is hopeless for selecting a government.”
I have been studying this issue since 2006, and I co-founded a non-profit organization called The Center for Election Science, which is headed by Warren D. Smith (creator of ScoreVoting.net, and protagonist of the William Poundstone book Gaming the Vote. All I can say is, years of research on this subject, including lots of research on real live human beings, massively refutes your position.
“(Range voting, which that site favours above all others, is rather better – that one actually works. But it’s not proportional. Let’s stick with STV)”
Well, Score Voting is not meant for multi-winner elections. It is useful for SINGLE-winner elections, such as mayor, governor, president, etc.
For proportionality in multi-member districts, there are two systems which are simpler than, and in some ways superior to STV.
http://ScoreVoting.net/RRV.html
http://ScoreVoting.net/Asset.html
Clay Shentrup
Secretary, Director
The Center for Election Science
San Francisco, CA, USA
206.801.0484
[email protected]
@Clay
I’ve not got that much free time today so I’m not going to check up on what criteria these various voting systems, though I would like to say that a voting system breaching the monotonicity criterion is much less of a big deal to me than a voting system breaching the majority criteria – AV/IRV (I prefer the term IRV) can breach monotonicity (which is not very common I have to add) a thousand times for each time approval voting breaches the majority criterion in my view! Plurality voting is completely rubbish but you’re trying to convince people who, by and large, would prefer IRV if they had to be stuck with a single winner method. Which also fixes up the spoiler effect quite nicely.
I also think there’s a lot of potential for sheer randomness in response (does this candidate get 6 or 7?) with range voting and I’d imagine there’d be “bullet voting” in approval voting – because putting two candidates down might hurt your favoured one and you can’t rank them there’d be an incentive to just put down one.
Some other concerns I have:
1) Simplicity is not, in itself, a virtue. Namely because the only truly simple system is just having closed party lists (which may be the third worst voting system, behind FPTP and the bloc vote), even FPTP is complicated because setting up constituencies is a complicated process. Most systems are more complicated than FPTP. Even approval voting. Definitely range voting.
Besides which, many people learn how to use their vote, even if they don’t know the mechanics inside out. The experiences of devolution show that even when people don’t know how STV or AMS works in detail they still figure out how to use their vote as best they can in them!
And most systems can be phrased confusingly or simply depending on whether you want to confuse the issue or actually explain it.
2) Even if you consider simplicity a virtue RRV seems no simpler to me than STV – certainly far more complex than IRV. Asset voting also seems marginally more complex (arithmetic rather than ranking) in terms of the actual ballot, even though the mechanics are simpler.
3) Many of the attacks used against AV in this referendum would apply equally or moreso to these voting systems. Their (stupid) tagline was – “keep one person one vote” implying (wrongly) that AV gives you multiple votes. With approval voting you have multiple votes, with range voting you don’t but as each point contributes to the score different people will contribute to the result more than others. This would be a genuine attack rather than just mud-slinging if these systems were proposed.
STV precludes this line of attack in that “single transferable vote” clearly states that everyone only has one vote.
4) Asset voting is quite clearly not going to go down well, the line of attack is too obvious!
Ultimately, the mathematics matter, but so do the politics. Approval voting will never fly, asset voting will definitely never fly, range voting probably wouldn’t either, IRV could have under different circumstances with a better run campaign but didn’t – those who were opposed to change were still opposed and those who wanted change thought it to be “tinkering at the edges” rather than actually solving the big problems with FPTP. STV and AMS could, eventually, fly in the UK.
Hah! Hardly. I’m a computer scientist but I follow several subfields of mathematics quite closely, including voting systems. Warren Smith is a noted researcher – but mostly in other branches of mathematics, and is no Nicolaus Tideman or Markus Schulze, and he’ll laugh at you if he finds out you wrote that.
This is not the appropriate forum for discussing how you have misinterpreted his research.
Painfully wrong; a voting method is a Condorcet method iff it always selects the Condorcet winner where one exists – that’s the definition of the term.
Only in the Nash equilibrium; in the case of ignorant voters (which is the case we are dealing with here) they do not – consider the case where each voter selects only their highest ranked candidate. Obviously Arrow’s theorem tells us that full IIA is expensive, but local IIA is readily achieved by several.
Please let him do the writing; you don’t understand the subject well enough to discuss it.
70:30 – I am sure its been commented before but that ratio is much the same as the ratio of those who voted labcon as opposed to all the rest in the last election. In other words those whose votes have more weight were (unsurprisingly) happy with the way FPTP works. Was it all as simple as that?
@DunKhan,
“I’d imagine there’d be “bullet voting” in approval voting – because putting two candidates down might hurt your favoured one and you can’t rank them there’d be an incentive to just put down one.”
This is one of the most irrational and counter-factual arguments we encounter. It’s actually quite backwards, as this problem is more severe for IRV than for Approval Voting.
I’ll distill that link into a few key points, for those who don’t want to click away.
Say you prefer LibDem>Labour>Conservative, but you currently vote tactically for Labour in your area, because LibDem has little chance of winning anyway. Then with Approval Voting, you would continue to do that, plus cast a sincere vote for LibDem. Then, if enough voters sincerely prefer LibDem to the perceived frontrunners, then LibDem can actually win. The appearance of being “unelectable” has a much smaller tendency to become a self-fulfilling prophecy.
But with IRV, if you wisely tactically rank Labour in first place, that necessarily prevents you from also ranking LibDem in place. Therefore you are “tactically forced” to rank LibDem in second place, and therefore LibDem can be prevented from winning a race it should have won, due to the mere appearance of not being likely to win.
Because Approval Voting satisfies the Favorite Betrayal Criterion, it is much less vulnerable to tactical voting than IRV and every other ranked method.
“I would like to say that a voting system breaching the monotonicity criterion is much less of a big deal to me than a voting system breaching the majority criteria”
Well, you don’t get to decide how much of a big deal it is. If non-monotonicity causes you to get a candidate you thought was a “7” instead of a candidate you thought was a “10”, you lost 3 points of utility, period. If getting a winner who was not the majority-favored candidate causes the same effect, you still lost the same amount of utility. It doesn’t matter why you got a less optimal outcome.
It’s sort of like arguing about the effectiveness of various race cars. I say mine is better because it has a bigger engine. You say yours is better because it has better aerodynamics. Someone else says his is better because it’s lighter. Well, your subjective evaluation of the relative importances of those factors means absolutely nothing. The only objective way to address the matter is to put the cars on the track and do timed trials until you get a statistically significant number of sample race times.
THAT is what Bayesian Regret is for voting methods. It reveals the combined anti-democratic effect of all property failures, simultaneously weighing their frequency times their severity. Bayesian Regret is far and away the most important concept for people to learn if they want to have a serious discussion of single-winner voting methods.
I repeat, if you want to have a scientifically valid and pragmatic discussion of voting methods, you have to stop focusing on properties/criteria, and start focusing on Bayesian Regret.
“AV/IRV (I prefer the term IRV) can breach monotonicity (which is not very common I have to add)”
Actually, the most recent evidence suggests it’s incredibly common. Here is a calculation of IRV pathology/paradox probabilities, performed by Warren Smith (again, a Princeton math Ph.D. who has studied this issue for over a decade).
If we look at the overall rate of non-monotonic elections, we get the 3 following values, depending on whether we use a REM, Dirichlet, or QUAS 1D model:
15.2305% 5.7436% 6.9445%
That does seem so concerning. But, if we restrict ourselves to those elections in which IRV made a difference vs. Plurality Voting, then the numbers become astonishingly large:
35.8569% 26.5477% 9.7221%
Usually it is just difficult to FIND these elections, because of a lack of full ballot sets. Of those elections in which full ballot data was available, non-monotonicity and other problems are actually quite large. For instance, the recent Frome election, and also the last IRV mayoral race in Burlington, Vermont, USA.
“… a thousand times for each time approval voting breaches the majority criterion in my view!”
Well, your view is just incorrect. It is mathematically proven that the majority winner, even when there is one, is not necessarily the favored candidate of the electorate.
Furthermore, your assessment fails to recognize comparably severe “problems” in various ranked methods. For instance, in that Burlington election, a majority of voters preferred the Democrat (the most centrist of the three major candidates) to the Progressive. Yet the Progressive won, because IRV mistakenly eliminated the Democrat due to his not having enough first-place rankings. The Democrat was the Condorcet winner, yet IRV did not elect him. If you are any more upset about failure to elect a majority winner than about failure to elect a Condorcet winner, then you are making a logical fallacy, by disregarding Independence of Irrelevant Alternatives. And IIA is not a subjective/arbitrary criterion like the Majority and Condorcet criteria. It is a logically forced result of axiomatic logic, as is any criterion based on consistency.
“Plurality voting is completely rubbish but you’re trying to convince people who, by and large, would prefer IRV if they had to be stuck with a single winner method. Which also fixes up the spoiler effect quite nicely.”
I’m in agreement with the former statement. In general, Top-Two Runoff is better than IRV is better than Plurality. So moving to IRV would be a small but significant upgrade. And IRV does mitigate the spoiler effect, although it does not fix it.
“I also think there’s a lot of potential for sheer randomness in response (does this candidate get 6 or 7?) with range voting”
But that is incredibly TINY in comparison to the data-deletion effect of ranking candidates. For instance, if I believe X=10, Y=8, Z=3, and you believe X=8,Y=4, Z=1, both of our ranked ballots look identical. All that preference data is discarded. By comparison to that, the random error introduced by people mistakenly putting down e.g. a 6 instead of a 7 is tiny. It is only a very small deletion of data. Furthermore, the same effect could cause someone to vote X=1st, Y=2nd, Z=3rd, when X=1st,Z=2nd, Y=3rd would have been accurate, had the person had more time to think about it or learn about the candidates and issues. It is not as if ranked methods are immune from this.
Moreover, Warren Smith’s Bayesian Regret calculations actually took this sort of thing into account by introducing random noise which simulated such ignorance. And after all that, Score Voting still came out far ahead of IRV. IRV is actually the worst of the major alternative systems tested.
“Simplicity is not, in itself, a virtue.”
Having fewer spoiled ballots, and thus more useful ballot data, is a virtue.
Having cheaper counting with results computed sooner is a virtue.
Having less risk of fraud, due to not requiring central counting is a virtue.
Having a system that is less likely to be used as justification for the adoption of fraud-prone electronic voting machines, is a virtue.
Having a system that produces outcomes voters can feel trust and confidence in, rather than confusion and/or suspicion, is a virtue.
“Namely because the only truly simple system is just having closed party lists (which may be the third worst voting system, behind FPTP and the bloc vote)”
That is not an argumente that simplicity is not a virtue. It is an argument that other properties can have negative effects which outweigh the simplicity virtue. I am not asserting that simplicity alone is the deciding factor. It is just one more benefit on top of numerous other benefits that Approval Voting has over the alternatives.
“Most systems are more complicated than FPTP. Even approval voting. Definitely range voting.”
For tabulation yes, but for actual voting, no. Voters make FEWER errant/spoiled ballots with Score and Approval Voting. And actually, Approval Voting is algorithmically simpler than Plurality Voting, in that you don’t have to check whether a ballot was invalidated by casting votes for more than one candidate per race.
Don’t believe that? Try writing the shortest program you can, in whatever language you prefer, which tabulates results for Plurality and Approval Voting, given a ballot array. The Approval Voting program can be made one step simpler.
And again, your only actual point here is that Plurality Voting’s simplicity does not alone make it superior. That does not mean that its simplicity is not one of its virtues. It means that it has other flaws which outweigh those virtues.
“The experiences of devolution show that even when people don’t know how STV or AMS works in detail they still figure out how to use their vote as best they can in them!”
We see a lot of empirical evidence that says that’s total nonsense. Most voters I talk to here in San Francisco have no idea how the tabulation process works. They tend to use the Naive Exaggeration Strategy, because it intuitively seems sensible (and actually is, for the most part). Did you read the conversation I had with my co-worker about IRV?
“And most systems can be phrased confusingly or simply depending on whether you want to confuse the issue or actually explain it.”
Yes. But some systems are more conducive than others to the problem of being confusingly explained. Whether or not that is the “fault” of the system, it is a pragmatic consideration. If IRV has properties which cause it to generally be explained in legalese which is confusing and verbose, whereas Approval Voting has properties which cause it to generally be explained in simple language that the average voter can more easily understand, then that is a significant practical matter.
“Even if you consider simplicity a virtue RRV seems no simpler to me than STV”
Well, it just is. It has lower Kolmogorov complexity. There are fewer calculations. You can write the code to perform both methods and see for yourself that it will be shorter for RRV, given that you apply equal attention to optimizing both programs. This is not subjective.
“Asset voting also seems marginally more complex (arithmetic rather than ranking) in terms of the actual ballot, even though the mechanics are simpler.”
The ballot and tabulation are radically simpler than IRV. There’s no question about that. What’s complex is the process where the politicians can exchange votes.
“Many of the attacks used against AV in this referendum would apply equally or moreso to these voting systems. Their (stupid) tagline was – “keep one person one vote” implying (wrongly) that AV gives you multiple votes.”
I would generally disagree with that first statement, although this particular “one person, one vote” argument, as horrendously flawed as it is, would seem to be easier to make against Approval Voting. I would stick with the point that it’s “one person, one ballot”, and that entire criticism is wrong for ANY system.
“Asset voting is quite clearly not going to go down well, the line of attack is too obvious!”
Well, the OPOV thing is not obvious at all. I would say that the people for whom that argument was effective with IRV are probably only slightly smaller in number than those for whom it would work with Approval or Asset. There are lots of other much bigger and more legitimate flaws with IRV/STV that are fixed by Approval/Score/RRV/Asset.
“Ultimately, the mathematics matter, but so do the politics.”
Yes.
“Approval voting will never fly”
I think it is vastly more likely to “fly” than IRV, because it is not prone to the sorts of severe problems that have caused IRV to be repealed in several places in the USA in the past couple of years.
“asset voting will definitely never fly, range voting probably wouldn’t either”
Well, you don’t have any evidence for that claim. STV is far more complicated than both systems, and it was in use in Australia back in the early 1900’s, before any computers or “voting machines”.
You’re treating your gut/intuition like it’s scientifically valid evidence here.
“Warren Smith is a noted researcher – but mostly in other branches of mathematics, and is no Nicolaus Tideman or Markus Schulze, and he’ll laugh at you if he finds out you wrote that.”
Nicolaus Tideman has made several severe flaws in his analysis which, in my view, indicate that his expertise in the field of election science is worse than most of the “amateur theorists” on our discussion list. Warren Smith addresses some of those flaws here (see the section called “Tideman’s ‘strategy resistance’ measure is flawed”).
Tideman’s measure was created by picking several arbitrary “strategy resistance criteria”, and giving them arbitrary significance. That is not objective or scientific. The way to be scientific is to perform Bayesian Regret experiments, to objectively see how well various systems perform, given that some fraction of the electorate will behave tactically.
Here is another conceptual tool for understanding the gist of Tideman’s logical flaws.
http://www.electology.org/tactical-voting
As for Markus Schulze, he is a member of our discussion list, so I have communicated with him quite a bit. From what I can tell, his contribution to the field of election science is minor in comparison to Smith’s. A few points:
– Smith co-authored a fairly revolutionary paper on secure voting with Ron “RSA” Rivest.
http://ScoreVoting.net/RivSmiPRshort.html
– Smith co-discovered a theorem that rivals the Gibbard-Satterthwaite Theorem (and therefore arguably exceeds Arrow’s Theorem) in significance
http://ScoreVoting.net/SimmonsSmithPf.html
– Smith performed the world’s most extensive Bayesian Regret calculations
http://scorevoting.net/BayRegDum.html
– Smith invented the Shortest Splitline Algorithm, to prevent Gerrymandering
http://scorevoting.net/GerryExamples.html
These are just a few significant contributions by Smith. I know of nothing even approximately as significant from Markus Schulze.
Furthermore, Schulze bizarrely advocates for the Condorcet method (or at least his specific implementation of it), even though Score Voting is simpler and objectively superior according to the world’s best Bayesian Regret calculations. And in spite of the fact that Score Voting is plausibly a better Condorcet method than ACTUAL Condorcet methods.
http://ScoreVoting.net/AppCW.html
“This is not the appropriate forum for discussing how you have misinterpreted his research.”
If you’re talking about Smith, misinterpretation would be difficult, considering we’ve worked together for almost five years, and that he is the CEO of a non-profit that we co-founded, called The Center for Election Science.
“Painfully wrong; a voting method is a Condorcet method iff it always selects the Condorcet winner where one exists – that’s the definition of the term.”
But Condorcet methods DO NOT elect the Condorcet winner, if there is enough tactical voting. As you would know had you read my link.
http://ScoreVoting.net/AppCW.html
http://ScoreVoting.net/CondBurial.html
http://ScoreVoting.net/DH3.html
>>”Score Voting and Approval Voting are the only systems which SATISFY Independence of Irrelevant Alternatives.
>Only in the Nash equilibrium; in the case of ignorant voters (which is the case we are dealing with here) they do not
This has nothing to do with Nash equilibria. Score Voting, and by extension Approval Voting, makes it impossible to change the winner by removing or adding a non-winning candidate from the cast ballots.
http://www.electology.org/spoiler
For those of us who are not intellectuals and academics on forms of PR with the exception of @Duncan who thinks the debate is over. Good News folks its not. Take Back Parliament have licked their wounds and are up and running again on twitter @takeback2010 or on Facebook http://www.facebook.com/takebackparliament.
Its time to start talking about where the strategy for achieveing Electoral Reform goes next. It is time for the Leaders of all the Political Parties in favour of PR and that may mean the Leader of Labour to have a meeting to thrash this out. If not then a Political Movement with possibly separate Party Candidates will have to develop to take this forward. Lib Dems its your choice.