What’s so precious about having an out-and-out winner to declare?

We know that there usually is an out-and-out winner to an election: in the sense of a party that can defeat every other party in a one-on-one fight (the Condorcet Winner). This is different from being able to get the majority of the vote (being able to defeat every one at once, which is a strange definition of a winner).

Under PR - if the Condorcet Winner gets less that 50% of the first preferences - then the parties that lost the election will always be able to form the government and will always be able to exclude the winning from power. In my view, this is an absolutely lethal criticism of PR.
# posted by  nik @ 19:12

 

Jarndyce,

I appreciate your courtesy in letting me know of this useful post, which is, as you say, mostly a handy compendium of the arguments you have deployed over at my blog (hope that link works!). My own comments on most of your points are on the record there, and there's no need to repeat them here. Nik's comment above is another, extremely elegant, nail in the PR coffin.

It's been, I think, a usefully mind-clearing debate, even if no-one on either side of it has been converted. And it's good to record that it has been notably good-humoured.

Signing off --

Brian
http://www.barder.com/ephems/
# posted by  Brian @ 21:45

 

Nik - er, yes, but FPTP doesn't guarantee the Condorcet winner a majority either. In fact, it's useless as a criterion to judge an FPTP vs PR debate. Unless you're happy to accept coalitions after the fact as representative proxies for decision-making and coalition forming, which I do.

It's also worth noting that the largest parties usually get to lynchpin coalitions, and under some systems it's a constitutional requirement or at least strong precedent. I realise that that might not be the Condorcet winner, but without preference votes to count, it's a decent proxy or guess.

Brian - thanks. I still don't think you've addressed the points I make satisfactorily. But I'm thinking you're thinking the same about me...
# posted by  Jarndyce @ 22:32

 

Yes, FPTP doesn't guarantee the Condorcet winner a majority - I'm not giving it unqualified support - but there are good reasons for supposing it comes closer than PR does, and that PR can lead to far more perverse situations than FPTP.

The mere fact that there's a coalition indicates that the Condorcet winner hasn't formed the government. Under PR you can only be sure of the winner forming the government in exceptional circumstances (they get >50% of 1st preferences).

I don't follow that Condorcet is a useless criterion to judge an FPTP vs PR debate, surely it's the only criterion if we want a representative government, though I think I'm perhaps missing your point - perhaps you could explain?
# posted by  nik @ 22:54

 

Have you seen the run off in Dresden - the city that would decide it all?

I dislike quoting directly from David's Medienkritik, but they have it in English and I can't be arsed to translate the ARD/ZDF site:

"The by-election in Dresden 1 (one of the two Dresden constituencies) Oct 2 has added one more seat to the conservative CDU/CSU ("Union") in the Bundestag (226 instead of 225 seats), giving the Union a 4 seats majority against chancellor Schroeder's SPD (222 seats).

Neiter of the two parties will be able to govern the country alone, though. The most likely coalition - as of today - is a "grand coalition" between Union and SPD. Schroeder probably won't stay as chancellor; Merkel (or some other Union candidate) will succeed him. Stay tuned for more news on Germany's government crisis...

The Dresden result sheds light on the wonders of Germany's byzantine election law.

1. While the CDU candidate won the Dresden 1 constituency (first vote), the SPD had the majority of the party votes (second votes). Had the CDU collected considerably more party votes (second votes) than the SPD (in Dresden I), and still won the constituency (first vote), the SPD would have gained one seat in the Bundestag. I'm not kidding you: had the CDU result (second votes) in the Dresden 1 election been considerable better (and correspondingly the SPD result worse), the SPD - and not the CDU - would have gained one additional seat in the Bundestag! It made sense for SPD sympathizers in Dresden 1 to vote for CDU (second vote) in order to secure an additional MP for the SPD. (Source)

2. The Dresden 1 election result changes the intra-party distribution of seats between German states. Quote from the official statement of the German Election Office: "As regards the intra-party distribution of seats won through the second votes, one CDU mandate will be shifted from the Land list of Nordrhein-Westfalen (Northrhine-Westphalia) to the Land list of Saarland..." Hmm... even though Dresden I is part of the state of Saxonia, the distribution of Union seats between the states of Northrhine-Westphalia and Saarland has changed as a consequence of the Dresden 1 vote"
# posted by  In Actual Fact @ 01:28

 

Okay, David's Medienkritik is right-wing, pro-Israel and pro-US (you can see that as bad or good), but the 1 directly-elected member, one party list vote is not condusive to strong government.
And yes, you can now argue whether strong government see Thatcher, Blair is a good thing.....
# posted by  In Actual Fact @ 01:33

 

Sorry, Nik, I wasn't clear. My point being that it's impossible to know how a Condorcet election would turn out unless you run one. So, projecting from, say, the 2005 UK election results doesn't give you any guide. Similarly, it's difficult to say what would happen to the UK party system under varieties of PR, etc.

A couple more points:

1. Condorcet is only really good at producing a winner. So, great for presidential elections. Not so good for legislatures, when surely the main (not only) criterion is representation. Yes, you could run 600+ mini-Condorcet elections (constituency-based), but you'd end up with similar problems (regional strengths over-represented, say) as under FPTP.

2. I still argue that coalitions can be seen as pseudo-Condorcet winners. If co-allied parties command between them >50% of first preference winners, and if we see MPs as elected proxies for voter preferences (which I do), then the winner has legitimacy and a mandate.

actual: note, I'm not defending the German system, with all its Ueberhangmandate weirdness. I do think it superior to FPTP, but not something I'd recommend. Note, though, that the closeness of the result is what's probably caused the chaos this time, not just the system.
# posted by  Jarndyce @ 08:47

 

Jarndyce: To answer your original points, as there has been a somewhat lower level discussion at The Sharpener:

1. Right, the governing majority to second placed parties is an artifact of the boundaries. Better boundary set-up would eliminate that problem. I think all FPTP advocates recognise this (and personally, I'd love to strip Scotland and Wales of loads of seats, for obvious reasons). On representativeness, well it depends how you define that. I'd argue that PR fails on this, but doubly so. Sure, you might get a legislature that matches the proportions of voters, but that legislature can't actually do anything unless coalitions form. Those coalitions are compromises. They don't even have to be between ideological bedfellows (see Germany...). They represent no-one. Having a winner in itself isn't the be all and end all, but strong government is important. It gets things done. And if it does the wrong things, we can turf it out and put another strong government in its place. Under FPTP, we can even opt for soggy coalition government, by electing a hung parliament.

2. Yes, but MPs aren't really proxies for public mood unless it's an election year. Cynical, sure, but basically true. Only the out-and-out rebels from each party show any genuine link to voter concerns, and then only because they've blown their career and it's the only way to maintain visibility. The difference between coalition agreements and trade deals is that one would expect to have an idea of what a governing party would do during a trade negotiation from the stance they take on other issues. In a coalition negotiation, you have no idea which policies will have to be binned to form the right compromise. Particularly in Germany, where the grand coalition could end up doing basically anything. To take a specific example, let's imagine in the next election we have PR, and the Lib Dems and Labour form a coalition. We're still in Iraq. What is the policy going to be on troop deployments? I couldn't tell you until chat-show Charlie leaves Number 10 with a grin or a grimace on his face.

3. I agree somewhat on this, but most of those are constitutional issues that we could solve by minor reforms (like a written constitution). It certainly wouldn't be fixed under PR, where if anything, party allegiance would become stronger, not weaker, as parties split and become more defined in their focus. On the powerful swing voters issue, this is as much a problem under PR, where the smallest minority party often ends up shoring up the coalition (although not always, I'll grant you). In all probability, these are much the same people anyway. Middle-class, middle-income, middle-England, soft-centrist, instinctive Lib-Dem voters.

God, it's good to discuss this intelligently.
# posted by  Andrew @ 10:40

 

Welcome back, Andrew. Okay...

If you accept MPs as proxies, though, then negotiating a coalition is perfectly acceptable, just the same as a trade deal (you have just as good an idea what the outcome wil be, anyway). If you don't accept them as proxies, then why do we have them in the first place? Why not referendum democracy, demand-revealing referenda, drawing of lots for representation in the House? To say the manifesto allows us to elect "strong" government with accountability says nothing at all about the representation element of democracy. Why not just vote every four years on a manifesto, and make it binding? Because, coming full circle, MPs and the legislature are proxies - and therefore coalition agreements are legitimate. It doesn't really matter that legislatures can't actually do anything unless coalitions form... that's just a fact of human existence, not something unique to government.

party allegiance would become stronger, not weaker

Not with multimember constituencies where several members of the same party were standing. Direct accountability to local voters would be increased, while still retaining party allegiance.

On the powerful swing voters issue, this is as much a problem under PR

But it isn't, though. There are no "swing voters". There may be a pivot party of two in the legislature, but that's an entirely different proposition - and ultimately an empirical question as how you view the outcomes of small-group bargaining. Swing voters are all about the watering down of policy levels to innocuousness, the rush to the diminishing centre ground. Big parties would have fewer inmcentives to behave like that in a whole variety of non-FPTP systems where every vote "counts".
# posted by  Jarndyce @ 11:30

 

If you accept MPs as proxies, though, then negotiating a coalition is perfectly acceptable, just the same as a trade deal (you have just as good an idea what the outcome wil be, anyway).

Not at all. Unless you know, up-front, what the only likely coalition is, PR is a lottery. When you have a case like the German elections, where 3 or 4 possible coalitions are touted, some of which have entirely opposed outcomes, the representativeness is gone. A trade deal has, yes, some element of chance because you're negotiating with outside parties, but at least you know what your party goes to the table with, and what it is not prepared to compromise on. And you can always walk away from a trade negotiation. You can't, in good faith, walk away from a chance to get into power in coalition under PR. The urge to compromise would be much stronger.

Why not referendum democracy, demand-revealing referenda, drawing of lots for representation in the House?

Because there seems to be a preference for representative government, and because it's probably the cheapest and easiest way to get things done without stepping over the line into outright dictatorship.

It doesn't really matter that legislatures can't actually do anything unless coalitions form...

It does, if you value representativeness as an ideal for a democratic system of government.

I'll take your word on multi-member constituencies - I don't really want to get into the nitty gritty of individual types of PR. I'm sure it can be structured such that parties are weak and people are strong. But vested interests, and the nature of politics and power will mean that you don't get to implement such a system.

There may be a pivot party of two in the legislature, but that's an entirely different proposition.

It really isn't. That pivot party will contain a core group of people with similar issues. They're no different to swing voters. They have to be pandered to in exactly the same way.

Big parties would have fewer incentives to behave like that in a whole variety of non-FPTP systems where every vote "counts".

No, they would water down their policies to suit the voters for whichever pivot party is likely to be able to get them into party. Every vote doesn't count equally. That's a myth.

To be clear, I don't think our system is perfect, but I don't see any great reason to change it, and throw the baby out with the bathwater.
# posted by  Andrew @ 12:22

 

"I still argue that coalitions can be seen as pseudo-Condorcet winners. If co-allied parties command between them >50% of first preference winners, and if we see MPs as elected proxies for voter preferences (which I do), then the winner has legitimacy and a mandate."

Let me demonstrate that this isn't the case with an example:

1/3rd of Voters: A>B>C
1/3rd of Voters: B>A>C
1/3rd of Voters: C>A>B

A beats B by 2:1, A beats C by 2:1, and B beats C by 2:1. The Condorcet Winner is A - the majority prefer them to any other party. There's a strong case they should form the government.

What happens in reality is you get AB, AC, or BC. In the case of AB the majority would prefer A to B - why is B included? In the case of AC - every candidate in the election beat C, why is he in government? And in the case of BC, the parties that lost exclude the party that won from power.

A representative assembly does not result in a representative government.
# posted by  nik @ 13:00

 

Andrew, I just don't accept your characterization of the workings of coalition politics. To me, MPs are just proxies. We're electing them to do what we believe they will probably do. We don't hold them directly responsible for every deviation from the manifesto, and we mandate them to behave as they judge best in situations that we cannot predict. As such, they are mandated (in my case) to negotiate a coalition government. Much better that they complete those negotiations than a small minority be given the handles of almost unlimited power for 5 years. You can do a lot of (irreversible) damage in half a decade. And if you spin it right, you can probably get more than one term without actually being competent.

And since we've been discussing Germany, neither of these assertions stand up to what's happened in this last month:

1. You can't, in good faith, walk away from a chance to get into power in coalition under PR

Well, the Greens could have joined a CDU/FDP coalition and haven't.

2. Middle-class, middle-income, middle-England, soft-centrist, instinctive Lib-Dem voters.

Not the Linkspartei, though...

We'll never agree, of course. I suspect our basic views on what a legislature is for are completely divergent. You think it's there to govern. I think that too, but primarily to represent. That's why they called it the Representation of the People Act.

Nik, thanks for the example. I do think Condorcet is interesting. But like Andrew above, I think you have a different view to me of what a legislature is for. Condorcet gives a clear (though quite complicated) winning mandate for governing, but how exactly to convert that into representation? Unlike direct presidential systems, we don't elect a government here, but a legislature. As I said above, running 600 mini-Condorcet elections for local MPs would have many (though not all) of the problems of FPTP, and anyway wouldn't necessarily produce a Condorcet-acceptable result when scaled up nationally. Even if you could draw the boundaries up acceptably, to produce an even weighting to votes all over the country.

(And to answer your direct question why is B included, B is there because A didn't manage a majority of first preferences and needs backup. With an AB coalition, government with a majoritarian mandate is possible, as long as you assume that parts of A and B's manifesto can be combined in an acceptable way. Which I do.)
# posted by  Jarndyce @ 13:43

 

"And to answer your direct question why is B included, B is there because A didn't manage a majority of first preferences and needs backup. With an AB coalition, government with a majoritarian mandate is possible, as long as you assume that parts of A and B's manifesto can be combined in an acceptable way."

What basis is there for presuming a government needs a majority of first preferences to be justified? I don't think there is one - mandates aren't additive in this manner. The example uses Condorcet to demonstrate this.

It's fair to say the majority would prefer an A to an AB coalition government. So I think it's valid to accuse a system that results in AB getting power of being undemocratic. (You don't have a comparable defence for an - equally likely - AC or BC government, do you?)

I'm not neccesarily suggesting running 600 mini-Condorcet elections for local MPs. As you say this wouldn't always result in the Condorcet Party getting in. (though, by defintion, it's an improvement on the current system which I'd support). I just think it's important that we have a system where the party that's the Condorcet Winner forms the government.
# posted by  nik @ 16:16

 

Well, the Greens could have joined a CDU/FDP coalition and haven't.

To be honest, I haven't followed it closely enough to know the ins and outs, but my take on that would be that it's the Green's loss. You can't do anything when you're not in power.

Middle-class, middle-income, middle-England, soft-centrist, instinctive Lib-Dem voters.

Not the Linkspartei, though...

Well, I was talking about PR in Britain there, really.

We'll never agree, of course.

Ah, but the sport is in trying to convince one another, of course.

I suspect our basic views on what a legislature is for are completely divergent. You think it's there to govern. I think that too, but primarily to represent.

Well, not completely divergent, but I'd guess we'd prioritise a list of what government is for differently. Your number 1 is representation. Mine is governing, but I see the value in representativeness, and other ideals as well. I just don't put them above the necessity for government.

That's why they called it the Representation of the People Act.

Well, the naming of an act isn't really a good description for what it actually does. Some of them are positively Orwellian - the Prevention of Terrorism Act, which does nothing of the sort?
# posted by  Andrew @ 19:04

 

Sorry I've missed out on this one... one quick point,

Andrew – "Right, the governing majority to second placed parties is an artifact of the boundaries. Better boundary set-up would eliminate that problem."

Given that you've responded to the Once More post about my post on the Tories and the electoral system, I assume you now know this to be bollocks... therefore, just checking :)
# posted by  Paul Davies @ 14:49

 

Caveat: pretty much bollocks. The changes to be introduced will make it harder for the 2nd party to win a majority, but not impossible. Much like a 10% Tory lead in votes - under some circumstances, could get a majority of seats, but most likely just a hung parliament.
# posted by  Paul Davies @ 14:50

 

Paul: I change my stance pretty quickly as I absorb new information, and your post was very helpful, but all it really did for me was to reconfirm old prejudices of mine. I've ranted at length over at Once More about the need for my party to become a broader church. Here though, I'm really just addressing the faults in the system, and for me, that's boundary set-up. The bias, as you call it, is for me just a failure of my party to capture a sufficiently widespread demographic. I don't see that as a fault of the system.
# posted by  Andrew @ 15:41

 

And also a caveat: As I understand it, the boundary changes don't enforce a uniform constituency size, particularly where a historical geographic boundary exists (Isle of Wight?), so I don't think the changes being imposed are that good, to be honest. I'd personally go further, although I guess in theory it would be possible for a party to win with a very small percentage of the overall vote, if they were sufficiently concentrated in the right seats - e.g. the Tories winning everything south of Birmingham? I don't think that's likely in practise, although I'd love to rule over the Scots with a pathetically weak mandate, if only to show them how it feels. ;)
# posted by  Andrew @ 15:45

 

"It's fair to say the majority would prefer an A to an AB coalition government."

Nik, you've missed one crucial factor out of your logic.

How do you know what they prefer unless you give them the option of voting for an AB coalition government versus an A government? The majority might prefer the coalition?

This seems to be the mistake FPTPers make, that each party is distinct on policy and each voter endorses ALL of their policies just because they vote for that party. We know that is not the case, in fact in ideological terms, Labour and Lib Dem voters see themselves as identical.

You only need to talk to voters to find out they agree with varying amounts of different parties' policies. It is the party that shares their values and maybe has the most attractive policies that gets their vote.

Couple this with what Jarndyce points out about electing proxies, and you are left with this choice.

A minority elected govt under FPTP or a majority elected govt under PR.

You know whats funny about the Tories fierce loyalty to FPTP for Westminster- its such a rubbish system they don't even trust it to elect their leader, they use STV.
# posted by  Neil Harding @ 04:17

 

"all it really did for me was to reconfirm old prejudices of mine...Here though, I'm really just addressing the faults in the system, and for me, that's boundary set-up."

well if even you admit that they are prejudices, then I can't really be arsed arguing against them :)

And the point with the boundaries is that even if they were all identically sized (which would be silly as it would do a lot to negate the fabled FPTP constit link as they would keep chopping and changing so much and across natural boundaries that any local sense of community would be soon abolished, or at least heavily negated) then due to other factors, the system would STILL shit on the Tories. This is obviously partly their fault, but also the fault of the system.

I'm sure you'd agree that Labour voters should not be deemed more important than Tory voters because of how their fellow peeps are spread across the country.

Sorry I won't be able to respond any more today - this is a quick missive from home, office is a bomb-site. Back Monday.
# posted by  Paul Davies @ 10:24

 

Nik, you've missed one crucial factor out of your logic... How do you know what they prefer unless you give them the option of voting for an AB coalition government versus an A government? The majority might prefer the coalition?

I didn't bring this up (to keep things simple) but the logic can be extended just fine. I'd first note that, although the majority might prefer a coalition, PR has no way of forcing a particularly parties to form particularly coalitions. The voters may want a Lib-Lab coalitions, but if Lib and Lab don't, the voters aren't going to get it.

I extend the preferences, above, so that voters can vote for coalitions:

1/3rd of Voters: A>AB>B>ABC>AB>BC>C
1/3rd of Voters: B>AB>A>ABC>BC>AC>C
1/3rd of Voters: C>AC>A>ABC>BC>AB>B

A is still the Condorcet Winner. A majority prefer A to AB and to any other possible government. I'd note that in this example the only government people want worse than BC, is C. But BC may well form the government.
# posted by  nik @ 14:14

 

Nik: in reality, though, do you think an election is ever going to break like that, or even close? Isn't it far more likely that the largest coalitionable party (i.e. not Communists or far-right) will get to form a broadly representative government with a suitable partner, and that their programs will meld in a way thought fairly acceptable to their voters?

Don't get me wrong, Condorcet is great. But I'm not sure it provides much real-world advantage over other fairly proportional preference voting systems like STV. It also remains silent on how we might convert a national vote to national representation on a macro (region/country) basis, not just micro (each constituency).
# posted by  Jarndyce @ 14:31

 

Nik, you have presented a conveniently worked unlikely hypothetical, but even the 'Condorcet winner' method can break down.

"A Condorcet winner may not exist, due to a fundamental paradox: It is possible for the electorate to prefer A over B, B over C, and C over A simultaneously. This is called a majority rule cycle, and it must be resolved by some other mechanism."

This is a quote from wikipedia.

As Jarndyce shows, the FPTP system doesn't even fulfil Condorcet criteria anyway because of the impossibility of fitting boundaries to voter preference geography.

What FPTP leaves you with, is minority rule determined by very dysfunctional quirks of the system, that work against the very principles of democracy, i.e. that votes should be of 'equal value'.
# posted by  Neil Harding @ 02:51

 

"I don't think the changes being imposed are that good, to be honest."

Would you prefer a half Isle-of-Wight plus a chunk of Southampton constituency?

And presumably you read the boundary bit (i.e. the Avon and Northamptonshire examples etc). Boundaries make bugger all difference in the grand scheme of things, even when they are the same size.
# posted by  Paul Davies @ 11:30

 

"I don't think the changes being imposed are that good, to be honest."

Would you prefer a half Isle-of-Wight plus a chunk of Southampton constituency?

And presumably you read the boundary bit (i.e. the Avon and Northamptonshire examples etc). Boundaries make bugger all difference in the grand scheme of things, even when they are the same size.
# posted by  Paul Davies @ 11:31

 

Would you prefer a half Isle-of-Wight plus a chunk of Southampton constituency?

Well, yes. Why not?

Boundaries make bugger all difference in the grand scheme of things, even when they are the same size.

Well, a small difference, but the rest of your analysis seems to boil down to:

Monopoly isn't fair to the Tories. Let's play Scrabble instead.

I'd rather get good at Monopoly. I like being the boot, amongst other things.
# posted by  Andrew @ 15:16

 

When asked about it, the good people of the IOW didn't agree with you. Damn democracy.

And politics is never going to be as classy as scrabble. Or as cool as monopoly. :)
# posted by  Paul Davies @ 15:40

 

When asked about it, the good people of the IOW didn't agree with you. Damn democracy.

Under what system of voting?
# posted by  Andrew @ 16:52

 

Can't remember to be honest, may just have been a poll, something that came up in conversation once...

Personally, as soon as you start to erode the biggest natural boundaries, you erode them all, and you simultaneously erode that precious FPTP version of the constit link
# posted by  Paul Davies @ 19:48

 

Personally, as soon as you start to erode the biggest natural boundaries, you erode them all, and you simultaneously erode that precious FPTP version of the constit link.

Erosion is better than destruction, but I don't accept this is true. An MP can work for a constituency as long as it is reasonably sensibly shaped. That doesn't have to conform to natural geographic features.

Under PR, how would the 1 UKIP MP and the 1 Green MP and the 1 BNP MP serve their constituents?
# posted by  Andrew @ 09:30