[EM] To Condorcetists:

Juho Laatu juho4880 at yahoo.co.uk
Thu May 17 02:18:30 PDT 2012


On 17.5.2012, at 0.41, Michael Ossipoff wrote:

> I liked Finland's elegant open list system when I read about it. 
> 
> But didn't I read that you use d'Hondt? That under-represents small parties.
> Sainte-Lague is more perfectly proportional and more fair.

Yes, Finland uses D'Hondt (and D'Hondt favours large parties when allocating the last fractional seats). Finland also allocates seats independently in each district. That actually favours large parties more than D'Hondt does. As you know having district size of 1 is quite radical from this point of view. In Finland district sizes are from 6 to 34, but still they they favour large parties. There was a reform proposal that would have counted the proportions at national level, but current government decided not to drive that proposal (that was already once approved earlier) forward.

> As you said
> small districts are ok if there's a "mixed-member system" in which national
> proportional results are used to top-up the parties' district seat totals.

You can do that also without a mixed-member system. The Finnish reform proposal first counted the proportions at national level and then forced all the districts to make their seat allocations so that the end result was in line with the agreed proportions.

> And anyway, as I said, I don't want
> parties that aren't good enough to win in single-winner elections to have
> Congressional seats.

Ok. But note that this approach allows minor parties whose supporters live in few hot spots to get seats, while parties of the same size but with even distribution of voters will not get any seats.

> How could using Approval instead of Plurality in our single-member districts
> be bad? I've talked about how Approval's results would differ from those of
> Plurality.

Proportional representation and two-party systems are two well known approaches. Approval with single winner districts is a new kind of a system, and that may bring surpises (I wrote about them before the referenced line). Also Approval method itself is not free of problems (my key concern is its strategic problems when there are more than two potential winners).

> You say that hasn't been discussed enough?
> Ok, shall we discuss the properties of the political system that would
> result from choosing what people actually like, when voters are free to
> indicate all the candidates that they like? How would it differ from now?

> If you're suggesting that there would be some drawback, disadvantage or bad
> result that could happen because we elect candidates and parties that are
> more liked than what Plurality elects, then please let's hear them.

I have now understood that your ideal (or actually best reachable) target system is a system that elects from few large parties, where few > 2. Technically multi-winner elections would use single-winner districts and Approval. Also the president could be elected with Approval.

At some point I thought that you might aim at electing good individuals without strong party affiliations, but maybe you are more party oriented that that. I assume that you expect most candidates to have a strong party affiliation.

One topic that may need further discussion is the dynamic behaviour of the proposed method. You seemd to assume that the method would converge towards electing candidates from few well known major parties. Could be but I'm not sure. People could also bullet vote (especially the old party supporters), and the old parties could still dominate (although less than before). There would be no alternating power balance anymore, which could mean that people could feel that they can not change the policy however they vote. Would the governments be minority governments or coalition governments? I mean that there could be need for further reforms. The problems of Approval with three or more potential winners might irritate people and change their voting behaviour. I'm not sure what would happen, but I expect this system to be at least in the beginning less predictable than the old well tested approaches.

I'm not saying that this system should not be tried. I'm just saying that you might get surprises too, and that the reform process might continue (or return back as in Burlington).

> you haven't talked with American
> favorite-buriers, and observed their voting, as I have.

I didn't observe any strong burial tendency in Burlington when I analyzed those votes. Normal voters do not know what FBC means, so I'd expect some burying to be present in Burlington if people have strong tendency to do so. My guess is that ranked votes of Condorcet elections would no be radically different. Maybe some activists would mention the theoretical strategic opportunities, but still I believe most voters would just rank as they would rank in IRV. Bullet voting is probably a more common deviation from sincere ranking than burial is.

> the C/D examples that I've given--my
> versions of the Approval bad-example (ABE).

Could you dig the key example up? I need a concrete set of votes since my claim was that in most situations the strategies don't work in real life as well as they seem to work on paper (under the contorol of one single stratgegist with ability change and not change the votes as he wishes).

> Ok, that sounds like a group of village elders (with no need to have party
> affiliations). That may be a working approach for some needs.
> 
> [endquote]
> 
> I don't understand what you mean by that.

I mean that since the system is not proportional and Condorcet and Approcal tend to elect centrists, the end result might be that the nicest person that has centrist opinions will be elected in each district, i.e. one that both Republicans and Democrats and also others can "approve". There would not necessarily be few parties that would alternate in power but rather a continuum of nice guys that can represent the viewpoints and can be trusted by all the parties. But now my guess is that you like more (and aim at) the alternating few parties approach (at district level).

> 2.  It isn't clear what you mean by "small". How "small" is a party that is
> considered alright by the most people, and given an "Approved" rating by the
> most people? A party that is hopefully-regarded by the most voters.

I talk about the small parties because U.S. people are used to being ruled by a party that had 50+% support in the last election. In sincere Approval the winner tends to be widely approved. In strategic Approval people might approve also lesser evils. A small party between Democrats and Republicans could get lots of approvals since it is consirered better than the "worst alternative". In any case, if you don't have coalition governments or other correspondn arrangements, parties with considerably less than 50% support would rule (or try to rule).

> 3. Or maybe you're saying that Approval would be worse than Plurality
> because it wouldn't elect the most favorite party, as you think that
> Plurality does. But Plurality doesn't elect the most favorite party, because
> millions of voters aren't voting for their favorite in Plurality.

In many of my comments I say that Approval + single-winner districts to elect representative bodies is an _unknown_ and therefore somewhat unpredictable method. I expect Approval to have also the typical Approval problems (and Plurality to have its own).

> Ok, but now I realize that your objection is against single-member districts
> in general, rather than against Approval in particular.

No objection, just observations on their properties. It would be an interesting experiment to try single-member districts with Condorcet or Approval (i.e. a compromise seeking single-winner method) to elect a representative body. I'm not ready to recommend any such methods to anyone since I'm not sure that they would converge towards some stable and well working state. Condorcet could be more interesting than Approval since I expect Approval to be more problemating from voting strategy point of view. I'd be happy if someone tried them.

> Yes, single-member districts could elect all of Congress from one party.
> That would be fine, if it's the most liked party (as it would be in
> Approval).

Sincere Approval would elect the most approved (=liked?) candidate, but I expect voters to be more strategic than that.

> What is this "alternating 2-party principle" that you advocate, and how do
> you justify it?

It is just one of the working (or semi-working) existing alternatives. It can be justified as a political system that alternates around the median opinion. But I don't advocate it, although it can be said to be one working form of democracy.

> parties chosen by corporate media and big money contributors

I'd say this problem is for the most part independent of the used election method.

> Are you saying that things
> In Washington are already set up for the Republicans and Democrats?

Yes, especially in the sense that after the presidential elections one 50+% party is supposed to take control of very many things. Naturally the system has adapted to the two-party structure during its long history.

> Such as the bribe arrangements?

Bribes are a different story. Also other political systems can have such problems.

> And you're afraid that we'd lose that, if we didn't elect the Democrats and Republicans?

I'm just saying that some redesign may be needed if you will give up the 50+% tradition.

> European countries likewise have to form new governments when the
> parliamentary coalition changes.

That's business as usual. I was concerned about the changes in the dynamics of the U.S. system that you proposed.

> On the contrary, as I said before, in Condorcet, if it's a u/a election
> (meaning that there are unacceptable candidates who could win), and if it
> appears as if Compromise is the only acceptable who can beat the
> unacceptable, then your best strategy is to bury your favorite by ranking
> Compromise alone at top.

Could you point out a concrete example with sincere preferences and strategic votes, so I can check if it is likely to work in a real life election. (The one example that I analyzed in the other mail didn't seem to be a threat in real life elections.)

Juho






More information about the Election-Methods mailing list