[EM] voter strategy & 2-party domination under IRV voting

[Juho Laatu]

Hello Rob, Dave and All,

On Aug 14, 2005, at 03:20, Rob Lanphier wrote:

> On Sat, 2005-08-13 at 18:48 -0400, Dave Ketchum wrote:
>> NOT at all clear that 2-party domination is as evil as some claim.
>
> This is a really good point to consider.  We probably need to discuss
> the specific characteristics of the two-party system we rail about in
> order to judge if a system is bad or good.

Yes.

Few random observations follow.

Two-party systems are democracies too, so I think it is a matter of 
taste if one wants to change them to something else (maybe to 
multiparty systems, maybe not to one-party systems ;-).

My philosophic view to two-party democracy is that the border line 
between the two parties is supposed to move when the opinions of the 
voters change. If party A wins several elections in sequence, party B 
must change its opinions closer to party A (or closer to the median 
voter). In a multiparty system parties may be more static, but the 
tendency to seek voters from the centre and to fish votes from the 
territories of other parties exists there too.

If we describe the opinions voters as linear values from 0 to 100, 
party A could be the "home" for voters with opinions from 0 to 49. 
Party B would then be the "home" of voters with opinions from 51 to 
100. An efficient strategy for party A when preparing for elections is 
to pretty much forget voters with opinions 0-40 since they will vote 
for party A in any case. If one wants to get more supporters from the 
ranks of party B one should offer something for them to. It may turn 
out that the best strategy for party A is to express opinions whose 
value is even higher than 50. Let's say 52. Voters know that these 
opinions are coming from a "0-49 party" and can see those 0-49 opinions 
behind the rhetoric. In a similar way party B could express opinions of 
value 48.

According to this viewpoint (which is not the only correct one) parties 
are thus not trying to split the world in two _separate_ camps but are 
trying to please the centrist voters and trying to play with the median 
voter border between the parties. The language of human beings often 
classifies the opinions using two extreme words like "conservative" and 
"liberal" but the actual distribution of opinions may often not be 
split in two separate camps (there may be a lot of voters in the 
middle, balancing between the slightly more or less 
conservative/liberal viewpoints).

Btw, after the last US presidential elections I heard comments that the 
Democrats forgot the conservatives of the midwest states (and thereby 
maybe lost touch to the median Americans of today) (9/11 surely has 
moved the median line, and Democrats may have missed that trend) (this 
is just theoretical speculation, so don't take this as a statement on 
the political situation/trends is US, just as an example).

In a multiparty system the same rules apply to the extreme parties 
(e.g. 0-10 and 90-100). They are likely to have better success if they 
express opinions that are close to or on the territory of one step more 
moderate parties. Parties closer to the middle may better seek new 
voters (and keep the old ones) in all political directions and 
therefore they do not have that big tendency to please only the 
"centrists". To some extent the biggest parties are competing between 
themselves fir the souls of "safety seeking big party voters", which 
makes them go for centrist values anyway (also in multiparty systems).

In theory in multiparty systems parties may however be more static and 
people may flow between them while in two-party systems the parties are 
forced to move when the median of the voters moves. Note that the 
political dimensions in a multiparty system may be more complex than 
the linear (one-dimensional) structure that I described above. Greens 
for example do not fit very well in the traditional 
liberal-conservative or left-right axis.

Now, how is this discussion about political systems linked to election 
methods. I think it easy to see multiparty and two-party systems as two 
separate kind of democratic systems. For a large part these systems are 
created by the election methods in use (typically e.g. use of one 
member districts leads to a two-party system). Trying to introduce 
additional candidates like Nader in a two-party system/elections is 
typically not successful in the sense that Nader would be elected. 
Maybe having Nader as a candidate and voting him is rather a protest 
against the dominance of the two major parties (and maybe a vote for 
changing the system one day). It seems that many of the discussions on 
this mailing list are directed towards trying to change the two-party 
system to something else one day (US and UK at least). But also 
improving the two-party systems (without breaking the two-party 
tradition) would be a valid topic of discussion. Trying to analyse 
cases that involve parts of both two-party and multiparty systems is 
interesting but maybe not the clearest area of study. Studying also 
pure two-party and multiparty systems would certainly provide some 
support to these mixed system and "system evolution" discussions too.

As a person coming from a multiparty country that has open party lists 
in the parliamentary elections multiparty looks like a natural way of 
organising the political arena. I remember being surprised when I (long 
time ago) learned that some countries have closed party lists and some 
have a two-party system (and they still call themselves democracies ;-) 
). Since then I have learned to respect also the good sides of 
two-party systems like stable governments and ability to drive clear 
policies. Living with two parties only may be boring when one learns to 
know the parties and their operation "too well" :-), but also that 
system works as we can see.

In the EM mailing list there also seems to be a lot of interest in 
systems that are "party-free", like STV. In addition to all these I'm 
interested in developing also the party based voting methods further. 
=> I'm hoping to see lots of discussion on these topics on this mailing 
list. I guess we are not limited to the single-winner discussions only.

> Regardless, I have yet to see a convincing case that Condorcet methods
> naturally lead to two-party duopoly.

Yes. I think the Condorcet method could rather be called an attempt to 
allow also other than main party candidates win if the voters like 
them.

BR, Juho