[EM] Election layering effect (or why election-method reform is important)

[Kristofer Munsterhjelm]

On 05/03/2012 09:29 AM, Richard Fobes wrote:

> I like this analogy. It does not amplify enough, yet it prompted me to
> think of this idea:
>
> We tend to think of politics as a pyramid that has our few-in-number
> leaders at the top, and the numerous voters at the bottom who support
> the leaders through voting.
>
> In contrast, an upside-down pyramid might be more realistic. Each layer
> in the pyramid corresponds to one of the layers mentioned above. At the
> bottom are the few voters who marked on their primary-election ballot
> support for the Congressmen who voted (as part of a majority) to pass a
> new law. I'm still working out how best to draw it, yet this seems like
> a useful path to clarify the importance of election-method reform.

I first found this pattern when considering forms of council democracy. 
In these types of democracy, you have local councils that appoint 
representatives from their number to form regional councils that appoint 
representatives... and so on up.

In the worst case, a bare majority at every level can control the whole 
system. In a one-level system, a majority suffices (which is much less 
than 100%); in a two-level system, a majority of a majority; in a three 
level system, a majority of a majority of a majority and so on.

Generally, if the councils are of size n, then a majority m is 
floor(n/2) + 1. Call the fraction required to get a majority, f. f = 
m/n, and this approaches 50% as n goes to infinity.

Then in the very worst case, f^(num levels) of the total population 
suffice to control the council democracy. In a primary system, it's 
worse since only a fraction of the population can vote in any given 
primary (excepting open/jungle primaries), and not all who can vote are 
going to.

> On 4/28/2012 10:52 AM, Stéphane Rouillon wrote:
>> ...
>> With an STV election, 3 seats in a single super-district, let's
> assume ...
>> ...
>> Typically STV produces a global individual satisfaction rates around
>> twice FPTP rates for the simulations
>> I have made yet...
>> ...
>> This does not covers the layering effect of multiple representative
>> levels, but it emphasizes the mismatch
>> between the will of electors and the results.
>>
>> Stéphane Rouillon
>
> Yes, proportional methods reduce the number of wasted votes (which can
> be defined in various ways). Yet, as you say, this does not address the
> layering effect. Nevertheless, thank you for your ideas.

In the council model, proportional representation does help, so to some 
extent it does alleviate the layering effect. If a council elects three 
instead of one to the next level, then in the worst case, the faction 
has to get all of them for enough councils to get a sufficient 
supermajority on the next level, and so on.

In a primary system, I think the most clear benefit is that it dissolves 
the problem. If you have proportional representation, there's no need 
for primaries - at least not for legislative elections. Any group that 
disagrees with the party can simply leave that party to form a party of 
its own.