[EM] IIAC. Juho: Census re-districting instead of PR for allocating seats to districts.

[Juho Laatu]

On 15.6.2012, at 1.41, Michael Ossipoff wrote:

> On Thu, Jun 14, 2012 at 5:37 PM, Juho Laatu <juho4880 at yahoo.co.uk> wrote:
>> On 14.6.2012, at 23.45, Michael Ossipoff wrote:
>> 
>>> If the district's population is off by one person,
>>> that's nothing compared to the amount by which even the best PR system
>>> will put it off, when allocating seats to fixed districts.
>> 
>> Could you give me an example (or a formula or some other exact definition on what goes wrong). Do you compare adjustable districts to fixed districts or single-member districts to multi-member districts?
> 
> I was comparing adjustable districts to fixed districts.

That explains it. PR is not bound to have fixed districts. Even if a multi-member district based (PR or other) system uses fixed districts, there is a balance point where its regional proportionality is at the same level with single-member adjustable districts. That balance point can be calculated if we know the sizes of the multi-member districts and the accuracy of size of the single-member districts.

> No one claims that PR gives exactly the same district representation
> per person, in each district.  ...Or exactly the same party
> representation per person to each party. That's common knowledge.

You can count (the ideal discrete) party proportionality at national level (and follow that when you allocate the seats in the districts).

> But, for exactly equal district
> represent ion per person, you'd need census-based re-districting,
> preferably by an automated un-gerrymanderable method.

Yes, you can get the best accuracy by using adjustable districts. Also large multi-member districts help. Gerrymandering related problems generally decrease when the number of members per distric grows (note however also the Sainte-Laguë example at the end of this mail).

> If you're bothered by that much difference in district representation
> per person, then don't you find it a little odd that you are ok with
> allocating seats to fixed districts by Largest-Remainder instead
> Sainte-Lague. Where is your passion for precision now?

I'm not bothered about the small differences between different methods. Different methods have different properties. Largest Remainder optimizes one criterion, but that need not be the right criterion for all needs.

> With Largest-Remainder, we're talking about some much bigger
> differences in district representation per person, especially among
> small districts.

Can you give an example where there is a difference between those methods? You talk about "differences in district representation per person". Does that refer to the Alabama paradox that causes movements up and down when the number of seats changes, or do you refer to changes in population/voters or in number of districts?

If you talked about fair allocation of seats to small districts, here's one possible example. Largest Remainder gives the first seat (out of 10 seats) to the second group with proportions 93-4-3. In Sainte-Laguë we need 93-5-2. What is your definition of ideal fair allocation? Should small parties be favoured more or less?

Is Sainte-Laguë's seat allocation 1-1-1 with proportions 57-23-20 ok? Note that this introduces an incentive to split a minority district (43) in two parts (23-20) to get majority of the seats. (may lead to gerrymandering)

Juho