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

[Michael Ossipoff]

Juho:

You'd said:

>>> I meant regional proportionality (not political proportionality), i.e. making the number of representatives per population equal in each area.
>>>
>>
>> In that case, then single member districts, drawn with equal
>> population, can have perfectly equal regional proportionality, as
>> exactly equal as desired. So your statement quoted at the top of this
>> post is incorrect. Equal district representation per person.
>
> In practice there will be some border drawing problems and some rounding errors. It is not easy to draw the border line e.g. in the middle of a twin bed > to make the districts equal in size :-).

Irrelevant evasion. 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. I've been
told that, in this country, districting involves "blocks" of 100
persons (or thereabout), for purposes of census privacy. Though I
don't understand whys that's needed, even that doesn't make
single-member redistricting have as great differences in district
representation per person as fixed district PR seat allocation would
have.

That said, I'll repeat that I of course fixed districts, and PR seat
allocation to them, make sense as the best solution when people want
to keep the historical districts.

And I don't criticize your preference for keeping Largest-Remainder,
for reasons of historical tradition, or even for the convenience of
not advocating, proposing and enacting a different PR allocation
method (Sainte-Lague) for districts.


You continued:

> Also if we assume that there are N seats and the population is N+1...

If so, then all but one of the districts contain only one person. And
the other district contains two people. That would indeed greatly
violate equal district representation per person. But, actually, our
districts are somewhat larger than that :-)

You continued:

> , one of the single-member districts will have only half of the "representation density" of all other districts.

Yes indeed. But, as I explained above, our single-member districts
don't contain only one or two people.

You continued:

But if we divide the same country in two  multi-member districts
(whose sizes could be just approximately similar) we will have better
regional proportionality (representation density is close to 1 seat
per 1 person in both districts).

[endquote]

Fine.But:

1) Our single member districts aren't as small as you think they are.
As of 2008, our House of Representatives Congressional districts each
contained about 699, 000 people.

2) The matter is irrelevant for another reason: Changing to
multimember districts would be a quite radical and unpopular
change--Just as changing your distsricts to automated redistricting
would be. Especially if the only justification you can come up with is
that
some districts have 699,000 and others have 699,001 :-)

In any case, we have much more important, and much more do-able,
considerations than changing to multimember districts. I'm referring
to the need to fix Plurality's forced-falsification problem, a reform
which is known as Approval Voting.

By the way, the District of Columbia has a representative in Congress,
with a population of of only about 592,000 in 2008. So you might say
they're over-represented. But their representative isn't allowed to
vote in Congress. So actually their voting representation is zero. But
that isn't a problem of the district system, it's just a problem of
how the District of Columbia is treated.

Mike Ossipoff