[EM] Goodbye Gerrymandering.

Curt Siffert siffert at museworld.com
Fri May 13 18:56:37 PDT 2005

I wonder about the agenda of this kind of partitioning scheme, though.

It seems to be that before partitioning scholars suggest partitioning 
strategies, they must agree on their objective.

Here in the US, a commonly held belief is that a fair district 
partioning must reflect the makeup of the region's voters.  In other 
words, if a state is split 55-45, then the districts must be formed in 
such a way so that the representatives are split about 55-45 as well.

But, if the 55-45 were instead evenly distributed amongst all 
districts, you'd get all districts with a 55-45 majority, or, unanimous 
representation from the 55 group.  This would be less representative 
than the most extreme gerrymanders.

I would bet that date-based partitioning would be extremely susceptible 
to that problem.


On May 13, 2005, at 11:25 AM, <stephane.rouillon at sympatico.ca> wrote:

> The Institut des Statistiques du Quebec provided me with the 
> repartition of people according to their date of birth.
> Using this information to build non-geographical districts based on 
> date of birth, month and some modulo of the year provided the 
> following results:
> for 125 seats (as in the current Assemblee Nationale), one can 
> obtained 125 consecutive districts with a +/- 10% variation. 
> Consecutive means that persons are regrouped using successive 
> birthdays (like january 1st, 2nd and 3rd). Using non-consecutive 
> birthdays to minimize the variations, one can obtain districts with 
> the same population size within a +/- 0.1% margin.
> Of course in both case a special treatment for people born a february 
> 29th puts them with christmas born persons which are less common.
> Birth dates are only one example of how to build non-geography based 
> districts, "astrological" districts in this case.
> Stéphane Rouillon.
> ----
> Election-methods mailing list - see http://electorama.com/em for list 
> info

More information about the Election-Methods mailing list