[EM] Gerrymandering and PR

[Adam Tarr]

I wrote and Alex responded:

> >if you have small districts you don't get real proportionality. In my
> >opinion, you have to have at least 5 or 6 seats in a district to get
> >acceptably proportional results. The more fractionalized the electorate
> >is, the more seats per district you need.
>
>I'm not interested in representing every portion of the electorate.  If the
>Communist is polling 1%, and the Transcendental Meditation candidate has
>2%, and the Fascist is getting 3%, well, sucks to be them.  On the other
>hand, factions with support in the ballpark of 15% or more are more worthy
>of representation.

My concern is not so much that the marginal parties get shoved out (I agree 
with you that this is not a big deal) but more that the represented parties 
might not get the right number of seats.  The smaller the districts, the 
greater the error between pure proportionality and the actual allocation in 
each district, and the greater the chance that the final legislature will 
hand a majority to the wrong party.  For a good example of this type of 
effect, see the electoral college the 2000 presidential election.

>If we set
>a quota of 16.6% (1 in 6, not unreasonable) then with a 1/(n+1) quota a
>district of 5 members works.  Maybe 6, 7, or 8 members is better.  But I
>don't see the need for going above that.

In methods other than STV, the quota is not a set number like that.  But 
your central point is valid; around 7 or 8 seats seems optimal to me 
too.  A good balance between proportionality and all the other concerns.

> >Ideally, I think that there should be no single-member districts for
> >legislative appointments. The disadvantages seem to outweigh the
> >advantages.
>
>I disagree.  I think the CA Senate should have PR to make sure minority
>views are represented in the more powerful body.  However, the smaller
>Assembly districts allow more attention to local concerns.  In Santa
>Barbara, where I live, most people, regardless of ideology, have certain
>environmental concerns.  In a multi-winner race, those local concerns may
>be overlooked in favor of ideological issues appealing to people across the
>entire (large) district.

It's true that they may be overlooked.  But in a seven-seat district, it 
seems more likely that candidates will still attempt to tap into an issue 
that is a make-or-break topic for 20% of the electorate.  This is 
especially true in a system like PAV where candidates from more than one 
party could tap into support from that area by addressing their specific needs.

By contrast, imagine that there are single-winner districts, but Santa 
Barbara has been split up among two or three districts, and the Santa 
Barbara element is a minority in each?  Now you could get zero 
representatives with your slant in stead of one or two.  You're counting on 
not getting disenfranchised by Gerrymandering.  Smart rules (which don't 
currently exist, mind you) can mitigate these effects, but they will never 
completely get rid of them in single winner districts.

-Adam