[EM] Re: STV-PR

[Dr.Ernie Prabhakar]

Hi James et al,

Thanks for the input; I've never considered PR seriously before, so I'm  
just starting to get the "Zen" of it.

On Aug 25, 2004, at 10:47 AM, James Gilmour wrote:

> Dr.Ernie Prabhakar  > Sent: Wednesday, August 25, 2004 4:30 PM
> Why have you got a district magnitude of 20?  If you are using (any  
> system of) PR why are you
> interested or concerned about "sub-districts"?  If you are looking at  
> an existing situation where
> you want to change from single-member districts to multi-member  
> districts for PR, reformers like me
> always advocate grouping the new multi-member districts around  
> "natural communities".  The scale of
> such communities will vary according to the scale of the whole  
> exercise.

Ah, okay, I think I'm getting it.   So, how the heck does one define  
natural communities in any sort of objective manner?   City boundaries?  
  Commute flows?  Geography?  And, isn't there some minimal district  
size where PR loses its value?  And maximum size where voters get  
overwhelmed? The optimal size Assembly is purportedly 300, which would  
mean 100 districts of average magnitude 3.

I'm particularly interested in the State of California.  One proposal  
that has been suggested is to create 'supercounties' that reflect major  
transportation grids boundaries, e.g., Los Angeles (six southern  
counties), the Bay Area (nine SF counties), central valley (Fresno &  
Yosemite), and upstate (Redding etc.).

Would the idea be to do PR in each of those four regions?  Is that  
still comprehensible to mere mortals, even if, say, the L.A. region has  
150 seats to be filled (more than the total entrants to our recall!)?  
Or do we draw the line at a certain size?  Does it matter at all if the  
different regions have different populations? (is that Tullymandering?)

> If locality matters to the voters, they will take this into account  
> when they mark their
> preferences.  So you will get PR of parties and PR of locality, both  
> as determined by the voters.

Ah, now I think I get the locality issue.   People can choose to make  
their top ranking based on locality or party, whichever is more  
important (in whatever combinations).  Thanks.  So, within a district  
locality is preserved by voter action.  But that still leaves the  
question of how to optimally draw districts.

> With all PR systems there is a trade-off between degree of PR obtained  
> and localness of
> representation.  To increase PR, increase district magnitude.  To  
> increase localness, reduce
> district magnitude.  Different people will have different views about  
> where the balance should be
> struck in making that trade-off.  Remember also that the law of  
> diminishing returns applies: see:
> http://homepages.phonecoop.coop/James.Gilmour/Representation_in_multi- 
> member_constituencies.gif

Interesting. So 20 actually is an interesting number, if one cares  
about 95% confidence levels. :-)   Adam's suggestion of seven (a nice  
number for cognitive scientists) puts us at around 90%, right?

> For discussion of these issues in a real political reform context, see:
> http://www.fairsharevoting.org/Implementing%20STV-PR%20- 
> %20Paper%20for%20SP%20LGC%201%20Dec%2002.pdf

Cool, thanks.  I note that they seem to have settled on flexible  
districts of size 4-10 or so, similar to Adam's recommendation.

>> P.S.  Does STV-PR actually allow voting for predefined lists > as  
>> well as
>> individual candidates?  I had trouble understanding the existing >  
>> articles on this point.
>
> Yes, this is allowed in the perverted version of STV-PR used in  
> Australia for Federal elections.
> The effect of "ticket voting" is such that STV-PR is reduced to  
> nothing more than a fancy version of
> party list PR.  This runs counter to the whole concept behind STV, ie  
> that the system is centred on
> the voters and not centred on the parties.

Um, I'm confused.   I understand that -requiring- ticket voting would  
be perverse.  But, I can easily imagine a voter wanting the  
simplification of saying, "I want Joe first, then any Green, then any  
Democrat."  Is that possible with "non-perverted" STV-PR?

On Aug 25, 2004, at 10:57 AM, Markus Schulze wrote:
> Here is my suggestion:
> http://lists.electorama.com/pipermail/election-methods-electorama.com/ 
> 2002-December/009102.html

I appreciate the flexibility, but I'm having hard time imagining the  
user interface.   I guess I see it more as a system for ranking  
individual candidates, but with a 'quick fill' button for each party  
that automatically ranks any unassigned candidates from that party  
below your existing picks.

Has anyone implemented/experimented with a UI for this?

-- Ernie P.