A computationally feasible method (algorithmic redistricting)

[greg]

Brian Olson suggests this approach for his anti-gerrymandering proposals.
http://bolson.org/dist/USIRA.html
and
http://bolson.org/dist/

I suggested a similar mathematical method for drawing Congressional
districts a few years ago. I like Brian's suggestion - and
particularly like his side-by-side state district maps, so I modified
See http://electionmathematics.org/ and click on "Drawing Districts".

I've added this link to my collection of algorithmic redistricting proposals.

I've been collecting references to algorithmic redistricting proposals
on my site on a post related to comments by the 'Rose Report' on the topic.

See:
http://allaboutvoting.com/2007/12/01/the-rose-report-on-algorithmic-redistricting/
and
http://allaboutvoting.com/category/algorithmic_redistricting/

-Greg Wolfe
http://AllAboutVoting.com
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Michael Rouse]

There was a discussion of district-drawing algorithms on the
election-methods list a few years back. I've always thought that taking
centroidal Voronoi cells with equal populations was an elegant way to do
it. Here's an example of standard Voronoi cells and the centroidal
version I pulled off of Google:
http://www.mrl.nyu.edu/~ajsecord/npar2002/html/stipples-node2.html

The other possibility I liked was allowing voters to vote for the
districts they wanted -- either for the next election, or more
entertainingly, the current one. People have a pretty good feel for what
mapping is compact and reasonable, and which ones are ridiculous,
especially if they can compare them. You could have certain criteria
that must be met -- like all districts must be contiguous -- and sort
the maps by some metric, like from shortest to longest aggregate
perimeter. You could have all qualifying parties submit a map, as well
as any group that gets above a certain number of signatures in a petition.

Michael Rouse

(The last time I posted, things like line breaks disappeared, so I
apologize if this happens again.)

Post by greg

Brian Olson suggests this approach for his anti-gerrymandering proposals.
http://bolson.org/dist/USIRA.html
and
http://bolson.org/dist/

I suggested a similar mathematical method for drawing Congressional
districts a few years ago. I like Brian's suggestion - and
particularly like his side-by-side state district maps, so I modified
See http://electionmathematics.org/ and click on "Drawing Districts".

I've added this link to my collection of algorithmic redistricting proposals.
I've been collecting references to algorithmic redistricting proposals
on my site on a post related to comments by the 'Rose Report' on the topic.
http://allaboutvoting.com/2007/12/01/the-rose-report-on-algorithmic-redistricting/
and
http://allaboutvoting.com/category/algorithmic_redistricting/
-Greg Wolfe
http://AllAboutVoting.com
----
Election-Methods mailing list - see http://electorama.com/em for list info

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

Post by Michael Rouse
There was a discussion of district-drawing algorithms on the
election-methods list a few years back. I've always thought that taking
centroidal Voronoi cells with equal populations was an elegant way to do it.
Here's an example of standard Voronoi cells and the centroidal version I
http://www.mrl.nyu.edu/~ajsecord/npar2002/html/stipples-node2.html

Is there an algorithm to take a discrete set of points (residents) and
find the set of district centres?

Post by Michael Rouse
The other possibility I liked was allowing voters to vote for the districts
they wanted -- either for the next election, or more entertainingly, the
current one.

It might be possible to achieve moderate PR with this method.

In principle, if each voter could decide what district he was in, no
voter would join a district where he was part of the weak minority.
Likewise, there is little point in joining a district where you are
part of a strong majority.

The effect should be competitive districts and then a few 'overflow'
ones which the party with a majority wins.

I wonder would non-geographic districts be allowed by the
constitution. Each voter would register for a numbered district and
could vote at any polling station. Once registered, a voter wouldn't
be allowed to change registeration unless moving to a district with a
lower population.

The polling stations would then announce the counts for each district,
which would be tallied centrally.

This might allow perfect gerrymandering though. If you had 55% of the
State population, your supporters might spread out evenly. Tactics
might be interesting, a minority might try to flood a district, so
that they get a majority and then the district can take no more new
people.

Even with perfect gerrymandering for the majority, what it would
prevent is a minority party using gerrymandering to maintain its
majority in the legislature (possible in theory even if they are only
supported by 26% of the populaton).

Also, it could be somewhat considered to break the secret ballot, as
part of your vote is in effect picking what district to stand in.

Another option is that each voter picks a party, and the algorithm
tries to create districts that will give PR (assuming that the party
support of the voters who just voted reflect actual party support).
Ofc, it would probably not be able to give any seats to a 3rd party.

They would have to focus on a specific area and build up their
popularity there. However, once they did that, they would get a
district centered on that location for as long as they managed 1 seats
worth of votes statewide and maintained sufficient support at that
location.

Post by Michael Rouse
You could have all qualifying parties submit
a map, as well as any group that gets above a certain number of signatures
in a petition.

That's another perhaps easier option. It is less likely that the
algorithm would go crazy.

I wonder if the voters would gerrymander. The voters might
effectively be given the option of a Republican or a Democrat
controlled House and then the actual election would just be a
rubberstamp.

Ofc, it is also possible that both parties would agree on a compromise
house that doesn't favour either party.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Kristofer Munsterhjelm]

Post by Michael Rouse
There was a discussion of district-drawing algorithms on the
election-methods list a few years back. I've always thought that taking
centroidal Voronoi cells with equal populations was an elegant way to do
it. Here's an example of standard Voronoi cells and the centroidal
http://www.mrl.nyu.edu/~ajsecord/npar2002/html/stipples-node2.html

To find the district centers (centroids), you have to do what's
effectively vector quantization. The voters make up the points, and you
want to choose n codebook points so that the average distance to the
closest codebook point, for all points, is minimized.

To my knowledge, optimal vector quantization is NP-hard. The good news
is that there are approximation methods that have proven worst-case time
complexity. However, they'll not give you the absolutely best possible
arrangement.

One such algorithm is described here:
http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.17.3529

A simpler algorithm, though one that doesn't give any worst-case bounds,
is Lloyd's algorithm. Start with random center points, then calculate
the centroids for the current Voronoi cells. Move the center points
towards the centroid (shifting the Voronoi cell somewhat) and repeat. Do
so until they settle. This can get stuck in a local optimum.

Post by Michael Rouse
The other possibility I liked was allowing voters to vote for the
districts they wanted -- either for the next election, or more
entertainingly, the current one. People have a pretty good feel for what
mapping is compact and reasonable, and which ones are ridiculous,
especially if they can compare them. You could have certain criteria
that must be met -- like all districts must be contiguous -- and sort
the maps by some metric, like from shortest to longest aggregate
perimeter. You could have all qualifying parties submit a map, as well
as any group that gets above a certain number of signatures in a petition.

Those maps could be pruned so that only the Pareto front remains. That
is, if there's some map that's worse on all metrics with regards to some
other map, then that first map isn't included. As long as there are
enough metrics to give a reasonable choice on the Pareto front, this
should exclude the worst of the gerrymandered proposals and keep the
voters from being swamped with millions of frivolous proposals.

I don't think it's necessary to make it that complex, though. If you
favor actual people doing the final choice, an independent commission
(like the redistribution commissions of Canada and Australia) could make
the choice of which nondominated map to use.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

Those maps could be pruned so that only the Pareto front remains. That is,
if there's some map that's worse on all metrics with regards to some other
map, then that first map isn't included. As long as there are enough metrics
to give a reasonable choice on the Pareto front, this should exclude the
worst of the gerrymandered proposals and keep the voters from being swamped
with millions of frivolous proposals.

Maybe, the best way would be to include the best maps via lots of
different measures.

I don't think it's necessary to make it that complex, though. If you favor
actual people doing the final choice, an independent commission (like the
redistribution commissions of Canada and Australia) could make the choice of
which nondominated map to use.

The real problem is that power that that commission has. If the
districting map has more influence on the result than the voters, then
there is a problem.

Multimember districts are much less effected by districting.

However, even then, disliked independents can be eliminated by drawing
a line right through their support base.

That is another possible way that voters could choose their districts.

Have each voter vote for 1 candidate. Have the lines drawn so that
they try to keep groups of voters that vote for the same candidate in
the same district.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Terry Bouricius]

I would argue that developing a genuinely neutral redistricting procedure
is of little value as long as single-seat winner-take-all election methods
are used. Although, it is easy to create redistricting plans that are
clearly UNfair, there is no such thing as a genuinely "fair" districting
plan.

Are folks on this list familiar with the Proportional Representation
Society of Australia's redistricting roulette wheel (you can find a
wonderful graphic example here http://www.fairvote.org/wheel/ )? Using
this wheel, you can draw compact, equal-population district boundaries, in
a "fair" way, and end up with any partisan balance desired.

After studying redistricting in the U.S., the Vermont League of Women
Voters adopted this position:

"The emphasis on geographic representation in legislative bodies in the
U.S. may be anachronistic. It is more important that voters be represented
by elected officials who reflect their political views, than happen to
live nearby. Single-seat winner-take-all elections, regardless of method
of redistricting, elevate the representation of geography above political
philosophy, and other priority voter self-identities.



"It is impossible to redistrict single-seat districts in such a way as to
promote BOTH competitive elections AND a highly representative delegation
(as these two priorities are in inherent conflict in single-seat
districts). Therefore,



"The League of Women Voters of Vermont supports the principle of
legislative districts using alternative voting methods, such as
proportional representation in multi-seat districts, as a way of achieving
both competitive elections and fair representation of both majorities and
minorities within a district."


-Terry Bouricius, Vermont USA


----- Original Message -----
From: "Kristofer Munsterhjelm" <km-***@broadpark.no>
To: "Michael Rouse" <***@mrouse.com>
Cc: <election-***@lists.electorama.com>
Sent: Monday, September 01, 2008 7:28 AM
Subject: Re: [EM] A computationally feasible method
(algorithmicredistricting)

Post by Michael Rouse
There was a discussion of district-drawing algorithms on the
election-methods list a few years back. I've always thought that taking
centroidal Voronoi cells with equal populations was an elegant way to do
it. Here's an example of standard Voronoi cells and the centroidal
http://www.mrl.nyu.edu/~ajsecord/npar2002/html/stipples-node2.html

To find the district centers (centroids), you have to do what's
effectively vector quantization. The voters make up the points, and you
want to choose n codebook points so that the average distance to the
closest codebook point, for all points, is minimized.

To my knowledge, optimal vector quantization is NP-hard. The good news
is that there are approximation methods that have proven worst-case time
complexity. However, they'll not give you the absolutely best possible
arrangement.

One such algorithm is described here:
http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.17.3529

A simpler algorithm, though one that doesn't give any worst-case bounds,
is Lloyd's algorithm. Start with random center points, then calculate
the centroids for the current Voronoi cells. Move the center points
towards the centroid (shifting the Voronoi cell somewhat) and repeat. Do
so until they settle. This can get stuck in a local optimum.

Post by Michael Rouse
The other possibility I liked was allowing voters to vote for the
districts they wanted -- either for the next election, or more
entertainingly, the current one. People have a pretty good feel for what
mapping is compact and reasonable, and which ones are ridiculous,
especially if they can compare them. You could have certain criteria
that must be met -- like all districts must be contiguous -- and sort
the maps by some metric, like from shortest to longest aggregate
perimeter. You could have all qualifying parties submit a map, as well
as any group that gets above a certain number of signatures in a petition.

Those maps could be pruned so that only the Pareto front remains. That
is, if there's some map that's worse on all metrics with regards to some
other map, then that first map isn't included. As long as there are
enough metrics to give a reasonable choice on the Pareto front, this
should exclude the worst of the gerrymandered proposals and keep the
voters from being swamped with millions of frivolous proposals.

I don't think it's necessary to make it that complex, though. If you
favor actual people doing the final choice, an independent commission
(like the redistribution commissions of Canada and Australia) could make
the choice of which nondominated map to use.
----
Election-Methods mailing list - see http://electorama.com/em for list info

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Brian Olson]

Post by Kristofer Munsterhjelm
A simpler algorithm, though one that doesn't give any worst-case
bounds, is Lloyd's algorithm. Start with random center points, then
calculate the centroids for the current Voronoi cells. Move the
center points towards the centroid (shifting the Voronoi cell
somewhat) and repeat. Do so until they settle. This can get stuck in
a local optimum.

I have implemented essentially that, and it pretty quickly gets pretty
good results as measured by the "distance per person to land-area-
center" test. I added one thing to deal with variations in population
density. Each district center also has an associated weight. Each
'voronoi cell'/district contains the people with a weighted distance
closest to that district center. So, a district center in a high
population density area doesn't have to reach very far to gather its
allotment of people.

There are some states where this doesn't work very well (CA, TX, NV,
and a couple others) and a more exhaustive solving method is needed. I
think one of the weaknesses of the voronoi method is that population
is not continuous, but in discrete census blocks of dozens to
thousands of people. The voronoi method can't necessarily tweak to
swap a block here and a block there to get within a reasonable margin
of equal-population.

I think it's worth repeating that I'd prefer to write the law with a
way of measuring what a good district mapping is, rather than writing
the law with a specific procedure for creating the district mapping.
Specific procedures get really complex, and I don't think in this case
that a fair procedure ensures a fair outcome. I used the voronoi
technique as just one possible way of writing a solver to get the
desired result. It could even wind up that the best way to solve for
the desired goals is to have a computer do 99% of the work and have a
human come back and clean up the edges. Since it's subject to the same
scoring in the end, the human couldn't deviate too far from the ideal
mapping.


Brian Olson
http://bolson.org/


----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

I have implemented essentially that, and it pretty quickly gets pretty good
results as measured by the "distance per person to land-area-center" test. I
added one thing to deal with variations in population density. Each district
center also has an associated weight. Each 'voronoi cell'/district contains
the people with a weighted distance closest to that district center. So, a
district center in a high population density area doesn't have to reach very
far to gather its allotment of people.

Your maps don't look like voronoi diagrams (edges aren't lines).

Do you assign based on

Assign voter to district with minimum

distance to district centre - Cd

Where Cd is some coefficient for the district d. Cd is set so as to
give equal populations.

If so, have you considered using squared distance instead of distance?

Assign voter to district with min

(distance to centre)^2 - Cd

This gives straight district edges (I think).

You might need to change your rule to

The best district map is the one where people have the lowest average
*squared* distance to the center of their district.

There are some states where this doesn't work very well (CA, TX, NV, and a
couple others) and a more exhaustive solving method is needed. I think one
of the weaknesses of the voronoi method is that population is not
continuous, but in discrete census blocks of dozens to thousands of people.
The voronoi method can't necessarily tweak to swap a block here and a block
there to get within a reasonable margin of equal-population.

When I was writing the software to implement the shortest splitline
for Rangevoting.org, my solution was to split the border block. (or
at least that was my plan ... I don't think I actually implemented
it).

This page has all the states (v1 of the software)
http://rangevoting.org/SplitLR.html

This page has the entire 'lower 48' all in one go (v2 of the software)
Loading Image...

Basically, the algorithm picks a slope to test and then sorts all the
blocks based on which line with that slope goes through them.

e.g. sort by Cn = Yn - m*Xn

It then finds the median block. This means that approx half the
population is on one side and half the population is on the other side
of the line.

(Blocks that have equal Cn are differentiated by Dn = m*Yn + Xn, this
means that there is only ever one block)

The block itself can be placed on one side or the other.

If it worked out

population above line: 649,700
population of block: 1000
population below line: 649,300

These blocks would be split into 2 sub-blocks of population 300 and
700. These blocks would still be considered to occur at the same
point.

The 300 would be assigned to the above the line district and the 700
to the below the line district.

The result from the algorithm could be

District A:
<list of blocks that are 100% in district>
partial blocks
300 of 1000 from tract X
100 of 800 from tract Y
150 of 1050 from tract Z

Unless the blocks are massive, there should only be a few partial
blocks, accounting for only a small fraction of the district's
population. The boundaries could be drawn by a judge or other group
by hand.

If you really wanted to be strict, they could be automatically split,
but I don't think it is worth the effort. They would be unlikely to
make much difference and thus it wouldn't be worth the effort
corrupting them.

I also added a rule which auto-split blocks prior to starting the
algorithm, based on a limit. For example, 650k blocks might be
converted into 750k by splitting the largest blocks.

If the limit was 500, then a 2250 block would be auto split into 5
blocks, 500, 500, 500, 500, 250.

I think it's worth repeating that I'd prefer to write the law with a way of
measuring what a good district mapping is, rather than writing the law with
a specific procedure for creating the district mapping. Specific procedures
get really complex, and I don't think in this case that a fair procedure
ensures a fair outcome.

Sounds reasonable. However, people don't really like randomness.
There is considerable resistance to using randomness in elections and
people might consider a system where the solution isn't well defined
as somewhat random.

"You mean that the best solution mightn't be picked because nobody
could find it?"

Also, there could be issues if a better solution is found after the
deadline, especially, if it favoured one party over another.

I used the voronoi technique as just one possible
way of writing a solver to get the desired result. It could even wind up
that the best way to solve for the desired goals is to have a computer do
99% of the work and have a human come back and clean up the edges. Since
it's subject to the same scoring in the end, the human couldn't deviate too
far from the ideal mapping.

Well a person is going to be unlikely to be able to beat a computer,
if the algorithm is reasonably optimised. You might need to allow the
human slightly degrade the performance slightly.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Kristofer Munsterhjelm]

Post by Raph Frank

I have implemented essentially that, and it pretty quickly gets pretty good
results as measured by the "distance per person to land-area-center" test. I
added one thing to deal with variations in population density. Each district
center also has an associated weight. Each 'voronoi cell'/district contains
the people with a weighted distance closest to that district center. So, a
district center in a high population density area doesn't have to reach very
far to gather its allotment of people.

Your maps don't look like voronoi diagrams (edges aren't lines).
Do you assign based on
Assign voter to district with minimum
distance to district centre - Cd
Where Cd is some coefficient for the district d. Cd is set so as to
give equal populations.
If so, have you considered using squared distance instead of distance?
Assign voter to district with min
(distance to centre)^2 - Cd
This gives straight district edges (I think).
You might need to change your rule to
The best district map is the one where people have the lowest average
*squared* distance to the center of their district.

The reasonable thing to use would be Euclidean distance, since that
makes sense, given the geometric nature of the districting problem. If
you want to be even more accurate, you can use great circle distance
instead to reflect that the districts are on the (near-)spherical Earth,
but at the scales involved, the difference would be slight (and so it
would be another "dotting the i-s" refinement :-)

Euclidean distance is simply, sqrt((x1-x2)^2 + (y1-y2)^2) for 2D. Since
the Voronoi check would simply have to compare distances
(nearest-neighbor), and the square root function is monotonic, one can
cut it out to get (x1-x2)^2 + (y1-y2)^2.

The non-squared alternative is |x1-x2| + |y1-y2|, the Manhattan
distance. It too would give straight lines (as opposed to curved ones),
but more of them. I think Warren has an example of a Voronoi diagram for
the Manhattan distance up on rangevoting somewhere (he used it to argue
Range would produce better results than Condorcet in the case where
voters have a preference distribution based on Manhattan distance). Some
other Manhattan distance Voronoi diagrams can be seen here:
http://www.nirarebakun.com/voro/eman.html

I may appear to be agreeing with you here, but the point of this
explanation is to show that I don't think the artifacts you're seeing is
from him using Manhattan distance. In order to get curved results, you
have to go to at least the L3 distance of (|x1-x2|^3 + |y1-y2|^3)^(1/3).
That's a fairly strange metric, so I'm not sure why he gets curved
district borders.

Also note that it's possible to find the borders of the Voronoi cells
(for the Euclidean metric, at least) much quicker than doing a
nearest-neighbor search on every single pixel. Quantization brought on
by the varying sizes of census blocks may complicate matters, though.


Your reference to a Cd variable to get population proportionality is
interesting. I think part of the problem that you're trying to fix here
comes from that clustering (such as by Lloyd's algorithm) optimizes for
energy instead of mass. Or in other words, it finds districts so that
the average distance for each voter to the closest center is minimized
(energy) rather than to find districts that have equal proportions of
people within their borders (analogous to equal mass, if each voter is a
point of unit mass).

Unless I'm mistaken, your generalization of Voronoi cells (with weights
linked to the center nodes) is called power diagrams. Not long ago, I
stumbled across a paper (draft?) about automatic redistricting involving
power diagrams. It's here:
http://www.stat.columbia.edu/%7Egelman/stuff_for_blog/fryer.pdf . The
authors state that for a certain given compactness measure, the
optimally compact district map will consist of power diagrams. They also
give an algorithm for finding these optimal (or near-optimal, since the
compactness measure is computationally intensive) power diagrams, but
it's a bit too complex for my understanding.

Post by Raph Frank

I think it's worth repeating that I'd prefer to write the law with a way of
measuring what a good district mapping is, rather than writing the law with
a specific procedure for creating the district mapping. Specific procedures
get really complex, and I don't think in this case that a fair procedure
ensures a fair outcome.

Sounds reasonable. However, people don't really like randomness.
There is considerable resistance to using randomness in elections and
people might consider a system where the solution isn't well defined
as somewhat random.
"You mean that the best solution mightn't be picked because nobody
could find it?"
Also, there could be issues if a better solution is found after the
deadline, especially, if it favoured one party over another.

Perhaps the randomness complaint could be diminished by having a default
map drawn according to some rules set in law. The redistricting law
could refer to that rule, where the rule is very simple - perhaps
splitline? Then a state commission or similar could draw the default map
so that one is ensured that the results won't be too bad if the
redistricting fails. As long as the rule is neutral, the state can't rig
the default map, either.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

On Tue, Sep 2, 2008 at 11:00 PM, Kristofer Munsterhjelm

The reasonable thing to use would be Euclidean distance, since that makes
sense, given the geometric nature of the districting problem. If you want to
be even more accurate, you can use great circle distance instead to reflect
that the districts are on the (near-)spherical Earth, but at the scales
involved, the difference would be slight (and so it would be another
"dotting the i-s" refinement :-)

My splitline software used a sphere (and gnomonic projection for the
lines). However, I think it wasn't necessary, but wanted it to be as
accurate as possible.

OTOH, I think that it might be better to define the map as a 2d map
using Mercator projection. The problem with the sphere is that each
point is a real number.

If the data is presented as (longitude, latitude), then the numbers
that are input into the algorithm are rational numbers as they are
given with a certain number of digits after the decimal point. This
allows an exact solution to be found. However, with reals, different
precisions could give different answers. Also, if there is a tie, it
may not be possible to determine that one occurs. This is less of a
problem if a specific algorithm for determining the result is used.

I think splitline can be solved exactly if the map is 2-d and all
coordinates in the input data are rational numbers.

Back to to the Voronoi diagrams, I think you may have misunderstood
what I meant.

The issue here is that ordinarily, it doesn't matter what power you use.

If you colour each pixel the colour of the nearest point, then you get
a standard Voronoi diagram.

Likewise, if you say that you colour each pixel the colour of the
point with the lowest (distance)^2, then you get the same diagram.

If you square all the distances, then the lowest distance is still the lowest.

All cells will have straight lines as their edges.

However, if I say that you should colour the pixel the colour of the
nearest point, but that the distance to point 1 is to be decreased by
100km, then you would expect that the cell with point 1 as its centre
would be increased in size as some pixels which went to other cells,
would now go to point 1's cell (as its distance has been decreased).

If you look at the results, then you would see that cell 1 no longer
has straight lines as its boundary.

Now, if instead, you assign the pixels to the cell with the nearest
distance squared and apply an offset to point 1, then you still
maintain straight line edges.

Also, it has the nice feature that you can work out the square distance as

(x0-x1)^2 + (y0-y1)^2 + Cn

You save a square root which takes a long time to calculate.

The non-squared alternative is |x1-x2| + |y1-y2|, the Manhattan distance. It
too would give straight lines (as opposed to curved ones), but more of them.
I think Warren has an example of a Voronoi diagram for the Manhattan
distance up on rangevoting somewhere (he used it to argue Range would
produce better results than Condorcet in the case where voters have a
preference distribution based on Manhattan distance). Some other Manhattan
http://www.nirarebakun.com/voro/eman.html

Cool, he also has the one I am talking about

http://www.nirarebakun.com/voro/epwvoro.html

The number beside each point is the weight

He also shows the result if you use distance instead of distance squared.

http://www.nirarebakun.com/voro/eawvoro.html

.. and weighting by multiplying

http://www.nirarebakun.com/voro/emwvoro.html

Actually, that site is pretty cool. :)

That last one might be ideal for districting. It would allow a city
to be a circle which surrounded by a rural district.

For example, if you had a State with 2 cities and 3 seats, it might be
best to split the state into 2 circle districts centred on the cities
and 1 rural district which is everyone else.

Most other methods can't handle having one district as an island
contained in another.

Also note that it's possible to find the borders of the Voronoi cells (for
the Euclidean metric, at least) much quicker than doing a nearest-neighbor
search on every single pixel. Quantization brought on by the varying sizes
of census blocks may complicate matters, though.

Yeah, most of the automatic methods assume that the 'population' is
uniform density.

What is nice about the reweighted version is that you can expand and
contract a region without having to move the points.

If you increase a region's weight, it gets smaller.

Your reference to a Cd variable to get population proportionality is
interesting. I think part of the problem that you're trying to fix here
comes from that clustering (such as by Lloyd's algorithm) optimizes for
energy instead of mass. Or in other words, it finds districts so that the
average distance for each voter to the closest center is minimized (energy)
rather than to find districts that have equal proportions of people within
their borders (analogous to equal mass, if each voter is a point of unit
mass).

Maybe both the Cd and the position of each point could be moved.

I think for (nearly) any given set of points, it is possible to create
a set of Cd values that will give near equal population.

The algorithm could be something like:

1) Pick initial points (some rule)
2) Determine set of Cds to give equal populations
3) Move points towards geographic centres of districts
4) goto 2) until points have converged

This probably doesn't always converge, so a method would need to be
created which always converges.

Unless I'm mistaken, your generalization of Voronoi cells (with weights
linked to the center nodes) is called power diagrams. Not long ago, I
stumbled across a paper (draft?) about automatic redistricting involving
http://www.stat.columbia.edu/%7Egelman/stuff_for_blog/fryer.pdf .

Yeah, though it isn't my generalization, Warren pointed it out to me.

The
authors state that for a certain given compactness measure, the optimally
compact district map will consist of power diagrams. They also give an
algorithm for finding these optimal (or near-optimal, since the compactness
measure is computationally intensive) power diagrams, but it's a bit too
complex for my understanding.

I only browsed the paper, just to find the definition of power diagram.

Perhaps the randomness complaint could be diminished by having a default
map drawn according to some rules set in law. The redistricting law could
refer to that rule, where the rule is very simple - perhaps splitline? Then
a state commission or similar could draw the default map so that one is
ensured that the results won't be too bad if the redistricting fails. As
long as the rule is neutral, the state can't rig the default map, either.

Perhaps, they would have to beat the default map by some measure.

One option would be to say that at least 80% of the residents assigned
to a given district must also be assigned to that district after the
commission has refined the map to take into account whatever
aesthetics are appropriate.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Kristofer Munsterhjelm]

Post by Raph Frank
On Tue, Sep 2, 2008 at 11:00 PM, Kristofer Munsterhjelm

The reasonable thing to use would be Euclidean distance, since that makes
sense, given the geometric nature of the districting problem. If you want to
be even more accurate, you can use great circle distance instead to reflect
that the districts are on the (near-)spherical Earth, but at the scales
involved, the difference would be slight (and so it would be another
"dotting the i-s" refinement :-)

My splitline software used a sphere (and gnomonic projection for the
lines). However, I think it wasn't necessary, but wanted it to be as
accurate as possible.
OTOH, I think that it might be better to define the map as a 2d map
using Mercator projection. The problem with the sphere is that each
point is a real number.
If the data is presented as (longitude, latitude), then the numbers
that are input into the algorithm are rational numbers as they are
given with a certain number of digits after the decimal point. This
allows an exact solution to be found. However, with reals, different
precisions could give different answers. Also, if there is a tie, it
may not be possible to determine that one occurs. This is less of a
problem if a specific algorithm for determining the result is used.
I think splitline can be solved exactly if the map is 2-d and all
coordinates in the input data are rational numbers.

If you use a Mercator projected map, you're just hiding the
quantization. All maps have some distortion, and since the map
projection uses trigonometric functions, you can just use the Haversine
distance directly. If you need the precision and an exact measurement of
error, you could use a rational number class with sine and cosine
approximation tables (or Taylor series), but I think real error like the
Earth not being perfectly spherical will get you before the rounding
errors do.

Post by Raph Frank
Back to to the Voronoi diagrams, I think you may have misunderstood
what I meant.
The issue here is that ordinarily, it doesn't matter what power you use.
If you colour each pixel the colour of the nearest point, then you get
a standard Voronoi diagram.
Likewise, if you say that you colour each pixel the colour of the
point with the lowest (distance)^2, then you get the same diagram.
If you square all the distances, then the lowest distance is still the lowest.
All cells will have straight lines as their edges.
However, if I say that you should colour the pixel the colour of the
nearest point, but that the distance to point 1 is to be decreased by
100km, then you would expect that the cell with point 1 as its centre
would be increased in size as some pixels which went to other cells,
would now go to point 1's cell (as its distance has been decreased).
If you look at the results, then you would see that cell 1 no longer
has straight lines as its boundary.
Now, if instead, you assign the pixels to the cell with the nearest
distance squared and apply an offset to point 1, then you still
maintain straight line edges.
Also, it has the nice feature that you can work out the square distance as
(x0-x1)^2 + (y0-y1)^2 + Cn
You save a square root which takes a long time to calculate.

I see. I thought you were talking about how to calculate the distance in
the first place. Since squaring and square roots are monotonic, if you
have squared distance = (x0-x1)^2 + (y0-y1)^2, picking the maximum or
minimum distance would be the same as picking the maximum or minimum
squared distance. Weighting would differ, of course, as you note.

Power diagrams on Euclidean distance are still convex polyhedra, though,
to my knowledge.

Post by Raph Frank
That last one might be ideal for districting. It would allow a city
to be a circle which surrounded by a rural district.

The surrounding district would score pretty badly on the all-pairs
distance measure, and also on a similar convexity measure (given as the
probability of the line between two random points being entirely inside
the district, or where it is outside of the district, being so only over
water or outside of the state).

Post by Raph Frank
For example, if you had a State with 2 cities and 3 seats, it might be
best to split the state into 2 circle districts centred on the cities
and 1 rural district which is everyone else.
Most other methods can't handle having one district as an island
contained in another.

If it's best, the earlier measures are not adequate to discover it.

Post by Raph Frank

Also note that it's possible to find the borders of the Voronoi cells (for
the Euclidean metric, at least) much quicker than doing a nearest-neighbor
search on every single pixel. Quantization brought on by the varying sizes
of census blocks may complicate matters, though.

Yeah, most of the automatic methods assume that the 'population' is
uniform density.
What is nice about the reweighted version is that you can expand and
contract a region without having to move the points.
If you increase a region's weight, it gets smaller.

Within limits, of course. If you have districts A, B, and C, next to
each other, then increasing the size of B will make it "eat into" both A
and C. Thus altering the size of B (to transfer population from A to B,
say) would also impact C. Trying to correct this by adjusting C's weight
could in turn impact a D district (and perhaps A as well, if D borders E
which borders... which borders Z which borders A).

Post by Raph Frank

Your reference to a Cd variable to get population proportionality is
interesting. I think part of the problem that you're trying to fix here
comes from that clustering (such as by Lloyd's algorithm) optimizes for
energy instead of mass. Or in other words, it finds districts so that the
average distance for each voter to the closest center is minimized (energy)
rather than to find districts that have equal proportions of people within
their borders (analogous to equal mass, if each voter is a point of unit
mass).

Maybe both the Cd and the position of each point could be moved.
I think for (nearly) any given set of points, it is possible to create
a set of Cd values that will give near equal population.
1) Pick initial points (some rule)
2) Determine set of Cds to give equal populations
3) Move points towards geographic centres of districts
4) goto 2) until points have converged
This probably doesn't always converge, so a method would need to be
created which always converges.

I did just that today with a toy program that "redistricts" the entire
world. It does seem to converge, but it's not good enough (to use for
redistricting purposes) if the courts demand +/- 1 proportionality of
the districts, as the fryer paper states (footnote 12).

Post by Raph Frank

Perhaps the randomness complaint could be diminished by having a default
map drawn according to some rules set in law. The redistricting law could
refer to that rule, where the rule is very simple - perhaps splitline? Then
a state commission or similar could draw the default map so that one is
ensured that the results won't be too bad if the redistricting fails. As
long as the rule is neutral, the state can't rig the default map, either.

Perhaps, they would have to beat the default map by some measure.
One option would be to say that at least 80% of the residents assigned
to a given district must also be assigned to that district after the
commission has refined the map to take into account whatever
aesthetics are appropriate.

The commission would be more robotic than that. The idea I considered
was to have the default map drawn in some simple manner - splitline
would work. Then that map becomes the reference for all future maps in
the "contest". Any map that doesn't strictly improve all relevant
measures with respect to the default map is excluded from consideration.
That way, the people can be sure that whatever happens, the map will be
at least no worse than the default.

The maximum change limitation would make sense, since it'd keep the
districts from wildly oscillating, but if you set the limit too high,
the districting process may get locked into a local optimum.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

On Wed, Sep 3, 2008 at 8:35 PM, Kristofer Munsterhjelm

If you use a Mercator projected map, you're just hiding the quantization.
All maps have some distortion, and since the map projection uses
trigonometric functions, you can just use the Haversine distance directly.
If you need the precision and an exact measurement of error, you could use a
rational number class with sine and cosine approximation tables (or Taylor
series), but I think real error like the Earth not being perfectly spherical
will get you before the rounding errors do.

True ofc. However, what I want is that the census bureau provides a
definition of the state.

For each block, there is a population and location.

Similarly, there is a boundary file that defines the boundary as lots
of line segments.

Somebody has to actually collect that info, and there will be errors.
State boundaries aren't actually made up of line segments and
populations are concentrated at the centres of their blocks.

However, once that info is provided, the ideal situation is that there
is a unique solution to the problem.

If you have to change projection, you convert rational numbers into
real numbers.

Hmm, perhaps if the original locations were given in gnomonic, I am
not sure if there would be a need for real numbers, I will need to
think about it (I have memories of a square root being required
somewhere anyway).

Ofc, then the boundary info would be great circles rather than line
segments, which shouldn't be a problem (and is probably more
accurate).

The surrounding district would score pretty badly on the all-pairs distance
measure, and also on a similar convexity measure (given as the probability
of the line between two random points being entirely inside the district, or
where it is outside of the district, being so only over water or outside of
the state).
If it's best, the earlier measures are not adequate to discover it.

Yeah, it is more an aesthetic thing. If there is 2 cities and 3
districts, I think it would be best to have 1 district per city and
the rest as a district.

I am not sure if that can be phrased in maths.

Another thought is that the map must have the shortest sum of
perimeters, but perimeters are weighted by population density. This
would run lines though low population regions.

Within limits, of course. If you have districts A, B, and C, next to each
other, then increasing the size of B will make it "eat into" both A and C.
Thus altering the size of B (to transfer population from A to B, say) would
also impact C. Trying to correct this by adjusting C's weight could in turn
impact a D district (and perhaps A as well, if D borders E which borders...
which borders Z which borders A).

Right, but the district having its weight adjusted either gains
territory or loses territory, it never ends up gaining some and losing
some elsewhere.

Another interesting point is that if you increase the weight of a
group of districts by the same amount, then it will never result in
any of the districts gaining or losing population from another member
of the group.

I think it should be possible to balance district populations to
within the minimum block size just by modifying the weights, no matter
how they are arranged.

Basically:

1) Set the threshold to the (max population + min population)/2

2) Split districts into 2 groups using 1) as threshold

3) Increase weight of all the low population group's districts by the
maximum (same for each district) amount such that (max-min) increases

4) goto 1)

The method of increasing the weight will depend on how the calculation
is performed. Either they will all have the same value
added/subtracted from them (if it is an additive weight) or they will
all be multiplied by a constant (if it is a multiplicative weight).

I did just that today with a toy program that "redistricts" the entire
world. It does seem to converge, but it's not good enough (to use for
redistricting purposes) if the courts demand +/- 1 proportionality of the
districts, as the fryer paper states (footnote 12).

Yeah, they are pretty amazing in their accuracy. If there is an
objective rule, they can comply with it, while still gerrymandering as
effectively as desired.

However, a simple rule probably couldn't match that. It would have to
be followed by a balancing step.

This would swap border block groups between districts in order to
balance the populations.

The commission would be more robotic than that. The idea I considered was to
have the default map drawn in some simple manner - splitline would work.
Then that map becomes the reference for all future maps in the "contest".
Any map that doesn't strictly improve all relevant measures with respect to
the default map is excluded from consideration. That way, the people can be
sure that whatever happens, the map will be at least no worse than the
default.

It might be defined as moving block groups from district to district
to balance population.

To replace the current 'champion' your map must

- have equal or better population balance (max pop - min pop)
- have a greater number of residents in the same district as the
original algorithm produced

In most cases, the population imbalance would end up as 1, i.e.
population in the largest district is one greater than the population
in the smallest district.

After that, you would need to try to come up with a solution that
matches the default map as closely as possible.

The maximum change limitation would make sense, since it'd keep the
districts from wildly oscillating, but if you set the limit too high, the
districting process may get locked into a local optimum.

Well, that idea was assuming that there was a pretty effective and
simple algorithm, but is sometimes made 'stupid' errors. If you look
at the splitline algorithm results on the rangevoting site, some of
the results have a tiny piece of land across a river as part of an
unconnected district.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Brian Olson]

Post by Kristofer Munsterhjelm
Your reference to a Cd variable to get population proportionality is
interesting. I think part of the problem that you're trying to fix
here comes from that clustering (such as by Lloyd's algorithm)
optimizes for energy instead of mass. Or in other words, it finds
districts so that the average distance for each voter to the closest
center is minimized (energy) rather than to find districts that have
equal proportions of people within their borders (analogous to equal
mass, if each voter is a point of unit mass).
Unless I'm mistaken, your generalization of Voronoi cells (with
weights linked to the center nodes) is called power diagrams. Not
long ago, I stumbled across a paper (draft?) about automatic
redistricting involving power diagrams. It's here: http://www.stat.columbia.edu/%7Egelman/stuff_for_blog/fryer.pdf
. The authors state that for a certain given compactness measure,
the optimally compact district map will consist of power diagrams.
They also give an algorithm for finding these optimal (or near-
optimal, since the compactness measure is computationally intensive)
power diagrams, but it's a bit too complex for my understanding.

I'm looking at fryer.pdf and maybe the notation is a little thick for
me but it looks like the pi(V_s) equation number 1 in section 3.2 is
summing, for each district, the distance between every voter in the
district and every other voter in that district. This is a little
counterintuitive to me since for so long I've been summing the
distance of voters to the center of their district. On the other hand,
it does kinda make sense too.

Anyway, what my pseudo-voronoi solver does is this:

Initialization:
• Assign district centers to random points (block coordinates).
• Assign census blocks to nearest district center.

Repeat until done:
• Move district centers towards center of population they were
actually assigned.
• Adjust district weight, nudging weight up or down if there are too
few or two many people assigned to it.
• (optional) Nudge district centers away from underpopulated district
centers and towards overpopulated district centers.
• For each census block, assign census block to the district with the
lowest ((distance to block from district center) * weightd)
• Fixup district contiguity (because straight-line-nearest can jump
across water and other gaps in odd ways)

(That's now enshrined here: http://code.google.com/p/redistricter/wiki/NearestNeighborSolver
)

Oh, and when it's done (an hour or two for most states), start it over
again with a different random starting state because it may have
gotten stuck in a local optimum.

Post by Kristofer Munsterhjelm

Post by Raph Frank

I think it's worth repeating that I'd prefer to write the law with a way of
measuring what a good district mapping is, rather than writing the law with
a specific procedure for creating the district mapping. Specific procedures
get really complex, and I don't think in this case that a fair procedure
ensures a fair outcome.

Sounds reasonable. However, people don't really like randomness.
There is considerable resistance to using randomness in elections and
people might consider a system where the solution isn't well defined
as somewhat random.
"You mean that the best solution mightn't be picked because nobody
could find it?"
Also, there could be issues if a better solution is found after the
deadline, especially, if it favoured one party over another.

Perhaps the randomness complaint could be diminished by having a
default map drawn according to some rules set in law. The
redistricting law could refer to that rule, where the rule is very
simple - perhaps splitline? Then a state commission or similar could
draw the default map so that one is ensured that the results won't
be too bad if the redistricting fails. As long as the rule is
neutral, the state can't rig the default map, either.

I guess my time in Computer Science land has left me pretty
comfortable with the idea that there are lots of problems that are too
hard to ever reliably get "the best solution". I don't know if there's
a short-short popularizing explanation of how finding a good solution
is Hard while measuring the quality of a solution is pretty quick.

If anybody asks and it's not the time, place, or audience for
discussing NP Hard problems, I will wave my hands and say, "Hey, look
over there! Good results, on my one puny computer! With more it'd only
get better!"


Brian Olson
http://bolson.org/


----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

I'm looking at fryer.pdf and maybe the notation is a little thick for me
but it looks like the pi(V_s) equation number 1 in section 3.2 is summing,
for each district, the distance between every voter in the district and
every other voter in that district. This is a little counterintuitive to me
since for so long I've been summing the distance of voters to the center of
their district. On the other hand, it does kinda make sense too.

That seems to be similar to finding the average resident location and
summing distance to that. It works on population location rather the
district shape.

• Assign district centers to random points (block coordinates).

I was thinking that a good way of picking the points would be to do
PR-STV between all the residents. Each resident's ranking would
effectively be to rank all other residents in terms of distance.

However, it turns out to be a killer to actually perform the election.
If there are 600k blocks, that is 600k rounds.

I am thinking that (an approx version of) CPO-STV could be used more
efficiently than full PR-STV.

1) Set the current district centres to a random set of points
2) set d to some initial value (maybe sqrt( State area ) )
3) For each of the current district centres pick a point inside the
State within d of it
4) Set the test district centres to the points from 2
5) Compare the test and current district centres using the CPO-STV comparison
6) If the test district is better,
d = d * 1.01
current points = test points
Otherwise
d = d * 0.99
7) goto 3, unless d < some value

Effectively, it tests lots of arrangements of centres. However, once
the current one starts being better on average than the ones being
tested against, it will start searching smaller regions around the
current best configuration. That should hopefully speed up
convergence.

The CPO comparison is actually simple when none of the points match.
You just assign each voter to the nearest district centre (testing
against both the current and test centres). There is no need to do
transfers as none of the 'candidates' are in both solutions.

You then just sum up the votes for all the current centre points and
if that is more than the sum for the test centre points, then the
current points are considered to have won.

Once that is done, an algorithm could be run to set the weights so as
to make the districts equal. This is pretty easy and (I think) gives
a unique result.

It would be possible to allow people to submit possible configurations
and then the CPO-STV winner of those submissions would be the final
set of points. It would require a condorcet completion method to be
decided.

• Assign census blocks to nearest district center.
• Move district centers towards center of population they were
actually assigned.

Is this guaranteed to improve the district quality?

In fact, do you only make an update if that improves quality by your measure?

• Adjust district weight, nudging weight up or down if there are too
few or two many people assigned to it.

Do you do this until it completely converges, or just 1 nudge?

• (optional) Nudge district centers away from underpopulated
district centers and towards overpopulated district centers.
• For each census block, assign census block to the district with
the lowest ((distance to block from district center) * weightd)

You would get straight edges if you used

lowest( (distance^2) - weight )

This also saves you having to do a square-root, though I guess your
method could be implemented as

lowest( (distance^2)*(weight)^2 )

This would speed up the calculations as there would be no need for
finding the square root of the distance, while still giving the same
result.

• Fixup district contiguity (because straight-line-nearest can jump
across water and other gaps in odd ways)

I think water counts as part of a district, so technically that isn't
a breech of contiguity.

What could be an issue if if a State has a concave boundary.

Btw, do you parse the block geography file to work out which blocks
border each other?

http://code.google.com/p/redistricter/wiki/NearestNeighborSolver
)
Oh, and when it's done (an hour or two for most states), start it over
again with a different random starting state because it may have gotten
stuck in a local optimum.

My splitline software was able to do all the States in one evening.
That is the benefit of a non-search algorithm.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Brian Olson]

Post by Raph Frank

Post by Brian Olson
• Assign district centers to random points (block
coordinates).
• Assign census blocks to nearest district center.
• Move district centers towards center of population they were
actually assigned.

Is this guaranteed to improve the district quality?
In fact, do you only make an update if that improves quality by your measure?

I think it's not guaranteed to be an improvement because the
simultaneous update of all the other district centers and weights
could wind up creating overshoots or other destructive interference. I
always make the update. It may actually not be terribly well thought
out beyond 'seems like an intuitively good idea' and 'seems to work
pretty well'.

Post by Raph Frank

Post by Brian Olson
• Adjust district weight, nudging weight up or down if there are too
few or two many people assigned to it.

Do you do this until it completely converges, or just 1 nudge?

1 nudge per cycle of all the "Repeat until done" stages. 10000 to
20000 cycles are needed to settle depending on the state.

Post by Raph Frank

Post by Brian Olson
• (optional) Nudge district centers away from underpopulated
district centers and towards overpopulated district centers.
• For each census block, assign census block to the district with
the lowest ((distance to block from district center) * weightd)

You would get straight edges if you used
lowest( (distance^2) - weight )
This also saves you having to do a square-root, though I guess your
method could be implemented as
lowest( (distance^2)*(weight)^2 )
This would speed up the calculations as there would be no need for
finding the square root of the distance, while still giving the same
result.

I checked my code and I'm not doing the expensive square root. It's
not quite the second though, it's actually:
((dx*dx + dy*dy) * weight)

The weight gets nudged by multiplying by 1.01 or 0.99. Squaring the
weight or not and how it's nudged is probably just an efficiency issue
and not a correctness issue, it should still get to the weight it
needs, just maybe along some suboptimal curve.

I think multiplicative weight makes more sense. No chance of
underflow, and might scale better between districts with different
densities.

Post by Raph Frank

Post by Brian Olson
• Fixup district contiguity (because straight-line-nearest can jump
across water and other gaps in odd ways)

I think water counts as part of a district, so technically that isn't
a breech of contiguity.
What could be an issue if if a State has a concave boundary.
Btw, do you parse the block geography file to work out which blocks
border each other?

Yes, the block shape file determines adjacency. They handily label
every line segment with what block is on either side of it, so I can
pretty quickly filter that to keep a list of all pairings I've seen.
Of course I do also wind up going back and rendering that geometry to
an actual map.

Post by Raph Frank

Post by Brian Olson
http://code.google.com/p/redistricter/wiki/NearestNeighborSolver
)
Oh, and when it's done (an hour or two for most states), start it over
again with a different random starting state because it may have gotten
stuck in a local optimum.

My splitline software was able to do all the States in one evening.
That is the benefit of a non-search algorithm.

I can get decent mappings for 90% of the states in a couple days. CA
and TX are tough. Lots of data and lots of of districts to balance.
There are some tunable parameters in the search process that I should
try adjusting specifically for solving those states.


Brian Olson
http://bolson.org/


----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

I checked my code and I'm not doing the expensive square root. It's not
((dx*dx + dy*dy) * weight)
The weight gets nudged by multiplying by 1.01 or 0.99. Squaring the weight
or not and how it's nudged is probably just an efficiency issue and not a
correctness issue, it should still get to the weight it needs, just maybe
along some suboptimal curve.

I think it is equivalent anyway. Well, you might have to multiple by
1.02 and 0.98 instead to get the same as weight squared.

I think multiplicative weight makes more sense. No chance of underflow, and
might scale better between districts with different densities.

Well, negative distances are fine too :).

However, it could mean that the point ends up outside its district.

Yes, the block shape file determines adjacency. They handily label every
line segment with what block is on either side of it, so I can pretty
quickly filter that to keep a list of all pairings I've seen.

Ahh, cool.

I might have a look at that file again. It just seemed more trouble
than it was worth.

In my software, I just assume that the people are compressed down to
the location point given in the block group table.

Of course I do
also wind up going back and rendering that geometry to an actual map.

So, the maps on your site draw the boundary as it using the
block-group boundary info?
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Jonathan Lundell]

I haven't been following this thread in great detail, but I do have a
question: what is the distance function actually trying to measure and
minimize? What exactly are we trying to optimize when we minimize
"distance", by whatever measure? I might be close in a crow's-flight
sense to a neighbor on the other side of a river or freeway with a
driving distance of several miles.

OK, we could solve that in principle (though not too quickly) by using
Google Maps driving time, or the like. But what does driving time have
to do with grouping voters (unless we're drawing a precinct and
measuring travel time t the polling place)? Maybe we mean
communication: two people with telephones are closer in this sense
than people without (and people with unlimited long distance are close
in this sense independent of cartesian distance).

I live in several voting districts where geographical lines make
sense--my local fire district is a case in point, as would water and
sewerage (except that I provide my own in those cases, as it happens).
But legislative districts? I can't think what I'm trying to optimize
for which straight-line vs manhattan distance would be relevant. I've
never met my representatives on business; I communicate with them via
email or telephone, and I neither know nor care where there local
offices are.

Post by Raph Frank

I checked my code and I'm not doing the expensive square root. It's not
((dx*dx + dy*dy) * weight)
The weight gets nudged by multiplying by 1.01 or 0.99. Squaring the weight
or not and how it's nudged is probably just an efficiency issue and not a
correctness issue, it should still get to the weight it needs, just maybe
along some suboptimal curve.

I think it is equivalent anyway. Well, you might have to multiple by
1.02 and 0.98 instead to get the same as weight squared.

I think multiplicative weight makes more sense. No chance of
underflow, and
might scale better between districts with different densities.

Well, negative distances are fine too :).
However, it could mean that the point ends up outside its district.

Yes, the block shape file determines adjacency. They handily label every
line segment with what block is on either side of it, so I can pretty
quickly filter that to keep a list of all pairings I've seen.

Ahh, cool.
I might have a look at that file again. It just seemed more trouble
than it was worth.
In my software, I just assume that the people are compressed down to
the location point given in the block group table.

Of course I do
also wind up going back and rendering that geometry to an actual map.

So, the maps on your site draw the boundary as it using the
block-group boundary info?
----
Election-Methods mailing list - see http://electorama.com/em for list info

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Kristofer Munsterhjelm]

Post by Jonathan Lundell
I haven't been following this thread in great detail, but I do have a
question: what is the distance function actually trying to measure and
minimize? What exactly are we trying to optimize when we minimize
"distance", by whatever measure? I might be close in a crow's-flight
sense to a neighbor on the other side of a river or freeway with a
driving distance of several miles.
OK, we could solve that in principle (though not too quickly) by using
Google Maps driving time, or the like. But what does driving time have
to do with grouping voters (unless we're drawing a precinct and
two people with telephones are closer in this sense than people without
(and people with unlimited long distance are close in this sense
independent of cartesian distance).
I live in several voting districts where geographical lines make
sense--my local fire district is a case in point, as would water and
sewerage (except that I provide my own in those cases, as it happens).
But legislative districts? I can't think what I'm trying to optimize for
which straight-line vs manhattan distance would be relevant. I've never
met my representatives on business; I communicate with them via email or
telephone, and I neither know nor care where there local offices are.

Proper redistricting has a few rules: the districts must be contiguous
(no empty spaces, exclaves or enclaves), and the districts must carry
similar population (otherwise "one person, one vote" suffers, and it
appears that United States court precedent have judged population
dissimilarity very strictly).

Minimizing over some distance is a side benefit. If you minimize over
Euclidean distance (or Haversine, if you want to be picky), the
districts are going to look compact - like squares or hexagons - rather
than the sort of contorted shape that "obviously gerrymandered"
districts have.

The underlying reason is that while there are very many ways of drawing
gerrymander-shaped districts, there are much fewer ways of drawing
districts that minimize with regard to some distance. Therefore,
gerrymandered districts (having no constraint on compactness) would
much more likely be gerrymander-shaped than compact. While there's no
direct reason to have districts that look nice on a map, it serves a
function similar to using expansions of pi or e in cryptographic
primitives: to show that no trickery is being done, because the
selection is so limited that finding trickery that's compatible with the
appearance (or the rules of the optimization) becomes nearly impossible.

By this reasoning, one could also turn your argument on its head. There
are various functions that leave too little to chance to be effectively
gerrymandered. If you pick Manhattan distance, fine enough; but if you
pick Euclidean or Haversine distance, then at least you have the added
improvement, however small, that if geographical lines *do* matter, then
the assignment optimizes for it, and that bonus is essentially free.

One could try to make districts that fit geographical lines even better
by weighting according to steepness (delta altitude; since it's hard to
cross mountains), or by using street distance (Google Maps driving
time). However, the methods discussed would need the distance metric to
be defined for all pairs of points, and I don't think there's street
distance data for that -- for instance, what's the driving time if you
start on a mountain peak?

(I wonder if candidates would try to make roads to nowhere to game the
metric :-)
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

Post by Jonathan Lundell
OK, we could solve that in principle (though not too quickly) by using
Google Maps driving time, or the like. But what does driving time have to do
with grouping voters (unless we're drawing a precinct and measuring travel
time t the polling place)?

I completely agree. Proportional representation is superior to single
seat districts. It in effect allows people to design their own
districts.

However, if you are going to have single seat districts, then you
might as well come up with an unbiased method for creating them.

The ideal algorithm is one that is simple and comes up with reasonable
districts. Also, it should have as few free parameters as possible.
This prevents politicians from abusing their power and designing the
districts to keep themselves in power.

However, even something like "they should be compact" favours some
people. If you are part of a group that is spread evenly, then you do
worse if the districts are compact. The problem is that philosophy
that districts should be geographically based.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Stéphane Rouillon]

Hello electorama fans,

regarding that last comment, I invite those interested in non-geographical
district
to consider astrological district. The idea is to obtain equivalent samples
of the electorate
in term of any distribution (age, geography, profession, language,
religion,...) like
poll survey use. For example, in Quebec with near 4 000 000 electors, we
could
obtain around 73 (73 x 5 = 365 days) of less than 55 000 electors each.
Thus electorate results could indicate a better performance from some
candidates
instead of reflecting the district bias produced by its design.
For example the first district could be formed with all Quebecors born
between
1st and 5th of january, the 2nd with Quebecors born between 6th and 10th of
january
and so on...

For more details of an electoral system using such "districts", search for
SPPA
(Scrutin Préférentiel, Proportionel et Acirconscriptif in french).
An english version is available on the electoral reform website
of the British-Colombia citizen assembly.

...

Post by Raph Frank
However, even something like "they should be compact" favours some
people. If you are part of a group that is spread evenly, then you do
worse if the districts are compact. The problem is that philosophy
that districts should be geographically based.
----
Election-Methods mailing list - see http://electorama.com/em for list info

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Allen Smith]

Post by Stéphane Rouillon
Hello electorama fans,
regarding that last comment, I invite those interested in non-geographical
district to consider astrological district.

How about self-chosen districts? One element of current geographical
districts is that people at least theoretically (and in some actual cases -
see Libertarians moving to New Hampshire, Alaska, etc for instance) can
choose to move to be more around people they agree more with; that even this
(frequently impractical) method would be impossible with such a system would
be one objection to it.

Post by Stéphane Rouillon
...

Post by Raph Frank
However, even something like "they should be compact" favours some
people. If you are part of a group that is spread evenly, then you do
worse if the districts are compact. The problem is that philosophy
that districts should be geographically based.

Yes.

-Allen

[Jonathan Lundell]

Post by Terry Bouricius
September

Post by Stéphane Rouillon
Hello electorama fans,
regarding that last comment, I invite those interested in non-
geographical
district to consider astrological district.

How about self-chosen districts? One element of current geographical
districts is that people at least theoretically (and in some actual cases -
see Libertarians moving to New Hampshire, Alaska, etc for instance) can
choose to move to be more around people they agree more with; that even this
(frequently impractical) method would be impossible with such a system would
be one objection to it.

STV-PR, baby.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Stéphane Rouillon]

Dear "baby" Jonathan,

STV-PR suffers from three principal problems that are exacerbated when
trying to push the proportionality limit. They are all caused by the large
number of candidates:
1) A pre-selection occurs within each party, in order for the star candidate
of each party to get elected, that star often tries to kill concurrency
having bad collegues running with him or none at all in order to increase
its own election probability;
2) It is hard to make fair debates when the number of candidates is huge and
they are not even the same for several parties: in the end the candidates
having the most means (money and visibility) have the opporunity of getting
heard and the others may simply not;
3) voters complain about the large number of names on the ballot adding
several undesirable behaviours like random completion or following a party
pre-selection.

Equivalent virtual districts have no such problems: they allow comparing all
candidates with every party proposing a unique candidacy per district. The
result is you can obtain PR results like using only one district for STV-PR,
without the previous problems.

Subject: Re: [EM] No geographical districts
Date: Wed, 3 Sep 2008 16:30:51 -0700

Post by Allen Smith

Post by Stéphane Rouillon
Hello electorama fans,
regarding that last comment, I invite those interested in non-
geographical
district to consider astrological district.

How about self-chosen districts? One element of current geographical
districts is that people at least theoretically (and in some actual cases
-
see Libertarians moving to New Hampshire, Alaska, etc for instance) can
choose to move to be more around people they agree more with; that even
this
(frequently impractical) method would be impossible with such a system
would
be one objection to it.

STV-PR, baby.
----
Election-Methods mailing list - see http://electorama.com/em for list info

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Jonathan Lundell]

Post by Stéphane Rouillon
STV-PR suffers from three principal problems that are exacerbated
when trying to push the proportionality limit. They are all caused
1) A pre-selection occurs within each party, in order for the star candidate
of each party to get elected, that star often tries to kill
concurrency having bad collegues running with him or none at all in
order to increase its own election probability;

That assumes a rather high degree of agenda manipulation by the
parties. Why need we assume that parties would be the exclusive
gatekeepers to an STV ballot? For that matter, behavior like that
seems like a good way to guarantee the formation of new parties that
don't kill their good candidates.

Post by Stéphane Rouillon
2) It is hard to make fair debates when the number of candidates is
huge and they are not even the same for several parties: in the end
the candidates having the most means (money and visibility) have the
opporunity of getting heard and the others may simply not;

In an STV election, though, candidates need not appeal to the entire
electorate. Sure, if we insist on running American-style campaigns,
money will play an undue role, and STV is not a magic bullet to fix
that problem.

Post by Stéphane Rouillon
3) voters complain about the large number of names on the ballot
adding several undesirable behaviours like random completion or
following a party pre-selection.

It's an issue. On the other hand, I recently voted in a FPTP election
with 135 gubernatorial candidates.

Post by Stéphane Rouillon
Equivalent virtual districts have no such problems: they allow
comparing all candidates with every party proposing a unique
candidacy per district. The result is you can obtain PR results like
using only one district for STV-PR, without the previous problems.

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

Post by Stéphane Rouillon
Dear "baby" Jonathan,
STV-PR suffers from three principal problems that are exacerbated when
trying to push the proportionality limit. They are all caused by the large
1) A pre-selection occurs within each party, in order for the star candidate
of each party to get elected, that star often tries to kill concurrency
having bad collegues running with him or none at all in order to increase
its own election probability;

This is somewhat true. This can be justified (to the party
leadership) on vote management grounds.

However, voters aren't necessarly bound to one party. Inter party
transfers can be significant. If a party runs useless candidates, it
tends to do badly.

Post by Stéphane Rouillon
2) It is hard to make fair debates when the number of candidates is huge
and they are not even the same for several parties: in the end the
candidates having the most means (money and visibility) have the opporunity
of getting heard and the others may simply not;

In Ireland, any debate on TV is held between the national parties.
For the last election, it was a debate between all the minor party
leaders (about 6) and then another debate the following week between
the 2 potential candidates for Taoiseach (Prime minister). There was
also some debates on other issues with each party represented by a
'spokesman'.

Local candidates are often on local radio stations though.

In the US, would the representatives from each district debate each
other on TV? Since your districts are so large 600k+, I can see that
working.

Post by Stéphane Rouillon
3) voters complain about the large number of names on the ballot adding
several undesirable behaviours like random completion or following a party
pre-selection.

This hasn't been an issue in Ireland, people love watching the count
and seeing where their vote went. Also, most people will just
truncate once they have no further opinion.

Post by Stéphane Rouillon
Equivalent virtual districts have no such problems: they allow comparing
all candidates with every party proposing a unique candidacy per district.
The result is you can obtain PR results like using only one district for
STV-PR, without the previous problems.

I don't think it gives PR at all. I think it ensures that the
majority party will win all the seats.

Also, with non-geographic districts, it means that local TV/radio
can't be used to host any debates.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[James Gilmour]

Stéphane Rouillon > Sent: Thursday, September 04, 2008 6:03 AM

Post by Stéphane Rouillon
STV-PR suffers from three principal problems that are exacerbated when
trying to push the proportionality limit.

Why would you want to try "to push the proportionality limit"? The law of diminishing returns applies to representation and
proportionality, as I said in a recent EM post under another topic heading. The available evidence from countries with a history of
FPTP elections from single-member districts (UK, USA, Canada) indicates that real electors would be very content with much less than
proportionality at the limit - indeed, they would demand such a trade-off in return for guaranteed local representation. In any
event, proportionality at the limit would bring its own political problems and so would be undesirable and unwanted for that reason
alone.

Post by Stéphane Rouillon
They are all caused by the large

Why need the numbers of candidates be large? Of course, to some electors who are used to single-member districts contested for
decades by only two parties, any number greater than two might be considered "large". In my experience such comments usually come
from those who are completely opposed to any reform of FPTP in single-member districts, but I know that group does not include
Stéphane. Such comments do, however, play into the hands of the anti-reformers.

Post by Stéphane Rouillon
1) A pre-selection occurs within each party, in order for the
star candidate of each party to get elected, that star often tries to kill concurrency
having bad collegues running with him or none at all in order
to increase its own election probability;

If the districts are of a size that would give an acceptable balance between proportionality and guaranteed local representation,
all the major parties would want to promote teams of candidates because they would have realistic chances of winning more than one
seat. So while the "star" might well want to behave like a prima donna, any party that allowed that to determine its team of
candidates would be heading for electoral disaster. Also, internal party democracy should prevail.

Post by Stéphane Rouillon
2) It is hard to make fair debates when the number of candidates is huge and
they are not even the same for several parties: in the end the candidates
having the most means (money and visibility) have the opportunity of getting
heard and the others may simply not;

Why would the numbers of candidates necessarily be "huge"? The issues arising from the availability of money have nothing to do
with the voting system. To level the democracy playing field, there clearly has to be reform (limitation) of the money that can be
spent by parties and by candidates during any election campaign - though I can see such limitation being extremely unpopular in
the USA. However, we could do much to reduce the impact of those differences by using a voting system that gave proportional
representation of the voters.

Post by Stéphane Rouillon
3) voters complain about the large number of names on the ballot adding
several undesirable behaviours like random completion or following a party
pre-selection.

Why would there be large numbers of names on the ballot? I agree that random selection might be considered an undesirable
behaviour (though some have suggested random selection would be better than selection by election!), but I don't think it is for you
or me to say that a voter who has knowingly followed his or her favoured party's selection has engaged in an undesirable behaviour.
Indeed, it could be said to be an extremely rational behviour.

Post by Stéphane Rouillon
Equivalent virtual districts have no such problems: they allow comparing all
candidates with every party proposing a unique candidacy per district. The
result is you can obtain PR results like using only one district for STV-PR,
without the previous problems.

Virtual districts may not have the three problems you specified, but they do have one real problem: real electors live (and vote)
in the real world - they do not live in a virtual world. They want their votes to reflect the real world in which they live, by
giving an acceptable balance between proportional representation of differing political viewpoints and local representation of the
geographically recognisable communities within which they live.

James Gilmour

No virus found in this outgoing message.
Checked by AVG.
Version: 7.5.526 / Virus Database: 270.6.16/1650 - Release Date: 03/09/2008 16:13


----
Election-Methods mailing list - see http://electorama.com/em for list info

[Stéphane Rouillon]

Hello James,

(for other readers , please let me state again that, in my humble opinion,
STV family of electoral systems are the best multi-winner methods "on the
market" actually. However, it should not stop us from criticizing aspects we
think could be enhanced. It is not because you have a radio and a
photographic apparel that you should not try to invent TV...).

Yes most electors would be very content with much less than proportionality
at the limit
because they are convinced they have to sacrifice something else to obtain
it
(local representation for some, government stability, access to a
representative,...).
No one has ever presented them with an option that conciliate all these
goals.
Several demand a guaranteed local representation (although this demand is a
lot stronger in rural areas not urban ones). But we all already have a
municipal level that is based on geographical management and can facilitate
access to a local representative.

Thus:
-I want to push the proportionnality limit to obtain a true representation
of the electorate.
I can cope with the stability problems that could occur using other tools,
like the crutch option for example.
- I want a large number of candidates because the more option you get as a
voter, the more ideas you can choose among and the more representative is
the result of the election.
-I want list built for a party from the electoral result of the voters would
expressed a preference for that party. Thus, no party oligarchy can impose a
hierarchy to its supporters and still individual imputability plays a role
because we would vote for somebody and not for a "cause".

STV or STV-PR with one big district would give me these properties, but I
try to get rid of the
problems it produces. Am I logical enough?

For the purpose of presenting different debates per district, I see very
well putting each debate onthe web, on a political youtube. Thus speech time
and not money would have to be fairly
attributed.

And finally about the real world issue, it seems it all comes to how we see
the world. I am very proud of having a municipal level which solves most of
my geographical problems (removing the snow, taking the garbagge, giving
access to a library, managing street parking,...). However,
my major concerns go through boundaries most of the time (SRAS, hurricanes,
acid rain, animals,
epidemia, religions, ideas, money, inflation, planes, violence on tv, bombs,
terrorists, free trade products, gay weddings, abortion, drug legalisation,
add as much actuality major issue you like...) . Virtual districts would fit
better the real world. I want my vote to reflect the real world in which I
live, by assessing that major debates are now some that concern the overall
planet.
Having representatives elected from non-geographical districts does not
remove their geographical link. We all come from somewhere we know well. It
just removes their geographical dependance about being elected or not.
Governing is taking decisions. We want defendors of the options and neutral
judges to choose between them like in a justice court. How can we expect our
current system to behave properly when MP's have to do both! Defendors
should be elected from a ground territory to defend the rights of the people
they know well. Judges should have no link to take decisions in the best
interest for all. Hence, decision takers (country representatives) should be
elected from non-geographical districts.

The real world now spreads ideas all around the globe in a minute like I am
doing right now.
Distance is less and less relevant. We should adapt our political structures
to this reality.

Thanks for taking the time to read,
Stéphane.

Subject: Re: [EM] No geographical districts
Date: Thu, 4 Sep 2008 12:06:10 +0100
Stéphane Rouillon > Sent: Thursday, September 04, 2008 6:03 AM

Post by Stéphane Rouillon
STV-PR suffers from three principal problems that are exacerbated when
trying to push the proportionality limit.

Why would you want to try "to push the proportionality limit"? The law of
diminishing returns applies to representation and
proportionality, as I said in a recent EM post under another topic heading.
The available evidence from countries with a history of
FPTP elections from single-member districts (UK, USA, Canada) indicates
that real electors would be very content with much less than
proportionality at the limit - indeed, they would demand such a
trade-off in return for guaranteed local representation. In any
event, proportionality at the limit would bring its own political problems
and so would be undesirable and unwanted for that reason
alone.

Post by Stéphane Rouillon
They are all caused by the large

Why need the numbers of candidates be large? Of course, to some electors
who are used to single-member districts contested for
decades by only two parties, any number greater than two might be
considered "large". In my experience such comments usually come
from those who are completely opposed to any reform of FPTP in
single-member districts, but I know that group does not include
Stéphane. Such comments do, however, play into the hands of the anti-reformers.

Post by Stéphane Rouillon
1) A pre-selection occurs within each party, in order for the
star candidate of each party to get elected, that star often tries to

kill concurrency

Post by Stéphane Rouillon
having bad collegues running with him or none at all in order
to increase its own election probability;

If the districts are of a size that would give an acceptable balance
between proportionality and guaranteed local representation,
all the major parties would want to promote teams of candidates because
they would have realistic chances of winning more than one
seat. So while the "star" might well want to behave like a prima donna,
any party that allowed that to determine its team of
candidates would be heading for electoral disaster. Also, internal party
democracy should prevail.

Post by Stéphane Rouillon
2) It is hard to make fair debates when the number of candidates is huge

and

Post by Stéphane Rouillon
they are not even the same for several parties: in the end the

candidates

Post by Stéphane Rouillon
having the most means (money and visibility) have the opportunity of

getting

Post by Stéphane Rouillon
heard and the others may simply not;

Why would the numbers of candidates necessarily be "huge"? The issues
arising from the availability of money have nothing to do
with the voting system. To level the democracy playing field, there
clearly has to be reform (limitation) of the money that can be
spent by parties and by candidates during any election campaign - though
I can see such limitation being extremely unpopular in
the USA. However, we could do much to reduce the impact of those
differences by using a voting system that gave proportional
representation of the voters.

Post by Stéphane Rouillon
3) voters complain about the large number of names on the ballot adding
several undesirable behaviours like random completion or following a

party

Post by Stéphane Rouillon
pre-selection.

Why would there be large numbers of names on the ballot? I agree that
random selection might be considered an undesirable
behaviour (though some have suggested random selection would be better than
selection by election!), but I don't think it is for you
or me to say that a voter who has knowingly followed his or her favoured
party's selection has engaged in an undesirable behaviour.
Indeed, it could be said to be an extremely rational behviour.

Post by Stéphane Rouillon
Equivalent virtual districts have no such problems: they allow comparing

all

Post by Stéphane Rouillon
candidates with every party proposing a unique candidacy per district.

The

Post by Stéphane Rouillon
result is you can obtain PR results like using only one district for

STV-PR,

Post by Stéphane Rouillon
without the previous problems.

Virtual districts may not have the three problems you specified, but they
do have one real problem: real electors live (and vote)
in the real world - they do not live in a virtual world. They want their
votes to reflect the real world in which they live, by
giving an acceptable balance between proportional representation of
differing political viewpoints and local representation of the
geographically recognisable communities within which they live.
James Gilmour
No virus found in this outgoing message.
Checked by AVG.
Version: 7.5.526 / Virus Database: 270.6.16/1650 - Release Date: 03/09/2008 16:13
----
Election-Methods mailing list - see http://electorama.com/em for list info

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Stéphane Rouillon]

Why not self-chosen districts ?
Because then the last half of voters would be able to pick
between district already composed of majoritarians ideologies.
Again the least organized and the smallest group would finish splitted
between several districts where they would be in minority.

Do you really believe that if you represent 1% of an ideology,
others political parties will keep the last of 100 seats for you?
With an FPTP system they would gang up as much districts
to 51% and let you be minoritarian in the district(s) you want.

You could argue that a proportional system would solve the representation
problem, whatever district definition used. But it would not solve
the ability that politicians have for "bribing" a geographical district,
simply by targeting an area that is an undecided district. With selfchosen
district,
anyone can say to teachers, automobile workers, or any other segment of the
population:
"gather in one district, mine, and I'll flood your field with fresh
investment"
With astrological districts, subventioning an hospital that would treat only
people
born between january 11th and january 16th would be:
1) complex to implement;
2) easy to be proven;
3) an obvious case of political bribery attempt.
It would bring regions to discuss between one another instead of confront
each other,
having all representatives of the whole country instead of each defending
its piece of cake.

But I concede: it would not stop a classical influence traffic which consist
of giving money
to a politician to tell him how to vote... That would still be a police job.

Subject: Re: [EM] No geographical districts
Date: Wed, 3 Sep 2008 18:51:28 -0400

Post by Stéphane Rouillon
Hello electorama fans,
regarding that last comment, I invite those interested in

non-geographical

Post by Stéphane Rouillon
district to consider astrological district.

How about self-chosen districts? One element of current geographical
districts is that people at least theoretically (and in some actual cases -
see Libertarians moving to New Hampshire, Alaska, etc for instance) can
choose to move to be more around people they agree more with; that even this
(frequently impractical) method would be impossible with such a system would
be one objection to it.

Post by Stéphane Rouillon
...

Post by Raph Frank
However, even something like "they should be compact" favours some
people. If you are part of a group that is spread evenly, then you do
worse if the districts are compact. The problem is that philosophy
that districts should be geographically based.

Yes.
-Allen
--
Allen Smith, Ph.D. http://cesario.rutgers.edu/easmith/
There is only one sound argument for democracy, and that is the argument
that it is a crime for any man to hold himself out as better than other men,
and, above all, a most heinous offense for him to prove it. - H. L. Mencken
----
Election-Methods mailing list - see http://electorama.com/em for list info

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

Post by Stéphane Rouillon
Why not self-chosen districts ?
Because then the last half of voters would be able to pick
between district already composed of majoritarians ideologies.
Again the least organized and the smallest group would finish splitted
between several districts where they would be in minority.
Do you really believe that if you represent 1% of an ideology,
others political parties will keep the last of 100 seats for you?
With an FPTP system they would gang up as much districts
to 51% and let you be minoritarian in the district(s) you want.

Self chosen districts can work in many ways. If they are completely
free, you are right. The optimal for a majority group is to spread
out evenly and win them all.

If you assign by birth date, you get near perfect spreading out by
default. A majority party will likely end up with a majority in every
district. It even neutralises any geographic variation in the State.

One option would be to allow the group itself decide who can and who
can't join their district. For example, it could be based on a
candidate or committee. They get to vet any potential new members.
Perhaps, any group of at least 11 people could form a committee and
register as a 'seed' district. Also, the might have the right to kick
members they don't like.

Unless another party managed to infiltrate the group so that a
majority of its members were for the other party, any infiltration
just means that they have less people in their own groups.

For example,

1) 6 months before election day, the N largest district groups are
'locked-in'.

2) For another 3 months people are allowed to register to change from
their district to another one, as long as their district is to large
and the target is to small (or they weren't in a district and want to
join a small one).

3) 3 months before the election, all the districts that are still to
large have random members reduced to make them the right size and
unassigned people are then randomly assigned to the small districts to
make them large enough.

4) Each district then selects a winner using approval voting

The randomness wouldn't necessarily have to be random. It could use
some fixed method to decide the ordering. For example, it could be
based on social security number and date of birth. You could pass
them into a function that re-orders them.

Also, the prcesses weakens the secret ballot as district formation
becomes part of the process and it isn't secret.

Post by Stéphane Rouillon
You could argue that a proportional system would solve the representation
problem, whatever district definition used. But it would not solve
the ability that politicians have for "bribing" a geographical district,
simply by targeting an area that is an undecided district.

It comes down to what you think representation is. That is
representing their district at the national level. It isn't bribing
anyone, it is getting them what they want. What if an area has no
hospital and nobody seems to care because they are concerned about the
national issues?

I think single seat districts aren't a good idea anyway. Most
modifications are just a hack to a broken system.

Post by Stéphane Rouillon
With astrological districts, subventioning an hospital that would treat
only people
1) complex to implement;
2) easy to be proven;
3) an obvious case of political bribery attempt.

It might be that people who meet the criteria get some benefit in any
hospital. Also, if 'everyone' is doing it, then it might be less
frowned upon.

Post by Stéphane Rouillon
It would bring regions to discuss between one another instead of confront
each other,
having all representatives of the whole country instead of each defending
its piece of cake.

Party list PR is designed to do this. Everyone represents the whole country.

It gives lots of power to the party leader though.

My thoughrs on the whole national representation thing is to have
multiple layers of representatives. In the US, it might work as

A) 30 elected in six districts (districts of ~50 million (10 mill per seat))
B) 100 elected in 20 districts (districts of ~15 million (3 mill per seat))
C) 300 elected in 60 districts (districts of ~5 million (1 mill per seat))

This districts would end up crossing State lines. One option is that
C is designed so that it doesn't. Small States would just have
smaller districts. A State with 2 Representatives would just hold a 2
seat election.

The A) and B) districts would need to cross State lines for sure.
This would require a constitutional amendment.

This creates a mix of national level, regional level and State level
representatives.

If you wanted to get fancy, you might give group A) Representatives 4
votes, group B) 2 votes and group C) 1 votes. This is to take into
account that an A) representative represents a larger number of
voters.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Fred Gohlke]

Good Afternoon, Stephane Rouillon

I, for one, find your suggestion original and elegant.

You have described a simple way of dividing the people into districts,
independent not only of their geographic location but of their
ideological predispositions, as well. Candidates who seek to represent
such 'districts' must attract the support of the entire community. They
must be truly representative of the district they seek to represent.
That is enormously different than the monstrosities we endure because of
partisan politics.

Thank you very much for a wonderful idea. It seeks to empower the
people rather than any group. I fear it bears little chance of adoption
because it protects no vested interest. The only way such a process
will ever be adopted is if the concept can be made a topic of
discussion, particularly among students interested in achieving a
righteous government.

Fred
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Juho]

Geographical proportionality is one specific dimension. Most other
dimensions could be called political dimensions. Also groupings that
do not live in any specific compact area could be called political
groupings. In principle they could form a party and that way get a
proportional number of representatives. (This is also in line with
the geographical proportionality related target of guaranteeing
representation from all _geographic_ areas.)

Many political systems have chosen geographical districts to be fixed
in the sense that people automatically "vote" for the district where
they live in. In the political dimension people are typically allowed
to pick the group that they want to represent them.

It is possible to have election methods that support multiple
dimensions, i.e. more than these two. One could e.g. simply have
multiple orthogonal "party" structures and then in the vote counting
process force the representatives to be elected so that
proportionality will be respected in all dimensions.

There could be also additional "fixed dimensions" like automatic
fixed sex or age based proportionality.

Some of the additional dimensions could also be "virtual districts"
in the sense that each voter would be registered in exactly one of
them, and probably also vote only for candidates that belong to one's
own "virtual district". I understood that you would use virtual
districts to replace the current geographical districts (and the
geographical proportionality that they represent).

The simplest (not necessarily optimal) approach to implement multiple
dimensions is one where you simply elect representatives starting
from the ones with strongest support (e.g. best candidate of the
largest party in the largest district), skip candidates that can not
be elected any more (e.g. district already full, party already full),
and continue until all seats have been filled. At some point in the
chain all "requirements" of all dimensions are met if they are strong
enough (and if there are suitable candidates left).

(Some dimensions could be one-directional in the sense that one would
aim at guaranteeing at least a proportional share of the seats but
would not limit them to this number. For example one could allow all
members of some minority to require proportional representation by
marking this in their ballot. Other voters would however not be
required to vote either for or against this minority. Any candidate
(or any party, of any regions etc.) could belong to this group. One
should however not allow these lists to overrule party
proportionality or other "complete dimensions" (to avoid riding under
two flags (party and "minority") and getting also corresponding
double representation).)

Small ad here too. Trees (hierarchical candidate lists) offer
multiple dimensions in a simplified framework, but with priorities
involved too. One can e.g. be a greenish red or a reddish green.

Juho

Post by Stéphane Rouillon
Hello electorama fans,
regarding that last comment, I invite those interested in non-
geographical district
to consider astrological district. The idea is to obtain equivalent
samples of the electorate
in term of any distribution (age, geography, profession, language,
religion,...) like
poll survey use. For example, in Quebec with near 4 000 000
electors, we could
obtain around 73 (73 x 5 = 365 days) of less than 55 000 electors each.
Thus electorate results could indicate a better performance from
some candidates
instead of reflecting the district bias produced by its design.
For example the first district could be formed with all Quebecors
born between
1st and 5th of january, the 2nd with Quebecors born between 6th and
10th of january
and so on...
For more details of an electoral system using such "districts",
search for SPPA
(Scrutin Préférentiel, Proportionel et Acirconscriptif in french).
An english version is available on the electoral reform website
of the British-Colombia citizen assembly.
...

Post by Raph Frank
However, even something like "they should be compact" favours some
people. If you are part of a group that is spread evenly, then you do
worse if the districts are compact. The problem is that philosophy
that districts should be geographically based.
----
Election-Methods mailing list - see http://electorama.com/em for list info

----
Election-Methods mailing list - see http://electorama.com/em for list info

___________________________________________________________
Try the all-new Yahoo! Mail. "The New Version is radically easier to use" – The Wall Street Journal
http://uk.docs.yahoo.com/nowyoucan.html

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

Post by Juho
The simplest (not necessarily optimal) approach to implement multiple
dimensions is one where you simply elect representatives starting from the
ones with strongest support (e.g. best candidate of the largest party in the
largest district), skip candidates that can not be elected any more (e.g.
district already full, party already full), and continue until all seats
have been filled. At some point in the chain all "requirements" of all
dimensions are met if they are strong enough (and if there are suitable
candidates left).

I would probably elect the weakest of each party's strongest
candidates, e.g. find the strongest candidate from each party and then
assign a seat to that weakest of them.

Once a party gets its allocation of seats, it can't be assigned any more.

This is to allow small parties fill in their seats before large
parties can lock them out.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Juho]

Post by Raph Frank

Post by Juho
The simplest (not necessarily optimal) approach to implement multiple
dimensions is one where you simply elect representatives starting from the
ones with strongest support (e.g. best candidate of the largest party in the
largest district), skip candidates that can not be elected any more (e.g.
district already full, party already full), and continue until all seats
have been filled. At some point in the chain all "requirements" of all
dimensions are met if they are strong enough (and if there are suitable
candidates left).

I would probably elect the weakest of each party's strongest
candidates, e.g. find the strongest candidate from each party and then
assign a seat to that weakest of them.

Why weakest? What is the "weakest of each party's strongest candidates"?

Juho

Post by Raph Frank
Once a party gets its allocation of seats, it can't be assigned any more.
This is to allow small parties fill in their seats before large
parties can lock them out.
----
Election-Methods mailing list - see http://electorama.com/em for list info

___________________________________________________________
All new Yahoo! Mail "The new Interface is stunning in its simplicity and ease of use." - PC Magazine
http://uk.docs.yahoo.com/nowyoucan.html

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

Post by Juho
Why weakest? What is the "weakest of each party's strongest candidates"?

It means find the candidate in each party with the highest vote.
These are the party's stongest candidates.

You then assign the seat to the weakest of them (but only if the party
is entitled to another seat).

My thinking was that if you assign them in a different order, you
could end up with a situation where a small party cannot be assigned a
seat that they are entitled to as they only ran in a few districts and
those districts have already been filled.

By assigning to the weakest party first, the chances of that are lessened.

A larger party would likely have someone running in all
constituencies, so should have someone to take their seats near the
end.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Stéphane Rouillon]

Hello Juho,

using age, gender or other virtual dimension to build virtual districts
replaces geographic antagonism by generation antagonism.
The idea is to get equivalent sample that are not opposed by intrinsec
construction.
Thus we may find neutral decision takers that will minimize the overall
bad impacts of a decision, thus maximize to the best of their knowledge
the decisions for all the electorate. If you split representative into
groups who have divergent opinions, the result will not optimize common
interest, it will only illustrate the "rapport de force"
(maybe translated as power struggle) between the representatives. Age
representatives would hardly stay neutral while deciding retirement fees and
pensions for example.

The Irish senate based on profession seems one step toward getting neutral
decision takers
for deciding the localization of projects for example.
I prefer equivalent samples of the entire electorate (phone numbers or hash
tables using names could work too, but it has some slight discrepancies and
problems...)

Subject: Re: [EM] No geographical districts
Date: Fri, 5 Sep 2008 00:13:34 +0300
Geographical proportionality is one specific dimension. Most other
dimensions could be called political dimensions. Also groupings that do
not live in any specific compact area could be called political groupings.
In principle they could form a party and that way get a proportional
number of representatives. (This is also in line with the geographical
proportionality related target of guaranteeing representation from all
_geographic_ areas.)
Many political systems have chosen geographical districts to be fixed in
the sense that people automatically "vote" for the district where they
live in. In the political dimension people are typically allowed to pick
the group that they want to represent them.
It is possible to have election methods that support multiple dimensions,
i.e. more than these two. One could e.g. simply have multiple orthogonal
"party" structures and then in the vote counting process force the
representatives to be elected so that proportionality will be respected in
all dimensions.
There could be also additional "fixed dimensions" like automatic fixed sex
or age based proportionality.
Some of the additional dimensions could also be "virtual districts" in the
sense that each voter would be registered in exactly one of them, and
probably also vote only for candidates that belong to one's own "virtual
district". I understood that you would use virtual districts to replace
the current geographical districts (and the geographical proportionality
that they represent).
The simplest (not necessarily optimal) approach to implement multiple
dimensions is one where you simply elect representatives starting from the
ones with strongest support (e.g. best candidate of the largest party in
the largest district), skip candidates that can not be elected any more
(e.g. district already full, party already full), and continue until all
seats have been filled. At some point in the chain all "requirements" of
all dimensions are met if they are strong enough (and if there are
suitable candidates left).
(Some dimensions could be one-directional in the sense that one would aim
at guaranteeing at least a proportional share of the seats but would not
limit them to this number. For example one could allow all members of some
minority to require proportional representation by marking this in their
ballot. Other voters would however not be required to vote either for or
against this minority. Any candidate (or any party, of any regions etc.)
could belong to this group. One should however not allow these lists to
overrule party proportionality or other "complete dimensions" (to avoid
riding under two flags (party and "minority") and getting also
corresponding double representation).)
Small ad here too. Trees (hierarchical candidate lists) offer multiple
dimensions in a simplified framework, but with priorities involved too.
One can e.g. be a greenish red or a reddish green.
Juho

Post by Stéphane Rouillon
Hello electorama fans,
regarding that last comment, I invite those interested in non-
geographical district
to consider astrological district. The idea is to obtain equivalent
samples of the electorate
in term of any distribution (age, geography, profession, language,
religion,...) like
poll survey use. For example, in Quebec with near 4 000 000 electors, we
could
obtain around 73 (73 x 5 = 365 days) of less than 55 000 electors each.
Thus electorate results could indicate a better performance from some
candidates
instead of reflecting the district bias produced by its design.
For example the first district could be formed with all Quebecors born
between
1st and 5th of january, the 2nd with Quebecors born between 6th and 10th
of january
and so on...
For more details of an electoral system using such "districts", search
for SPPA
(Scrutin Préférentiel, Proportionel et Acirconscriptif in french).
An english version is available on the electoral reform website
of the British-Colombia citizen assembly.
...

Post by Raph Frank
However, even something like "they should be compact" favours some
people. If you are part of a group that is spread evenly, then you do
worse if the districts are compact. The problem is that philosophy
that districts should be geographically based.
----
Election-Methods mailing list - see http://electorama.com/em for list
info

----
Election-Methods mailing list - see http://electorama.com/em for list
info

___________________________________________________________ Try the all-new
Yahoo! Mail. "The New Version is radically easier to use" – The Wall Street
Journal http://uk.docs.yahoo.com/nowyoucan.html
----
Election-Methods mailing list - see http://electorama.com/em for list info

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

On Fri, Sep 5, 2008 at 2:00 AM, Stéphane Rouillon

Post by Stéphane Rouillon
Hello Juho,
using age, gender or other virtual dimension to build virtual districts
replaces geographic antagonism by generation antagonism.
The idea is to get equivalent sample that are not opposed by intrinsec
construction.

A simple option would be to convert the date of birth into a number,
but have the year the the least significant part..

16-04-82 would become 160,482

The public could then be sorted by those numbers. In effect, you are
splitting people by the day of the month they are born on, if there is
a tie, you use month and only use year at the end.

This would give a mix of ages, genders and any other measure in each district.

It is pretty much equivalent to just randomly distributing the voters
between the districts, but unlike a random system, it is harder to
corrupt.

Post by Stéphane Rouillon
Thus we may find neutral decision takers that will minimize the overall
bad impacts of a decision, thus maximize to the best of their knowledge
the decisions for all the electorate.

You make a good point. It would reduce the pork issue, but it gives
minorities no representation. A group with a majority will probably
win all the seats.

The probability of a group with 55% of the votes not getting a
majority in all the districts would be tiny due to the law of large
numbers.

If that group is geographically concentrated, you are back where you started.

Post by Stéphane Rouillon
The Irish senate based on profession seems one step toward getting neutral
decision takers
for deciding the localization of projects for example.

Professionals are also a defined group.

However, I like your idea to use a group that is non-local to decide
localisation issues.

What about having 2 houses. The geographic house is elected by
PR-STV. The national one is elected using your method.

The geographic house might decide that a hospital needs to be build,
but the national house would then decide where.

Ofc, if the country was ethnically divided and the majority ethnic
group lived in the East, then the national house would probably direct
most projects in that direction.

Btw, the Irish Senate looks (somewhat) good in theory, but doesn't
actually work that way in practice. The nominating boards (which
represent different professions) have very little power. The county
councillors are the ones who actually vote for the Senators. It is a
secret ballot, but most councillors vote for their party (or as part
of a voting pact). This means that the Senate elections tend to
follow the distribution of county councillor seats.

The exception is the university seats, they are elected by graduates
of certain universities (but not all ... grrr).

Also, the Taoiseach (PM) gets to appoint a few. The combination of
the county councillor (the governing coalition should have at least a
strong minority of those seats) and the fact that the Taoiseach gets
to appoint some mean that generally the Government has an easy
majority in the Senate.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Stéphane Rouillon]

Dear Raph,

your understanding is perfect.
Of course using still FPTP with virtual districts would typicaly produce an
assembly with all the
seats of the same party. It was designed to be used with an open list
system, as much proportional as possible (to the integrality limit). The
list is filled from individual support gathered from
each candidates, having equivalent sample of the electorate let us suppose
that the one which have the best results promote the best ideas. SPPA
provides some other details like the possibility to vote None and an option
to garantee an almost majoritarian government at most for stability purpose.

http://www.citizensassembly.bc.ca/public/get_involved/submission/R/ROUILLON-65
You are welcome to comment. At least I hope you have fun reading it if you
find the time.

Subject: Re: [EM] No geographical districts
Date: Fri, 5 Sep 2008 02:25:24 +0100
On Fri, Sep 5, 2008 at 2:00 AM, Stéphane Rouillon

Post by Stéphane Rouillon
Hello Juho,
using age, gender or other virtual dimension to build virtual districts
replaces geographic antagonism by generation antagonism.
The idea is to get equivalent sample that are not opposed by intrinsec
construction.

A simple option would be to convert the date of birth into a number,
but have the year the the least significant part..
16-04-82 would become 160,482
The public could then be sorted by those numbers. In effect, you are
splitting people by the day of the month they are born on, if there is
a tie, you use month and only use year at the end.
This would give a mix of ages, genders and any other measure in each district.
It is pretty much equivalent to just randomly distributing the voters
between the districts, but unlike a random system, it is harder to
corrupt.

Post by Stéphane Rouillon
Thus we may find neutral decision takers that will minimize the overall
bad impacts of a decision, thus maximize to the best of their knowledge
the decisions for all the electorate.

You make a good point. It would reduce the pork issue, but it gives
minorities no representation. A group with a majority will probably
win all the seats.
The probability of a group with 55% of the votes not getting a
majority in all the districts would be tiny due to the law of large
numbers.
If that group is geographically concentrated, you are back where you started.

Post by Stéphane Rouillon
The Irish senate based on profession seems one step toward getting

neutral

Post by Stéphane Rouillon
decision takers
for deciding the localization of projects for example.

Professionals are also a defined group.
However, I like your idea to use a group that is non-local to decide
localisation issues.
What about having 2 houses. The geographic house is elected by
PR-STV. The national one is elected using your method.
The geographic house might decide that a hospital needs to be build,
but the national house would then decide where.
Ofc, if the country was ethnically divided and the majority ethnic
group lived in the East, then the national house would probably direct
most projects in that direction.
Btw, the Irish Senate looks (somewhat) good in theory, but doesn't
actually work that way in practice. The nominating boards (which
represent different professions) have very little power. The county
councillors are the ones who actually vote for the Senators. It is a
secret ballot, but most councillors vote for their party (or as part
of a voting pact). This means that the Senate elections tend to
follow the distribution of county councillor seats.
The exception is the university seats, they are elected by graduates
of certain universities (but not all ... grrr).
Also, the Taoiseach (PM) gets to appoint a few. The combination of
the county councillor (the governing coalition should have at least a
strong minority of those seats) and the fact that the Taoiseach gets
to appoint some mean that generally the Government has an easy
majority in the Senate.

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

Post by Stéphane Rouillon
http://www.citizensassembly.bc.ca/public/get_involved/submission/R/ROUILLON-65
You are welcome to comment. At least I hope you have fun reading it if you
find the time.

Your implementation of IRV is non-standard (though I agree with the
none of the above change and it is needed for your 2nd step).

Perhaps, votes would only go to none if the voter actually ranks none.

E.g.

A 1
B
C
None 2

would transfer to none if A is eliminated, but would be exhausted if
the voter didn't vote for None. However, that would mean that the
voter has no effect on the PR stage, so probably a bad idea.

it might be better to use a divisor method (d'Hondt or Webster's
methods) to share out the seats. This has some advantages in terms of
resistance to weird effects
(See http://en.wikipedia.org/wiki/Alabama_paradox#Alabama_Paradox ).
Your method looks like the largest remainder method.

I also don't agree with independents getting a seat with 0.51 seats
worth of votes, but that is another discussion (see previous PR
thread)

Also, you could include the None votes as a None-party. This party
could be assigned seats. They would be left vacant. My opinion is
that they should be considered to vote against every bill (though with
the crutch option, it probably doesn't matter). Alternatively, they
could just trigger a by-election.

For example, of all the valid arrangements (each party has the correct
number of seats), find the one where the elected representatives have
the highest total votes.

I am not sure I agree with the 'crutch' but, if you must have it, I
would suggest a modification.

After the election, the new parliament tries to nominate a PM.

First they vote to keep the current PM. If he fails to get a
majority, he becomes a caretaker-PM and he and the cabinet can't
excersise major powers, without consent of the House.

They can then try to nominate others in the ordinary way.

If after 1 month, no nomination motion has obtained majority support,
they can attempt to use the crutch rule.

Any member of the House may stand as a candidate.

The House holds three sequential votes.

In the first round, an approval ballot is used. For the last 2, each
member votes for 1 candidate.

The candidate who wins the plurality of the vote is considered
nominated to be PM (with crutch support).

The additional candidates are added, but they only remain members of
the House for as long as the PM remains PM.

Also, they cannot vote in motions which nominate someone else to be PM
(or motions which declare that the PM has gained support of the
majority of the House). They also cannot vote in motions which call
for a new election to be held.

The term limit is reduced on a pro-rata basis. If the PM is PM for 1
year and then they manage to get a majority, then under your example,
that would count as 7/4 * 1 year = 1.75 years used. Thus the term
would have 2.25 years left (of 4).

This system has the advantage that it allows the smaller parties the
option to try to form a coalition. If the split was 40%, 30%, 30%,
under your rule, the 40% party could declare a minority government
against the wishes of the 60% of the other 2 parties.

It also allows 2 parties which has a larger vote between them access
to the crutch system. For example, if the split was 40%, 30%, 15%,
9%, 6%. It would allow a coalition of the 30%+15% parties to form the
minority government as it has more votes than the 40% party.

In addition, it allows a moverment by the House back to a majority
government. In the above example, if the 40% party finally managed to
get the 9% and 6% parties into a coalition, they could vote to
nominate the leader of 40% party as PM. The additional members due to
the crutch would not be allowed to vote for this motion, so the new
coalition would win by 55% to 45%.

It might also be worth allowing them to switch crutch-coalition. For
example, once the 40% party gets support of the 6% party, it now has
the plurality of the vote.

It might also be worth having a rule that within 1 year of the end of
term, the crutch rule cannot be used.

I was thinking that a better way of having Heterogeneous districts
would be to based them on polling stations. If each district needed
10 polling stations, then you could randomly allocate polling stations
to districts. This doesn't get perfect blinding, but is much easier
to implement.

This might be better than social security number as the districts
could be completely different each election. Also, social security
numbers are static (and set by the government), so districts might
have overlap from the previous election.

If you waited until the last possible moment to decide the polling
stations, then a politician wouldn't know where to direct pork and in
any case, it would be hard to direct it at a specific polling station
area.

I don't know if a party 'front-bench' member would actually be willing
to stand against the cabinet. Even if more than 1 of them can be
elected from the same district, the odds of more than 1 are reasonably
low. I think the strategy would actually be to try to find a district
that nobody 'important' was standing in.

It does mean that a cabinet member would need to make sure his
policies have the support of the majority. I actually don't know if
this is a good thing. Sometimes, it is helpful to implement unpopular
policies (e.g. a 'basket' of policies might have 70% support, even
though some individual policies have support of less than the a
majority.)

Your system has an unusual blend. Each party gets a PR share of the
seats, but a random sample of the voters get to decide which party
members actually get elected.

I think this may lead to the more off-centre parties only nominating a
fixed number of candidates. Otherwise, the majority might elect all
their inexperienced members.

It might even be worth running as independents for non-centerist
parties. This is especially true since your method favours
independents. This means that a known party member gets elected based
on the party's vote. Ofc, that prevents the party from combining its
vote over multiple districts, so may be not.

Also, an actual independent is not likely to win. He would need to
have large support nationwide. However, if that is the case, it would
be expected that he would be in one of the mainstream parties.

Btw, have you considered Fair Majority Voting. This also elects
candidates based on a combination of PR and districts. Unlike, MMP,
it doesn't have top-up seats.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Kristofer Munsterhjelm]

Post by Raph Frank
On Fri, Sep 5, 2008 at 2:00 AM, Stéphane Rouillon

Post by Stéphane Rouillon
Hello Juho,
using age, gender or other virtual dimension to build virtual districts
replaces geographic antagonism by generation antagonism.
The idea is to get equivalent sample that are not opposed by intrinsec
construction.

A simple option would be to convert the date of birth into a number,
but have the year the the least significant part..
16-04-82 would become 160,482
The public could then be sorted by those numbers. In effect, you are
splitting people by the day of the month they are born on, if there is
a tie, you use month and only use year at the end.
This would give a mix of ages, genders and any other measure in each district.
It is pretty much equivalent to just randomly distributing the voters
between the districts, but unlike a random system, it is harder to
corrupt.

It could have a similar result to having alphabetic ranked ballots, only
with birthdays instead of last names. The selection would be biased in
the direction of those that are born close to January. It may not
matter, but it would appear unfair.

If you have computers, you could just sort by SHA512(name concatenated
with birthdate concatenated with the year of the election). That's
probably overkill (since even if you could break SHA-512, which would be
a feat by itself, you'd have to convince the favored member to change
his name to something suitable), but then there'd be a sufficient margin
of safety. Randomness without randomness.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Juho]

Post by Stéphane Rouillon
Hello Juho,
using age, gender or other virtual dimension to build virtual
districts
replaces geographic antagonism by generation antagonism.

Ok, also that may happen. Each society should pick dimensions that
suit them best. (I'm just listing different options.)

Post by Stéphane Rouillon
The idea is to get equivalent sample that are not opposed by
intrinsec construction.

If the intention is not to divide people to groups that defend the
interests of that group (or just feel like being part of the same
group), then one could get rid of the districts altogether and use
only one nation wide district.

Post by Stéphane Rouillon
Thus we may find neutral decision takers that will minimize the overall
bad impacts of a decision, thus maximize to the best of their
knowledge
the decisions for all the electorate. If you split representative
into groups who have divergent opinions, the result will not
optimize common interest, it will only illustrate the "rapport de
force"
(maybe translated as power struggle) between the representatives.
Age representatives would hardly stay neutral while deciding
retirement fees and pensions for example.
The Irish senate based on profession seems one step toward getting
neutral decision takers
for deciding the localization of projects for example.
I prefer equivalent samples of the entire electorate (phone numbers
or hash tables using names could work too, but it has some slight
discrepancies and problems...)

Yes, if one wants to avoid any groupings (like age groups, regional
groups, races, political parties) then maybe electing a random set of
citizens (trying to avoid giving them the opportunity to organize
themselves) could be the best approach. Some groupings are however
likely to emerge afterwards even if we would elect the
representatives by some random selection method.

Juho

Post by Stéphane Rouillon

Subject: Re: [EM] No geographical districts
Date: Fri, 5 Sep 2008 00:13:34 +0300
Geographical proportionality is one specific dimension. Most
other dimensions could be called political dimensions. Also
groupings that do not live in any specific compact area could be
called political groupings. In principle they could form a party
and that way get a proportional number of representatives. (This
is also in line with the geographical proportionality related
target of guaranteeing representation from all _geographic_ areas.)
Many political systems have chosen geographical districts to be
fixed in the sense that people automatically "vote" for the
district where they live in. In the political dimension people
are typically allowed to pick the group that they want to
represent them.
It is possible to have election methods that support multiple
dimensions, i.e. more than these two. One could e.g. simply have
multiple orthogonal "party" structures and then in the vote
counting process force the representatives to be elected so that
proportionality will be respected in all dimensions.
There could be also additional "fixed dimensions" like automatic
fixed sex or age based proportionality.
Some of the additional dimensions could also be "virtual
districts" in the sense that each voter would be registered in
exactly one of them, and probably also vote only for candidates
that belong to one's own "virtual district". I understood that
you would use virtual districts to replace the current
geographical districts (and the geographical proportionality that
they represent).
The simplest (not necessarily optimal) approach to implement
multiple dimensions is one where you simply elect representatives
starting from the ones with strongest support (e.g. best
candidate of the largest party in the largest district), skip
candidates that can not be elected any more (e.g. district
already full, party already full), and continue until all seats
have been filled. At some point in the chain all "requirements"
of all dimensions are met if they are strong enough (and if there
are suitable candidates left).
(Some dimensions could be one-directional in the sense that one
would aim at guaranteeing at least a proportional share of the
seats but would not limit them to this number. For example one
could allow all members of some minority to require proportional
representation by marking this in their ballot. Other voters
would however not be required to vote either for or against this
minority. Any candidate (or any party, of any regions etc.) could
belong to this group. One should however not allow these lists to
overrule party proportionality or other "complete dimensions" (to
avoid riding under two flags (party and "minority") and getting
also corresponding double representation).)
Small ad here too. Trees (hierarchical candidate lists) offer
multiple dimensions in a simplified framework, but with
priorities involved too. One can e.g. be a greenish red or a
reddish green.
Juho

Post by Stéphane Rouillon
Hello electorama fans,
regarding that last comment, I invite those interested in non-
geographical district
to consider astrological district. The idea is to obtain
equivalent samples of the electorate
in term of any distribution (age, geography, profession,
language, religion,...) like
poll survey use. For example, in Quebec with near 4 000 000
electors, we could
obtain around 73 (73 x 5 = 365 days) of less than 55 000
electors each.
Thus electorate results could indicate a better performance from
some candidates
instead of reflecting the district bias produced by its design.
For example the first district could be formed with all
Quebecors born between
1st and 5th of january, the 2nd with Quebecors born between 6th
and 10th of january
and so on...
For more details of an electoral system using such "districts",
search for SPPA
(Scrutin Préférentiel, Proportionel et Acirconscriptif in french).
An english version is available on the electoral reform website
of the British-Colombia citizen assembly.
...

Post by Raph Frank
However, even something like "they should be compact" favours some
people. If you are part of a group that is spread evenly, then
you do
worse if the districts are compact. The problem is that philosophy
that districts should be geographically based.
----
Election-Methods mailing list - see http://electorama.com/em
for list info

----
Election-Methods mailing list - see http://electorama.com/em for list info

___________________________________________________________ Try
the all-new Yahoo! Mail. "The New Version is radically easier to
use" – The Wall Street Journal http://uk.docs.yahoo.com/
nowyoucan.html
----
Election-Methods mailing list - see http://electorama.com/em for list info

___________________________________________________________
All new Yahoo! Mail "The new Interface is stunning in its simplicity and ease of use." - PC Magazine
http://uk.docs.yahoo.com/nowyoucan.html

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Juho]

I like natural districts, so one approach would be to let people say
and let history decide. The reason why I find "natural" districts
natural in politics is that when people feel like they are part of
some community it is easier to find consensus and cooperate within
that community. And of course the border lines will then follow
whatever natural dividing lines there are.

One simple approach would be to ask the voters directly about the
(physical/mental) distances. The answers could be of e.g.
Village1>Village2>Village3>... There could be more villages on the
questionnaire than there will be districts. Also the home coordinates
of each voter would be known. With these values you can then draw the
districts so that you serve the voters/inhabitants as well as you can
an let them belong to the districts that they like (because of
driving distance or historical or whatever reason).

One could e.g. first pick n random villages as the district centres
(voters may or may not need to travel there to vote), and then
optimize the border lines so that more and more people are happier
and happier with their district, and occasionally change also the
chosen villages to check if that could lead to better results. Also
some basic rules needed to keep the districts compact. This is thus a
general optimization problem (and sufficiently "computationally
feasible" in line with the theme of this mail stream).

If you have single seat districts you have to force the districts to
be about equal in size. This will work against having natural border
lines, but at at least we can force the results to be as natural as
possible.

There's some risk of someone trying to gerrymander the border lines
by giving guidance on how to vote. But this may not be a practical
strategy.

Juho

Post by Jonathan Lundell
I haven't been following this thread in great detail, but I do have
a question: what is the distance function actually trying to
measure and minimize? What exactly are we trying to optimize when
we minimize "distance", by whatever measure? I might be close in a
crow's-flight sense to a neighbor on the other side of a river or
freeway with a driving distance of several miles.
OK, we could solve that in principle (though not too quickly) by
using Google Maps driving time, or the like. But what does driving
time have to do with grouping voters (unless we're drawing a
precinct and measuring travel time t the polling place)? Maybe we
mean communication: two people with telephones are closer in this
sense than people without (and people with unlimited long distance
are close in this sense independent of cartesian distance).
I live in several voting districts where geographical lines make
sense--my local fire district is a case in point, as would water
and sewerage (except that I provide my own in those cases, as it
happens). But legislative districts? I can't think what I'm trying
to optimize for which straight-line vs manhattan distance would be
relevant. I've never met my representatives on business; I
communicate with them via email or telephone, and I neither know
nor care where there local offices are.

Post by Raph Frank

I checked my code and I'm not doing the expensive square root. It's not
((dx*dx + dy*dy) * weight)
The weight gets nudged by multiplying by 1.01 or 0.99. Squaring the weight
or not and how it's nudged is probably just an efficiency issue and not a
correctness issue, it should still get to the weight it needs, just maybe
along some suboptimal curve.

I think it is equivalent anyway. Well, you might have to multiple by
1.02 and 0.98 instead to get the same as weight squared.

I think multiplicative weight makes more sense. No chance of
underflow, and
might scale better between districts with different densities.

Well, negative distances are fine too :).
However, it could mean that the point ends up outside its district.

Yes, the block shape file determines adjacency. They handily
label every
line segment with what block is on either side of it, so I can pretty
quickly filter that to keep a list of all pairings I've seen.

Ahh, cool.
I might have a look at that file again. It just seemed more trouble
than it was worth.
In my software, I just assume that the people are compressed down to
the location point given in the block group table.

Of course I do
also wind up going back and rendering that geometry to an actual map.

So, the maps on your site draw the boundary as it using the
block-group boundary info?
----
Election-Methods mailing list - see http://electorama.com/em for list info

----
Election-Methods mailing list - see http://electorama.com/em for list info

___________________________________________________________
All new Yahoo! Mail "The new Interface is stunning in its simplicity and ease of use." - PC Magazine
http://uk.docs.yahoo.com/nowyoucan.html

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Jonathan Lundell]

Post by Juho
I like natural districts, so one approach would be to let people say
and let history decide. The reason why I find "natural" districts
natural in politics is that when people feel like they are part of
some community it is easier to find consensus and cooperate within
that community. And of course the border lines will then follow
whatever natural dividing lines there are.

That seems wrong to me, but I don't have anything but subjective
impressions. Certainly for my local city council and school board the
community has no more consensus (and perhaps less) than one finds at
the state level. If anything, differences become more pronounced
locally.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Juho]

Post by Jonathan Lundell

Post by Juho
I like natural districts, so one approach would be to let people
say and let history decide. The reason why I find "natural"
districts natural in politics is that when people feel like they
are part of some community it is easier to find consensus and
cooperate within that community. And of course the border lines
will then follow whatever natural dividing lines there are.

That seems wrong to me, but I don't have anything but subjective
impressions. Certainly for my local city council and school board
the community has no more consensus (and perhaps less) than one
finds at the state level. If anything, differences become more
pronounced locally.

What I meant was that electing a school board that is responsible for
the local school is more natural than electing a school board that is
responsible for 50% of one school and 50% of another school. In
geographical districts that would mean border lines that follow the
rivers, highways, traditional understanding etc.

People tend to be "patriotic" and they share and defend the common
"values" of units that they consider to be "their own". Having such
natural units in politics could be considered positive as long as
people don't get too patriotic (and e.g. start "wars" against the
"competing units").

I agree that local matters are often more difficult to handle
peacefully than some more general matters. But natural districting
may still be considered a positive thing also at local level.

Juho






___________________________________________________________
To help you stay safe and secure online, we've developed the all new Yahoo! Security Centre. http://uk.security.yahoo.com

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Jonathan Lundell]

Post by Juho

Post by Jonathan Lundell

Post by Juho
I like natural districts, so one approach would be to let people
say and let history decide. The reason why I find "natural"
districts natural in politics is that when people feel like they
are part of some community it is easier to find consensus and
cooperate within that community. And of course the border lines
will then follow whatever natural dividing lines there are.

That seems wrong to me, but I don't have anything but subjective
impressions. Certainly for my local city council and school board
the community has no more consensus (and perhaps less) than one
finds at the state level. If anything, differences become more
pronounced locally.

What I meant was that electing a school board that is responsible
for the local school is more natural than electing a school board
that is responsible for 50% of one school and 50% of another school.
In geographical districts that would mean border lines that follow
the rivers, highways, traditional understanding etc.
People tend to be "patriotic" and they share and defend the common
"values" of units that they consider to be "their own". Having such
natural units in politics could be considered positive as long as
people don't get too patriotic (and e.g. start "wars" against the
"competing units").
I agree that local matters are often more difficult to handle
peacefully than some more general matters. But natural districting
may still be considered a positive thing also at local level.

I agree that there are some districts that are "naturally" local. I
mentioned fire districts earlier, as well as certain kinds of utilities.

School boards are a case in point. In California, local school boards
have very little to say about school policy, but they do have some
inherently local responsibilities, like hiring and firing the
superintendent/CEO, siting schools, etc.

These local districts, though, are already multi-member, and amenable
to PR within the district.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[James Gilmour]

Jonathan Lundell > Sent: Thursday, September 04, 2008 10:52 PM

Post by Jonathan Lundell

Post by Juho
I like natural districts, so one approach would be to let people say
and let history decide. The reason why I find "natural" districts
natural in politics is that when people feel like they are part of
some community it is easier to find consensus and cooperate within
that community. And of course the border lines will then follow
whatever natural dividing lines there are.

That seems wrong to me, but I don't have anything but subjective
impressions. Certainly for my local city council and school board the
community has no more consensus (and perhaps less) than one finds at
the state level. If anything, differences become more pronounced
locally.

It is not (or should not be) a question of whether or not there is a consensus at any particular geographical level of community.
The defining factors for "the geographical community" should be the level at which the electors can engage with the particular issue
and the level at which something can actually be done. In all cases the objective should be to ensure that the various "assemblies"
elected to deal with the issues are properly representative of those they are elected to serve. For city-wide issues, the
"geographical community" is the whole city. For issues affecting only my local school, the "geographical community" is the area of
the city served by that school - but if there are no fixed geographies associated with the various schools in the city, the
appropriate community for the school board is the families whose children attend the local school.

The physical geography of the real world (mountains, rivers, roads) and the social geography (travel to work patterns, traditional
administrative centres, major and minor shopping centres) are also important factors that should be taken into account in devising
electoral districts that "make sense" to local electors. In many cases these physical and social features define the communities
(at various levels) to which the electors feel they belong. Those who would devise new schemes of electoral districts, especially
to go with new voting systems, ignore the reality of these various "geographically defined communities" at their peril.

James

No virus found in this outgoing message.
Checked by AVG.
Version: 7.5.526 / Virus Database: 270.6.16/1650 - Release Date: 03/09/2008 16:13


----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

Post by James Gilmour
It is not (or should not be) a question of whether or not there is a consensus at any particular geographical level of community.
The defining factors for "the geographical community" should be the level at which the electors can engage with the particular issue
and the level at which something can actually be done. In all cases the objective should be to ensure that the various "assemblies"
elected to deal with the issues are properly representative of those they are elected to serve. For city-wide issues, the
"geographical community" is the whole city. For issues affecting only my local school, the "geographical community" is the area of
the city served by that school - but if there are no fixed geographies associated with the various schools in the city, the
appropriate community for the school board is the families whose children attend the local school.

This is called subsidiarity. It is (in theory) the guiding principle
when deciding if the EU as a whole or the individual members should
handle an issue.

It is a good idea. However, who gets to decide what is the correct
level. Often, it is the larger assembly that gets to decide if power
should be delegated to a smaller area.

In the US, the federal government decides to a certain extent what
power the States should have.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[James Gilmour]

From: Raph Frank > Sent: Friday, September 05, 2008 12:17 AM
On Thu, Sep 4, 2008 at 11:34 PM, James Gilmour

Post by James Gilmour
It is not (or should not be) a question of whether or not there is a
consensus at any particular geographical level of community. The
defining factors for "the geographical community" should be the level
at which the electors can engage with the particular issue and the
level at which something can actually be done. In all cases the
objective should be to ensure that the various "assemblies" elected to
deal with the issues are properly representative of those they are
elected to serve. For city-wide issues, the "geographical community"
is the whole city. For issues affecting only my local school, the
"geographical community" is the area of the city served by that school
- but if there are no fixed geographies associated with the various
schools in the city, the appropriate community for the school board is
the families whose children attend the local school.

This is called subsidiarity. It is (in theory) the guiding
principle when deciding if the EU as a whole or the
individual members should handle an issue.

With all due respect, what I was writing about was not subsidiarity. Nor has subsidiarity (senu stricto) anything to do with the
proposal for how the EU and its Member States should deal with issues, despite the abuse of the term "subsidiarity" in this context.
The EU proposals are all about devolution, i.e. handing down. (Never forget: "Power devolved is power retained.") Subsidiarity, on
the other hand, is about building decision-making structures from the bottom up, i.e. a lower (smaller, more local) group
voluntarily giving power to a higher (larger, more widespread) group only because the required decision can be made only at that
higher level or because the decision will be better made at that higher level.

One could image (NB imagine) subsidiary operating in the "schools case". Decisions affecting only our local school should be and
would be made within the school community, perhaps through the mechanism of a school board. Issues affecting school education
across the whole city can be made only at the whole city level, so they are remitted up to a wider geographical unit. And so on up.
But that is NOT what is actually in operation. Certain educational requirements are set in state law (Scotland in my case). The
state (Parliament and Government) has devolved the operational decision-making (and some policy making) to the 32 local authorities
(elected Councils). Some local authorities have devolved some (minor) aspects of decision-making to individual schools. So the
existing structure is a top-down one that has come about by a process of enforced centralisation followed by varying degrees of
devolution. It bears no resemblance to subsidiarity in origin, legal basis or operation.

It is a good idea. However, who gets to decide what is the
correct level. Often, it is the larger assembly that gets to
decide if power should be delegated to a smaller area.

If the larger assembly is deciding if power should be DELEGATED, it is devolution that is in operation, not subsidiarity.

In the US, the federal government decides to a certain extent
what power the States should have.

This is devolution, not subsidiarity. It may have started out as subsidiarity, i.e. the States agreed to give certain powers to the
federal centre, but that's not how it is today.

James

No virus found in this outgoing message.
Checked by AVG.
Version: 7.5.526 / Virus Database: 270.6.16/1652 - Release Date: 04/09/2008 18:54


----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

Post by James Gilmour
With all due respect, what I was writing about was not subsidiarity. Nor has subsidiarity (senu stricto) anything to do with the
proposal for how the EU and its Member States should deal with issues, despite the abuse of the term "subsidiarity" in this context.
The EU proposals are all about devolution, i.e. handing down. (Never forget: "Power devolved is power retained.") Subsidiarity, on
the other hand, is about building decision-making structures from the bottom up, i.e. a lower (smaller, more local) group
voluntarily giving power to a higher (larger, more widespread) group only because the required decision can be made only at that
higher level or because the decision will be better made at that higher level.

Erm, that *is* what the EU is about. The EU member countries are
giving up some of their power (the so-called 'pooled' sovereignty), so
that decisions on those issues can be handled at the EU level.

I did say in theory, because, it isn't really followed. The EU wants
as much power as it can get its hands on irrespective of the optimal
level of making decisions.

In any case, it can only take new powers if all countries agree.

However, the EU doesn't mandate that members actually implement
subsidiarity in their own countries.

Post by James Gilmour
If the larger assembly is deciding if power should be DELEGATED, it is devolution that is
in operation, not subsidiarity.

I guess it depends on how you want do define the term. I don't think
subsidiarity is determined by actual power, it is determined by where
the decisions are made.

Post by James Gilmour
From wiki

"Subsidiarity is an organizing principle that matters ought to be
handled by the smallest, lowest or least centralized competent
authority."

For example, if the UK Parliament passed a bill setting up the
structure you recommend for schools. Then there would be
subsidiarity, even though Parliament would retain the power to reverse
its decisions and take direct rule again.

If Parliament decides the levels that each type of decision is made,
then you would expect that they would assign more powers to larger
councils than is optimal (i.e. decide more centrally than is optimal).

OTOH, if all powers are given to the individual school boards, then
you would expect that powers would decided at smaller councils than
is optimal.

Generally, political bodies will keep powers to themselves, if at all
reasonable.

Post by James Gilmour

Post by Raph Frank
In the US, the federal government decides to a certain extent
what power the States should have.

This is devolution, not subsidiarity. It may have started out as subsidiarity, i.e. the States agreed to give certain powers to the
federal centre, but that's not how it is today.

I would say that the federal government is still somewhat restricted,
but certainly not as restricted as it was. It has large authority,
but not unlimited.

Btw, on the schools issue, how would you decide who can vote. Would
voters get to vote for the nearest school? Alternatively, only
parents with children in the school might be allowed to vote (or maybe
parents who have expressed an interest in sending their children to
that school).

Where should the decision on the question "Should parents be allowed
to send their children to any school that they want to?" be made?

Also, what is optimal for "Should we use subsidiarity to make decisions?".
----
Election-Methods mailing list - see http://electorama.com/em for list info

[James Gilmour]

Raph Frank > Sent: Friday, September 05, 2008 12:35 PM

Post by Raph Frank

Post by James Gilmour
With all due respect, what I was writing about was not subsidiarity.
Nor has subsidiarity (senu stricto) anything to do with the proposal
for how the EU and its Member States should deal with issues, despite
the abuse of the term "subsidiarity" in this context. The EU
"Power devolved is power retained.") Subsidiarity, on the other
hand, is about building decision-making structures from the bottom up,
i.e. a lower (smaller, more local) group voluntarily giving power to
a higher (larger, more widespread) group only because the required
decision can be made only at that higher level or because the
decision will be better made at that higher level.

Erm, that *is* what the EU is about. The EU member countries
are giving up some of their power (the so-called 'pooled'
sovereignty), so that decisions on those issues can be
handled at the EU level.

I don't agree with your analysis of what is really happening (or not happening) in the EU. Indeed, you say as much yourself in your
next paragraph.

Post by Raph Frank
I did say in theory, because, it isn't really followed. The
EU wants as much power as it can get its hands on
irrespective of the optimal level of making decisions.

What has to be remembered is that "the EU" is not a simple political structure. Instead there are three (unequal) parts to its
political structure: the Commission (like a very political civil service), the European Parliament (directly elected, but with less
power), and the Council (comprising ministers from the governments of the Member States). The real power lies with the Council, a
body that until very recently met in secret. In a UK context, it often suits UK Government ministers to allow blame to be directed
at "Brussels" (seat of the Commission) for unpopular EU-wide decisions, when the reality is that UK government ministers approved
those decision when they met in the EU Council. The Council, and the government ministers it comprises, are most reluctant to give
up any of the power they have.

< cut>

Post by Raph Frank

Post by James Gilmour
If the larger assembly is deciding if power should be DELEGATED, it is
devolution that is in operation, not subsidiarity.

I guess it depends on how you want do define the term. I
don't think subsidiarity is determined by actual power, it is
determined by where the decisions are made.

The power lies where the decisions are made, so I cannot see the distinction you are trying to make.

Post by Raph Frank
From wiki
"Subsidiarity is an organizing principle that matters ought
to be handled by the smallest, lowest or least centralized
competent authority."

That sounds fairly neutral, but there is something missing if that definition allows you to adduce the example below as an
illustration of "subsidiarity".

Post by Raph Frank
For example, if the UK Parliament passed a bill setting up
the structure you recommend for schools. Then there would be
subsidiarity, even though Parliament would retain the power
to reverse its decisions and take direct rule again.

This is not subsidiarity - this is devolution. The decision-make here was not bottom-up, it was top-down. There is a fundamentally
different philosophy in subsidiarity, even though the fully developed political structure may look nearly identical to one derived
by devolution.

Post by Raph Frank
If Parliament decides the levels that each type of decision
is made, then you would expect that they would assign more
powers to larger councils than is optimal (i.e. decide more
centrally than is optimal).
OTOH, if all powers are given to the individual school
boards, then you would expect that powers would decided at
smaller councils than is optimal.
Generally, political bodies will keep powers to themselves,
if at all reasonable.

Certainly in the UK, the history of political development over many centuries has been one of struggle to wrest power progressively
from the centre, originally from the monarch, but now from Parliament at Westminster (and similarly all the way down the various
levels of government).

< cut >

Post by Raph Frank
Btw, on the schools issue, how would you decide who can vote.
Would voters get to vote for the nearest school?
Alternatively, only parents with children in the school might
be allowed to vote (or maybe parents who have expressed an
interest in sending their children to that school).

I would say it depends on the responsibilities of the elected school board. If those responsibilities extend only to the education
of the children who are within the school, I would consider it reasonable that the electorate should comprise only the parents of
children at that school (one child = one vote). [You may also want to allow the children to elect their own representatives, but
that's a separate issue.] If the responsibilities extend to the provision of school education within a defined geographical area, I
would consider it reasonable for all electors within that geographical area to have a vote (one person = one vote).

Post by Raph Frank
Where should the decision on the question "Should parents be
allowed to send their children to any school that they want
to?" be made?

I don't think there is one simple answer to this question, except that it should not be made at the level of the individual school
because it affects a wider community. Given the existing political hierarchy, one could remit this decision up to the city level,
the state level or even the federal level. Views on the appropriate level will differ, but if we applied the original principles of
subsidiarity strictly, we should remit the decision up only to the lowest level above the individual school at which such a decision
could be made. That would suggest at the city level in the hierarchy I gave above. Others however would want uniformity on such
policy across the whole state or even across the whole federal country.

Post by Raph Frank
Also, what is optimal for "Should we use subsidiarity to make decisions?".

I don't think this question can be answered as you have asked it. Perhaps it would be more appropriate to ask "Do we want our
decision-making to be based on the principles of subsidiarity (i.e. bottom-up)?" There are also problems with regard to "optimal".
A benign dictator might make a more "optimal" decision than any democratic group, but on the other hand you could say that that
could never be "optimal" because it was not democratic. [As Professor Joab (BBC Brains Trust - long ago!) might have said: "It all
depends on what you mean by 'optimal'"!!]

James

No virus found in this outgoing message.
Checked by AVG.
Version: 7.5.526 / Virus Database: 270.6.16/1652 - Release Date: 04/09/2008 18:54


----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

Post by James Gilmour

Post by Raph Frank

Post by James Gilmour
If the larger assembly is deciding if power should be DELEGATED, it is
devolution that is in operation, not subsidiarity.

I guess it depends on how you want do define the term. I
don't think subsidiarity is determined by actual power, it is
determined by where the decisions are made.

The power lies where the decisions are made, so I cannot see the distinction you are
trying to make.

I am trying to split the decison about what level a particular power
is exercised and the power to actually make the decision.

For example, imagine you had 2 independent branches of government.

The first brach is hierarchical. There is parliament which covers the
whole territory and then progressively smaller subdivisions right down
to say areas with around 1000 people.

There is a second branch which has no power, other than the power to
decide at what level individual decisions should be taken at.

The would also need to be some mechanism to change the constitutional
arrangement.

The second branch could decide on a policy of centralisation or subsidiarity.

Maybe the second branch would operate like a court and apply well
defined rules to decide the issue In any case, I see subsidiarity as
a policy, not as a definition of power distribution.

Giving the smaller councils the right to block powers from moving
upwards is potentially just as bad as giving parliament the power to
block powers from moving downwards.

Maybe the second branch above would decide where a power rests but is
constrainted between two limits. It cannot be higher than the
top-down decision or lower than the bottom up decision.

Post by James Gilmour
Certainly in the UK, the history of political development over many centuries has been
one of struggle to wrest power progressively
from the centre, originally from the monarch, but now from Parliament at Westminster
(and similarly all the way down the various
levels of government).

Maybe I should have said 'will try' to keep any powers that they have.
They give them away only when they have little other choice.

Also, as you point out with the powers devolved are powers retained,
Parliament in the UK skill does retain all the devolved powers. The
powers taken from the King were taken, not given.

Scottish independance would be a time where power wasn't devolved.

Post by James Gilmour
That would suggest at the city level in the hierarchy I gave above.
Others however would want uniformity on such
policy across the whole state or even across the whole federal country.

Is this decision made by all the school boards? How is it handled, by
majority of the school boards or unanimous decision? Another option
would be a majority of all the members of the school boards and a
majority of all the school boards.

Post by James Gilmour
I don't think this question can be answered as you have asked it. Perhaps it would be
more appropriate to ask "Do we want our
decision-making to be based on the principles of subsidiarity (i.e. bottom-up)?"

Your proposed subsidiarity power process does follow from the concept
that each person controls themselves. All powers not delegated to the
group are retained. Similarly, powers that are not delegated by the
group to larger groups are retained by the group.

Post by James Gilmour
There are also problems with regard to "optimal".
A benign dictator might make a more "optimal" decision than any democratic group,
but on the other hand you could say that that
could never be "optimal" because it was not democratic. [As Professor Joab (BBC
Brains Trust - long ago!) might have said: "It all
depends on what you mean by 'optimal'"!!]

I think there can be optimality issues. Deciding where decisions
should be taken can have an effect on outcomes. Some of these
decisions would result in improvements that can be measured.

In any case, you get back to the circular question about who gets to
decide who gets to decide. This needs to be decided outside the
system in order to prevent the circle.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[James Gilmour]

Raph Frank > Sent: Friday, September 05, 2008 4:31 PM

I am trying to split the decision about what level a
particular power is exercised and the power to actually make
the decision.
In any case, you get back to the circular question about who
gets to decide who gets to decide. This needs to be decided
outside the system in order to prevent the circle.

If I have understood the principles of subsidiarity correctly, the basic power of decision-making ("sovereignty") rests at the
lowest level. It is at that level the decision was taken to remit some decision-making up to a higher level, for necessity or
convenience. So it is, or should be, at that lowest level that the fundamental decisions about "who decides" are made. The people
at the lowest level gave their power up (in both senses), so they should be able to take it back again. That would be the logical
inverse of devolution.

I am not aware of any community or hierarchy of communities that is actually organised according to the principles of subsidiarity.
I do, however, live in a community where there is something very similar: there is a upward hierarchy of decision-making that it is
not devolution, in that a higher level cannot impose any decision on a lower level when that decision properly belongs to the lower
level, nor can the higher level take away the power to make those decisions. This structure was not agreed among the participants
(who might spontaneously have embraced subsidiarity!), but was imposed by a (benign?) dictator - the property developer who built
the blocks of flats. The owners of the 146 dwellings in this development are bound by a Deed of Conditions that is part of the
property title. I am sovereign in all decisions affecting the inside of my flat. The 21 owners of flats in my block are sovereign
in all decisions affecting the common areas and structure of our block. The 62 owners of flats in our building (comprising 4
blocks) are sovereign in decisions that would affect the whole building, principally the common roof. The common areas of the whole
development (gardens, car parking bays, etc) are owned in common by all 146 proprietors and so sovereignty for decisions affecting
those common areas is vested in all 146 owners. If we wish, we can remit upwards, but no-one can impose downwards, nor can they
take any powers away.

James

No virus found in this outgoing message.
Checked by AVG.
Version: 7.5.526 / Virus Database: 270.6.16/1652 - Release Date: 04/09/2008 18:54


----
Election-Methods mailing list - see http://electorama.com/em for list info

[Kristofer Munsterhjelm]

Post by James Gilmour
Raph Frank > Sent: Friday, September 05, 2008 12:35 PM

Post by Raph Frank
Also, what is optimal for "Should we use subsidiarity to make
decisions?".

I don't think this question can be answered as you have asked it.
Perhaps it would be more appropriate to ask "Do we want our
decision-making to be based on the principles of subsidiarity (i.e.
bottom-up)?" There are also problems with regard to "optimal". A
benign dictator might make a more "optimal" decision than any
democratic group, but on the other hand you could say that that could
never be "optimal" because it was not democratic. [As Professor Joab
(BBC Brains Trust - long ago!) might have said: "It all depends on
what you mean by 'optimal'"!!]

By the standard of freedom, (real) subsidiarity seems only logical. If
you're an individual and you do something that has no impact on the rest
of the world, there would appear, by that measure, to be little point
for the rest of the world to interfere. If we make this more general,
then if you're a group and you do something that has no impact on anyone
but your group, then there would be little point for those outside to
interfere.

There are two problems with this, though. Some may use logic similar to
the potential way of arguing for a temporary regime of enforced equality
of minorities or sexes in an assembly: to interfere, if shortly, with
the dynamics of the system, one can quickly redirect it towards a
direction that it would naturally find but that would take significantly
longer had you not interfered.

The second is that, for some actions, it's not easy to say who will be
influenced. All may be, because we are social animals. Another argument
I've heard, regarding public services, is that if you do something that
would usually only affect yourself, but that you need a public service
because of this, then that's a concern of all. For instance, in a nation
with public healthcare, some may argue that if you do risky things, that
concerns society in general since you could end up getting hurt and need
the public health service, which the entire society maintains.


In conclusion, it appears that optimization for any given programmatic
sense of optimality can just as well expose edge or corner cases that
one didn't think of, as give what was intended. Still, as long as one
keeps the above in mind, I think subsidiarity is a good idea.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Juho]

Some notes on devolution and subsidiarity.

Maybe one could classify this so that if the real power is and
ultimate decision making happens at some level, then levels above
this level get their decision making power from below, and levels
below this level get their decision making power from above.

In different questions the decision making power may lie at different
levels. Schools may be quite autonomous low level entities. (One
could extend this also beyond politics, e.g. companies and families
that make their own decisions.) Countries typically decide on war and
peace. Larger structures like EU may decide on common rules for trade
in the unified market area.

However also in cases where power is at different levels on different
questions one may still see the semi-permanent decision making power
of various levels to be derived from some single strongest level that
has the ultimate decision making power in all questions. In our world
that is typically a country. Delegation of power upwards or downwards
is seldom irrevocable (to a country).

In democratic systems one can also say that the voters are the
ultimate decision makers that will make the decisions, and that the
country is just their slave. Here the country should however be seen
as the collection of its citizens. Individual citizens tend to be
still subjects of their country (could e.g. be thrown to jail when
needed). In this sense the state/country is the ultimate decision
maker even if it is seen as the sum of the opinions of its citizens.

What is then the optimal level for decision making? My basic guess is
that although in many questions it would be sensible to do the
decisions at the level that is the lowest that can handle that
question independently (=no external impact) it is typical that all
units with power have some tendency of grabbing more of that power.
The bigger entities typically win in this game. One should be quite
careful and monitor where the decision making power will move in time
(hopefully intentionally, not just as driven by individuals fighting
for power or believing in their own role too much).

Democracy is one invention that tends to balance the situation (when
compared to laws of jungle, dictators etc.). We can however still
improve a lot.

Since power tends to concentrate (move up) it makes sense to monitor
the process and lean al the time slightly towards defending the
rights of the small (individual citizens and lower decision making
levels).

Natural border lines are important. Countries that are not stable
typically have some problems with unnatural border lines and possibly
also with decision making structures that some people are not
comfortable with. I'm not that much in favour of artificially created
identities (maybe created with good intention to reduce conflicts,
since artificial unification may also cause conflicts, and especially
since individual and group differences are also a richness (that we
should try to benefit of instead of trying to eliminate them).

Juho

Post by James Gilmour

From: Raph Frank > Sent: Friday, September 05, 2008 12:17 AM
On Thu, Sep 4, 2008 at 11:34 PM, James Gilmour

Post by James Gilmour
It is not (or should not be) a question of whether or not there is a
consensus at any particular geographical level of community. The
defining factors for "the geographical community" should be the level
at which the electors can engage with the particular issue and the
level at which something can actually be done. In all cases the
objective should be to ensure that the various "assemblies"
elected to
deal with the issues are properly representative of those they are
elected to serve. For city-wide issues, the "geographical
community"
is the whole city. For issues affecting only my local school, the
"geographical community" is the area of the city served by that school
- but if there are no fixed geographies associated with the various
schools in the city, the appropriate community for the school board is
the families whose children attend the local school.

This is called subsidiarity. It is (in theory) the guiding
principle when deciding if the EU as a whole or the
individual members should handle an issue.

With all due respect, what I was writing about was not
subsidiarity. Nor has subsidiarity (senu stricto) anything to do
with the
proposal for how the EU and its Member States should deal with
issues, despite the abuse of the term "subsidiarity" in this context.
The EU proposals are all about devolution, i.e. handing down.
(Never forget: "Power devolved is power retained.") Subsidiarity, on
the other hand, is about building decision-making structures from
the bottom up, i.e. a lower (smaller, more local) group
voluntarily giving power to a higher (larger, more widespread)
group only because the required decision can be made only at that
higher level or because the decision will be better made at that higher level.
One could image (NB imagine) subsidiary operating in the "schools
case". Decisions affecting only our local school should be and
would be made within the school community, perhaps through the
mechanism of a school board. Issues affecting school education
across the whole city can be made only at the whole city level, so
they are remitted up to a wider geographical unit. And so on up.
But that is NOT what is actually in operation. Certain educational
requirements are set in state law (Scotland in my case). The
state (Parliament and Government) has devolved the operational
decision-making (and some policy making) to the 32 local authorities
(elected Councils). Some local authorities have devolved some
(minor) aspects of decision-making to individual schools. So the
existing structure is a top-down one that has come about by a
process of enforced centralisation followed by varying degrees of
devolution. It bears no resemblance to subsidiarity in origin, legal basis or operation.

It is a good idea. However, who gets to decide what is the
correct level. Often, it is the larger assembly that gets to
decide if power should be delegated to a smaller area.

If the larger assembly is deciding if power should be DELEGATED, it
is devolution that is in operation, not subsidiarity.

In the US, the federal government decides to a certain extent
what power the States should have.

This is devolution, not subsidiarity. It may have started out as
subsidiarity, i.e. the States agreed to give certain powers to the
federal centre, but that's not how it is today.
James
No virus found in this outgoing message.
Checked by AVG.
Version: 7.5.526 / Virus Database: 270.6.16/1652 - Release Date: 04/09/2008 18:54
----
Election-Methods mailing list - see http://electorama.com/em for list info

___________________________________________________________
The all-new Yahoo! Mail goes wherever you go - free your email address from your Internet provider. http://uk.docs.yahoo.com/nowyoucan.html

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Jonathan Lundell]

Post by James Gilmour
Those who would devise new schemes of electoral districts, especially
to go with new voting systems, ignore the reality of these various
"geographically defined communities" at their peril.

There's a proposition on the November ballot in California that would
establish an independent commission to draw (single-member) state
legislative districts (the legislature draws them now). The list might
be interested in the criteria specified for use by the commission.
Sorry, I only have a scan and don't feel like typing it in.

[Jonathan Lundell]

Post by James Gilmour
Those who would devise new schemes of electoral districts, especially
to go with new voting systems, ignore the reality of these various
"geographically defined communities" at their peril.

There's a proposition on the November ballot in California that
would establish an independent commission to draw (single-member)
state legislative districts (the legislature draws them now). The
list might be interested in the criteria specified for use by the
commission. Sorry, I only have a scan and don't feel like typing it
in.

To the list admin, fyi: this message was received by the list's
servers on 9/4, and sent on today (a second time; it was also sent on
9/4):

Received: from frida.dreamhost.com (frida.dreamhost.com
[66.33.206.23]) by

;

Tue, 30 Sep 2008 00:01:08 -0400 (EDT)
Received: from diego.dreamhost.com (diego.dreamhost.com
[66.33.216.104]) by
frida.dreamhost.com (Postfix) with ESMTP id 7961317FE3; Mon, 29 Sep 2008
21:00:51 -0700 (PDT)
Received: from diego.dreamhost.com (localhost [127.0.0.1]) by
diego.dreamhost.com (Postfix) with ESMTP id 251E948044; Mon, 29 Sep 2008
21:00:51 -0700 (PDT)
Received: from swarthymail-mx1.g.dreamhost.com
(swarthymail-mx1.g.dreamhost.com [208.113.200.31]) by
diego.dreamhost.com
(Postfix) with ESMTP id DC1B148044 for
<election-***@lists.electorama.com>; Thu, 4 Sep 2008 16:21:55 -0700
(PDT)
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Brian Olson]

Post by James Gilmour
Those who would devise new schemes of electoral districts, especially
to go with new voting systems, ignore the reality of these various
"geographically defined communities" at their peril.

There's a proposition on the November ballot in California that
would establish an independent commission to draw (single-member)
state legislative districts (the legislature draws them now). The
list might be interested in the criteria specified for use by the
commission. Sorry, I only have a scan and don't feel like typing it
in.
<pastedGraphic.jpg>----

California tried this a few years ago, and a few years before that,
and probably several times before that. They change around little
things about how the commission is formed or the guidelines for what
the commission should produce, but I think they always fail because:
1. The commission is ultimately unaccountable. The procedure for
making it is undemocratic, and while they're not beholden to the
powers we're trying to challenge, they're not accountable to the
citizens either. If we get random people on the commission we could
get random maps out, or maps actually designed by establishment
bureaucrats who know what they're doing.
2. The guidelines are vague and probably unenforceable. I'd bet
there's a good chance of a court challenge for any map that comes out
of the commission because it wouldn't be compact somewhere or it would
break some community of interest and any group that happened to or any
other group with an axe to grind with the map that came out could
challenge based on that. At worst the map would get redrawn by that
court case, possibly only to be left in a configuration that could
again be challenged, ad nauseum.

So, when I lived in CA, I voted against this nonsense. Oh, last time
they also included the bitter pill of immediate mid-term
redistricting, and it was being backed by a Republican governor, and
the specter of the Texas mid-term-gerrymander-House-theft loomed large.

And back then I decided the proposed system was crazy, and that I
could do better, and thus was born:
http://bolson.org/dist/CA/
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Stéphane Rouillon]

Can't the municipal level respond to the preocupations of these
"geographically defined communities" and the national/provincial/federal
level answer to other considerations?

Subject: Re: [EM] Geographical districts
Date: Thu, 4 Sep 2008 16:21:49 -0700

Post by James Gilmour
Those who would devise new schemes of electoral districts, especially
to go with new voting systems, ignore the reality of these various
"geographically defined communities" at their peril.

There's a proposition on the November ballot in California that would
establish an independent commission to draw (single-member) state
legislative districts (the legislature draws them now). The list might be
interested in the criteria specified for use by the commission. Sorry, I
only have a scan and don't feel like typing it in.
<< pastedGraphic.jpg >>
----
Election-Methods mailing list - see http://electorama.com/em for list info

----
Election-Methods mailing list - see http://electorama.com/em for list info

[James Gilmour]

Stéphane Rouillon > Sent: Tuesday, September 30, 2008 5:27 PM
Can't the municipal level respond to the preocupations of these
"geographically defined communities" and the
national/provincial/federal
level answer to other considerations?

This reflects a delightfully detached view of politics and the political system. The realities of the concerns of electors are
rather different in my experience.

I see from my original post (4/9/2008) that I was writing about:
"The physical geography of the real world (mountains, rivers, roads) and the social geography (travel to work patterns, traditional
administrative centres, major and minor shopping centres) are also important factors that should be taken into account in devising
electoral districts that "make sense" to local electors. In many cases these physical and social features define the communities
(at various levels) to which the electors feel they belong."

In the Canadian context, I suspect the physical and topographical features define "communities" at all levels below federal, and I
suspect the others have effects at a variety of different scales, so that they contribute to the definition of "community" at the
municipal, national and provincial levels.

James

No virus found in this outgoing message.
Checked by AVG.
Version: 7.5.526 / Virus Database: 270.7.5/1698 - Release Date: 29/09/2008 19:25


----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

Post by Jonathan Lundell
That seems wrong to me, but I don't have anything but subjective
impressions. Certainly for my local city council and school board the
community has no more consensus (and perhaps less) than one finds at the
state level. If anything, differences become more pronounced locally.

I think there is to a certain extent a critical mass to democracy.
Resident's associations can be a serious issue, even if democratic.

Ofc, town meetings allow more direct democracy. This may be more to
do with the lack of representation though.

I wonder would resident's associations work better if there was proxy
voting allowed.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Raph Frank]

Post by Juho
One simple approach would be to ask the voters directly about the
(physical/mental) distances. The answers could be of e.g.
Village1>Village2>Village3>... There could be more villages on the
questionnaire than there will be districts.

I suggested something similar a while back. Basically, you assume
that each resident ranks all the residents in order of distance.

You then hold a PR-STV election with 'seats' equal to the number of districts.

The winners then become the centres of the districts.

Ofc, under that system, the voters don't actually get to vote.

If voters were going to vote, maybe they could have less accuracy as
they go further away (e.g. the nearest 5 villages appear but
eventually it is only towns, then only cities). Also, it would be a
nightmare to count without computers.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Juho]

Post by Raph Frank

Post by Juho
One simple approach would be to ask the voters directly about the
(physical/mental) distances. The answers could be of e.g.
Village1>Village2>Village3>... There could be more villages on the
questionnaire than there will be districts.

I suggested something similar a while back. Basically, you assume
that each resident ranks all the residents in order of distance.
You then hold a PR-STV election with 'seats' equal to the number of districts.
The winners then become the centres of the districts.
Ofc, under that system, the voters don't actually get to vote.
If voters were going to vote, maybe they could have less accuracy as
they go further away (e.g. the nearest 5 villages appear but
eventually it is only towns, then only cities). Also, it would be a
nightmare to count without computers.

My plan was that the voters would actually vote, and not only to
indicate physical distances but also other preferences. That would
mean that village that is near but on the other side of the river
(and without a bridge) would not be ranked high. There would be
probably good correlation with driving time but there are also many
other factors. Maybe one favours the direction where one works, where
one's relatives live, where people speak the same language or have
the same religion, historically associated areas, areas with similar
needs, areas with similar minded people etc.

Use of computers in the vote counting phase should nowadays be ok.

Juho






___________________________________________________________
All New Yahoo! Mail – Tired of ***@gr@! come-ons? Let our SpamGuard protect you. http://uk.docs.yahoo.com/nowyoucan.html

[Kristofer Munsterhjelm]

I guess my time in Computer Science land has left me pretty comfortable
with the idea that there are lots of problems that are too hard to ever
reliably get "the best solution". I don't know if there's a short-short
popularizing explanation of how finding a good solution is Hard while
measuring the quality of a solution is pretty quick.
If anybody asks and it's not the time, place, or audience for discussing
NP Hard problems, I will wave my hands and say, "Hey, look over there!
Good results, on my one puny computer! With more it'd only get better!"

I think puzzles and games make good examples of NP-hard problems.
Sokoban is PSPACE-complete, and it's not that difficult to show people
that there are puzzles (like ciphers) where you know if a solution is
right, but it takes effort to find the solution. That's pretty much the
point of a puzzle, after all (although not all puzzles are NP-hard; they
can be fun even if they're not, as long as they do something for which
it's challenging to find a solution).
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Juho]

Post by Kristofer Munsterhjelm

Post by Brian Olson
I guess my time in Computer Science land has left me pretty
comfortable with the idea that there are lots of problems that are
too hard to ever reliably get "the best solution". I don't know if
there's a short-short popularizing explanation of how finding a
good solution is Hard while measuring the quality of a solution is
pretty quick.
If anybody asks and it's not the time, place, or audience for
discussing NP Hard problems, I will wave my hands and say, "Hey,
look over there! Good results, on my one puny computer! With more
it'd only get better!"

I think puzzles and games make good examples of NP-hard problems.
Sokoban is PSPACE-complete, and it's not that difficult to show
people that there are puzzles (like ciphers) where you know if a
solution is right, but it takes effort to find the solution. That's
pretty much the point of a puzzle, after all (although not all
puzzles are NP-hard; they can be fun even if they're not, as long
as they do something for which it's challenging to find a solution).

Puzzles and ciphers are good examples of cases where general
optimization may typically fail to find even a decent answer (well,
in these example cases the solution must be 100% good or it is no
good at all). My assumption was that in the area of voting methods it
would be typical that general optimization methods are sufficient and
will with good probability lead to good enough results. Are there and
counterexamples to this?

Juho






___________________________________________________________
The all-new Yahoo! Mail goes wherever you go - free your email address from your Internet provider. http://uk.docs.yahoo.com/nowyoucan.html

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Kristofer Munsterhjelm]

Post by Juho

Post by Kristofer Munsterhjelm
I think puzzles and games make good examples of NP-hard problems.
Sokoban is PSPACE-complete, and it's not that difficult to show people
that there are puzzles (like ciphers) where you know if a solution is
right, but it takes effort to find the solution. That's pretty much
the point of a puzzle, after all (although not all puzzles are
NP-hard; they can be fun even if they're not, as long as they do
something for which it's challenging to find a solution).

Puzzles and ciphers are good examples of cases where general
optimization may typically fail to find even a decent answer (well, in
these example cases the solution must be 100% good or it is no good at
all). My assumption was that in the area of voting methods it would be
typical that general optimization methods are sufficient and will with
good probability lead to good enough results. Are there and
counterexamples to this?

That gets harder, since most puzzles are of the "right or not" quality.
Games could count, but it muddies the situation because if games are
*-complete, they're usually PSPACE (because you have to come up with
something that works for all possible replies).

However, games that have rules governing the dynamics could be used. For
instance, finding out where to put the pieces in a Tetris game in the
absolutely best manner possible is NP-complete. Still, people manage to
play Tetris, because their approximations are good enough (or not, in
which case they usually lose at higher levels) and because the
situations aren't critical.

The problem with using these examples is that you lose the explaining
power. If you tell someone that placing pieces optimally in Tetris is
NP-complete, he won't get it, since to him Tetris is easy up to a
certain point (assume this is someone who knows how to play it), and the
reason he loses at a sufficiently level is because he can't approximate
fast enough, which probably has very little to do with asymptotic
complexity and rather much to do with the constants in the term.

So you'd have to use puzzles to explain the all-or-nothing version and
only then go on to the optimization version.

Note that I fudged my own explanation a bit here: optimal anything isn't
NP-complete, it's NP-hard. It's decision problem ("is there a way of
doing it better than x by some given measure") that's NP-complete, and
you can find the optimum (best value of x possible) by doing a binary
search.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[Juho]

The traditional algorithm complexity research covers usually only
finding perfect/optimal result. I'm particularly interested in how
the value of the result increases as a function of time. It is
possible that even if it would take 100 years to guarantee that one
has found the best solution, it is possible that five minutes would
be enough to find a 99% good solution with 99% probability.

Decrypting ciphered text does not work this way (the results could
still be worth 0 after 10 years with good probability). But solving
e.g. CPO-STV may well behave more this way (probably one can find an
80% good solution in one minute). Good performance in value/time
means that general optimization works (and the method can be
considered feasible in practice despite of being theoretically
infeasible).

Juho

Post by Kristofer Munsterhjelm

Post by Juho

Post by Kristofer Munsterhjelm
I think puzzles and games make good examples of NP-hard problems.
Sokoban is PSPACE-complete, and it's not that difficult to show
people that there are puzzles (like ciphers) where you know if a
solution is right, but it takes effort to find the solution.
That's pretty much the point of a puzzle, after all (although not
all puzzles are NP-hard; they can be fun even if they're not, as
long as they do something for which it's challenging to find a
solution).

Puzzles and ciphers are good examples of cases where general
optimization may typically fail to find even a decent answer
(well, in these example cases the solution must be 100% good or it
is no good at all). My assumption was that in the area of voting
methods it would be typical that general optimization methods are
sufficient and will with good probability lead to good enough
results. Are there and counterexamples to this?

That gets harder, since most puzzles are of the "right or not"
quality. Games could count, but it muddies the situation because if
games are *-complete, they're usually PSPACE (because you have to
come up with something that works for all possible replies).
However, games that have rules governing the dynamics could be
used. For instance, finding out where to put the pieces in a Tetris
game in the absolutely best manner possible is NP-complete. Still,
people manage to play Tetris, because their approximations are good
enough (or not, in which case they usually lose at higher levels)
and because the situations aren't critical.
The problem with using these examples is that you lose the
explaining power. If you tell someone that placing pieces optimally
in Tetris is NP-complete, he won't get it, since to him Tetris is
easy up to a certain point (assume this is someone who knows how to
play it), and the reason he loses at a sufficiently level is
because he can't approximate fast enough, which probably has very
little to do with asymptotic complexity and rather much to do with
the constants in the term.
So you'd have to use puzzles to explain the all-or-nothing version
and only then go on to the optimization version.
Note that I fudged my own explanation a bit here: optimal anything
isn't NP-complete, it's NP-hard. It's decision problem ("is there a
way of doing it better than x by some given measure") that's NP-
complete, and you can find the optimum (best value of x possible)
by doing a binary search.

___________________________________________________________
Now you can scan emails quickly with a reading pane. Get the new Yahoo! Mail. http://uk.docs.yahoo.com/nowyoucan.html

----
Election-Methods mailing list - see http://electorama.com/em for list info

[Kristofer Munsterhjelm]

Post by Juho
The traditional algorithm complexity research covers usually only
finding perfect/optimal result. I'm particularly interested in how the
value of the result increases as a function of time. It is possible that
even if it would take 100 years to guarantee that one has found the best
solution, it is possible that five minutes would be enough to find a 99%
good solution with 99% probability.
Decrypting ciphered text does not work this way (the results could still
be worth 0 after 10 years with good probability). But solving e.g.
CPO-STV may well behave more this way (probably one can find an 80% good
solution in one minute). Good performance in value/time means that
general optimization works (and the method can be considered feasible in
practice despite of being theoretically infeasible).

In computer science, you have something called polynomial time
approximation schemes (or PTASes). If a problem has a PTAS, then it's
possible to get within a factor e (epsilon) of being optimal, in
polynomial time, where the solution time for a fixed e is polynomial
(but need not be polynomial with respect to e). An approximation scheme
that runs in time n^(1 + 1/e) is a PTAS, although it's not polynomial
with respect to e.

The equivalent group for probabilistic algorithms is called "polynomial
time randomized approximation schemes", and give a result within e of
optimal in polynomial time /with high probability/.
----
Election-Methods mailing list - see http://electorama.com/em for list info

[James Gilmour]

Jonathan > Sent: Friday, September 05, 2008 12:11 AM

There's a proposition on the November ballot in California that would
establish an independent commission to draw (single-member) state
legislative districts (the legislature draws them now). The list might
be interested in the criteria specified for use by the commission.

In the UK, parliamentary and local government boundaries have been drawn by independent commissions for several decades. The
commissions make proposals which are open for public consultation and formal local inquiry if there are specified types of
objections to the proposals, and then the commission makes it recommendations. Because we operate a parliamentary system of
government, the commission's recommendations have to be given effect by an Order made by a government minister, subject to approval
by the relevant Parliament. So politicians can interfere in the process, but the last one who did so to any significant extent got
his just deserts at the following general election when his party lost every one of its MP seats in Scotland!

The Boundary Commission for Scotland was established in its current form by the Parliamentary Constituencies Act 1986. It
determines the boundaries for the UK Parliamentary constituencies (single-member FPTP) and for the constituencies and electoral
regions for the Scottish Parliament (regionalised MMP). A statutory review of the boundaries for the Scottish Parliament is in
progress.

The Local Government Boundary Commission for Scotland was created by the Local Government (Scotland) Act 1973. It carried out a
major review in 2006 to create the multi-member wards (local government electoral districts) required for the STV-PR elections held
in May 2007.

The different Commissions have to operate under slightly different rules.

These are the rules for the constituencies of the Scottish Parliament:

Rule 1
(1) So far as is practicable, regard must be had to the boundaries of local authority areas.
(2) “Area” and “local authority” have the same meanings as in the Local Government (Scotland) Act 1973.
Rule 2
(1) The electorate of a constituency must be as near the electoral quota as is practicable, having regard to Rule 1.
(2) The Boundary Commission for Scotland may depart from the strict application of Rule 1 if they think that it is desirable to do
so to avoid an excessive disparity between the electorate of a constituency and the electoral quota or between the electorate of a
constituency and that of neighbouring constituencies.
(3) The electoral quota is the number obtained by dividing the total electorate by 71.
(4) The electorate of a constituency is the aggregate of the persons falling within paragraphs (5) and (6) below.
(5) A person falls within this paragraph if his name appears on the register of local government electors in force on the
enumeration date under the Representation of the People Acts for a local government area which is situated wholly in the
constituency.
(6) A person falls within this paragraph if his name appears on the register of local government electors in force on the
enumeration date under the Representation of the People Acts for a local government area which is situated partly in the
constituency and his qualifying address is situated in the constituency.
(7) The total electorate is the total number of persons whose names appear on the registers of local government electors in force on
the enumeration date under the Representation of the People Acts for all of the local government areas in Scotland (except the local
government areas of Orkney and Shetland).
(8) The enumeration date is, in relation to a report of the Boundary Commission for Scotland, the date on which notice with respect
to the report is published in the Edinburgh Gazette in accordance with paragraph 7(1) above.
(9) “Qualifying address” and “local government area” have the same meanings as in the Representation of the People Act 1983 (c. 2).
Rule 3
The Boundary Commission for Scotland may depart from the strict application of Rules 1 and 2 if they think that special geographical
considerations (including in particular the size, shape and accessibility of a constituency) render it desirable to do so.
Rule 4
The Boundary Commission for Scotland need not aim at giving full effect in all circumstances to Rules 1 to 3 but they must take
account (so far as they reasonably can) —
(a) of the inconveniences attendant on alterations of constituencies other than alterations made for the purposes of Rule 1, and
(b) of any local ties which would be broken by such alterations.


These are the rules for the local government wards:

1 (1) This Schedule applies to the consideration by the Secretary of State or the Boundary Commission of the electoral arrangements
for election of councillors of local government areas.
(2) Having regard to any change in the number or distribution of electors of a local government area likely to take place within
the period of five years immediately following the consideration, the number calculated by dividing the number of local government
electors in each electoral ward of that local government area by the number of councillors to be returned in that ward shall be, as
nearly as may be, the same.
(3) Subject to sub-paragraph (2) above, in considering the electoral arrangements referred to in sub-paragraph (1) above regard
shall be had to —
(a) the desirability of fixing boundaries which are and will remain easily identifiable;
(b) any local ties which would be broken by the fixing of any particular boundary,
but if, in any case, there is a conflict between those criteria, greater weight shall be given to the latter.
2 The strict application of the rule stated in paragraph 1(2) above may be departed from in any area where special geographical
considerations appear to render a departure desirable.


One of the key differences between these two sets of rules is the extent to which the commissions can take identifiable
"communities" into account, to avoid splitting them artificially to meet the equality of numbers criteria. The Local Government
Boundary Commission for Scotland went much further in respecting communities that many people thought it would, but its options were
severely constrained by the requirement for all wards to return either three or four councillors. Many of us argued for greater
flexibility in ward sizes when the legislation was being debated in the Scottish Parliament, but we lost that argument, so we are
stuck with 3s and 4s for the time being.

James

No virus found in this outgoing message.
Checked by AVG.
Version: 7.5.526 / Virus Database: 270.6.16/1652 - Release Date: 04/09/2008 18:54


----
Election-Methods mail