Typically uninformed review of the Shortest Splitline Algorithm

[Clay Shentrup]

http://www.dailykos.com/story/2015/05/12/1384062/-Compactness-is-a-terrible-standard-for-redistricting-and-determining-if-maps-were-gerrymandered

Thus it makes no sense to combine areas that have nothing in common except that they fit neatly into a square. Districts should be united by common factors such as demography, culture, socioeconomic class, and geography.

You might think these neat lines above would produce fair outcomes and protect the process from bias resulting from gerrymandering. However, all it does is needlessly and unproductively split communities, cities, and counties. Just as importantly, it violates the Voting Rights Act and denies black voters the ability to elect the candidate of their choice.

[Dominic P]

I'm disappointed he only barely touches on proportional representation. The shortest split line algorithm shows it's quality when considered along side other voting and electoral reforms.

[Clay Shentrup]

I mean, look at that map. How can you deny it's a huge improvement over the status quo?

[Bruce Gilson]

On Fri, May 15, 2015 at 12:00 AM, Clay Shentrup <cl...@electology.org> wrote:

I mean, look at that map. How can you deny it's a huge improvement over the status quo?

​It may be an improvement over the (gerrymandered) status quo in some respects, but the article is on point, I think, in saying that splitting communities, putting rural and urban voters together, and such makes for representatives that do not have the ability to represent.

 

​I have never liked the SSA, and have said so, because of some of the very reasons he cites.​

[Warren D Smith]

Stephen Wolf's piece didn't make much sense to me.
It's basically whining -- kind of like a baby screaming --
not an actually coherent statement of anything, nor expressing any
actually clear claim. Well, there was one actually-clear claim:
the pictured splitline districting of N.Carolina violates the voting rights act.

Wolf makes that claim with no evidence or rationale whatever, and
the full text of the voting rights act (before much of it was invalidated
by the supreme court) is here:
http://www.rangevoting.org/VRAtext.html
I fail to see the wording within it which makes splitline map illegal.



--
Warren D. Smith
http://RangeVoting.org <-- add your endorsement (by clicking
"endorse" as 1st step)

[Markus Schulze]

Hallo,

I see the following problems with the
Shortest Splitline Algorithm:

1) When the distribution of the population is uneven,
then the shortest cut is not necessarily a straight line.
It could also be a curve.

2) Even when the population is even, it is not guaranteed
that you get the most compact districting by cutting
successively the electorate into halves.

Markus Schulze

[Warren D Smith]

On 5/15/15, Markus Schulze <markus....@alumni.tu-berlin.de> wrote:
> Hallo,
>
> I see the following problems with the
> Shortest Splitline Algorithm:
>
> 1) When the distribution of the population is uneven,
> then the shortest cut is not necessarily a straight line.
> It could also be a curve.

--true. (Which I already knew.)
Incidentally, the curve is always an arc of
a circle if the two populations on each side of it each are uniform
(but at two different densities).

> 2) Even when the population is even, it is not guaranteed
> that you get the most compact districting by cutting
> successively the electorate into halves.

--also true. (Also already knew.)
In fact, the "least cutting" optimum districting is usually not the same as the
splitlined map. However
(a) splitlining empirically usually does well (even though I can invent
artificial examples where it does poorly compared to optimum)
(b) optimum map is NP-hard to find, but splitlining is algorithmically easy.

Open question: find polytime algorithms to find maps with provable
approximation guarantees, such as within a constant factor of optimum
cutting length.

[Markus Schulze]

Hallo,

I wrote (15 May 2015):

> I see the following problems with the
> Shortest Splitline Algorithm:
>
> 1) When the distribution of the population is uneven,
> then the shortest cut is not necessarily a straight line.
> It could also be a curve.

Example: A state has to be divided into two districts.
One half of the population lives in a metropolitan (and
geographically small) area in the middle of this state.
The other half is distributed evenly across the rest of
this state.

Then the most compact districting would be to say that
this metropolis forms the one district and the rest of
this state forms the other district.

However, the Shortest Splitline Algorithm would
necessarily cut this metropolis into halves, because
this algorithm always uses straight lines.

Warren D. Smith replied (15 May 2015):

> --true. (Which I already knew.)

Thank you for acknowledging this problem.
But how do you then claim that the Shortest Splitline
Algorithm is still a good districting method?

Markus Schulze

[Clay Shentrup]

On Saturday, May 16, 2015 at 1:03:25 AM UTC-7, Markus Schulze wrote:

 > --true. (Which I already knew.)

Thank you for acknowledging this problem.
But how do you then claim that the Shortest Splitline
Algorithm is still a good districting method?

How is it a "problem"? You use the shortest line *that divides the two halves into the right population ratio*. Who cares if this isn't the actual shortest line? 

[Warren D Smith]

Markus, see
1. http://rangevoting.org/TheorDistrict.html
2. http://rangevoting.org/SplitLR.html

Is splitline a "good" districting method? Partly a matter of opinion,
partly testable.
Ivan Ryan ran it on all 50 US states in (2) and it seemed like the
splitline maps
were superior to the official maps in either 49 or all 50 cases. The
exception was Colorado.

If you can solve the "open mathematical question" I said last post, please do.
I had some interesting ideas, but did not succeed. I suspect any
algorithm that does solve
it efficiently (if possible) will be very complicated.

The DailyKos idiot Stephen Wolf proclaimed splitline "made no sense"
(whatever that meant). He claimed splitline was illegal under the VRA
and that the VRA requires a plurality-black district to exist. Both
are false statements; I have read the full text of the VRA,
and also there exist many US states right now with zero
plurality-black districts.
He also failed to mention much of the VRA has been voided by the
supreme court anyhow.

Wolf had the idea maps drawn by well meaning people that he approved
of, would be superior to splitlining. That is precisely the same
stance that every gerrymanderer in the history of the universe, has
taken.

He then exhibited two maps that he claimed were better than splitlining.
It looked like they used county borders as district borders, in which
case his two
maps are definitely illegal because their districts are not equipopulous.
He presented zero evidence for equipopulousness.

He also contended his map was better because it "made sense" to group
people he considered alike, together in districts. Since he figured
all blacks are alike, he figured all the blacks had to be packed into
a district. Since he figured all urbanites are alike, and all
educated people are alike, he tried to pack them into single districts
too.

Because that "made sense."

(That happens to be very convenient for the Republican party because
those types of
people currently tend to vote Democrat, so they'll all be packed into
a few districts
thus assuring Republican domination of the rest of the state.)