[EM] [CES #3089] Re: Theoretical Issues In Districting

[Dave Ketchum]

A bit of thinking, and a bit of personal history.

I see no value in splitline.
.     It happily mixes city and rural and suburbs - city and rural  
each should be kept together, as should suburbs, though suburbs fit  
with either of the first two.
.     It happily mixes new collections of people, giving them little  
opportunity to get together and work together.

1990 - NY-28 includes Kimgston on the Hudson, Ithaca on the Finger  
Lakes, and Owego where I live.  FAR from compact.

1992 - NY-26 inherits above NY-28 description.  Assemblyman Hinchey  
from near Kingston is completing 18 years in Albany and gets elected  
to Congress.

2002 - NY-22 inherits above NY-28 description.  How tightly can a  
waist be bound?  Near Nichols NY-22 northern boundary, on the  
Susquehanna River, is less than 5 miles from PA.
.     Congressman Hinchey, completing 10 years, is reelected.

2012 -  Hinchey is completing 20 years.  NY will have two less  
congressmen.  NY's habit is to keep current districts, amended as  
needed for census results, so what to do?
.     NYC area needs to lose one and a scandal leaves nothing to save  
in NY-9 - so dump that one.
.     NY-26 is having a special election, so that seems like a good  
prospect.  Hochul's win makes her deserve a full term, so look  
elsewhere.

Dave Ketchum

On Jun 14, 2011, at 10:33 AM, Warren D. Smith (CRV cofounder, http://RangeVoting.org 
) wrote:

> I think Justin Levitt's view of optimal districting, is basically
> this.
> (Although perhaps this is a caricature? I'm not trying to caricature,
> I'm just trying to present just an honest picture of what, as far as I
> can tell JL thinks -- but I'm only going by his emails, not his paper
> "weighing the potential of citizen redistricting" which he emailed me
> the pdf of 2 times, but both times my computer refused to open it
> claiming file was invalid/corrupted etc. Can anybody else obtain/read
> that paper?  Perhaps if you can convert it to postscript it'd fix it?)
>
> Justin Levitt's view as described by WDS:
> There should be some committee of beneficent people, unbiased by party
> politics, who draw the districts in such a way as to help everybody,
> because they have beneficent "purposes" in mind.  These people should
> not care about how the map looks, they should care about what purposes
> it accomplishes. (JL made the analogy of "Susan Boyle," a singer who,
> he claimed, did not have a very good visual appearance, but sung well,
> and, JL said, that proves appearance does not matter, what matters is
> results.)   JL disparages mathematical approaches, because with them
> the human element is sacrificed, and because they concentrate on
> appearance, not -- what really matters -- results.
> These beneficent people need to cluster people with common interests
> into common districts, so that their representatives will be able "to
> know what they represent."   But what exactly is a "common interest?"
> What qualifies, and what does not?  Does "lovers of feathered animals
> who also like
> mining gravel" count as a "common interest"?  Does "likes reality TV
> shows" count? And what if you are BOTH Black, AND a Commie Sympathizer
> (2 "interests" simultaneously) but can only be located in one
> district?  Then what? Well, the beneficent people will decide those
> things.   They're kind of like your big brother, helping everybody to
> overcome those annoying real-world problems to get good results.
>
> What will be the net effect  of this?   Well, it will be essentially
> this.  That committee
> will decide (a) what are the top issues of the day and (b) who wins on
> each issue.  But they will
> not have total power on (b) because gerrymandering is only capable of
> making a 26% minority win a 2-way choice, not a 24% minority. So
> subject to those limitations they'll effectively BE the government.
> So then the question arises: how are they to be elected, or appointed,
> or randomly chosen, or what?  It's a bit difficult to elect them,
> because almost all people do not even know who even a single such
> committee member is, and also do not know what each one did and how
> each one affected the district maps, and even with maximum possible
> effort to make the process transparent (which, as far as I know, has
> never happened in the prior history of the universe, but I suppose it
> could) it would still be very hard for Joe Voter to understand+know
> that.   They could be appointed, in which case you can be damn sure
> the appointer will have a pretty good idea how each appointee will
> behave, and now this appointer will effectively be the government.  Of
> course the committee-candidates could try to overcome that by lying to
> him.  Finally, they could be randomly chosen, in which case the main
> decisions made by our government will basically be decided by dice
> rolls.  Perhaps the best such system would be something like the way
> juries are selected -- random selection followed by a deterministic
> winnowing conducted by the legislature. In that case I daresay the
> committee would be biased to try to help some legislative majority
> keep their seats, but not as biased as if the legislature itself did
> the districting; and there still will be a great deal of randomness
> involved.
> A committee member like Justin Levitt himself might have enormous
> influence due to (let's say)
> his comparative expertise at using computer district-drawing systems,
> and his philosophy which
> might differ considerably from the others' philosophies.
>
> All this would, compared to the splitline algorithm, produce districts
> at enormously higher cost in time and money (but still small compared
> to the cost of the whole government) and based on
> almost all prior history also usually worse by various mathematical
> compactness measures.  Some US states might get districted in quite
> different ways than other states. As far as I know, JL concedes those
> points, but still says the results will be better his way.
>
> In particular, I think JL is under the impression that the civil
> rights advances during the 1960s were due to gerrymandering intended
> to create majority-black districts.