[EL] Manchin

[Doug Spencer]

Doug,

It's true that for years the MCMC algorithms developed by academics and the
code used by experts (sometimes these same academics, but not always) were
not disclosed. But things have begun to change. As a starting point, it's
helpful to note that (1) Markov Chains have only recently been adopted in
the redistricting space even though scholars have been using computers
since the 1960s to explore the implications of randomized redistricting,
and (2) not all MCMC algorithms are the same. There is a short and
accessible summary of the development and use of Markov Chains in this
piece that I wrote with Moon Duchin
<https://www.yalelawjournal.org/forum/models-race-and-the-law> a few months
ago (see pp. 767-772). There is a somewhat longer, but I think equally
accessible, summary in this forthcoming book chapter
<https://arxiv.org/abs/2011.09504> by Justin Solomon (MIT) and Amy Becker
(Brown).

To more directly respond to your query, there are now a handful of
open-source algorithms:

"GerryChain <http://github.com/mggg/gerrychain>" is a Python code base for
the "Recombination" method developed by the MGGG team at Tufts/MIT. The
package also runs flip chains and IID methods as well. (There is a version
in Julia <http://github.com/mggg/gerrychainjulia> which runs faster).

"Merge-Split <http://git.math.duke.edu/gitlab/gjh/mergesplitcodebase>" is
another Python code base for running the "merge-split" algorithm developed
by Greg Herschlag and Jonathan Mattingly at Duke.

"redist <https://alarm-redist.github.io/redist/>" is an R package that runs
ReCombination, flips and a hierarchical IID method, developed by a team at
Harvard.

Finally, for a sample implementation, note that Jowei Chen (frequent
redistricting expert) posted the code---a mix of Python, Java, and
R---behind a recent YLJ article
<https://www.yalelawjournal.org/article/the-race-blind-future-of-voting-rights>
with Nick Stephanopoulos. See here:
http://www-personal.umich.edu/~jowei/race/


Like you, I'm hopeful that the use of these algorithms in litigation during
this next redistricting cycle will be more transparent than it has been in
previous years. To the extent that experts use the algorithms/stat packages
above, there is no reason why they shouldn't be.

Doug

-----
*Douglas M. Spencer*
*Professor of Law and Public Policy*
University of Connecticut
65 Elizabeth Street
Hartford, CT 06105
(773) 633-5196
http://www.dougspencer.org


On Thu, Jun 17, 2021 at 10:19 AM Douglas Johnson <djohnson at ndcresearch.com>
wrote:

> Is there open source code anywhere for a Markov Chain or other
> redistricting algorithm? All I have seen so far are proprietary programs
> that were kept secret even during trial. Any links to sample programs would
> be appreciated.
>
> - Doug
>
> On Thu, Jun 17, 2021 at 9:11 AM Steve Hoersting <hoersting at gmail.com>
> wrote:
>
>> The Manchin approach is brilliant. I would be surprised if he isn’t
>> submitting the proposal on behalf of a wise tactician.
>>
>> … a very wise tactician: The People must believe in elections, after all,
>> else what’s the point? (And H.R. 1 is — in the public mind —on that metric,
>> now a nonstarter).
>>
>> I.D. will go National, in form or another, in the near term. And will be
>> available to all, regardless of status. Team Manchin would know that.
>> (National Security, is as good a rationale as any). Thus keeping a certain
>> amount of … well, fluidity, in the system.
>>
>> And maintenance of the rolls will not be a barrier once new arrivals get
>> status. How far off can that be?
>>
>> The Manchin compromise isn’t as brilliant as Ellen Weintraub’s October
>> 2015 NYT op-Ed — which I praised to the hilt for its strategic genius, at
>> the time. But it’s in the same vein.
>>
>> The larger problem will remain — for followers of Ellen … and of Senator
>> Manchin:
>>
>> What to do when the new arrivals deploy Newton’s Scientific Method — as a
>> certain percentage of humanity must — and find themselves wanting, for
>> themselves, independently premised economic prosperity — no less than do a
>> goodly portion of the 330 Million of us already here — those already voting
>> in the Several States… after reasonably free speech campaigns… on
>> reasonably maintained voter rolls?
>>
>> All the best,
>>
>> Steve
>>
>> Sent from my iPhone
>> _______________________________________________
>> Law-election mailing list
>> Law-election at department-lists.uci.edu
>> https://department-lists.uci.edu/mailman/listinfo/law-election
>
>
>
> --
> Douglas Johnson
> National Demographics Corporation
> djohnson at NDCresearch.com
> phone 310-200-2058
>
> _______________________________________________
> Law-election mailing list
> Law-election at department-lists.uci.edu
> https://department-lists.uci.edu/mailman/listinfo/law-election
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://webshare.law.ucla.edu/Listservs/law-election/attachments/20210617/29c932f3/attachment.html>