Redistricting Algorithms Micah Altman <[email protected]> Director of Research -- MIT Libraries, Massachusetts Institute of Technology Non-Resident Senior Fellow, Brookings Institution Prepared for International Seminar on Electoral Boundaries Delimitation Instituto Federal Electoral Mexico City, November 2012
Redistricting Algorithms
May 09, 2015
During this presentation I was able to reflect on the interplay of algorithms and public participation. And it became even clearer to me that applications like DistrictBuilder exemplify the ability of information science to improve policy and politics.
Redistricting in Mexico is particularly interesting, since it relies heavily on facially neutral geo-demographic criteria and optimization algorithms -- which represents a different sort of contribution from information science. Thus, it was particularly interesting to me to consider the interplay between algorithmic approaches to problem solving and "wisdom of crowd" approaches -- especially for problems in the public sphere.
It's clear that complex optimization algorithms are an advance in redistricting in Mexico, and have an important role in public policy. However, they also have a number of limitations:
Algorithmic optimization solutions often depend on a choice of (theoretically arbitrary) 'starting values' from which the algorithm starts its search for a solution
Quality algorithmic solutions typically rely on accurate input data
Many optimization algorithms embed particular criteria or particular constraints into the algorithm itself
Even where optimization algorithms are nominally agnostic to the criteria used for the goal, some criteria are more tractable than others; and some are more tractable for particular algorithms
In many cases, when an algorithm yields a solution, we don't know exactly (or even approximately, in any formal sense) how good that solution is.
I argue that explicitly incorporating a human element is important for algorithmic solutions in the public sphere. In particular:
Use open documentation and open (non-patented, or open-licensed) to enable external replication of algorithms
Use open source to enable external verification of the implementation of particular algoritms
Incorporate public input to improve the data (especially describing local communities and circumstances) in algorithm driven policies.
Incorporate crowd-sourced solutions as candidate "starting values" for further algorithmic refinement
Subject algorithmic output to crowd-sourced public review to verify the quality of the solutions produced
Redistricting in Mexico is particularly interesting, since it relies heavily on facially neutral geo-demographic criteria and optimization algorithms -- which represents a different sort of contribution from information science. Thus, it was particularly interesting to me to consider the interplay between algorithmic approaches to problem solving and "wisdom of crowd" approaches -- especially for problems in the public sphere.
It's clear that complex optimization algorithms are an advance in redistricting in Mexico, and have an important role in public policy. However, they also have a number of limitations:
Algorithmic optimization solutions often depend on a choice of (theoretically arbitrary) 'starting values' from which the algorithm starts its search for a solution
Quality algorithmic solutions typically rely on accurate input data
Many optimization algorithms embed particular criteria or particular constraints into the algorithm itself
Even where optimization algorithms are nominally agnostic to the criteria used for the goal, some criteria are more tractable than others; and some are more tractable for particular algorithms
In many cases, when an algorithm yields a solution, we don't know exactly (or even approximately, in any formal sense) how good that solution is.
I argue that explicitly incorporating a human element is important for algorithmic solutions in the public sphere. In particular:
Use open documentation and open (non-patented, or open-licensed) to enable external replication of algorithms
Use open source to enable external verification of the implementation of particular algoritms
Incorporate public input to improve the data (especially describing local communities and circumstances) in algorithm driven policies.
Incorporate crowd-sourced solutions as candidate "starting values" for further algorithmic refinement
Subject algorithmic output to crowd-sourced public review to verify the quality of the solutions produced
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Redistricting Algorithms
Micah Altman <[email protected]> Director of Research -- MIT Libraries,
Massachusetts Institute of TechnologyNon-Resident Senior Fellow,
Brookings Institution
Prepared for International Seminar on Electoral Boundaries Delimitation
Instituto Federal ElectoralMexico City, November 2012
Collaborators*
[11/9/2012]Redistricting Algorithms2
Michael P. McDonaldAssociate ProfessorDepartment of Public and International AffairsGeorge Mason UniversityWeb: http://elections.gmu.edu
Karin Mac DonaldDirector, Statewide Database & Election Administration Research CenterU.C. BerkeleyWeb: http://swdb.berkeley.edu/
Research SupportThanks to the Sloan Foundation, the Joyce Foundation,
National Science Foundation * And co-conspirators
Related Work
[11/9/2012]Redistricting Algorithms3
Reprints available from: http://micahaltman.com
M. Altman, (1997). "Is Automation the Answer? The Computational Complexity of Automated Redistricting", Rutgers Computer and Technology Law Journal 23 (1).
M. Altman, (1998). "Modeling the Effect of Mandatory District Compactness on Partisan Gerrymanders", Political Geography 17 (8): 989-1012.
M. Altman, (2002). "A Bayesian Approach to Detecting Electoral Manipulation" Political Geography 22(1).
M. Altman, K. Mac Donald, and M. P. McDonald, (2005). "From Crayons to Computers: The Evolution of Computer Use in Redistricting" Social Science Computer Review 23(3).
M. Altman, K. Mac Donald, and M. P. McDonald, (2005). "Pushbutton Gerrymanders", in Party Lines: Competition, Partisanship, and Congressional Redistricting Thomas E. Mann and Bruce E. Cain (eds), Brookings Press.
M. Altman & M.P. McDonald. (2010) “The Promises and Perils of Computer Use in Redistricting”, Duke Constitutional Law and Policy Journal, 5(69).
M. Altman & M.P. McDonald. (2011). "BARD: Better Automated Redistricting." Journal of Statistical Software 42(4).
M. Altman, & M. P. McDonald, (2012). ”Technology for Public Participation in Rdistricting", in Redistricting and Reapportionment in the West, G. Moncrief (ed.), Lexington Press.
How are computers used in redistricting?
[11/9/2012]Redistricting Algorithms4
Evolution of Computing for Redistricting
[11/9/2012]Redistricting Algorithms5
6
Automated Redistricting Is Easy-- If you ignore solution quality
[11/9/2012]Redistricting Algorithms
Choose a starting point Examine local trades
– choose one that yield most improvement Repeat until no further improvement is
possible
The Quality Problem – Local Optimum
Automated redistricting algorithms yield a solution
Practical algorithms not guaranteed to yield best solution
[11/9/2012]Redistricting Algorithms7
Automated Redistricting is Fundamentally Difficult
[11/9/2012]Redistricting Algorithms8
Why not look at all possible solutions?
(TOO MANY)
Redistricting using common criteria is NP-complete [Altman 1997; Puppe & Tasnadi 2008, 2009]
Not mathematically possible to find optimal solutions to general redistricting criteria!
State of the Art -- Exact solutions
Enumeration – [30-50 geographical units] Explicit enumeration intractable even for small #’s
of units Early work with implicit enumeration (branch and
cut) yielded solutions for 30-50 units [E.g. Mehohtra, et. al 1998]
Integer Programming – [100’s of units] School districting problem solved for < 500 units.
[Caro et. al 2004] Integer programming applied to < 400 units, but
used early termination, rendering solution non-exact. [Shirabe 2009]
[11/9/2012]Redistricting Algorithms9
State of the Art – Non-Exact Algorithms “Redistricter” [Olson 2008]
Specialized for compactness and population uses kmeans with ad-hoc refinements (including annealing) to solve Using 500000 census blocks can find solutions within 1% of population
General Metaheuristics [Altman & McDonald 2010] Framework for multiple metaheuristics & criteria
iRedistrict [Guo 2011] General criteria Tabu search, agglomeration, enhanced by connected-components trading Successful for 1000’s of units
IFE System [Trelles 2007] Complete GIS interface for redistricting – not just an optimization algorithm Successfully used for automated redistricting of 1000’s of units in Mexico
Other notable algorithms Q State Pott’s Model [Chou and Li 2007] Shortest Split-line [Kai et al 2007] Ad Hoc Greedy Heuristics [Sakguchi and Wado 2008] Genetic Algorithm w/TSP Encoding [Forman and Yu 2003] Annealing [Andrade & Garcia 2009] Tabu Seach [Bozkaya et. al 2003] Weighted Voronoi Diagrams [Ricca, et. al 2008]
[11/9/2012]Redistricting Algorithms10
ConclusionsAlgorithms are an advance in redistricting – and they are a part of the
solution
Solutions depend on starting valuesSolutions depend on good dataSome algorithms assume particular criteriaSome criteria are more tractable to optimizationAnd it is difficult to answer the question how good is this solution?
Some implications
Algorithms matter-- Same criteria + same data + different algorithm = different result
Code Matters-- Difficult to externally verify implementation of a complex algorithm
Transparency and public participation matters Open documentation allows for external replication of algorithms Open source allows external verification of implementation of algorithms Public input provides local community data for use in algorithmic redistricting Publicly submitted plans can provide good starting points for algorithmic refinement Public review of algorithmically created plans can help verify the quality of the solution
[11/9/2012]Redistricting Algorithms11
Additional References J. Aerts, C.J.H,. Erwin Eisinger,Gerard B.M. Heuvelink and Theodor J. Stewart, 2003. “Using Linear Integer Programming for Multi-Site
Land-Use Allocation”, Geographical Analysis 35(2) 148-69. M. Andrade and E. Garcia 2009, “Redistricting by Square Cells”, A. Hernández Aguirre et al. (Eds.): MICAI 2009, LNAI 5845, pp. 669–679,
2009. J. Barabas & J. Jerit, 2004. "Redistricting Principles and Racial Representation," State and Politics Quarterly¸4 (4): 415-435. B. Bozkaya, E. Erkut and G. Laporte 2003, A Tabu Search Heuristic and Adaptive Memory Procedure for Political Districting. European
Journal of Operational Re- search 144(1) 12-26. F. Caro et al . , School redistricting: embedding GIS tools with integer programming Journal of the Operational Research Society (2004)
55, 836–849 PG di Cortona, Manzi C, Pennisi A, Ricca F, Simeone B (1999). Evaluation and Optimization of Electoral Systems. SIAM Pres, Philadelphia. J.C. Duque, 2007. "Supervised Regionalization Methods: A Survey" International Regional Science Review, Vol. 30, No. 3, 195-220 S Forman & Y. Yue 2003, Congressional Districting Using a TSP-Based Genetic Algorithm Guo D. and H. Jin (2011). "iRedistrict: Geovisual Analytics for Redistricting Optimization", Journal of Visual Languages and Computing,
doi:10.1016/j.jvlc.2011.03.001 P. Kai, Tan Yue, Jiang Sheng, 2007, “The study of a new gerrymandering methodology”, Manuscript. http://arxiv.org/abs/0708.2266 J. Kalcsics, S. Nickel, M. Schroeder, 2009. A Geometric Approach to Territory Design and Districting, Fraunhofer Insititut techno und
Wirtshaftsmethematik. Dissertation. A. Mehrotra, E.L. Johnson, G.L. Nemhauser (1998), An optimization based heuristic for political districting, Management Science 44,
1100-1114. B. Olson, 2008 Redistricter. Software Package. URL: http://code.google.com/p/redistricter/ C. Puppe,, Attlia Tasnadi, 2009. "Optimal redistricting under geographical constraints: Why “pack and crack” does not work", Economics
Letter 105:93-96 C. Puppe,, Attlia Tasnadi, 2008. "A computational approach to unbiased districting", Mathematical and Computer Modeling 48(9-10),
November 2008, Pages 1455-1460 F. Ricca, A. Scozzari and B. Simeone, Weighted Voronoi Region Algorithms for Political Districting. Mathematical Computer Modelling
forthcoming (2008). F. Ricci, C, Bruno Simeone, 2008, "Local search algorithms for political districting", European Journal of Operational Research189, Issue
3, 16 September 2008, Pages 1409-1426 T. Shirabe, 2009. District modeling with exact contiguity constraints, Environment and Planning B (35) 1-14 Trelles, A. 2007. Electoral Boundaries, The Contribution of Mexico´s Redistricting Model to California. Mexico DF: ITAM. S. ,Toshihiro and Junichiro Wado. 2008, "Automating the Districting Process: An Experiment Using a Japanese Case Study" in Lisa
Handley and Bernard Grofman (ed.) Redistricting in Comparative Perspective, Oxford University Press D.H. Wolpert, Macready, W.G. (1997), "No Free Lunch Theorems for Optimization," IEEE Transactions on Evolutionary Computation 1, 67 N. Xiao, 2003. Geographical Optimization using Evolutionay Alogroithms, University of Iowa. Dissertation
[11/9/2012]Redistricting Algorithms12
Related Documents
