9

L C” Q 3 5 if 2 6

r?i fD
I li„L.

STATE OF NORTH CAROLINA IN THE GENERAL COURT OF JUSTICE

202} NOV lb P IjS^epRIOR COURT DIVISION
COUNTY OF WAKE CVS
NORTH CAROLINA LEAGUE OF C(i>f3jSERVATIONC. S
VOTERS, INC.; HENRY M. MICHAUX/JR.; DANDR^^
LEWIS; TIMOTHY CHARTIER; TALIA FERNOS; '
KATHERINE NEWHALL; JASON PARSLEY; EDNA ^
SCOTT; ROBERTA SCOTT; YVETTE ROBERTS; ^
JEREANN KING JOHNSON; REVEREND REGINALD
WELLS; YARBROUGH WILLIAMS, JR.; REVEREND AFFIDAVIT OF
DELORIS L. JERMAN; VIOLA RYALS FIGUEROA; and DR. MOON DUCHIN
COSMOS GEORGE,

Plaintiffs.

V.

REPRESENTATIVE DESTIN HALL, in his official capacity

as Chair of the House Standing Committee on Redistricting;
SENATOR WARREN DANIEL, in his official capacity as Co-
Chair of the Senate Standing Committee on Redistricting and
Elections; SENATOR RALPH E. HISE, JR., in his official
capacity as Co-Chair of the Senate Standing Committee on
Redistricting and Elections; SENATOR PAUL NEWTON, in
his official capacity as Co-Chair of the Senate Standing
Committee on Redistricting and Elections;
REPRESENTATIVE TIMOTHY K. MOORE, in his official
capacity as Speaker of the North Carolina House of
Representatives; SENATOR PHILIP E. BERGER, in his
official capacity as President Pro Tempore of the North
Carolina Senate; THE STATE OF NORTH CAROLINA; THE
NORTH CAROLINA STATE BOARD OF ELECTIONS;
DAMON CIRCOSTA, in his official capacity as Chairman of
the North Carolina State Board of Elections; STELLA
ANDERSON, in her official capacity as Secretary of the North
Carolina State Board of Elections; JEFF CARMON III, in his
official capacity as Member of the North Carolina State Board
of Elections; STACY EGGERS fV, in his official capacity as
Member of the North Carolina State Board of Elections;
TOMMY TUCKER, in his official capacity as Member of the
North Carolina State Board of Elections; and KAREN
BRINSON BELL, in her official capacity as Executive Director
of the North Carolina State Board of Elections,

Defendants.
 I, Dr. Moon, Duchin, having been duly sworn by an officer authorized to administer oaths,
depose and state as follows:
1. I am over 18 years of age, legally competent to give this Affidavit, and have personal
knowledge of the facts set forth in this Affidavit.

2. All of the quantitative work described in this Affidavit was performed by myself with the
support of research assistants working under my direct supervision.

Background and qualifications

3. I hold a Ph.D. and an M.S in Mathematics from the University of Chicago as well as an A.B.
in Mathematics and Women’s Studies from Harvard University.

4. I am a Professor of Mathematics and a Senior Fellow in the Jonathan M. Tisch College of

Civic Life at Tufts University.
5. My general research areas are geometry, topology, dynamics, and applications of mathe-
matics and computing to the study of elections and voting. My redistricting-related work
has been published in venues such as the Election Law Journal, Political Analysis, Founda-
tions of Data Science, the Notices of the American Mathematical Society, Statistics and
Public Policy, the Virginia Policy Review, the Harvard Data Science Review, Foundations
of Responsible Computing, and the Yale Law Journal Forum.
6. My research has had continuous grant support from the National Science Foundation
since 2009, including a CAREER grant from 2013–2018. I am currently on the editorial
board of the journals Advances in Mathematics and the Harvard Data Science Review. I
was elected a Fellow of the American Mathematical Society in 2017 and was named a
Radcliffe Fellow and a Guggenheim Fellow in 2018.
7. A current copy of my full CV is attached to this report.
8. I am compensated at the rate of $400 per hour.

2
 Analysis of 2021 enacted redistricting plans
in North Carolina
Moon Duchin
Professor of Mathematics, Tufts University
Senior Fellow, Tisch College of Civic Life

November 16, 2021

1 Introduction
On November 4, 2021, the North Carolina General Assembly enacted three districting plans:
maps of 14 U.S. Congressional districts, 50 state Senate districts, and 120 state House dis-
tricts. This affidavit contains a brief summary of my evaluation of the properties of these
plans. My focus will be on the egregious partisan imbalance in the enacted plans, following a
brief review of the traditional districting principles.
Because redistricting inevitably involves complex interactions of rules, which can create
intricate tradeoffs, it will be useful to employ a direct comparison to an alternative set of
plans. These demonstrative plans illustrate that it is possible to simultaneously maintain or
improve metrics for all of the most important redistricting principles that are operative in North
Carolina’s constitution and state and federal law. Crucially, this shows that nothing about the
state’s political geography compels us to draw a plan with a massive and entrenched partisan
skew.
To this end, I will be comparing the following plans: the enacted plans SL-174, SL-173,
and SL-175 and a corresponding set of alternative plans labeled NCLCV-Cong, NCLCV-Sen, and
NCLCV-House (proposed by plaintiffs who include the North Carolina League of Conservation
Voters).

SL-174 SL-173 SL-175

NCLCV-Cong NCLCV-Sen NCLCV-House

Figure 1: The six plans under discussion in this affidavit.

3
2 Traditional districting principles
Principles that are relevant to North Carolina redistricting include the following.

• Population balance. The standard interpretation of One Person, One Vote for Congres-
sional districts is that districts should be fine-tuned so that their total Census population
deviates by no more than one person from any district to any other.
There is more latitude with legislative districts; they typically vary top-to-bottom by no
more than 10% of ideal district size. In North Carolina, the Whole County Provisions make
it very explicit that 5% deviation must be tolerated if it means preserving more counties
intact.
All six plans have acceptable population balance.

Population deviation

Max Positive Deviation District Max Negative Deviation District

SL-174 0 (eight districts) −1 (six districts)
NCLCV-Cong 0 (eight districts) −1 (six districts)
SL-173 10,355 (4.960%) 5 −10,434 (4.997%) 13,18
NCLCV-Sen 10,355 (4.960%) 5 −10,427 (4.994%) 15
SL-175 4250 (4.885%) 18 −4189 (4.815%) 112
NCLCV-House 4341 (4.990%) 82 −4323 (4.969%) 87

Table 1: Deviations are calculated with respect to the rounded ideal district populations of
745,671 for Congress, 208,788 for Senate, and 86,995 for House.

• Minority electoral opportunity. Minority groups’ opportunity to elect candidates of

choice is protected by both state and federal law. A detailed assessment of opportu-
nity must hinge not on the demographics of the districts but on electoral history and
an assessment of polarization patterns. That is not the focus of the current affidavit.
Instead we make the brief note that it is important to avoid the conflation of majority-
minority districts with effective districts for a minority group. An involved analysis of
voting patterns—necessarily incorporating both primary and general elections to ensure
that candidates of choice can be successfully nominated and elected—will frequently re-
veal that districts can be effective at demographic levels well below 50% of voting-age
population or citizen voting-age population (VAP and CVAP, respectively). For instance,
in [3], my co-authors and I drew an illustrative plan for Texas congressional districting in
which some parts of the state had districts that were shown to reliably elect Black candi-
dates of choice with BCVAP as low as 28.6%; by contrast, there are other parts of Texas
where a 40% BCVAP district is less consistently effective. In a Louisiana case study, we
found somewhat different patterns of human and political geography, producing numer-
ous examples of Congressional-sized districts with 55% BCVAP in some parts of the state
that are nonetheless marginal in terms of opportunity for Black voters to elect candidates
of choice.
In North Carolina, taking the crossover voting patterns of White, Latino, and Asian voters
into account, I note that a district with BCVAP in the low to mid 30s can often be effective
for Black voters—but there is no demographic shortcut to a full examination of primary
and general election history.
• Contiguity. All six plans are contiguous; for each district, it is possible to transit from
any part of the district to any other part through a sequence of census blocks that share
boundary segments of positive length. As is traditional in North Carolina, contiguity
through water is accepted.

4
• Compactness. The two compactness metrics most commonly appearing in litigation
are the Polsby-Popper score and the Reock score. Polsby-Popper is the name given in
redistricting to a metric from ancient mathematics: the isoperimetric ratio comparing a
region’s area to its perimeter via the formula 4πA/ P2 . Higher scores are considered more
compact, with circles uniquely achieving the optimum score of 1. Reock is a different
measurement of how much a shape differs from a circle: it is computed as the ratio of a
region’s area to that of its circumcircle, defined as the smallest circle in which the region
can be circumscribed. From this definition, it is clear that it too is optimized at a value of
1, which is achieved only by circles.
These scores depend on the contours of a district and have been criticized as being
too dependent on map projections or on cartographic resolution [1, 2]. Recently, some
mathematicians have argued for using discrete compactness scores, taking into account
the units of Census geography from which the district is built. The most commonly cited
discrete score for districts is the cut edges score, which counts how many adjacent pairs
of geographical units receive different district assignments. In other words, cut edges
measures the "scissors complexity" of the districting plan: how much work would have to
be done to separate the districts from each other? Plans with a very intricate boundary
would require many separations. This score improves on the contour-based scores by
better controlling for factors like coastline and other natural boundaries, and by focusing
on the units actually available to redistricters rather than treating districts like free-form
Rorschach blots.
The alternative plans are significantly more compact than the enacted plans in all three
compactness metrics.

Compactness

block cut edges average Polsby-Popper average Reock

(lower is better) (higher is better) (higher is better)
SL-174 5194 0.303 0.381
NCLCV-Cong 4124 0.383 0.444
SL-173 9702 0.342 0.402
NCLCV-Sen 9249 0.369 0.423
SL-175 16,182 0.351 0.419
NCLCV-House 13,963 0.414 0.456

Table 2: Comparing compactness scores via one discrete and two contour-based metrics.

• Respect for political subdivisions. For legislative redistricting, North Carolina has one
of the strongest requirements for county consideration of any state in the nation. In my
understanding, courts have interpreted the Whole County Provisions as follows.

– First, if any county is divisible into a whole number of districts that will be within ±5%
of ideal population, then it must be subdivided accordingly without districts crossing
into other counties.
– Next, seek any contiguous grouping of two counties that is similarly divisible into a
whole number of districts.
– Repeat for groupings of three, and so on, until all counties are accounted for.

A complete set of solutions is described in detail in the white paper of Mattingly et al.—
though with the important caveat that the work "does not reflect... compliance with the
Voting Rights Act" [4]. Absent a VRA conflict, the 2020 Decennial Census population data
dictates that the North Carolina Senate plan must be decomposed into ten single-district
fixed clusters and seven multi-district fixed clusters (comprising 2, 2, 3, 3, 4, 6, and 6

5
 districts, respectively). It has four more areas in which there is a choice of groupings. In
all, there are sixteen different possible clusterings for Senate, each comprising 26 county
clusters. The House likewise has 11 single-district fixed clusters and 22 multi-district
fixed clusters (with two to thirteen districts per cluster), together with three more areas
with a choice of groupings. In all, the House has only eight acceptable clusterings, each
comprising 40 county clusters. Again, it is important to note that VRA compliance may
present a compelling reason to select some clusterings and reject others.
Once clusters have been formed, there are more rules about respecting county lines
within clusters. The legal language is again explicit: "[T]he resulting interior county lines
created by any such groupings may be crossed or traversed in the creation of districts
within said multi-county grouping but only to the extent necessary" to meet the ±5%
population standard for districts. To address this, I have counted the county traversals in
each plan, i.e., the number of times a district crosses between adjacent counties within a
grouping.
Table 3 reflects the county integrity metric that is most relevant at each level: the enacted
congressional plan splits 11 counties into 25 pieces while the alternative plan splits 13,
but splits no county three ways. (The enacted plans unnecessarily split three counties
into three pieces.) In the legislative plans, the law specifies traversals as the fundamental
integrity statistic.
The alternative plans are comparable to the enacted plans, or sometimes far superior, in
each of these key metrics regarding preservation of political boundaries.

County and municipality preservation

# county pieces # traversals

SL-174 25 SL-173 97
NCLCV-Cong 26 NCLCV-Sen 89
SL-175 69
NCLCV-House 66

# municipal pieces
SL-174 90
NCLCV-Cong 58
SL-173 152
NCLCV-Sen 125
SL-175 292
NCLCV-House 201

Table 3: Comparing the plans’ conformance to political boundaries.

I will briefly mention several additional redistricting principles.

• Communities of interest. In North Carolina, there was no sustained effort by the state
or by community groups to formally collect community of interest (COI) maps, to my
knowledge. Without this, it is difficult to produce a suitable metric.
• Cores of prior districts. In some states, there is statutory guidance to seek districting
plans that preserve the cores of prior districts. In North Carolina, this is not a factor in the
constitution, in statute, or in case law. In addition, attention to core preservation would
be prohibitively difficult in the Senate and House because of the primacy of the Whole
County Provisions, which forces major changes to the districts simply as a consequence
of fresh population numbers.

6
 • Incumbent pairing. In 2017, the North Carolina legislative redistricting committee
listed "incumbency protection" as a goal in their itemization of principles. In 2021, this
was softened to the statement that "Member residence may be considered" in the draw-
ing of districts. I have counted the districts in each plan that contain more than one
incumbent address; these are sometimes colorfully called "double-bunked" districts. For
this statistic, it is not entirely clear whether a high or low number is preferable. When a
plan remediates a gerrymandered predecessor, we should not be surprised if it ends up
pairing numerous incumbents.

Double-bunking

# districts pairing incumbents

SL-174 3
NCLCV-Cong 1
SL-173 6
NCLCV-Sen 9
SL-175 7
NCLCV-House 15

Table 4: For Congress and Senate, the enacted and alternative plans are comparable; at
the House level, the alternative plan has more double-bunking. Note: These numbers were
calculated using the most accurate incumbent addresses that have been provided to me.

3 Partisan fairness
3.1 Abstract partisan fairness
There are many notions of partisan fairness that can be found in the scholarly literature and in
redistricting practitioner guides and software. Most of them are numerical, in the sense that
they address how a certain share of the vote should be translated to a share of the seats in a
state legislature or Congressional delegation.
The numerical notions of partisan fairness all tend to agree on one central point: an elec-
toral climate with a 50-50 split in partisan preference should produce a roughly 50-50 repre-
sentational split. North Carolina voting has displayed a partisan split staying consistently close
to even between the two major parties over the last ten years, but the plans released by the
General Assembly after the 2010 census were very far from realizing the ideal of converting
even voting to even representation. This time, with a 14th seat added to North Carolina’s
apportionment, an exactly even seat outcome is possible. But the new enacted plans, like the
plans from ten years ago, are not conducive to even representation.

3.2 Geography and fairness

However, some scholars have argued that this ideal (that even vote preferences should trans-
late to even representation) ignores the crucial political geography—the location of votes for
each party, and not just the aggregate preferences, has a major impact on redistricting out-
comes. In [5], my co-authors and I gave a vivid demonstration of the impacts of political
geography in Massachusetts: we showed that for a ten-year span of observed voting patterns,
even though Republicans tended to get over one-third of the statewide vote, it was impossible
to draw a single Congressional district with a Republican majority. That is, the geography of
Massachusetts Republicans locked them out of Congressional representation. It is therefore
not reasonable to charge the Massachusetts legislature with gerrymandering for having pro-
duced maps which yielded all-Democratic delegations; they could not have done otherwise.

7
 In North Carolina, this is not the case. The alternative plans demonstrate that it is possible
to produce maps that give the two major parties a roughly equal opportunity to elect their
candidates. These plans are just examples among many thousands of plausible maps that
convert voter preferences to far more even representation by party. In Congressional redis-
tricting, the geography is easily conducive to a seat share squarely in line with the vote share.
In Senate and House plans, even following the strict detail of the Whole County Provisions,
there are likewise many alternatives giving a seat share for each party that falls, in aggregate,
within a few percentage points of the vote share across a large set of elections.
The clear conclusion is that the political geography of North Carolina today does not ob-
struct the selection of a map that treats the parties equally and fairly.

3.3 Translating votes to seats

The enacted plans behave as though they are built to resiliently safeguard electoral advantage
for Republican candidates. We can examine this effect without invoking assumptions like
"uniform partisan swing" that impose counterfactual voting conditions; instead, we will use
the rich observed dataset of 52 statewide party-ID general elections in North Carolina in the
last ten years. 29 of these are elections for Council of State (ten offices elected three times,
with the Attorney General race uncontested in 2012), three presidential races, three for U.S.
Senate, and 17 judicial races since mid-decade, when those became partisan contests. See
Table 6 for more detail on the election dataset.
I will sometimes focus on the smaller set of better-known "up-ballot" races: in order, the first
five to appear on the ballot are the contests for President, U.S. Senator, Governor, Lieutenant
Governor, and Attorney General. Together these occurred 14 times in the last Census cycle.

Up-ballot generals (14) All generals (52)

D vote share D seat share D vote share D seat share
SL-174 .2908 .3118
.4883 .4911
NCLCV-Cong .4796 .4931
SL-173 .3957 .4065
.4883 .4911
NCLCV-Sen .4557 .4592
SL-175 .3994 .4080
.4883 .4911
NCLCV-House .4649 .4684

Table 5: Comparing overall fidelity of representation to the voting preferences of the elec-
torate. Vote shares are reported with respect to the major-party vote total.

To understand how the enacted plans create major shortfalls for Democratic representa-
tion, we will overlay the plans with voting patterns from individual elections in the past Census
cycle. As we will see, the enacted Congressional plan (SL-174) shows a remarkable lack of
responsiveness, giving 10–4 partisan outcomes across a wide range of recent electoral condi-
tions, meaning that 10 Republicans and only 4 Democrats would represent North Carolina in
Congress. The alternative plan (NCLCV-Cong) is far more faithful to the vote share, far more
responsive, and tends to award more seats to the party with more votes.
The top of Figure 2 shows this dynamic in the three Presidential contests in the last Census
cycle, with a Democratic vote share (pink box) between 48% and 50% of the major-party total
each time. For a contest that is so evenly divided, we would expect a fair map to have 6, 7,
or 8 out of 14 districts favoring each party. The alternative Congressional map NCLCV-Cong
does just that, while the enacted plan SL-174 has just 4 out of 14 Democratic-majority districts
each time (green and maroon circles). The alternative plan is far more successful at reflecting
the even split of voter preferences. Below the initial explainer, simplified versions of the same
type of graphic are presented for all five up-ballot races. Figure 3 compares legislative maps in
the same fashion. Next, Figure 4 returns to the full 52-election dataset to give the big picture
of entrenched partisan advantage in the enacted plans.

8
 Congressional plan comparison in Presidential elections
Does even voting translate to even representation?
Dem seats

Dem vote share

8 Alternative plan
48.96% Dem seat share
7/14 7
49.31% Enacted plan
48.09%
6 Dem seat share

4/14 4 4 4

2012 2016 2020

Congressional plan comparison across up-ballot races

President Senate Governor

2012 2016 2020 2014 2016 2020 2012 2016 2020

Lieutenant Governor Attorney General

2012 2016 2020 2016 2020

Figure 2: For up-ballot general election contests across the previous Census cycle, we can
compare the seat share under the enacted Congressional plan SL-174 (maroon) and the seat
share under the alternative Congressional plan NCLCV-Cong (green) to the vote share (pink) for
Democratic candidates. At top is a detailed look at the presidential contests; this is repeated
below, alongside the other four up-ballot offices. The 50% line is marked each time.

9
 State Senate plan comparison across up-ballot races

President Senate Governor

25

16

2012 2016 2020 2014 2016 2020 2012 2016 2020

Lieutenant Governor Attorney General

25

16

2012 2016 2020 2016 2020

State House plan comparison across up-ballot races

President Senate Governor

60

40

2012 2016 2020 2014 2016 2020 2012 2016 2020

Lieutenant Governor Attorney General

60

40

2012 2016 2020 2016 2020

Figure 3: Legislative plans tested against voting patterns from up-ballot elections. The enacted
plans SL-173 and SL-175 are shown in maroon. The alternative plans NCLCV-Sen and NCLCV-
House, in green, have seat shares tracking much closer to the nearly even voting preferences.

10
 Congress
11

Democratic Seats 10

3
.42 .44 .46 .48 .5 .52 .54 .56 .58
Votes

state Senate state House

90
35
80
30 70

25 60

20 50

40
15
30

.44 .47 .5 .53 .56 .44 .47 .5 .53 .56

Figure 4: On a seats-vs.-votes plot, the election results for the six maps are shown for 52
general election contests in the last decade; each colored dot is plotted as the coordinate pair
(vote share, seat share). The diagonals show various lines of responsiveness that pivot around
the central point of fairness: half of the votes securing half of the seats. The Congressional
comparison is at top, followed by Senate and House. The enacted plans are shown in maroon
and the alternative plans in green.

11
3.4 Swing districts and competitive contests
Another way to understand the electoral properties of districting plans is to investigate how
many districts always give the same partisan result over a suite of observed electoral condi-
tions, and how many districts can "swing" between the parties. Figure 5 compares the six plans
across the up-ballot elections. The enacted plans lock in large numbers of always-Republican
seats. In the Senate and House, nearly half the seats are locked down for Republicans. In the
Congressional plan, it’s well over half. This provides another view from which the NCLCV plans
provide attractive alternatives.

9 Always R 1 Swing 4 Always D

SL-174

5 Always R 5 Swing 4 Always D

NCLCV-Cong

24 Always R 13 Swing 13 Always D

SL-173

22 Always R 13 Swing 15 Always D

NCLCV-Sen

57 Always R 27 Swing 36 Always D

SL-175

52 Always R 27 Swing 41 Always D

NCLCV-House

Figure 5: These visuals show the breakdown of seats that always have a Republican winner,
always have a Democratic winner, or are sometimes led by each party across the 14 up-ballot
elections over the previous Census cycle. The 50-50 split is marked.

One more measure of partisan fairness, frequently referenced in the public discourse, is
the tendency of a districting plan to promote close or competitive contests. We close with a
comparison of the enacted and alternative plans that displays the number of times across the
full dataset of 52 elections that a contest had a partisan margin of closer than 10 points, 6
points, or 2 points, respectively. This can occur up to 14 · 52 = 728 times in Congressional
maps, 50 · 52 = 2600 times in state Senate maps, and 120 · 52 = 6240 times in state House

12
maps. The figures below show horizontal rules at every 10% interval of the total number of
possible competitive contests; we can see, for instance, that the alternative Congressional
plan has contests within a 10-point margin more than 40% of the time.

Competitive contests in the Congressional plans

296

192 187

92
56
25

≤ 10 points ≤ 6 points ≤ 2 points

Senate plans House plans

566
1182 1184
454
390
297 674 703

167
113 214 233

≤ 10 points ≤ 6 points ≤ 2 points ≤ 10 points ≤ 6 points ≤ 2 points

Figure 6: These bar graphs show the number of competitive contests for the enacted plans
(maroon) and the alternative plans (green). In each plot, we consider increasingly restrictive
definitions of "competitive" from left to right, counting districts in which the major-party vote
split is closer than 45-55, 47-53, and 49-51, respectively.

4 Conclusion
North Carolina is a very "purple" state. In 38 out of the 52 contests in our dataset, the
statewide partisan outcome is within a 6-point margin: 47-53 or closer. We can make a striking
observation by laying our six plans over the vote patterns.

13
 D Vote Share SL-174 NCLCV-Cong SL-173 NCLCV-Sen SL-175 NCLCV-House
GOV12 0.4418 4 4 16 18 41 44
AGC16 0.4444 4 4 17 17 40 42
LAC16 0.4475 4 5 18 20 42 45
JHU16 0.4563 4 5 18 19 42 49
AGC20 0.4615 3 4 17 19 40 51
JZA16 0.4619 4 5 19 21 43 50
JDI16 0.4653 4 6 19 21 44 53
LTG16 0.4665 4 6 19 21 44 54
LAC12 0.4674 4 5 20 20 44 51
AGC12 0.4678 4 5 18 18 43 50
SEN16 0.4705 4 6 19 21 43 55
TRS16 0.4730 4 6 19 21 45 53
TRS20 0.4743 4 6 17 20 45 51
JA620 0.4806 4 7 17 21 46 55
PRS16 0.4809 4 7 19 22 48 56
JA420 0.4822 4 7 17 22 47 56
INC20 0.4823 4 7 18 23 47 56
LTG20 0.4836 4 7 18 21 46 55
JA720 0.4842 4 7 17 22 48 56
SUP20 0.4862 4 7 19 23 49 56
JA520 0.4874 4 7 18 22 49 57
JA218 0.4876 4 7 18 22 45 55
JS420 0.4879 4 7 19 24 49 56
J1320 0.4885 4 7 19 23 49 56
PRS12 0.4897 4 6 20 21 46 55
SEN20 0.4910 4 7 20 24 48 56
LAC20 0.4918 4 8 21 25 51 58
SEN14 0.4919 4 6 20 22 46 52
PRS20 0.4932 4 8 20 25 50 60
JS220 0.4934 4 8 21 24 51 59
SUP16 0.4941 4 6 22 23 49 57
JS118 0.4955 4 7 20 25 50 58
INC16 0.4960 4 6 22 22 50 57
JST16 0.4976 4 7 21 23 50 58
LTG12 0.4992 5 7 22 22 50 58
JS120 0.5000 4 8 22 27 52 60
AUD16 0.5007 5 8 22 23 51 56
GOV16 0.5011 4 7 20 27 50 58
ATG20 0.5013 4 8 21 25 51 58
ATG16 0.5027 4 7 20 23 50 57
JA118 0.5078 4 8 22 26 51 58
AUD20 0.5088 4 8 24 28 54 61
JA318 0.5091 4 8 21 26 52 59
SOS20 0.5116 5 8 24 28 53 62
JGE16 0.5131 5 8 22 25 52 59
INC12 0.5186 5 8 22 22 55 61
SOS16 0.5226 5 9 24 24 57 62
GOV20 0.5229 4 8 23 27 58 63
AUD12 0.5371 8 9 27 28 61 65
SOS12 0.5379 7 9 26 26 59 63
TRS12 0.5383 7 9 25 24 59 65
SUP12 0.5424 8 9 28 28 61 66

Table 6: 52 general elections, sorted from lowest to highest Democratic share. Election codes
have a three-character prefix and a two-digit suffix designating the office and the election
year, respectively. AGC = Agriculture Commissioner; ATG = Attorney General; AUD = Au-
ditor; GOV = Governor; INC = Insurance Commissioner; LAC = Labor Commissioner; PRS =
President; SEN = Senator; SOS = Secretary of State; SUP = Superintendent of Schools; TRS
-=Treasurer. The prefix JA* refers to judicial elections to the Court of Appeals (so that, for
instance, JA118 is the election to the Seat 1 on the Court of Appeals in 2018), those beginning
with JS* refer to elections to the state Supreme Court. All other J* prefixes refer to an election
to replace a specific judge on the Court of Appeals.

14
 The three enacted plans combine with those 38 relatively even vote patterns to produce
114 outcomes. Every single pairing of an enacted plan with a close statewide contest—a
complete sweep of 114 opportunities—gives an outright Republican majority of seats. All
three enacted plans will lock in an extreme, resilient, and unnecessary advantage for one
party.
By every measure considered above that corresponds to a clear legal or good-government
redistricting goal or value, the alternative plans meet or exceed the performance of the en-
acted plans. It is therefore demonstrated to be possible, without any cost to the redistricting
principles in play, to select maps that are far fairer to the voters of North Carolina.

References
[1] Assaf Bar-Natan, Lorenzo Najt, and Zachary Schutzmann, The gerrymandering jumble:
map projections permute districts’ compactness scores. Cartography and Geographic In-
formation Science, Volume 47, Issue 4, 2020, 321–335.
[2] Richard Barnes and Justin Solomon, Gerrymandering and Compactness: Implementation
Flexibility and Abuse. Political Analysis, Volume 29, Issue 4, October 2021, 448–466.
[3] Amariah Becker, Moon Duchin, Dara Gold, and Sam Hirsch, Computational redistricting
and the Voting Rights Act. Election Law Journal.
Available at https://www.liebertpub.com/doi/epdf/10.1089/elj.2020.0704
[4] Christopher Cooper, Blake Esselstyn, Gregory Herschlag, Jonathan Mattingly, and Re-
becca Tippett, NC General Assembly County Clusterings from the 2020 Census.
https://sites.duke.edu/quantifyinggerrymandering/files/2021/08/countyClusters2020.pdf

[5] Moon Duchin, Taissa Gladkova, Eugene Henninger-Voss, Heather Newman, and Hannah
Wheelen, Locating the Representational Baseline: Republicans in Massachusetts. Elec-
tion Law Journal, Volume 18, Number 4, 2019, 388–401.

15
Moon Duchin moon.duchin@tufts.edu - mduchin.math.tufts.edu
Mathematics · STS · Tisch College of Civic Life | Tufts University

Education
University of Chicago MS 1999, PhD 2005
Mathematics
Advisor:
. Alex Eskin Dissertation: Geodesics track random walks in Teichmüller space

Harvard University BA 1998

Mathematics and Women’s Studies
.
Appointments
Tufts University
Professor of Mathematics 2021—
Assistant Professor, Associate Professor 2011–2021
Director | Program in Science, Technology, & Society 2015–2021
(on leave 2018–2019)
Principal Investigator | MGGG Redistricting Lab 2017—
Senior
. Fellow | Tisch College of Civic Life 2017—

University of Michigan
Assistant Professor (postdoctoral) 2008–2011

.University of California, Davis

NSF VIGRE Postdoctoral Fellow 2005–2008

.
Research Interests
Data science for civil rights, computation and governance, elections, geometry and redistricting.
Science, technology, and society, science policy, technology and law.
Random walks and Markov chains, random groups, random constructions in geometry.
Large-scale geometry, metric geometry, isoperimetric inequalities.
Geometric group theory, growth of groups, nilpotent groups, dynamics of group actions.
Geometric topology, hyperbolicity, Teichmüller theory.

Awards & Distinctions

Research Professor - MSRI Program in Analysis and Geometry of Random Spaces Spring 2022
Guggenheim Fellow 2018
Radcliffe Fellow - Evelyn Green Davis Fellowship 2018–2019
Fellow of the American Mathematical Society elected 2017
NSF C-ACCEL (PI) - Harnessing the Data Revolution: Network science of Census data 2019–2020
NSF grants (PI) - CAREER grant and three standard Topology grants 2009–2022
Professor of the Year, Tufts Math Society 2012–2013
AAUW Dissertation Fellowship 2004–2005
NSF Graduate Fellowship 1998–2002
Lawrence and Josephine Graves Prize for Excellence in Teaching (U Chicago) 2002
Robert Fletcher Rogers Prize (Harvard Mathematics) 1995–1996

1
Mathematics Publications & Preprints
The (homological) persistence of gerrymandering
Foundations of Data Science, online first. (with Thomas Needham and Thomas Weighill)
You can hear the shape of a billiard table: Symbolic dynamics and rigidity for flat surfaces
Commentarii Mathematici Helvetici, to appear. arXiv:1804.05690
(with Viveka Erlandsson, Christopher Leininger, and Chandrika Sadanand)
Conjugation curvature for Cayley graphs
Journal of Topology and Analysis, online first. (with Assaf Bar-Natan and Robert Kropholler)
A reversible recombination chain for graph partitions
Preprint. (with Sarah Cannon, Dana Randall, and Parker Rule)
Recombination: A family of Markov chains for redistricting
Harvard Data Science Review. Issue 3.1, Winter 2021. online. (with Daryl DeFord and Justin Solomon)
Census TopDown: The impact of differential privacy on redistricting
2nd Symposium on Foundations of Responsible Computing (FORC 2021), 5:1–5:22. online.
(with Aloni Cohen, JN Matthews, and Bhushan Suwal)
Stars at infinity in Teichmüller space
Geometriae Dedicata, Volume 213, 531–545 (2021). (with Nate Fisher) arXiv:2004.04321
Random walks and redistricting: New applications of Markov chain Monte Carlo
(with Daryl DeFord) For edited volume, Political Geometry. Under contract with Birkhäuser.
Mathematics of nested districts: The case of Alaska
Statistics and Public Policy. Vol 7, No 1 (2020), 39–51. (w/ Sophia Caldera, Daryl DeFord, Sam Gutekunst, & Cara Nix)
A computational approach to measuring vote elasticity and competitiveness
Statistics and Public Policy. Vol 7, No 1 (2020), 69–86. (with Daryl DeFord and Justin Solomon)
The Heisenberg group is pan-rational
Advances in Mathematics 346 (2019), 219–263. (with Michael Shapiro)
Random nilpotent groups I
IMRN, Vol 2018, Issue 7 (2018), 1921–1953. (with Matthew Cordes, Yen Duong, Meng-Che Ho, and Ayla Sánchez)
Hyperbolic groups
chapter in Office Hours with a Geometric Group Theorist, eds. M.Clay,D.Margalit, Princeton U Press (2017), 177–203.
Counting in groups: Fine asymptotic geometry
Notices of the American Mathematical Society 63, No. 8 (2016), 871–874.
A sharper threshold for random groups at density one-half
Groups, Geometry, and Dynamics 10, No. 3 (2016), 985–1005.
(with Katarzyna Jankiewicz, Shelby Kilmer, Samuel Lelièvre, John M. Mackay, and Ayla Sánchez)
Equations in nilpotent groups
Proceedings of the American Mathematical Society 143 (2015), 4723–4731. (with Hao Liang and Michael Shapiro)
Statistical hyperbolicity in Teichmüller space
Geometric and Functional Analysis, Volume 24, Issue 3 (2014), 748–795. (with Howard Masur and Spencer Dowdall)
Fine asymptotic geometry of the Heisenberg group
Indiana University Mathematics Journal 63 No. 3 (2014), 885–916. (with Christopher Mooney)
Pushing fillings in right-angled Artin groups
Journal of the LMS, Vol 87, Issue 3 (2013), 663–688. (with Aaron Abrams, Noel Brady, Pallavi Dani, and Robert Young)
Spheres in the curve complex
In the Tradition of Ahlfors and Bers VI, Contemp. Math. 590 (2013), 1–8. (with Howard Masur and Spencer Dowdall)

2
 The sprawl conjecture for convex bodies
Experimental Mathematics, Volume 22, Issue 2 (2013), 113–122. (with Samuel Lelièvre and Christopher Mooney)
Filling loops at infinity in the mapping class group
Michigan Math. J., Vol 61, Issue 4 (2012), 867–874. (with Aaron Abrams, Noel Brady, Pallavi Dani, and Robert Young)
The geometry of spheres in free abelian groups
Geometriae Dedicata, Volume 161, Issue 1 (2012), 169–187. (with Samuel Lelièvre and Christopher Mooney)
Statistical hyperbolicity in groups
Algebraic and Geometric Topology 12 (2012) 1–18. (with Samuel Lelièvre and Christopher Mooney)
Length spectra and degeneration of flat metrics
Inventiones Mathematicae, Volume 182, Issue 2 (2010), 231–277. (with Christopher Leininger and Kasra Rafi)
Divergence of geodesics in Teichmüller space and the mapping class group
Geometric and Functional Analysis, Volume 19, Issue 3 (2009), 722–742. (with Kasra Rafi)
Curvature, stretchiness, and dynamics
In the Tradition of Ahlfors and Bers IV, Contemp. Math. 432 (2007), 19–30.
Geodesics track random walks in Teichmüller space
PhD Dissertation, University of Chicago 2005.

Science, Technology, Law, and Policy Publications & Preprints

Models, Race, and the Law
Yale Law Journal Forum, Vol. 130 (March 2021). Available online. (with Doug Spencer)
Computational Redistricting and the Voting Rights Act
Election Law Journal, Available online. (with Amariah Becker, Dara Gold, and Sam Hirsch)
Discrete geometry for electoral geography
Preprint. (with Bridget Eileen Tenner) arXiv:1808.05860
Implementing partisan symmetry: Problems and paradoxes
Political Analysis, to appear. (with Daryl DeFord, Natasha Dhamankar, Mackenzie McPike, Gabe Schoenbach, and
Ki-Wan Sim) arXiv:2008:06930
Clustering propensity: A mathematical framework for measuring segregation
Preprint. (with Emilia Alvarez, Everett Meike, and Marshall Mueller; appendix by Tyler Piazza)
Locating the representational baseline: Republicans in Massachusetts
Election Law Journal, Volume 18, Number 4, 2019, 388–401.
(with Taissa Gladkova, Eugene Henninger-Voss, Ben Klingensmith, Heather Newman, and Hannah Wheelen)
Redistricting reform in Virginia: Districting criteria in context
Virginia Policy Review, Volume XII, Issue II, Spring 2019, 120–146. (with Daryl DeFord)
Geometry v. Gerrymandering
The Best Writing on Mathematics 2019, ed. Mircea Pitici. Princeton University Press.
reprinted from Scientific American, November 2018, 48–53.
Gerrymandering metrics: How to measure? What’s the baseline?
Bulletin of the American Academy for Arts and Sciences, Vol. LXII, No. 2 (Winter 2018), 54–58.
Rebooting the mathematics of gerrymandering: How can geometry track with our political values?
The Conversation (online magazine), October 2017. (with Peter Levine)
A formula goes to court: Partisan gerrymandering and the efficiency gap
Notices of the American Mathematical Society 64 No. 9 (2017), 1020–1024. (with Mira Bernstein)
International mobility and U.S. mathematics
Notices of the American Mathematical Society 64, No. 7 (2017), 682–683.

3
Graduate Advising in Mathematics
Nate Fisher (PhD 2021), Sunrose Shrestha (PhD 2020), Ayla Sánchez (PhD 2017),
Kevin Buckles (PhD 2015), Mai Mansouri (MS 2014)
Outside committee member for Chris Coscia (PhD 2020), Dartmouth College

Postdoctoral Advising in Mathematics

Principal supervisor Thomas Weighill (2019–2020)
Co-supervisor Daryl DeFord (MIT 2018–2020), Rob Kropholler (2017–2020), Hao Liang (2013–2016)

Teaching

Courses Developed or Customized

Mathematics of Social Choice | sites.tufts.edu/socialchoice
Voting theory, impossibility theorems, redistricting, theory of representative democracy, metrics of fairness.

History of Mathematics | sites.tufts.edu/histmath

Social history of mathematics, organized around episodes from antiquity to present. Themes include materials and
technologies of creation and dissemination, axioms, authority, credibility, and professionalization. In-depth treatment
of mathematical content from numeration to cardinal arithmetic to Galois theory.

Reading Lab: Mathematical Models in Social Context | sites.tufts.edu/models

One hr/wk discussion seminar of short but close reading on topics in mathematical modeling, including history of
psychometrics; algorithmic bias; philosophy of statistics; problems of model explanation and interpretation.

Geometric Literacy
Module-based graduate topics course. Modules have included: p-adic numbers, hyperbolic geometry, nilpotent
geometry, Lie groups, convex geometry and analysis, the complex of curves, ergodic theory, the Gauss circle problem.

Markov Chains (graduate topics course)

Teichmüller Theory (graduate topics course)
Fuchsian Groups (graduate topics course)
Continued Fractions and Geometric Coding (undergraduate topics course)
Mathematics for Elementary School Teachers

Standard Courses
Discrete Mathematics, Calculus I-II-III, Intro to Proofs, Linear Algebra, Complex Analysis, Differential Geometry,
Abstract Algebra, Graduate Real Analysis, Mathematical Modeling and Computation

Weekly Seminars Organized

- Geometric Group Theory and Topology
- Science, Technology, and Society Lunch Seminar

4
Selected Talks and Lectures

Distinguished Plenary Lecture June 2021

75th Anniversary Meeting of Canadian Mathematical Society, Ottawa, Ontario online (COVID)

BMC/BAMC Public Lecture April 2021

Joint British Mathematics/Applied Mathematics Colloquium, Glasgow, Scotland online (COVID)
AMS Einstein Public Lecture in Mathematics [March 2020]
Southeastern Sectional Meeting of the AMS, Charlottesville, VA postponed

Gerald and Judith Porter Public Lecture

AMS-MAA-SIAM, Joint Mathematics Meetings, San Diego, CA January 2018
Mathematical Association of America Distinguished Lecture
MAA Carriage House, Washington, DC October 2016

American Mathematical Society Invited Address

AMS Eastern Sectional Meeting, Brunswick, ME September 2016

Named University Lectures

- Parsons Lecture | UNC Asheville October 2020
- Loeb Lectures in Mathematics | Washington University in St. Louis [March 2020]
- Math, Stats, CS, and Society | Macalester College October 2019
- MRC Public Lecture | Stanford University May 2019
- Freedman Memorial Colloquium | Boston University March 2019
- Julian Clancy Frazier Colloquium Lecture | U.S. Naval Academy January 2019
- Barnett Lecture | University of Cincinnati October 2018
- School of Science Colloquium Series | The College of New Jersey March 2018
- Kieval Lecture | Cornell University February 2018
- G. Milton Wing Lectures | University of Rochester October 2017
- Norman Johnson Lecture | Wheaton College September 2017
- Dan E. Christie Lecture | Bowdoin College September 2017

Math/Computer Science Department Colloquia

- Reed College Dec 2020 - Université de Neuchâtel Jun 2016
- Georgetown (CS) Sept 2020 - Brandeis University Mar 2016
- Santa Fe Institute July 2020 - Swarthmore College Oct 2015
- UC Berkeley Sept 2018 - Bowling Green May 2015
- Brandeis-Harvard-MIT-NEU Mar 2018 - City College of New York Feb 2015
- Northwestern University Oct 2017 - Indiana University Nov 2014
- University of Illinois Sept 2017 - the Technion Oct 2014
- University of Utah Aug 2017 - Wisconsin–Madison Sept 2014
- Wesleyan Dec 2016 - Stony Brook March 2013
- Worcester Polytechnic Inst. Dec 2016

5
 Minicourses
- Integer programming and combinatorial optimization (two talks) | Georgia Tech May 2021
- Workshop in geometric topology (main speaker, three talks) | Provo, UT June 2017
- Growth in groups (two talks) | MSRI, Berkeley, CA August 2016
- Hyperbolicity in Teichmüller space (three talks) | Université de Grenoble May 2016
- Counting and growth (four talks) | IAS Women’s Program, Princeton May 2016
- Nilpotent groups (three talks) | Seoul National University October 2014
- Sub-Finsler geometry of nilpotent groups (five talks) | Galatasaray Univ., Istanbul April 2014

Science, Technology, and Society

- The Mathematics of Accountability | Sawyer Seminar, Anthropology, Johns Hopkins February 2020
- STS Circle | Harvard Kennedy School of Government September 2019
- Data, Classification, and Everyday Life Symposium | Rutgers Center for Cultural Analysis January 2019
- Science Studies Colloquium | UC San Diego January 2019
- Arthur Miller Lecture on Science and Ethics | MIT Program in Science, Tech, and Society November 2018

Data Science, Computer Science, Quantitative Social Science

- Data Science for Social Good Workshop (DS4SG) | Georgia Tech (virtual) November 2020
- Privacy Tools Project Retreat | Harvard (virtual) May 2020
- Women in Data Science Conference | Microsoft Research New England March 2020
- Quantitative Research Methods Workshop | Yale Center for the Study of American Politics February 2020
- Societal Concerns in Algorithms and Data Analysis | Weizmann Institute December 2018
- Quantitative Collaborative | University of Virginia March 2018
- Quantitative Social Science | Dartmouth College September 2017
- Data for Black Lives Conference | MIT November 2017

Political Science, Geography, Law, Democracy, Fairness

- The Long 19th Amendment: Women, Voting, and American Democracy | Radcliffe Institute Nov–Dec 2020
- "The New Math" for Civil Rights | Social Justice Speaker Series, Davidson College November 2020
- Math, Law, and Racial Fairness | Justice Speaker Series, University of South Carolina November 2020
- Voting Rights Conference | Northeastern Public Interest Law Program September 2020
- Political Analysis Workshop | Indiana University November 2019
- Program in Public Law Panel | Duke Law School October 2019
- Redistricting 2021 Seminar | University of Chicago Institute of Politics May 2019
- Geography of Redistricting Conference Keynote | Harvard Center for Geographic Analysis May 2019
- Political Analytics Conference | Harvard University November 2018
- Cyber Security, Law, and Society Alliance | Boston University September 2018
- Clough Center for the Study of Constitutional Democracy | Boston College November 2017
- Tech/Law Colloquium Series | Cornell Tech November 2017
- Constitution Day Lecture | Rockefeller Center for Public Policy, Dartmouth College September 2017

Editorial Boards
Harvard Data Science Review
Associate Editor since 2019

Advances in Mathematics
Member, Editorial Board since 2018

6
Selected Professional and Public Service
Amicus Brief of Mathematicians, Law Professors, and Students 2019
principal co-authors: Guy-Uriel Charles and Moon Duchin
Supreme Court of the United States, in Rucho v. Common Cause - cited in dissent

Committee on Science Policy 2020–2023

American Mathematical Society

Program Committee 2020–2021

Symposium on Foundations of Responsible Computing

Presenter on Public Mapping, Statistical Modeling 2019, 2020

National Conference of State Legislatures

Committee on the Human Rights of Mathematicians 2016–2019

American Mathematical Society

Committee on The Future of Voting: Accessible, Reliable, Verifiable Technology 2017–2018

National Academies of Science, Engineering, and Medicine

Visiting Positions and Residential Fellowships

Visiting Professor Department of Mathematics Fall 2021
Boston College | Chestnut Hill, MA
Fellow Radcliffe Institute for Advanced Study 2018–19
Harvard University | Cambridge, MA
Member Center of Mathematical Sciences and Applications 2018–19
Harvard University | Cambridge, MA
Visitor Microsoft Research Lab 2018–19
MSR New England | Cambridge, MA
Research Member Geometric Group Theory program Fall 2016
Mathematical Sciences Research Institute | Berkeley, CA
Research Member Random Walks and Asymptotic Geometry of Groups program Spring 2014
Institut Henri Poincaré | Paris, France
Research Member Low-dimensional Topology, Geometry, and Dynamics program Fall 2013
Institute for Computational and Experimental Research in Mathematics | Providence, RI
Research Member Geometric and Analytic Aspects of Group Theory program May 2012
Institut Mittag-Leffler | Stockholm, Sweden
Research Member Quantitative Geometry program Fall 2011
Mathematical Sciences Research Institute | Berkeley, CA
Postdoctoral Fellow Teichmüller "project blanc" Spring 2009
Agence Nationale de la Recherche (Collège de France) | Paris, France