“I propose that we draw the maps to give a partisan advantage to 10 Republicans and three Democrats,” said North Carolina Rep. David Lewis, in 2016 in a legislative committee hearing, “because I do not believe it’s possible to draw a map with 11 Republicans and two Democrats.”

This might be the most infamous gerrymandering confession ever — a rare moment of clarity exposing the GOP’s ambition to control the state’s map through redistricting. This year, Lewis’s statement was debated in the Supreme Court; in 2017, it was even cited by comedian John Oliver. And this week, the admission that state lawmakers intended to give an advantage to the Republican Party was at the root of a decision by a North Carolina state court that invalidated the state’s congressional map ahead of 2020.

But even Lewis’s bold acknowledgement understated the GOP’s redistricting capabilities, new documents obtained by The Intercept suggest. In fact, GOP strategists did prove that it was possible to draw a map with 11 or 12 Republican congressional seats — in one case, a partisan gerrymander that could have feasibly elected an all-white slate.

In 2016, Thomas Hofeller, the veteran redistricting mastermind, was tapped by Lewis and the North Carolina legislature to craft the state’s congressional lines. The original map, drawn in 2011 redistricting, had been thrown out for racially discriminatory intent. During that process, Hofeller created drafts of maps that would give Democrats only one or two seats in this competitive purple state. These never-before-seen draft maps are among the more than 70,000 previously unpublished documents and emails from Hofeller’s hard drives, obtained and reviewed by The Intercept. And there is evidence that he talked them over with Lewis, who chaired the state House of Representatives redistricting committee from 2011 until 2018.

New documents suggest that even Lewis’s bold acknowledgement understated the GOP’s redistricting capabilities.

Lewis has maintained that his specific instructions to draw Republicans a partisan advantage were meant to make clear that the maps were motivated by politics, not race. At the time, GOP lawmakers believed that the partisan gerrymander would be legal. But on Monday, the court ruled that it violates the state’s constitutional protections of both fair elections and equal protection “beyond a reasonable doubt.” (Lewis did not reply to a request for comment for this story.)

The decision to invalidate the state’s current maps throws the voting process into disarray just months before both parties must nominate congressional candidates from districts that now don’t exist. The judges indicated that any Republican appeal would be unlikely to succeed. Democrats control the state Supreme Court. So the most likely road now is that state legislators will need to quickly draw — and win court approval — of new maps.

These documents, though never enacted, could now be crucial to any Republican appeal of the state court decision on Monday, as well as the debate over how a new map should be drawn. They provide extra evidence of what the court called the “partisan intent and the intended partisan effect” of the 2016 GOP-drawn map. And they represent an important warning about how Republican mapmakers can secretly calibrate maps to have a deep partisan bias, even when they look compact and follow rules about not splitting counties into separate districts.

Lewis and other GOP leaders, apparently concerned that an 11-2 map might spread Republican votes too thin, selected a map that was partisan but not as risky. (Maryland Democrats made a similar calculation, drafting a map that gave them all eight of the state’s congressional seats, but settling on a safer, incumbent friendly 7-1 breakdown.)

Indeed, the careful bet paid off: The congressional maps returned the 10-3 advantage Lewis sought in 2016 and again in 2018, cushioning the GOP against the blue wave and leaving the party in control of 10 of 13 seats even as Democratic candidates won more statewide votes. Democrats won their three races with upwards of 70 percent of the vote. Every Republican won with a more efficient percentage in the 50s, except for the one incumbent who ran unopposed.

These maps not only twisted representation inside North Carolina by all but guaranteeing Republicans more than 70 percent of the state’s seats in Congress, even with less than 50 percent of the statewide vote, but had a considerable national impact as well. North Carolina’s new 11th District, for example, created by cracking liberal Asheville in two and diluting the votes of Democrats, safely elected Rep. Mark Meadows, the former chair of the House Freedom Caucus. This map also cracked the city of Greensboro, not only dividing this majority-minority city among two conservative districts, but also cleaving the historically black campus of North Carolina A&T State University almost exactly in half, scattering seven dorms into one district and six in another.

In February 2016, when a panel of three federal district court judges struck down Hofeller’s original North Carolina map, drawn in 2011, as a racial gerrymander in violation of the 14th Amendment’s “one person, one vote” protections, it didn’t take long for Hofeller to get back to work. The judges ruled — and the U.S. Supreme Court later upheld — that Hofeller impermissibly used race data when drawing a majority of black voters into two congressional districts, the 1st and 12th.

Hofeller explored the outer limits to see if an even larger GOP edge might be possible.

During the first and second week of February, according to depositions, Lewis and state Sen. Robert Rucho, the other legislative co-chair, visited Hofeller’s home in a modest Raleigh retirement community on multiple occasions. A different, if still crafty, plan came together. After watching their 10-3 congressional map overturned for overreliance on race data, they understood that no U.S. Supreme Court standard existed to prevent them from maximizing political data. Maybe they couldn’t get away with a racial gerrymander, but they could partisan gerrymander to their hearts’ content. Hofeller devised an algorithm based on the results of seven recent statewide races that he believed would perfectly predict the outcome of any North Carolina race.

Hofeller got back to work on new maps the same week that the court’s decision arrived. Republicans held 10 of the state’s 13 congressional seats on the now-invalid map. Hofeller knew he could create another that provided the same result; indeed, Lewis, Rucho, and the state legislature actually enacted criteria that would require that outcome. But Hofeller also explored the outer limits to see if an even larger GOP edge might be possible.

Turns out, it was. A Hofeller map labeled Plan 17A created 11 GOP districts, guaranteed Democrats just one, and included one toss-up seat that remained 49.5 percent Republican.

“It’s just an experiment to see what the possibilities could be,” Hofeller said of this map, in a previously unreported deposition during Rucho v. Common Cause, also found in his files. (A less detailed version of this map, though not any of these supporting materials, surfaced during discovery for that trial, which ended last spring at the U.S. Supreme Court.)

A second Hofeller map, this one named Plan ST-B, advantaged Republicans in at least 10 seats, locked in two for the Democrats, and also included one competitive toss-up district with a 49.5 percent GOP population.

Comparison-of-2016-to-2011-withPlan-17A-and-STB-1572460872

Document: Files of Thomas Hofeller

Gerrymandering tends to work like this: A mapmaker packs as many opposing voters as possible into one or two districts, then sprinkles the remaining voters as thinly as possible across the remaining seats. That creates landslide districts for the other side, won with over 65 or 70 percent of the vote, and uses one’s own voters efficiently to win more seats by a smaller margin, usually in the mid-to-high 50s. A mapmaker especially confident in the voter algorithm might feel comfortable narrowing that margin into the low 50s edge in order to go for even more seats.

Hofeller did not lack confidence, and Plan 17-A might have been Hofeller’s master stroke: a map that spread the GOP vote across North Carolina with tremendous precision in search of one or two extra seats.

Map-2-Plan-17-A-1572460874

Document: Files of Thomas Hofeller

Plan 17-A packs most North Carolina Democrats into a single seat, a radically reimagined NC-01 that essentially corrals liberal Chapel Hill and Durham with the urban core of Raleigh. According to Hofeller’s tabulations, that seat would be 72.1 percent Democratic and just 27.9 percent Republican. This area is currently represented by a white Democrat, veteran lawmaker David Price.

Democratic voters would only hold a majority in one other district on this map, the second, and just barely. That district, centered in rural, eastern North Carolina counties such as Pitt, Johnston, and Greene, would be 50.5 percent Democratic and 49.5 Republican. The counties in the imagined NC-02 here are currently scattered among four other congressional districts.

The uber-aggressive drafts are a frightening illustration of what the 2021 redistricting cycle might look like,  in an ever-more-polarized nation, with ever-more-sophisticated mapmaking software.

It’s a wildly but elegantly gerrymandered map: Asheville, a liberal bastion in Buncombe County in western North Carolina, is cracked in half and divided between two Republican districts. Hofeller’s reimagined 12th District dances narrowly along the South Carolina border just long enough to lasso the whitest, wealthiest neighborhoods in Charlotte. It envelops one of the only majority-minority districts, currently represented by Alma Adams. Adams would have been relocated to the 6th District, an overwhelmingly Republican seat. It splits Charlotte’s county, Mecklenberg, three times. And if you wanted to drive the 45 miles from Charlotte to White Store, both in the same district, you’d cross over two other districts before arriving back in the 12th.

In five districts, the GOP edge was under 52 percent — meaning that these districts would be 51/49 nail-biters. Only one district, at 57 percent Republican, could be seen as safe for the GOP. In a table named “2016 Congressional Plan Compared to Three Additional Plans,” Hofeller presented only that single district as safely Republican, with 10 others that leaned Republican, and then one safely Democrat and another leaning blue.

If Plan 17-A had been enacted, and the university towns in NC-01 elected a white Democrat, it’s easy to imagine that a map enacted to cure a racial gerrymander might have actually elected 13 white members of Congress.

Hofeller’s files also include a second plan, this one named ST-B. This map, while just as gerrymandered, shored up Republican margins. It created two packed Democratic districts instead of one: Democratic voters dominated the first 72 percent to 28 percent, and also the 12th, 69.7 percent to 30.3 percent. That allowed Hofeller to reinforce the GOP seats: Six of these seats were safely above 55 percent Republican, with four additional seats leaning their way. Once again, Hofeller created a district that was 50.5 percent Democratic and 49.5 Republican. Plan ST-B, then, played safer for the GOP: Republicans began with a 6-2 edge, and would be favored in four of the other five districts. Adams, once again, would have been forced to run in the new 6th District, a majority Republican seat.

According to depositions during Rucho v. Common Cause, the two politicians visited Hofeller, and Hofeller shared various maps and the math behind them. Ultimately, the legislature settled on the 10-3 map; in 2018, Democrats won their districts with 69.9, 75.1, and 73.1 percent of the vote, while every Republican running a contested race landed safely in the 50s. Democrats won more total votes, but the map spread the Republican votes more effectively and efficiently.

A document from October 2017, “Placement of Incumbents,” suggests why Lewis, Rucho, and the GOP leaders made that calculation. Hofeller’s 11-2 map would have required “double-bunking” — pitting two incumbents against one another — several Republican members.

Table-XX-Placement-of-Incumbents-1572460877

Document: Files of Thomas Hofeller

In another previously unreported deposition discovered inside Hofeller’s files, Lewis conceded that as the leaders chose between Hofeller’s various plans, protecting Republican incumbents was very much on their mind. “I’m certain that was one of the criteria that we talked about,” Lewis said. “I know we talked about if we could (at) all avoid — we didn’t want to place two incumbent members in the same seat. I know we talked about that.”

While the uber-aggressive plans Hofeller drew ultimately weren’t used, they’re a frightening illustration of what the 2021 redistricting cycle might look like — and just how far redistricting can go in an ever-more-polarized nation, with ever-more-sophisticated mapmaking software.

Hofeller’s zealously partisan maps, after all, appear reasonable: They hold counties together, appear contiguous, and score well on compactness tests. Documents in his files show that Hofeller tested them to be sure.

Table-XX-Comparison-of-Compactness-Scores-1572460876

Document: Files of Thomas Hofeller

Democratic strongholds in Charlotte, Asheville, Durham, Chapel Hill, and Raleigh are artfully “cracked” or “packed,” but it’s only the Charlotte-area districts that resemble “funny-looking” gerrymanders. In other words, the maps are carefully designed to pass both the eye test and state legal standards, while still giving Republicans as many as 11 or 12 reliable victories.

These draft maps are also important because they foreshadow the dangers for minority representation in explicitly partisan maps. North Carolina notwithstanding, the Supreme Court decision this summer to make partisan gerrymandering the jurisdiction of states will likely embolden future state legislatures to tilt maps in a partisan direction. What North Carolina makes clear is that allowing partisan gerrymanders could potentially undermine protections against racial gerrymandering. In states where black and other minority voters are closely aligned with the Democratic Party, a partisan gerrymander can become a legal way to draw a racial gerrymander.

Correction: October 31, 2019.
A previous version misstated that Alma Adams was North Carolina’s only black member of Congress. Adams represents one of the only majority-minority districts in North Carolina.