Are you against gerrymandering? Of course you are! You've laughed at the shapes of districts with nicknames like the Praying Mantis, the Steam Shovel, and Goofy Kicking Donald Duck. Like almost everyone who follows politics, you agree that it's wrong to fiddle with legislative maps to help a favored party or candidate.
Or do you? To test your commitment, here's a composite example from a fictional 51st state of the union we'll call the State of Madison.
Public opinion in Madison is strongly opposed to extreme partisan gerrymandering, the sort where the more powerful of two major parties redraws the map to hurt the other. The leadership of the state legislature has taken this to heart and entrusted the task of drawing the next set of district lines to a bipartisan commission. It's split half and half between the two major parties; the tiebreaker is a genial retired lawmaker who gets along with everyone. True, there are no Libertarians or Greens on the panel, nor even any registered independents. But that's understandable—isn't it?—since voters have not chosen to elect anyone from those groups to the legislature.
And there's more good news. Some were worried that the majority party, which got about 54 percent of the vote and 56 percent of the seats last time around, would engineer matters so as to grab many more safe seats. Not so. When seasoned political analysts look at the lines that were drawn, they can predict exactly which party is going to win nearly every seat next time, and they say hardly any will change hands. There had been some grumbling about how only three seats were competitive in the last general election; the commission must have been listening, because this time there will be four competitive seats instead of three—not enough to tip any balance, but at least enough to provide some interest come November.
One curious thing about those four competitive-in-November seats: They're all open seats where an incumbent is retiring. That's because none of the incumbents who planned to run again volunteered their districts to be made into the competitive ones. In fact, when you look more closely, many incumbents got their districts snipped here and expanded there so as to make them safer, not just in the general election but also—this will be less obvious, except to the well-informed—in the primaries. Sometimes the voters in a town never really warm up to you, in which case the best course is to pass that town on to the next lawmaker over.
Once you look more closely, you see that many of the districts have shapes that are a little more stretchy and boundary lines that are a little more jiggly than they would strictly need to be. Also, they crisscross county and city borders more than they have to. In two or three instances, you notice a thin peninsula of land that juts out from the main body of a district to capture a remote neighborhood. You ask an insider, who explains that those fingers are meant to connect the home residence of some lawmaker with the district he or she wishes to represent. In fact, the starting point for most of the districts had simply been to ask incumbent members of both parties how they wanted their districts to look. To paraphrase a famous line: Officials had succeeded in picking their voters, rather than letting voters pick their officials.
In one case (and only one), the new map throws two incumbents from the same party into a single district. Your insider friend explains that that was to get rid of a member of the majority party who just caused trouble all the time—making noise about supposed scandals, never cooperating with more senior colleagues. No one likes him, really. Or at least no one in the leadership does. Without this disruptive personality, the next legislature will be more collegial and less polarized. And that's to the good, right? Also, seeing what happened to this troublemaker, none of the members are going to think about crossing the leadership next term.
Everyone Wants Reform—But What Kind?
I call this kind of arrangement a "buddymander." Many people who hate partisan gerrymandering hate it too, but others are willing to let it slide or even are fine with it. The animating logic is: We'll protect our guys and you can protect yours. It's outwardly different from extreme partisan gerrymandering, since the main goal is not to take away seats from the opposition. But the two spring from the same underlying temptation: When the system gives insiders wide discretion over line drawing, they are apt to use it to advance their own interests.
The main task of redistricting reform is to confine that discretion. To appreciate the difficulty of that task, let's switch for the moment to a seemingly remote question: Why allow any discretion in drawing legislative maps at all?
When you serve on a redistricting commission, as I have now done in Maryland twice, that's one of the most common questions you get: Why can't we turn the whole thing over to a computer algorithm? At its simplest, this can take the form of proposing that the state simply be divided into districts of equal population (as the courts require) by some brute method. Thus a state might be divided among the proper number of congressional districts by drawing vertical lines dividing it into strips of varying widths.
To spend a few minutes with such a map is to grasp its flaws. Even in a conveniently rectangular state like Colorado, districts would end up comprising unrelated communities separated from each other by mountains and long distances. Coherent communities would be split, perhaps multiple ways, to no good purpose. Before long, you will have rediscovered some of the basic keys to good districting, namely: compactness, with districts looking more like turtles than snakes or octopuses; practical contiguity, meaning that all sections of a district are accessible by road connections without having to leave the district; and congruence with the boundaries of other political subdivisions, such as counties and cities.
Happily, each of these three desirable features can be translated into formulas in algorithm-friendly ways. While experts have devised many mathematical formulas to score compactness, picking any one of them will help curtail the worst gerrymanders. Likewise, a formula can keep track of the number of county splits (lower is better). Further prescriptions can install a decision mechanism such as splitting more populous counties before those that are less populous.
Unfortunately, algorithms are far less adept at incorporating formulas for a fourth aspect of good districting, one that has been called intelligibility. People want at least a fighting chance to describe their district in words, and to guess correctly whether someone lives in it based on knowing where his or her residence is. Curved and diagonal lines usually don't register as intelligible, while "east of the River" or "south of I-70" may work fine. And while some neighborhoods may need to be split to make the numbers come out evenly, intelligibility is lost if a district line heedlessly splits every neighborhood it hits rather than finding the boundaries between them.
Because few of us are willing to jettison intelligibility entirely, fully algorithmic districting is unlikely to arrive anytime soon. But the impulse at least deserves respect, since it stands for the right goal: to confine the role of discretion. And it points the way to what is probably the most promising use of mathematics in districting, which is to pair quantifiable formulas with a band of discretion within which mapmakers are asked to pursue intelligibility. For example, it might be proposed that a lawful map must attain a compactness score no more than 20 percent worse than the most compact map taken under consideration, or that it must inflict no more than two more than the minimum attainable number of county splits.
It's unsurprising and true: States that have enacted clear, objective rules to guide mapmakers on topics like compactness and congruence tend to have far less of a gerrymandering problem than those that have not. The same kinds of rules also provide a firm basis for judicial review. While it is troublesome to give judges themselves massive discretion in line drawing—for one thing, it risks magnifying the role of politics in judicial selection, already a problem in many states—it is much less dangerous to assign them the quintessentially judicial task of holding others to clear and specific marching orders.
For an example of a redistricting criterion that is anything but clear and objective, consider the notion that district lines should follow so-called communities of interest. No one can pin down what this means to general satisfaction. Should a town that is suburban, industrial, and coastal be grouped with other areas that are suburban? Industrial? Coastal? It's a recipe for arbitrariness, disagreement, and manipulability. The same is true if a court is instructed to apply that same vague standard later on.
The toughest question—on which it is hard to offer more than speculation—is that of who, if not political insiders, should draw the lines. The currently popular plan, adopted in such places as Arizona and California, is that of the independent citizen volunteer commission. Experience with this innovation has been mixed so far, with much depending on the details of how a given law is drawn. Over time, interest groups will probably attempt to influence, or even infiltrate, the citizen commission. At the same time, the new blueprints for citizen redistricting include powerful measures to shake up the old way of doing things, such as rules forbidding commissions to take into account the residence of any incumbent or the voter registration or voting history of any community.
Public submission of maps, enabled by open databases and the availability of free or cheap software, holds great promise as well. For one thing, courts are more likely to provide effective judicial review if multiple maps are made available for comparison.
Yet another reform blueprint is to turn over the task to a legislative services bureau bound by strong impartiality norms. The results have been applauded in Iowa, a state known for relatively clean politics in which the two main political parties are approximately equally matched. But legislative service bureaus in other states might prove less robust in resisting political influence.
In the background are widening differences over what the goal of good districting should be in the first place. Those of us on the classical liberal side are likely to be inspired by the ideals of neutrality, impartiality, and objectivity. But some of the academics and commentators drawn to the controversy want a rough match between seat strength and voter strength—implicitly, a criterion of proportional representation, so that if a state is 40 percent Republican, say, somewhere around 40 percent of its seats will go to Republicans. A second group of thinkers takes the view that the chief evils to be fought are those of polarization and the alienation that arises from feeling one's vote doesn't matter; they thus advocate conscious efforts to create more competitive districts, especially ones that are competitive in general elections. (As Charles Blahous of the Mercatus Center at George Mason University and others have shown, the trend has been for more districts to become competitive in party primaries, even as fewer remain competitive in the general.)
These two latter schools of thought—proportional representation and a preference for the creation of more competitive districts—are in practical terms at odds with each other. When many districts are drawn to be competitive, relatively modest swings in voter sentiment can lead to large swings in seat control.
The polarization issue is also more complicated than it may look. It is true that party positions in the U.S. House and many state legislatures have grown more polarized in recent decades, with a winnowing out of conservative Democrats and liberal Republicans. But the same dynamic can be seen in the U.S. Senate, and no one now alive plays any role in drawing that body's boundaries. The best guess is that several forces are contributing to polarization, with gerrymandering one part of the mix.
Proponents of proportional outcomes sometimes seem to be fighting a losing war against the inherent nature of America's "first-past-the-post" electoral system, which has always tended to generate major gaps between voter strength and seat strength. (This faction might be better advised to throw its support behind a ranked-choice, Australian-ballot, or European-style system, each of which is meant to avoid this outcome.) At any rate, the result-minded thinkers in both schools have something important in common, which is that both are obliged to resort to line drawing that is intensely conscious of voters' political leanings. It's hard for either to get on board with the California idea of blinding mapmakers to political data about registration, voting history, and politicians' residences.
Indeed, once you accept a goal of corralling voters into patterns judged to yield good electoral outcomes, you may even grow cool (as some contemporary academics are) toward traditional neutral-impartial-objective criteria such as compactness and avoiding county splits. They just get in the way of reaching the right results.
Redistricting Reform Returns From the Dead
After many years of back-burner status, interest in partisan gerrymandering began mounting rapidly around 2015 for two reasons. First, constitutional litigators had a case they hoped to win. Second, the issue got pulled into the ceaseless noise machine of Red Team/Blue Team warfare, because (as had not been the case over long historical stretches) one party was now doing significantly better from gerrymandering than the other.
For years, the only hope of getting the Supreme Court to recognize a constitutional remedy for gerrymandering turned on the cooperation of Justice Anthony Kennedy. With Kennedy's tenure on the bench nearing what was to prove to be its end in 2018, a search went out for a case that might tempt him. That search failed. The "efficiency gap" test proffered in a case out of Wisconsin failed to persuade him. Kennedy then retired, after which the necessary votes weren't there. In 2019, the Court used two cases—Rucho v. Common Cause and Lamone v. Benisek—to rule, 5–4, that there was no constitutional remedy to be had in federal court over partisan gerrymandering. But in the meantime, the usual publicity apparatus deployed for big Supreme Court cases had done its thing, and the issue had risen in public awareness.
A brighter-than-usual spotlight on this issue also followed the 2010 wave election, in which Republicans ousted Democrats from 680 state legislative seats in the biggest such partisan pickup in history, flipping no fewer than 20 state legislative chambers. In what became an oft-told tale, the GOP carefully deployed its new power using a program called REDMAP, which helped in devising exquisitely detailed gerrymanders that enabled the party to push its advantage further against Democrats in state after state. It helped that database and geographic information system technologies were improving constantly so as to allow super-fine-grained assemblage of districts on the fly. Legislators were able to sort local voters by political preference down to individual blocks, buildings, and households—a far cry from the old days, when pulling off a gerrymander might require weeks amid maps and awkward printouts of voter data.
Republicans enjoyed one other systemic advantage as well: Their objectives often meshed nicely with those of the federal Voting Rights Act (VRA). That law sanctions and even encourages—though the courts have had trouble sorting out exactly to what extent—the creation and maintenance of race-conscious districts in which minority voters hold a majority big enough to elect a candidate of their choice. It's an open secret that maps that result in significant black representation are often also maps where Republicans do well, since VRA districts funnel one of the most loyal Democratic voting groups off from the rest of the map. A Republican strategist could simply approach black legislators and ask them to draw their "perfect district." Wildly noncompact districting was accepted as legitimate in many VRA situations; indeed, it's not uncommon for districts that show up on lists of the worst partisan gerrymanders to have been created by legislators (or even suggested by judges) under a VRA rationale.
Notwithstanding what happened in 2010, there has been little over the longer term to mark out gerrymandering as a distinctively Republican practice. In 1986, for example, officials of both major parties took positions more or less the opposite of their 2019 ones. In Davis v. Bandemer, a high-profile Supreme Court case from that year, the Republican National Committee filed an amicus brief in favor of strong Court intervention to correct partisan gerrymanders (a stance requiring it to argue against its own Indiana state party, which had engaged in the practice in the case at hand). As one of the brief's co-authors explained the following year, Democrats had just pulled off a massive and successful gerrymander in the state of California, and Republican leaders foresaw the same thing happening in many other states, "since Democrats control considerably more state legislative houses than do Republicans."
Meanwhile, the chairman of the Democratic National Committee disparaged the idea that the federal courts should be in any hurry to jump in, saying that Republicans, having failed "to win control of more legislatures," were now seeking "a quick fix" to make up their losses.
In the longer run, unease at California gerrymanders did help touch off what became the most notable development in redistricting reform: the rise of independent citizen commissions, typically propelled by the ballot initiative process. Arizona went first with Proposition 106 in 2000, followed by California with Proposition 11 in 2008 and Proposition 20 in 2010. (Prominent California Democrats, including once-and-future Speaker Nancy Pelosi, quietly worked to sabotage the latter effort and keep the electeds in control.)
Map-related misconduct was indeed a bipartisan affair. A 2006 report from Azavea, a geographical software applications firm, listed the 10 most gerrymandered states. At the time the maps were drawn, four were controlled by Democratic legislatures, five were controlled by Republican legislatures, and one was split. Of the 10 most gerrymandered districts, four were in states that had Democratic legislatures at the time of drawing, three were in states controlled by Republicans, and three were split. Illinois Democrats in 2016 managed to kill a referendum backed by 500,000 petition signers, just as Michigan Republicans have lately fought a pitched legal battle to foil a voter-backed plan for an independent commission. Maryland Democrats behave much like Texas Republicans, and so on.
What Comes Next?
Some of today's momentum will continue, come what may. While the federal courts may have bolted their doors against gerrymander challenges, their state counterparts are still capable of surprises, especially when they draw on the language of state constitutions, as Pennsylvania's high court did in striking down that state's congressional map in 2018.
A deeper problem for reformers is that they are beginning to run out of states with strong ballot-initiative and referendum provisions. Only 18 states allow voters to initiate laws or constitutional amendments directly, and the practical number is a few less than that, since several of the states make the process quite hard to use. Most districting reform successes in recent years have come in states where advocates either ran a ballot initiative or credibly threatened to do so, starting with Western states and more recently extending to Ohio, Michigan, and Missouri.
With Democratic fortunes beginning to revive in the 2018 midterms, the party will soon face decisions about whether to pursue gerrymanders in Virginia and other newly consolidated states at the cost of giving up some of the moral high ground the party has briefly occupied on the issue. Meanwhile, Republicans, who have ceded so much of that same ground in the scramble for the imagined cartographic Ring of Power, will have to decide whether it's worth trying to reclaim any. Barack Obama, who since his presidency ended has sometimes spoken out against gerrymandering, has now thrown in with the National Democratic Redistricting Committee (NDRC), an official Democratic activist organization, which may limit his maneuvering room to act in ways the party perceives as adverse to its interests.
Wouldn't it be neat, though, if there were a neutral reform that would do a powerful lot of good on a national basis, was plainly consistent with the U.S. Constitution, promised to work as intended with few or no unintended effects, and was easy to explain to boot? One that could slay buddymanders as well as gerrymanders of the extremely partisan kind?
Good news: There is! What's more, it's been hiding in plain sight all the while. Charles Blahous describes it in a valuable 2019 paper for the Mercatus Center.
It begins with the Elections Clause—Article I, Section 4—in which the Constitution grants Congress an express role in overseeing the elections states hold for the House of Representatives. The wording makes clear that it is allocating power so as to give state legislatures the lead but not the final word: "The Times, Places and Manner of holding Elections for Senators and Representatives, shall be prescribed in each State by the Legislature thereof; but the Congress may at any time by Law make or alter such Regulations."
Congress has used its enumerated powers in this area for well over 150 years. For example, it has at various points (including the present) required that states elect House members from single- rather than multi-member districts. It also began requiring states to divide population equally among House districts long before the Supreme Court began interpreting the Constitution to require as much.
Less well known is that for about 30 years a century ago, Congress extended its oversight to include other good districting practices. The Apportionment Act of 1901, whose relevant terms remained in effect until 1929, stated that districts must be made up of "contiguous and compact territory and containing as nearly as practicable an equal number of inhabitants." There is no reason why such rules could not be re-enacted today, updated (as Blahous persuasively argues) to specify a quantitative test of the sort that political scientists regularly employ. In measuring compactness across states, it makes sense to disregard elements of noncompactness that derive from the irregularity of states' external outlines, since Florida cannot help being more elongated than South Carolina, for example. A fairer comparison can be obtained by focusing on the length of district lines that are interior to the silhouette.
Depending on how much pressure it wishes to apply against gerrymandering, Congress would be free to make an overall compactness standard easier or tougher. For example, of the 18 states that already have a legal compactness requirement for House districts on their books, none currently has any districts with a "G score" (a metric that adjusts for exterior state boundaries) above 150. If Congress set the threshold at that level, it would render just 5 percent of current districts illegal, but those would include most if not all of what are known as the most egregious gerrymanders. If it proceeded to a somewhat tougher standard of 125, it would make 8 percent of current districts illegal.
Either way, we'd finally be rid of those oddly shaped, colorfully nicknamed monsters whose habitat is our electoral maps—districts like the Duck, the Snake by the Lake, the Broken-Winged Pterodactyl. And we wouldn't miss them.