Make state legislatures bigger.
This week, the Supreme Court agreed to take up the case of Gill v. Whitford, whose plaintiffs claim the electoral boundaries drawn up by the GOP-controlled Wisconsin legislature constitute a gerrymander so extreme as to deny them their full ability to vote. The maps in my home state were engineered with merciless precision to ensure a lasting Republican majority in a place whose recent voting practices have oscillated within a narrow purple band. This is surely unfair; the courts will decide whether it’s unconstitutional as well.
It won’t be easy. Redistricting is an unholy snarl of law, politics, and math, and there’s probably no single person who understands it from top to bottom. (But some of us are trying: A group of researchers at Tufts University is organizing a series of conferences in which citizens and scientists are hashing some of these issues out; I’m hosting one in Madison, Wisconsin, in October.)
Here’s one reason redistricting is such a difficult problem. In Wisconsin, two of the major population centers, Madison and Milwaukee, are dominated by Democrats, while only one (centered on, yes, crucial Waukesha County) is similarly Republican-heavy. The rest of the state is more moderate but leans mildly Republican overall. It’s not totally crazy, then, for a legislative map to consist of a few heavily Democratic districts in the big cities and many slightly Republican ones elsewhere. Such a map wouldn’t necessarily be evidence of biased, partisan gerrymandering—in some measure, it’s just a consequence of the state’s geography.
Wisconsin claims its map is just such an organic, free-range gerrymander. University of Michigan political scientist Jowei Chen sees it differently. Chen generated 200 random maps algorithmically, using party-neutral methods. Not one of them was as GOP-friendly as the one Wisconsin is using—not even close.
In the computer-generated maps, the number of districts that supported Romney over Obama in 2012 ranged from 38 to 47 out of 99. (Obama got 52.8 percent of the vote statewide, while his Republican challenger got 45.9 percent.) In the actual Wisconsin map, however, 56 of the 99 districts contain a Romney-backing majority. The GOP-drawn map also scores badly on traditional measures of quality. Every computer-generated map kept at least 18 counties entirely within a single district, and some preserved as many as 24. Wisconsin’s map leaves only 14 counties intact.
While we’re not about to give over our districting process to Professor Chen’s laptop, the auto-generated maps serve as a useful benchmark for human-made districts. This figure from Chen’s paper shows just how much of an outlier Wisconsin’s current map really is.
Is there a way to solve the problem of gerrymandering without getting the courts involved? I think there is, and it’s pretty simple: We ought to have a lot more state legislators.
State legislatures are like the restaurant dinner in the old joke—awful, and the portions are much too small. The Wisconsin Assembly has 99 members, each representing around 60,000 people. That’s just two more legislators than we had in 1860, when the state’s population was about an eighth as large. While most states have big districts like Wisconsin’s, there are exceptions. West Virginia, a third of Wisconsin’s size, has 100 members in its lower house. North Dakota has 94 representatives for a population a little bigger than Milwaukee’s. And New Hampshire, the champ of them all, has a House of Representatives with 400 members, each one representing just more than 3,000 people.
Why would bigger legislatures blunt the power of gerrymandering? Chen has a paper about this, too. Think about the extremes. If Wisconsin had the maximal possible number of legislative districts—namely, one for each Wisconsinite—the partisan composition of We-the-Legislature would, by definition, match the partisan composition of the state. Gerrymandering would be impossible. At the other extreme, what if there was only one district, comprising the whole state? In a swing state like Wisconsin, legislative control would be up for grabs. In a reliably Democratic or Republican state, the opposition party would have zero representation in the legislature, probably forever—the ultimate gerrymander.
If the largest possible districts yield the strongest gerrymandering and the smallest possible ones allow for no gerrymandering at all, you might guess that making districts smaller makes gerrymandering harder. Chen’s data suggests that, in swing states at least, this guess is right: States with more legislators per capita, like Minnesota, display less evidence of gerrymandering than big-district states like Michigan.
Smaller legislative districts would have other healthy effects, too. The average Wisconsin Assembly candidate raised $33,000 in campaign money in 2014—sofa-cushion change by federal election standards but enough to deter many potential candidates without rich-person connections. If our state legislature had 300 assembly reps instead of 99, every town of 20,000 could have its own legislator. An ordinary person with some standing in the community would have a real chance, and, if she won, that ordinary person could spend her two-year term legislating instead of frantically scrabbling for re-election funds. If more people have the chance to vote for someone who lives in their own town, you’d expect higher turnout, too.
Enlarging state legislatures obviously wouldn’t do anything to solve the problem of gerrymandering U.S. congressional seats (though lots of people think the House of Representatives, too, should be a lot larger). There are also real obstacles even if we do stick to the state level. In some states, Wisconsin among them, increasing the size of the legislature would require amending the state constitution. Incumbent legislators also have an obvious disincentive to dilute their own power by crowding the chamber. Not least, a 300-person Wisconsin Assembly would be expensive and complicated to run.
And what about the big states? To have one representative for each 3,000 Californians, the legislature would need 13,000 members. Guess what—there are people who think that’s not insane! John Cox, a San Diego businessman and conservative gubernatorial candidate, is pushing a statewide ballot measure to create an 8,000-member assembly. Each of the 80 current districts would have 100 representatives, who would select one of their number to go to Sacramento and be a full-time legislator. But all bills would be voted on by all 8,000 representatives. And if the group of 100 soured on their chosen captain, they could call for a recall at any time by majority vote. There are probably a million reasons this California plan is a lousy idea, but the scientist in me would love to see the experiment carried out.
There’s one more problem with the big-statehouse solution to gerrymandering. Under the current system, the legislature can dilute the electoral power of a college town by, for instance, pasting on an adjacent chunk of mining country that’s just large enough to give the district a Republican lean. Let’s say you break the college town off from the miners; that gives you two districts, one Democratic and one Republican, each one rock solid for its party. Keep going, and the original big magenta district becomes a patchwork of red and blue tracts, like a blurry image that resolves into focus as you load it.
Gerrymandering requires competitive districts; you pack the opposition voters into a few districts where they have a supermajority, leaving yourself a small edge in the ones that remain. What Chen finds is that bigger legislatures protect against gerrymandering precisely because smaller districts are more homogeneous, and thus more likely to be uncompetitive. As civic-minded people, we want lots of vigorously contested elections, district maps that don’t rig the game in favor of one party, and elections carried out without the constant intervention of the courts. We might have to settle for two out of three.