This case has its roots in the framing of the Constitution.  The original structure of the British parliament awarded a certain number of seats to each incorporated borough (town) and to each shire (county).  When combined with the fact that only freeholders (property owners) had the right to vote, by the middle of the Eighteenth Century, there were boroughs that were very small with only a handful of voters (so-called “rotten boroughs”).   The non-representative nature of the British Parliament was one of the reasons why colonists did not accept the argument that they were represented by the British Parliament.  In drafting the U.S. Constitution, at least for the House of Representatives, the framers decided that representation in Congress would depend upon population based on a decennial census.

By requiring that representation in the House would be based on representation, the Constitution created a de facto requirement that states draw new congressional districts (at least when a state’s representation changed).  Some, but not all, states also based representation in state legislatures on population — again requiring periodic redistricting.  In simply requiring redistricting, the U.S. Constitution was ahead of its time.  Now, most countries that use a first-past-the-post system also have periodic redistricting.  The vast majority of them also use a non-partisan commission with specific criteria to draw fair and competitive district lines.  The framers, however, did not have the extra two centuries of seeing what works and what doesn’t work in the redistricting process.  And it is some of what happened next in the U.S. that has led the other countries to have neutral agencies handle redistricting.

In 1812, after the adoption of a new state constitution, Massachusetts had to draw new district lines.  And with one party (the Democratic-Republicans) controlling the legislature and the Governor’s mansion, the new lines were drawn to maximize the number of legislative seats that would favor the Democratic-Republicans and to minimize the number of seats that would favor the Federalists.  The weird shape of one of the districts was noted by a Federalist newspaper (back in the days when media was actually biased) and the term gerrymander (after then-governor and soon-to-be-Vice-President Elbridge Gerry) was born.

The essence of the gerrymander has not changed over time.  Every gerrymander does two things:  it packs the opposing party into a handful of safe districts and it cracks the opposing party out of a potential swing district into neighboring safe districts that favor the majority.  For example, imagine a region with four hundred people that need to be split into four districts.  This district has 180 people who vote for Party A, 180 people who vote for Party B, and 40 swing voters.  “Natural” district lines would give you a Western district that favors Party A (say 70-20-10), an Eastern District that favors Party B (again say 70-20-10) and two central swing districts (say the Western one favors Party A by 48-42-10 and the Eastern one favors party B by 42-48-10).    A pro-Party A gerrymander would redraw the lines so that District 1 (the westernmost district) favored them by only 60-25-15 and District 2 (the central western district) favored them by 58-37-5 — cracking Party B and swing voters out of district 2 and placing them into a district that was safe for Party B, turning 1 safe and 1 swing district into 2 safe districts.  At the same time, District 4 (the Eastern District) is redrawn to make it 80-5-15 giving Party A the majority (57-38-5 ) in District 3, making District 4 even safer for Party B and turning District 3 from a swing seat into a safe seat for Party A by packing Party B voters into District 4.

While the essence of gerrymandering has not changed, what has changed is the ability to manipulate data.  In the past, gerrymandering involved looking at precinct/voting district level results.  If you split a precinct between two districts, you basically assumed that all voters in the precinct were interchangeable.  Now, computers can make highly accurate guesses based on census block level data and data mining as to where the Republican and Democratic voters in a precinct live.  More importantly, computers can quickly run multiple alternative maps to maximize a party’s performance.   Once the majority party defines the parameters for the program (e.g., priority one — maximize the number of seats in which the opposing party gets more than 70% of the vote; priority two — maximize the number of seats in which your party gets between 57% and 62% of the vote; priority three — a certain number of seats that in which minorities have at least 40% of the vote; priority four — a certain score for compactness, etc.), it can then have the computer run ten or twenty maps and then opt for the one that has the most safe districts for the majority party.  The end result is a legislative map in which the minority party needs to get 55% or more of the statewide vote to even have a shot at winning control of the legislature.

Several times in the past, the Supreme Court has taken a look at partisan gerrymandering.  Each time, the majority has seen that such a claim is theoretically valid.  However, there has never been a majority that could agree on a standard.  The last time that the Supreme Court took a look at this issue, the split was 4-4-1.  Chief Justice Rehnquist, Justice Scalia, Justice Thomas, and Justice O’Connor did not think that there was a valid claim.  Justices Stevens, Souter, Breyer, and Ginsburg thought that there was a claim but could not agree on the proper standard for such a claim.  Justice Kennedy thought that there might be a claim but that none of the proposed standards was workable (in the sense of drawing a clear line between a legitimate map and an invalid map — in other words how far was too far).  Given Justice Kennedy’s position, for the last ten years, litigants have been looking for tools and standards that would establish a sufficiently clear line.

Wisconsin (and those who support Wisconsin on this appeal) basically attempt to mischaracterize the challengers as proposing a standard that requires proportional representation.  That is completely false.  The issue is not exact proportionality.  The issue is whether the lines give voters a realistic opportunity to change which party governs.  Fair lines should have the tipping point (the point at which control changes) be close to 50% of the vote.  How close the rest of the districts are to the median district is distinct from whether the tipping point is close to 50% of the statewide vote.

For the tipping point to be at 50%, the two-party vote — by percent — in the median district (i.e. the district that is exactly in the middle if you ranked the district by the percentage of the two-party vote that one of the parties gets) should be almost exactly the same as the statewide two-party vote — by percent.  Which leads naturally to the first part of any standard — do the district lines inherently favor one party,   There are lots of tools for measuring whether district lines are “unfair”  — comparing the vote in the median district to the state-wide numbers, the efficiency gap (does one party “waste” more votes than the other party), looking at the results in multiple elections (including elections for state-wide offices) to see if one party’s apparent advantage is real or merely a fluke caused by the candidates who ran.

Some suggest that the standard should require that the advantage be persistent.  One of the reasons for redistricting is to reflect changes in population over a decade; so it is expected that the impact of the lines will change.  More significantly, it is difficult to show that lines create a persistent disadvantage without allowing the unfair lines to be used in several elections.  I think it should be enough for the challengers to the district lines to use the last two cycles (state-wide vote) before redistricting (and any cycles that occur under the new lines before any trial) to show that the lines do create a clear advantage (make it difficult for the other party to get a majority of seats with a majority of the statewide vote) for the governing party under current conditions.

The other issue is one of intent — is the reasons for the lines to create that clear advantage.  Admittedly, the best theoretical argument for those opposing recognizing partisan gerrymanders is a claim that the apparent lopsided results are the result of geography.  Both in Wisconsin and in other states, however, there is evidence to support a finding of intent, and the intent standard in a partisan gerrymander case is not that different from a case involving racial gerrymander.  Back to the computer programs discussed above.  In the example used, the top two priorities were about partisan advantage for the majority — packing Democratic voters into ultra-safe Democratic districts and cracking them out of lean-Republican and swing districts to create as many safe Republican districts as possible.  It is possible to give the redistricting program strictly non-partisan criteria — compact districts, equal population, VRA compliant (setting a certain number of minority influence districts at 40% of the population), respecting precinct and other political boundaries, etc. — and have the computer generate a large number of maps.  The partisan nature of those maps can be compared to the map chosen.   In particular, each of those maps will generate a median district with a measurable partisan composition, and the set of maps will generate a mean, median, and mode for the partisan composition of the median district of each map along with a standard deviation from the mean and median.  If the map chosen is more favorable to one party by several standard deviations than the typical map, then that is strong evidence of intent.  Additionally, if there are other actions — such as the Wisconsin Republicans starting with a very slanted map and then looking to make adjustments to make it even more slanted — revealing a desire to create an unfair map, then you have a pretty good case that the map is the product of an improper intent.

The ultimate question is whether it is proper for legislators to be creating maps designed to give a partisan advantage.  The answer to that should ultimately be no.  The Constitution provides for a republican form of government — one in which the legislative branch is responsive to and fairly represents the will of the majority.  If it is acceptable for legislatures to draw district lines to frustrate the will of the majority, we have deviated from a republican form of government.  It may, of course, be too much to ask for legislatures to ignore their partisan desire (and their desire to run for higher office), but the Supreme Court can give us a legal rule that limits the ability of legislators to depart from partisan neutrality in drawing legislative and congressional maps.  Maybe, if the test has enough bite, more states will adopt the non-partisan commission as the instrument for drawing legislative maps.  As noted at the start, by early next year, we will know whether legislators in 2021 will have to worry about courts stopping them from using the redistricting process to prevent the voters from controlling who governs or if legislators will continue to be able to choose their voters in a way that strips all power away from the electorate.