Is it possible to fairly re-district?

Redistricting will always end in gerrymanders

By Samuel G. Howard in UpRiseRI

“...Would it not in that case be simpler

for the government

To dissolve the people

And elect another?”

-Bertolt Brecht, “Die Lösung” (The Solution) (1953)

Every ten years, following a Census, we engage in a battle over redistricting.

Redistricting (or reapportionment) is the process through which maps of voters are divided up into sections for the purpose of electing politicians. 

Because this task is typically placed in the hands of legislatures, it is often remarked that if elections are when voters choose their politicians, redistricting is when politicians choose their voters.

At its fundamental heart, this is a battle about fairness. When a group perceives a redistricting process to be unfair, they call it “gerrymandering” – there are more than enough explainers about where the name comes from, so I won’t rehash it here. 

Gerrymandering comes in many forms; there’s racial gerrymandering where racial groups are excluded or included. There’s political gerrymandering, where voters are included or removed based on their voting history. A city can be gerrymandered so it isn’t in the same district as towns it might share county borders with. The term is used across such a wide variety of situations that it’s fairly difficult to define in a way that covers all situations.

Redistricting is a Competition of Values

In 2014, political scientist Jonathan Ladd argued gerrymandering is a useless concept. Ladd instead lays out six values that come into play during redistricting, citing David Butler and Bruce Cain’s Congressional Redistricting:

1.  Equal Population

2. Matching Natural and Political Boundaries

3. Compactness and Contiguity

4. Party Fairness (an unbiased seats-to-votes curve between the parties)

5. Ethnic Fairness (substantial numbers of minority ethnic group members elected)

6. Party Competition (close elections and party alternation in a substantial number of districts)

Crucially, Ladd argues, it is impossible to satisfy all these values across all districts. Some of these must be prioritized over others.

Under current precedent and law, there are two values that need to take priority over others. 

First, Reynold v. Sims (1964) requires that state legislative districts have equal population (political scientist Adam Myers, in my view, persuasively argues that this decision is responsible for the collapse of the RI Republican Party as a serious political force). 

Second, under the Voting Rights Act (especially since the 1986 Thornburg v. Gingles case), districts cannot break up sizeable, cohesive blocs of minority voters who, in a single district, could elect a candidate (an example of a district that satisfies this requirement is State Senate District 6, which connects Black communities from Providence’s Mount Hope and South Side into a single district).

So any redistricting plan in the United States starts with its first two values set. First, districts must be of equal population. Second, they must not diminish the political power of minority voters.

Problems then arise as we try to satisfy the other values (and whatever further values we tack on). For instance, as we see partisan polarization along geographic lines (i.e., Democrats in the cities, Republicans in the rural areas), it becomes difficult to create party fairness and still respect natural and political boundaries, as you have to stitch disparate voters from across municipalities into districts with their co-partisans. 

Something similar comes into play if you want competitive elections between parties. In Rhode Island, to maximize competitive districts that would give both parties a fair shot at control of the legislature, you’d likely have to create long strip districts that would reach from Providence to the Connecticut border.

These are the problems that any redistricting process must navigate. But many prefer to pretend that these tradeoff aren’t serious concerns, or can be overlooked. This is not helped by the conflation of lack of “compactness” and “gerrymandering” – people respond negatively to wriggling districts that bestow gerrymandering its name.

One example of a “solution” that is algorithmic redistricting (such as this example from researchers at Brown), using an algorithm to draw the line. Proponents of such proposals generally highlight its removal of human biases, but this is a false assumption. 

Instead, the algorithms come pre-biased, often discarding all other requirements in favor of two: compactness and equal population. The results are districts that are often nonsensical in terms of natural and political geography, do nothing to address problems of partisan gerrymandering, and pose substantial violations of the Voting Rights Act.

The simple fact of the matter is that we can’t just set aside the other values that arise as part of the redistricting process. Emphasizing compactness and equal population alone means overlooking racial discrimination or partisan advantage. 

One need not have set out to make a racially discriminatory map to develop one that would do so in effect. Ignoring those values is a political act, with real consequences. Thus, would-be redistricting plans need to be explicit in how they work through those values.

Declare Your Values

Legislation about redistricting this legislative, fortunately, does declare its values. So let’s walk through the pieces of legislation that are before the General Assembly.

H6222A/S0852A (Shekarchi-Ruggerio Plan)

The Shekarchi-Ruggerio Plan is introduced by General Assembly leadership and is the most likely to pass (each chamber has passed their respective bills). It creates an 18-member commission, which is made up of six senators, six representatives, and “six members of the general public.” The speaker and the senate president each get to appoint four of the members from their respective chambers, and then half of the members of the general public. The House and Senate minority leaders each get to appoint two members from their chambers and zero members of the general public. The end result is a 14-4 Democratic to Republican commission.

Shekarchi-Ruggerio lists the following values (in this order):

1. Districts must be single-member districts.

2. They must respect the VRA and any other federal law or court precedent.

3. They must be as near to equal population as possible (with ≤1% deviation for US Congressional districts and ≤5% for state legislative districts).

4. Use 2020 Census data.

5. Districts should be compact, “reflect natural, historical, geographical, municipal and other political lines and communities of interest, as well as the right … to fair representation and equal access to the political process.”

6. They must be as contiguous as possible.

7. They must attempt to ensure precincts have more than 100 voters by not breaking up overlapping districts (US congressional, state senate, and state representative) too much.

So, Shekarchi-Ruggerio is a mix of political values with practical demands. Of the values it can choose from, it seems to order them to emphasize compactness, respecting existing boundaries, and contiguity. It sets aside questions of partisan fairness or competition. It would be surprising then if this commission drew maps that resulted in the RI GOP winning closer to a third of the General Assembly seats instead of the 12-13% they have now.

The part about single-member districts is important, but I’ll address it later in this piece.

H5868 (Cortvriend Plan)

The Cortvriend Plan also establishes an 18-member commission, but it switches the ratio of General Assembly members general public. There are four representatives and senators, with three of each appointed by the respective chamber’s presiding officer and one from each chamber’s minority leader. The remaining ten members are members of the general public, each chamber’s presiding officer appointing three and each chamber’s minority leader appointing two. 

It also lays out the requirement that a certain proportion of general public selections must not be “registered member of the two (2) largest political parties in Rhode Island.” This results in an on-paper more balanced 8-4-6 ratio of Democrats to Republicans to (likely nominal) Independents for the commission. Given who makes the appointments though, I would say this is actually a 12-6 Democratic to Republican commission.

It lists the following values (in this order):

1. Districts must be single-member districts.

2. They must respect the VRA and any other federal law or court precedent.

3. They must be as near to equal population as possible (with ≤1% deviation for US Congressional districts and ≤5% for state legislative districts) according to the 2020 Census data.

4. Shall not dilute or diminish racial and linguistic minorities’ electoral power.

5. Respect the integrity of communities of interest (defined as “including, but not limited to, racial, ethnic, economic, social, cultural, geographic, or historic identities”).

6. Districts should be compact, “reflect natural, historical, geographical, municipal and other political lines and communities of interest, as well as the right … to fair representation and equal access to the political process.”

7. They must be as contiguous as possible.

8. They must attempt to ensure precincts have more than 100 voters by not breaking up overlapping districts (US congressional, state senate, and state representative) too much.

9. Not unduly favor or disfavor any political party.

So Cortvriend incorporates all of Shekarchi-Ruggerio, but goes further by explicitly mandating minority protections and defining communities of interest, while also including a provision to prevent one party gaining partisan advantage over the other. However, this value is the last listed value, and by the time the others are gone through, it would likely be difficult to prove intentional partisanship in the way the lines are drawn.

H6009 (Newberry Plan)

The Newberry Plan creates a seven-member commission, with a commissioner each appointed by the Speaker, the Senate President, the House Minority Leader, and the Senate Minority Leader. Two members are appointed by the Ethics Commission and must not be members of the two largest parties. 

Finally, a retired judge from either the state Supreme or Superior Courts is also appointed by the Ethics Commission. The result would likely result in a 3-2-2 Democratic to Republican to (nominal) Independent commission, but the majority could switch around a lot, based on who the Ethics Commission appoints.

It lists the following values (in this order):

1. Districts must be single-member districts.

2. Be contiguous.

3. Comply with federal laws.

4. May use the most recent Census data.

5. Cannot favor a political party or incumbent.

6. Cannot use partisan data or voting history data.

7. Cannot intentionally dilute the representation of communities of interest.

8. Can only preserve “the cores” of existing districts if all other requirements on this list are met.

9. Be as equal as practicable.

10. Not deviate more than 5% from each other in population size.

11. Take tribal governments into account.

12. Avoid splitting government subdivisions (e.g., municipalities and counties).

13. Pick the squarest or roundest districts with the shortest perimeters (i.e., the most compact).

This is basically the Republican proposal (though I do want to mention that both the Cortvriend Plan and the Newberry Plan share supporters, with both bills’ sponsors signed on as co-sponsors of the other bill); so it is the most with preventing the consideration of political concerns (notably at the expense of the Reynolds-required equal-sized districts). 

It also explicitly mandates that incumbents can’t be protected, which is a favored use of redistricting. It kind of ignores the VRA and the relevant court precedents, never mentioning either, just referring to “federal laws.” 

However, it also takes into account the existence of Native governments, and also moves to keep smaller governments intact. It doesn’t care about small voting precincts. Notably, it has working definitions of what “compact” means, but it’s basically the last thing checked for.

One other aspect is that the commission actually selects three to five maps that go to the chambers to be voted on, and notes which one bests meets the stated values. The legislature then gets to vote on which of the selected plans it likes best, and if can’t decide, the commission’s “best” plan is submitted to the governor for approval.

S0864 (Euer Amendment)

The Euer Amendment enshrines in the state constitution an Independent Redistricting Commission. The commission is made up of ten voters who are registered affiliates of the two largest parties in the state (five apiece), and five members who are not members of those parties. The Secretary of State is develops an application process. 

The Secretary then gets rid of any applicants who haven’t voted in two of the last three elections or haven’t been a resident of the state for at least three years; and isn’t an elected or appointed official, family to an elected or appointed official, or staff or paid consultant for any legislature or political party or a lobbyist for the last five years. 

The Secretary then randomly selects two each from the applicant pools of the largest party, the second largest party, and the rest, and then these six commissioners select three of each from the applicants from each partisan pool from those a majority feel are qualified (as long as there’s support from at least one representative of each partisan pool).

The Euer Amendment lists the following values (in this order):

1. Districts must be single-member districts.

2. Comply with the US Constitution.

3. Be equal in population size except where it would conflict with the VRA. Population size will be based off of the most recent Census.

4. Comply with the VRA and not dilute or diminish racial and linguistic minorities’ electoral power.

5. Be contiguous.

6. Have partisan fairness (defined as “be able to translate [parties’ and non-affiliated voters’] popular support into representation in an elected body”)

7. Respect geographic integrity of any city, town, village, local neighborhood, or community of interest (defined as “a contiguous population which shares common social and economic interests”) except where this would conflict with the previous values.

8. Be compact and attempt to include nearby population areas over more distant ones.

9. Not take an incumbent’s or a candidate’s place of residence into account.

10. Not favor an incumbent, a candidate or a particular political party.

11. Not dilute the voting strength of any linguistic or ethnic group.

Once it’s drawn its maps (and given substantial time for public input), the commission then votes and accepts its maps.

The Euer Amendment is interesting in that it enshrines this process in the State Constitution. That’s good for achieving some semblance of a resolution to the redistricting battles, but it also makes it so that if our values as a state change, we can’t as easily adjust them. 

The Euer Amendment is also the only of the four plans here that explicitly demands that parties be able to translate their voting strength into seats in the General Assembly. That comes at the cost of compactness, which is one of the last values to be considered. 

Finally, it also places redistricting entirely outside of legislature’s power. It was also intended to be voted on in 2020, so that it would be in place for redistricting in 2022. As it is proposed now, it would have to be approved in 2022 and only come into play in 2032.

No Easy Answers to the Redistricting Question

What you think of these plans depends entirely on what you value. For instance, if you believe that partisanship shouldn’t be taken into account when the lines are drawn, then the Newberry Plan is for you (Shekarchi-Ruggerio doesn’t mention it, but that would seem to me to be a permissive silence). 

If you go the other way, and believe that parties need to be able to be able to turn voting strength into political power, then the Euer Amendment would provide stronger protections.

You might also have views about who should be in control of the redistricting process itself, whether it should be legislatures or citizens and residents. The most legislative-supreme is the Shekarchi-Ruggerio Plan. The most pro-resident plan is the Euer Amendment.

None of this is to criticize any of these plans for redistricting commissions. But it’s important if we want to discuss these that we lay out the values they offer. There are valid things to criticize about the way these commissions are set up. 

In her rebuke of the Shekarchi-Ruggerio Plan, Senator Mendes rightly notes it concentrates appointment power in the hands of legislative leadership. To a lesser extent, the same complaint is valid about the Cortvriend Plan.

What these plans should prompt is a discussion of how we want our maps drawn, what values should be emphasized, and who we want drawing these maps. These are difficult questions to answer, and yet, every ten years, we basically skate past them and answer almost none of them that have not already been answered at a federal level.

We Can Evaluate Districts Better Than Ever Before

One incredible aspect that has changed over the past ten years is that we, the people, have more tools to evaluate districts than ever before.

Free redistricting tools exist. Districtr allows users to map districts and then evaluate them based on racial & ethnic makeup of the population and the voting age population as well as compare against presidential, senate, and gubernatorial voting to get a sort of partisan score. 

It also allows you to redistricting Providence’s wards in addition the RI General Assembly and the US Congressional districts. It’s easy to use, but has a major drawback on that its most recent elections are from 2018, and it relies on 2010 data.

Another free redistricting tool is Dave’s Redistricting. You can draw your maps for US Congress and both chambers of the RI General Assembly and then Dave’s will evaluate your plan along five metrics: how proportional seats are to parties’ vote share, how many competitive seats there are, how well it provides for minority representation, how compact districts are, and how well plans do at not splitting counties (which really shouldn’t be very relevant in RI). 

Dave’s has an advantage over Districtr in that it uses 2020 precincts, and more recent data from the 2019 ACS estimates (which unfortunately undercounted Rhode Island by a whole 38,000 residents), and composites election results from 2012, 2016, and 2018.

PlanScore will evaluate district maps for partisan bias, using three different measures of partisan gerrymandering: the efficiency gap (how many votes parties “waste”), partisan bias (how many seats parties might be expected to win in a perfect 50%-50% election), and mean-median difference (the difference between the average vote share a party receives across all districts and the median result; if these aren’t close, the districts are likely biased against one party).

None of these are perfect tools, and they all have their own biases baked into them. But even in 2012, this sort of mapping and access to data was still the stuff of specialists. In 2022, as redistricting occurs, residents will be be better able to evaluate and understand redistricting proposals, and even develop their own.

The Conversations About Elections We Aren’t Having

Redistricting is a thorny conversation, but the conversation we have about it isn’t adequate to address all of these thorns. All of these plans, whether it’s Shekarchi-Ruggerio or the Euer Amendment for an Independent Redistricting Commission essentially ensure that gerrymandering will persist. 

There is simply no way to draw single-member districts that will not require sacrificing one or more of the values. Draw districts for partisan fairness, and we’ll see complaints about respecting natural and political boundaries and separating communities of interest. If you overemphasize compactness, you can violate the VRA very easily.

The end result is that any plan will draw accusations of gerrymandering of some sort. I said at the outset of this that gerrymandering is a form of redistricting where the districts are perceived to be unfair. You have to be unfair to one value or another when you redistrict.

And yet, all of the plans mandate single-member districts. The Euer Amendment bakes them into the state constitution, effectively meaning we will be constitutionally required to gerrymander. No matter how independent the redistricting commission, it will have to create some level of unfairness in its maps, simply to do its job. Each of these redistricting plans ensures we are destined to fight over this problem a decade from now.

As long as we start from the assumption that this is the only way to have representative government, this is going to repeat. It is setting ourselves up for failure.

And yet, there are solutions. One way we could sidestep the gerrymandering conversation entirely would be by adopting mixed-member proportional (MMP) as an electoral system (used in Germany and New Zealand). 

What MMP does is assign seats in a legislature by votes for party in addition to having seats elected from districts (typically, it’s half selected from lists of parties’ candidates and half from districts). If a party earns 40% of the statewide vote for parties, but only 20% of the seats, MMP gives them more seats in the legislature until they have 40% of the total seats. 

This effectively eliminates the incentive for partisan gerrymandering, since parties are almost always guaranteed a fair amount of seats. That leaves us free to focus on things like balancing compactness with the requirements of the VRA. Lists also allow us to place mandates on parties, like requiring that candidates from lists result in a group of elected officials that is reflective of the state’s population (e.g., if you elect only men from the districts, requiring that you achieve gender parity from the candidates who advance from the list).

Another option is multi-member districts. These can be voted for in multiple manners (the most common version is single transferable vote, which sets a quota of votes candidates need to clear, asks voters to rank their choices and then transfers surplus votes for candidates who exceed the quota over to other candidates), but even just the basic every voter gets one vote works relatively fine (or, at least, better than single-member districts). 

Multi-member districts tend to work well for protecting minority rights, since it’s relatively easy for candidates to clear the quota, and thus it’s harder to bury a particular racial or ethnic group under the weight of another group’s votes.

Finally, we could just eliminate districts altogether and switch to a purely proportional system where candidates are elected from lists. 

While (as I mentioned above) lists can have mandates placed on them to ensure demographic representation, a purely proportional system can also work by allowing parties to form which explicitly intend to represent minority groups’ interests. This competition for votes can also force other parties to respond by specifically addressing those groups’ concerns. One drawback of these systems is that they often use a threshold system. 

Set too low, and you get a fragmented political system where it’s hard for anyone to govern (see Netherlands, Belgium, Israel). Set too high, and you can end up throwing out almost 20% of the votes (if this is a shocking number, consider that we effectively waste a similar amount of votes under our current system).

I realize this is beyond where a lot of people want to go. Such a conversation is greatly destabilizing to the way we conceive of elections and politics. Very few people in Rhode Island politics, even those who would stand to gain, want a more proportional electoral system that has space for other parties to rise and fall. Everyone understands the system we have now, and there’s almost no appetite to learn something new.

But the reality is that Rhode Island is poorly served by its current electoral system. Democrats are in a constant civil war about access to party resources. Republican caucuses in the General Assembly are less than half the size they should be based on the statewide vote for Republican candidates. 

Voters who voted against their representatives effectively go unrepresented. And much of the legislature never faces a general election opponent, leading to reduced turnout and legislators who are insulated from the consequences of bad behavior.

If we want to actually change this, we need to start talking about more than the idea of “who draws the lines?” and start asking if we want to draw lines at all.