If you are active on the political side of Twitter, every now and then you may see a post protesting partisan gerrymandering. These posts would often contain the phrases “fair maps”, or “let the voters pick their politicians, and not the other way around”. In fact, gerrymandering has been a point of discussion, and even litigation, for decades, as politicians in power have sought to retain their positions and the people have fought back against them.
Although it’s certainly fair to criticize this behavior, it’s also incredibly easy. A power-hungry politician drawing districts for his/her own benefit makes for a good boogeyman, but looking at the bigger picture, solving the issue of gerrymandering is only one step in allowing voters to gain equal representation in their governments. The fact of the matter is that gerrymandering itself is a feature of winner-take-all elections, and it can’t happen without them.
Leaving the issue of gerrymandering to one side, let me explain why winner-take-all elections are a problem for democracy. As almost anyone would be able to tell from looking at a typical election map, voters are geographically divided. Urban, densely populated areas typically have heavy concentrations of Democratic voters, while Republicans outnumber Democrats in most sparse rural areas.
This, in itself, isn’t a problem. Voters can live wherever they want, and vote however they choose. But it’s very easy to look at a map of electoral districts, at the areas represented by Democrats and Republicans, and think that they are fair because they reflect the political geography. But it completely ignores the millions of Democratic voters living in Republican dominated areas, and the millions of Republican voters living in Democratic dominated areas. That’s the key underlying problem of single-member districts.
The fair representation model
One of the systems I’ve devised to solve this problem is a state legislative model. The state legislature would be bicameral, with a House of Representatives and a Senate, similar to Congress and most current state legislatures. A single 35-district plan would be drawn by an independent commission (in other words, by me) for both chambers, and the chambers would differ in how they elect their members.
The Senate would elect 1 member per district using a form of Ranked Choice Voting (RCV). This would be something of a status-quo chamber, with a figurehead in each district with an incentive to work for all his/her constituents. The difference would be that, under RCV, third-party, independent, and intra-party challengers would have more of an opportunity to compete in these races without the fear of vote-splitting, which could produce Senators which better represented the districts.
The House would elect 7 members per district using a form of Single Transferable Vote (STV). STV is similar to RCV, in that voters can rank their candidates, and their votes (or parts of votes) can be transferred between candidates. The distinction from RCV is that multiple candidates are elected rather than just one, and the threshold for winning a seat would be 12.5% (with seven winners) rather than 50%. This would allow smaller communities of interest to elect their Representatives, and would facilitate proportionality, not just with respect to party ID, but also to ideology (beyond partisanship), race, income, and other demographic aspects. Votes for third-party and independent candidates would not only not be wasted, but would have a significantly higher chance of helping elect them, and as such more of these candidates would win seats.
If I were to explain in detail the intricacies of RCV and STV, it could take me a while. But the main benefit of these voting systems is that it expands the opportunity of voters to express their full preferences, which changes the way their vote counts to better reflect these preferences. If you’re curious, I’d recommend reading up on it.
Ideally, elections would be non-partisan in both chambers. Non-partisan primaries may be held to winnow the field of candidates to avoid voter confusion, but the balance of power would fundamentally shift from the internal structures of the dominating parties to the general election voters. They would no longer be limited to the hand-picked candidates of the two main parties, and they would be given the opportunity to elect more suitable candidates and hold them accountable more easily.
North Carolina
To demonstrate this model, I applied it to North Carolina, but it can be applied to any state. Democrats and Republicans hold roughly equal numbers here, with a slight Republican edge, and it is very diverse, both ethnically and politically.
This is what an example of a 35-district plan would look like in North Carolina, courtesy of Dave’s Redistricting App:
I took 2 primary measurements from each district as an indicator of partisanship in the state. The first is the Cook Partisan Voter Index (PVI), which takes the results of the last 2 presidential elections in each district, measures the lean compared to the national popular vote and averages them. The second is the 2016 Voter Preference Index (VPI), which adds up raw vote totals of the 2016 races for President, Governor, Senator and Attorney General. To view the exact district-by-district statistics, including individual race results, click here.
I decided to map two projections of what a potential North Carolina General Assembly would look like using this model if results reflected these two partisan metrics. If they reflected PVI, Republicans would lead the statewide vote 53.0-47.0, and if they reflected VPI, Republicans would lead 49.9-47.7.
Starting with the Senate, which would have 35 members:
In each of these scenarios, Republicans would hold a 20-15 advantage. Just 2 seats would change hands between these scenarios. The 2nd district is an Obama-Clinton district and was won by Democrats in the VPI, but as a result of narrow margins it still remains Republican leaning relative to the nation. President Barack Obama carried the ancestrally Democratic 21st district in 2012 by a large margin, enough to make the PVI lean Democratic overall, but swung very heavily to Republicans in 2016, the only Democrat to win the district that year being Attorney General Josh Stein.
On top of these districts, Democrats could also make gains in the 13th and 29th districts, both of which Governor Roy Cooper won in 2016, as well as the consistently close 5th and 19th districts. Alternatively, Republicans could sway suburban swing voters in the 8th, 10th and 27th districts, all three of which Mitt Romney won in 2012. Overall, there would be a large battleground which would vary as a result of the political environment, giving opportunities for both parties to make gains.
Next, the 245-member House of Representatives:
Now these maps are a little bit more interesting.
While the Senate map was shaded in dark blue and dark red only, you can see that the House map contains varying shades of blue and red, which represent the partisan compostion within each district. In order to measure the number of seats each party would win, I automatically gave four seats to any district a party was leading in, then gave a fifth seat if they were leading by over 25 points, and a sixth seat if they were leading by over 50. This is consistent with the 12.5% threshold I established in order to win a seat.
Overall Republicans would have a lead of 129-116 in the PVI map, and 126-119 in the VPI map. Democrats would have more seats in the VPI scenario as a result of gaining an extra seat in the 2nd, 12th, 17th and 25th districts, while losing one in the 21st, relative to the PVI scenario.
But the numbers get fascinating when you look locally. See that big chunk of blue around Raleigh, Durham and Chapel Hill? This area would elect 14-15 Republicans. The four districts in Mecklenburg County would elect 10. While these numbers fall short of the number of seats that Democrats would get in these areas, it’s still a large number of seats, and without them they would be hard pressed to find a path to a majority. Similarly, the 6 districts west of Mecklenburg, most of which are heavily Republican, would elect a combined total of 14 Democrats.
This effect is not because of some cryptic electoral nuance, but because this is how things operate in elections. Democrats need voters in urban areas in order to win, and Republicans need voters in rural areas. But if either party focuses solely on the regions they dominate in, and ignore the rest of the state, they’re in for a rough surprise, because that’s not how you win statewide. And this electoral system reflects that.
Competitivity
I’ve discussed the set of potential competitive districts in the Senate, and to put a number on it, there are 5 seats in both scenarios that are within a margin of 5 points. You can consider these to be “highly competitive” seats. You would expect these Senate seats to change hands frequently, and respond to partisan environments. So a majority would never be unbreakable, but it would never be too breakable (a small lead resulting in a large majority as a result of excess competitivity). Sounds ideal, right? But this system would put extra focus on the competitive districts, at the expense of others. Faux fair representation, if you like.
But it’s a completely different story in the House. While the 5 districts mentioned above would still be competitive, with either party safely holding three out of seven seats in each district and battling it out for a fourth, there would also be an expanded list of battleground seats elsewhere. Republicans in the majority-black 4th district may seek to deprive Democrats of their fifth seat there and take a third for themselves. Democrats may do the same in the Union County based 25th. Even in the 17th, which holds some of the reddest areas in the country, Democrats could utilize campaign resources to get the 25% required to expand their one seat delegation into two seats.
In total, 14 out of 35 districts contain seats which would be within the previously defined “highly competitive” threshold of flipping to the other party in the PVI simulation. In the VPI simulation this number goes up to 19. Expanding the threshold to a single digit (10 point) margin would classify almost all seats as competitive.
This number isn’t as high as it is in the Senate as a proportion of total seats (in fact it’s about half), but the fact that there are three times as many voters in competitive districts allows for a chamber which is truly more reflective of the general population, by allowing voters in each district to have a more balanced impact on the final results.
Demographic proportionality
As mentioned above, not only parties would be represented proportionally in a House of Representatives with STV. Theoretically any community which could gather 12.5% of the vote in a district could have their interests represented. College students in Orange County could elect a member. Conservative Democrats in Appalachia could have their say. The Lumbee tribe in Robeson County could easily elect two of their own to make laws. The opportunities are plentiful.
Let’s take a dive into how this would work, and look at race. This is a bit more difficult to quantify, but I used the D’Hondt method to allocate seats. In reality the electoral system may produce different results, but it’s worth looking into. Here’s a graphical representation:
By my calculations, the House of Representatives would be composed of 184 White Representatives dominating in most parts of the state, 53 Black Representatives concentrated in the eastern part of the state and in urban areas, 6 Hispanic Representatives scattered around the state with no more than one per district, and 2 Native Representatives both representing the aforementioned Lumbee tribe in the 21st district.
While the current system requires that majority-minority districts be drawn to allow minorities to have their representation, a proportional system would allow this to happen naturally.
Ideal bicameralism
It may be that a system of proportional representation is the best way forward. It may be that it will help make our elected lawmakers look more like us, rather than have legislative bodies producing gridlock and not allowing anything meaningful to get done. What I’m proposing is a simple compromise: one chamber with the usual winner-take-all elections, albeit with slightly different electoral rules, and another with a radical reform that just may drastically benefit the way our governments operate, and let all voters have their say.