Redistricting for Dummies: How to End the Gerrymander


English: Gerrymander diagram for four sample d...
English: Gerrymander diagram for four sample districts. Created in Adobe Illustrator by Jeremy Kemp. (Photo credit: Wikipedia)


Every ten years, based on the latest census data, the states receive new apportionments of seats in the US House of Representatives. The state legislatures begin mapping out revised districts to accommodate changes in population, population distribution, and increases or decreases in the number of seats.

And five years later, some state legislatures (at the moment, Florida, North Carolina and Virginia top the list) are still fighting over how to divide the spoils. Districts are re-drawn to protect powerful incumbents, give each major party at least token representation, and preserve the political power of labor lobbies, racial and ethnic communities, and other special interests. Each redistricting scheme ends up in court with multiple trips back to the drawing board.

This process is called “gerrymandering,” after Massachusetts governor Elbridge Gerry, who in 1812 signed a state senate redistricting map in which one district resembled a salamander.

It’s an ugly process. Everybody claims to hate it. But nobody seems interested in ending it, though it would be simple to do so.

Judge Robert Bork, later a failed nominee to the US Supreme Court, was once tasked with submitting a redistricting plan. He suggested starting in one corner of the state and drawing roughly square districts by population, without regard to special interest factors. His suggestion wasn’t accepted. But it would be easy to implement. Just plug the map and census data into a computer program and voila — uniform districts, fairly drawn.

Even better, why not transition to “at-large” elections for all US Representatives?

The district concept was implemented before the invention of the telegraph, at a time when most Americans got their news from a local paper and never strayed more than 50 miles from their birthplaces. Local elections made sense then. Today we cross the continent in hours and read worldwide news seconds after it happens (or watch it AS it happens).

Why not just have a statewide election for (for example) five seats, in which the five top vote-getters are elected? This would not eliminate sectional interests, pork barrel earmarks and other maladies of supposedly representative government entirely, but it would make members of Congress more accountable to large, mixed constituencies and less beholden to the insular coalitions controlling gerrymandered districts.

Switching to “at-large” elections might also mitigate the power of the two-party “duopoly” in favor of more proportional representation, especially if  better voting systems — approval voting, single transferable vote and instant run-off are three interesting ideas — were implemented as well.

Which is why it will ever happen. The American political system is brittle. Our politicians would rather break it than bend their will to ours. So maybe we should instead start thinking about what comes next.

Thomas L. Knapp is director and senior news analyst at the William Lloyd Garrison Center for Libertarian Advocacy Journalism ( He lives and works in north central Florida.




  • JdL

    But it would be easy to implement. Just plug the map and census data into a computer program and viola — uniform districts, fairly drawn.

    I’m a computer programmer, and I’ve got to say that those sentences could only be written by someone who is not. 🙂 A program of that sort, taking in not just a map to cut up, but a map with continuously varying population densities, would be quite a challenge to write, if we want it to do a good job. Possible, definitely. “Easy”, no.

    There’s something seriously fishy about the graphic at the top. If all four districts represent equal population in both maps (which is a prime requirement, is it not?), the purported outcome is impossible. Say there are 10,000 voters. The map on the left claims to divide them into four districts, each of which is balanced. So there must be 5,000 voters in each party. But the second map says there are three districts with 10% of Party 2, and one district with 90% of Party 2. That adds up to 3,000 members of Party 2 compared to 7,000 in Party 1.

    The real strategy of gerrymandering, I think, would be achieved by drawing three districts just slightly tilted to Party 1, and one overwhelmingly aligned to Party 2. For example, 3 districts with 1,500 Party 1 and 1,000 Party 2, and one with 500 Party 1 and 2,000 party 2. Now Party 2’s presence is “wasted” on just a single seat, while Party 1’s slight predominance carries them to victory in three seats.

    • JdL,

      It’s true that I am not, at present, a computer programmer. I was an amateur programmer in the early-to-mid-1980s; these days anything more complex than a little PHP scripting is beyond me.

      Nonetheless, the task is mathematical and can be expressed. And once it’s been expressed, it can be coded for. I didn’t say it would be SIMPLE (not complicated). I said it would be EASY (there are people who know how to do math, and people who know how to code; hire them, pay them, and it WILL get done).

      As far as the graphic is concerned, you’re right. I didn’t make that graphic myself. I had my super-duper “find related graphics” service (Zemanta) search for graphics related to “gerrymander” and picked the first one that was free for use and that scaled reasonably well to the space I needed it to fill (400 pixels wide to meet Flipboard standards, and not so tall as to be overly intrusive).

      • JdL

        I didn’t say it would be SIMPLE (not complicated). I said it would be EASY

        Hmmm, I see the potential for a semantic wrangle, but will avoid the temptation. 🙂

        Actually, such a programming task appeals to me even if I took it on just for fun. But here’s an interesting question: exactly what would the program try to optimize? I presume it would not consider past voter preferences, but only population distribution? Yet a case could be made that it would be more fair to consider past preferences (representing, say, rural vs. city interests) and to try to draw boundaries where the majority shifts from one to the other.

        What else? Intuitively, we think of trying to make each district tend toward round rather than elongated. But is that really a pressing consideration? (Aside: if it is, we should tend toward hexagons, not Bork’s squares).

        Certainly, going without rigidly-imposed districts makes sense, as you say. In effect it lets people choose their own “district” in a candidate who focuses on issue(s) they are passionate about. I also like letting people split their vote as they please among one or several candidates.

        • JdL,

          Well, “sectional interests,” as one of the founders who opposed political parties call them (I think it was Madison, but I don’t recall for sure), are part of the whole problem with gerrymandering.

          It’s not obvious that because you and I live next door to each other, or because we both live in the city or in the country, etc., that our interests will be identical or even close.

          But the gerrymander is specifically a result of catering to such “sectional interests.”

          Black civil rights leaders assert that black people don’t have “enough” representation relating solely to skin color, so districts are drawn like snakes with multiple heads to encompass black neighborhoods and elect black political representatives.

          Labor leaders assert that they need good union folks in the legislature, so working class districts get gerrymandered up while posh neighborhoods do too, so that one rep will be a union guy and the other rep will be a corporate executive.

          IF districts are going to be based on geographic contiguity, they should be drawn up without reference to whatever “sectional interests” happen to have a clout to get their own salamander-shaped fiefdoms.

          A better way would be to just start at some arbitrary or random point and have an algorithm draw up contiguous districts, each with a population of X.

          An even better way would be to elect “at-large” instead of in districts, or in some kind of “choose your own rep, more people choosing that rep gives him more voting power in the body” scheme, so that people who care about gun rights, labor issues, race, whatever can get together THEMSELVES to elect people who represent their ACTUAL priorities instead of being shunted into artificial voting ghettos that represent this or that ASSUMED priority.

          And of course the best way would be to abolish the state altogether 😀

          • JdL

            But the gerrymander is specifically a result of catering to such “sectional interests.”

            I’m not sure exactly what you mean by “catering to”. Gerrymandering is about diluting the voting power of certain sectional interests in favor of other(s), as in my example above.

            In any case, aren’t at-large elections actually catering to sectional interests? In effect they let each interest move to a corner of the state and draw a district around themselves. Which, I think you and I might agree, is not a bad thing.

            And of course the best way would be to abolish the state altogether 😀


  • bilejones

    It should be a balance between keeping smaller political units: Boroughs, Townships, villages whole and minimizing the length of the border.

    • Why should it be assumed that someone who lives just inside a village border has different representational interests at the state level than someone who lives just outside that border?