How do we get rid of Gerrymandering?

What to do about Gerrymandering?A new follower of this blog from Texas wrote me recently asking what he could do to help get rid of Gerrymandering. It is a really good question- one that I wish I could answer better. But let me take a stab at it….

As frequently happens in politics, the broad consensus among thoughtful Americans that Gerrymandering is wrong is not matched with a consensus view of what would be right in its stead. I’ve made the case here that a simple “hands-off” algorithmic approach to defining geographic Congressional districts is the best way forward, but I also admit I’m sometimes reticent to make the argument too strenuously. Other approaches being promoted in different states would be vastly preferable to the partisan carve-up that happens now, and there’s no time and precious little political capital to spend on arguing over different replacements for partisan Gerrymandering.

Handicapping the alternatives, picking a horse

I think good policy with regard to banishing Gerrymandering comes down to pragmatics, which means answering these two questions:

  1. What are the alternative approaches that both improve upon partisan political redistricting AND are achievable?
  2. What political processes are underway in my state (referenda, citizen’s investigatory commissions, federal lawsuits underway, etc.) and which of these offer an opportunity to place one of the alternatives on the table?

From my research, it seems like there are exactly three alternatives to partisan redistricting that satisfy question #1:

hands-off redistrictingSome form of “hands-off” approach like OPRA or Brian Olson‘s algorithm. This is still geographic redistricting as called for in the Constitution, but doesn’t allow any meaningful human intervention.

non-partisan commissionCreating a statewide redistricting commission. This is still geographic redistricting as called for in the Constitution, but relies on the non-partisan nature of a commission rather than the partisan state legislature.

proportional representationElecting representatives proportionally per party slate, based on the parties’ popularity at the polls. There is no geographical component- representatives can come from anywhere in the state.


Granting that all three of these approaches are better than the current system most states use, it is still fair to ask which is the best. The basic choice in terms of broad outcomes is that #1 and #2 maintain the notion of neighbors voting together, and of assuring that regions within each the state have their own point of view represented. That seems valuable to me and was unmistakably part of the original constitutional plan.

On the other hand, #3 Proportional representation will tend to break the U.S. out of this two-party power condominium we live under today and invite more minority parties into the mix. I’ve lived under this in Europe and I think it works very well. I think it could also take a lot of the crudity out of our bi-polar, total war, take-it-or-leave-it brand of present-day American politics. When I am dissatisfied, I want to be able to vote for something other than against whomever I’m mad at. It would be great to see a ballot, for once, with meaningful options for Greens, Libertarians, Socialists or targeted protest parties on it, where I know my vote might take the minority part from say 3 to 4 seats (or to achieve a minority threshold, where it essentially goes from say 0 to 3 seats) rather than sending an impotent message that I’m mad that I know nobody is listening to.

The major problem with Proportional Representation, in my opinion, is a practical one. In most states there is some kind of opportunity to propose a different process for drawing districts, but that still results in geographical districts. If as frequently happens a judge has found the districts were drawn improperly and is looking for alternatives, friends of the court can suggest Hands-Off redistricting or Non-Partisan Commission redistricting as an obviously neutral alternative, or even present specific redistricting schemes generated by these methods for consideration. Similarly if there is a mechanism for a citizen’s referendum, these improved methods of geographical redistricting can be proposed. On the other hand, it is a much bigger step to propose an entirely new mechanism for how to represent the citizenry, and much easier to imagine a court ruling that even if Proportional Representation is arguably an excellent system, it is not the system called out for in the Constitution. The recent Arizona Redistricting case at the Supreme Court made it clear that states can select alternative methods of drawing districts without action by the legislature, but the idea that states can select an entirely new mode of representation is obviously a much tougher legal row to hoe.

Practicality also has a very important time component: a quick solution that can be felt starting in 2020 after the next census is vastly preferable to one that can’t arrive until after then, because for most states their districts will be set in stone immediately thereafter for the remainder of the decade.

So, finally, what should you do?

  1. Get “social” and play your part to keep the information flowing…
    • Follow this blog (instructions here) and let me know what is going on in your state and anything I can highlight or should write about.
    • Follow Gerrymandering – Let’s Get Serious on Facebook, an excellent source for news on the issue and a good place to discover what’s going on in your state.
    • Follow the Brennen Center for Justice, which tracks redistricting issues thoroughly.
    • Check out for a comprehensive view of how proportional representation might work.
    • Give me other suggestions, and encourage your friends to do the same! (If you have good websites/blogs/organizations to add to the list let me know.)
  2. Look for opportunities to insert Hands-Off, Non-Partisan Commissions, or Proportional Representation  into your state’s conversation. Look for citizen’s groups who are focused on improving things in your state.
  3. Be pragmatic! if you’re a fan of Non-Partisan Commissions but Proportional Representation has an opportunity to work, become a fan of Proportional Representation!
This entry was posted in Brennan Center, Brian Olson, Politics of Redistricting, Proportional Representation and tagged , , , , , . Bookmark the permalink.

8 Responses to How do we get rid of Gerrymandering?

  1. realrepresentation says:

    A couple of bones to pick.

    “Proportional Representation: Create state-wide slate by parties. There is no geographical component- representatives can come from anywhere in the state.”

    While that is one form of proportional representation, perhaps the European system where you lived, it is not what proposes. Except in the smallest states, the Fairvote proposal retains a geographic component. In Texas for example, Fairvote proposes combining five single-member districts into a five-member super-district. Instead of the existing 36 single member districts, Texas would have 8 multi-member districts.

    In addition, Fairvote proposes ranked choice voting which gives more emphasis to the candidate than his party. Voters rank candidates in order of preference, putting a “1” by their first choice a “2” by their second choice, and so on. Voters can rank as few or as many candidates as they wish, knowing that a lower choice will never count against the chances of a higher choice.

    “. . . even if Proportional Representation is arguably an excellent system, it is not the system called out for in the Constitution.”

    There is NOTHING in the Constitution which prevents the adoption of proportional representation by the states. In fact, while they didn’t employ proportional representation which had yet to be invented, most of the original 13 states had multi-member districts instead of single member districts.

    The current requirement of single member districts is by legislation occasioned by the concern that Southern states would enact winner-take-all at large multi-member districts in order to dilute the black vote.

    The Fairvote proposals are the best IMO.

    • noahkennedy says:

      I stand corrected, RealRep. I’ll add to my most recent post wrt suggested steps to take to get rid of Gerrymandering. (I had trouble getting through to the website when I first did my post.)

      I still think it will be much more difficult to get a proportional representation scheme in place than merely finding a better way to do strictly geographical redistricting. But as I mentioned in the post, I think it is much more important to be pragmatic. If there’s a way to get any of the three alternative approaches implemented anywhere where partisan legislatures are drawing district boundaries, I think it is smart for concerned citizens to get behind whatever might work. Just my opinion…

  2. realrepresentation says:

    Noah, do you ever give school presentations about gerrymandering? You could do a really great class.

    • noahkennedy says:

      that’s a high compliment, RealRep, thanks. I certainly wouldn’t mind, but I also have to wonder how well it would be received as a single speaker. Perhaps a muti-speaker format would come off as more fair.

      I could imagine an exercise that goes like this…

      1. Everyone name your favorite pop artist
      2. Oh, by the way, I will pay anyone who says their favorite pop artist is Merle Haggard $20.
      3. Now I will divide the class into 5 districts, and then we will vote on who the class’s favorite pop artist is. EVeryone who likes Merle Haggard please keep your hand up while I’m redistricting the class.
      4. Now the vote is 3:2 that Merle Haggard is this class’s favorite pop star. (Oh and You’re not getting the $20)

      Its a crude and imprecise metaphor but you get the picture.

  3. There IS an algorithm that does more than just geographic, AND allows for meaningful human intervention. So there is a 4th option. I am speaking of: I wrote it, so if you have any questions, ask away. I would love an interview. 😉

    The meaningful human intervention is: you can control how each factor is weighted, and how much the map is changed, and you can choose “communities of interest” to lock together.

    When you make a community of interest, if you have the fairness criteria packing, the program will try to neutralize the negative impact that that vote packing has on the community’s preferred political party.

    BTW it can also do multi-member districts (such as STV) like FairVote is pushing for.

  4. Oh and i’ve done an analysis showing that the informed approach (what my program does) is much better than the data-blind approach (what my program does with the fairness rules off, or what a purely compactness algorithm does) analysis here:

    if the compactness only approach is “the best way forward”, well then the approach i suggest, which drawfs the compactness approach, is at least as good as “the best way forward”.

    my software has been show to legislatives and political scientists and the like, and i’ve implemented every feature requests and resolved every criticism.

    with one exception, i suppose: some want it to be able to do racial vote packing under the pretense of the VRA. that’s not going to happen. I state my opinion about that pretty clearly right here: i hope that won’t prevent people from using it.

    • noahkennedy says:

      “HJ27”- thanks very much for this, very interesting work you’re doing there. Let’s talk more and I’ll give you a better write-up, but for now readers can click over to your site on their own.

      At first blush, I have a temptation to be a purist about the “blind” approach, largely because I think creating districts with an intention to serve “communities” is more profoundly wrong-headed than it may appear at first. Beyond the abstract philosophy about that, I think we are seeing that corralling “communities of interest” together creates a intra-district partisanship that is destructive. I say slug out your differences with fairly diverse districts, and therefore we’ll get (in aggregate) a higher number of more moderate representatives who can’t afford to pander to one POV or they’ll lose at the polls next time.

      The other thing I’d like to think about (and ask you about) are the genetic algorithm modes that equate success with voting patterns, which at a glance are red/blue. What a tool like this should never, ever do IMO is create any sort of balance between Republicans and Democrats, or mediate in any way along that axis. Dem/Rep binary politics, again IMO, is merely evidence of a very inefficient market for ideas. A system that enshrines Dems/Reps even more into the architecture of government for the purpose of “eliminating” gerrymandering is a bad move!

      BUT- this is very first-blush and at worst I think I’m quibbling about which sliders in the UI are good and which ones might be not-so-good. Thanks very much again- fantastic looking work!

      • You touch on a lot of the nuances there. I’ll try to answer each one briefly. (Can chat more over email.)

        Regarding the blind approach vs. the informed approach, I posted a brief empirical comparison here: The informed approach does much better, the short answer as to why is that demographics tend to cluster, and to varying degrees, consequently applying a compactness or minimize splits rule tends to favor some while disfavoring others. so you need to add fairness (equal voting power / competitiveness) rules to counteract the biases produced by the compactness rule.

        I agree that “communities of interest” is profoundly wrongheaded. It’s thinly veiled vote-packing and does a disservice to precisely those communities it pretends to serve. Only reason I have it in my program is because from talking to some politicians and polisci’s, it seemed not having it was a deal-breaker. I still have much reservation about it and long for the day when communities no longer elect to shoot themselves in the foot.

        And finally your red/blue comment – I get what you’re asking and the answer is no – it isn’t a “false balance” thing. it’s to make the outcome proportional to the vote. so it doesn’t go to 50-50, it goes to most closely match the popular vote.

        There’s a caveat with winner-take-all elections (the model most u.s. states use) – single winner elections cannot be made proportional, to do that you have to have multi-member proportional districts. Single winner elections give a sigmoidal (s-shaped) seats-votes curve. So for them you can’t enforce a diagonal seats-votes curve (which would be proportional). The best you can do is make it “symmetric”. That means that e.g. if a popular vote of 60% democrat resulted in 70% democratic candidates elected, then a popular vote of 60% republican would result in 70% republican candidates elected. Note that if one party’s sure voters are vote-packed a lot more than the others’, it won’t be symmetric like that, it’ll tend to give the less-packed party extra seats. That’s another way to think of the seats-votes curve: it shows the difference in vote-packing. The “partisan symmetry” slider enforces a symmetric seats-votes curve, that is it makes the amount of vote packing even between the two parties, while the competitiveness slider reduces the overall amount of vote packing, so they kind of work together.

        In multi-member proportional districts, you’d use the “proportional” slider instead of the “partisan symmetry” slider, because you’d have a diagonal seats-votes curve instead of a sigmoidal one.

Leave a Reply

Your email address will not be published. Required fields are marked *