Miscalibrated Internet Receptor Stalks

I Solved Gerrymandering … sorta!

Okay, so not really. When you wake up tomorrow, gerrymandering districts will still be a thing, but there is something we can do! It won't be easy, it won't be quick … but we can do this by 2020! YAY!

This might be a bit long for some folks, so I've kept the titling somewhat instructive. Peruse at your discretion. The TL/DR is simply this: limit the magnitude of gerrymandering without getting rid of the process entirely. Skip down to see my proposal.

Needless to say, this is very U.S. focused and my proposal has to take into account the political landscape in the U.S. While I admire some the different voting schemes in other countries (proportional representation), I doubt that our country will change how we elect congressmen since we can't even convert to the metric system.


Gerrymandering is … what again?

So what is gerrymandering, you may ask? And I answer; gerrymandering is the act of defining a congressional district such that it maximizes a particular party's political power. By our constitution, every ten years state legislatures have the opportunity to redraw districts to matchup with new census data. New census data can change the number of districts a state gets as demographics in the United States shift decade to decade.

The word gerrymander comes from the Boston Gazette hating on 1800s Massachusetts Governor Gerry for creating a district it thought looked like a salamander.


I'm not sure how Salamanderish it looks, but hey, its 1812. The Gazette created this portmanteau because celebrity couple names have been popular forever … and I guess Governor Gerry had a salamander mistress or something. I don't want to judge.

The Governor changed the districts shape in hopes of solidifying the power of his Democratic-Republican party in the next election (ah, the good old days when Democrats and Republicans were so tight, they hyphenated). It worked. Though he lost his governorship, the Democratic-Republicans kept control of the State Senate.


How does it work? Well imagine you live in state with ten districts. You have a diverse populace with about 50/50 split party identification. You want your party to win more seats than the other guy, so what do you do? Well, if you're smart you put all of your opponents into one district and spread the rest out over the other districts. Suddenly you lose one seat 100 to 0 percent, but win all the other 56 to 44. Voilà, equal distribution of voter ID still results in you winning 9 seats to the other party's 1 seat.

Not clear … here's a video with cute cartoon animals that will explain it with fun pictures.

This process is how Illinois ends up with its 4th district. I mean … what the ever loving crap Illinois? That's literally going over voter rolls street by street. That district by the way … that's where all the Hispanics in Chicago live.


So what? Everyone does it. Welp.

Guys, Guys … Gerrymandering is Totally Bad

First let's all get outraged together. This is not a partisan issue. Both Democrats and Republicans do this whenever possible and it is a very bad thing. It makes it impossible to throw out congressmen no matter how much everyone hates them. And we hate them. It's the only thing we as a country agree on anymore, we really hate Congress, but gerrymandering means we probably love our own congressmen.


Gerrymandering takes a 13% approval of Congress and turns it into a 90% reelection rate for incumbents. Let me repeat. Despite only 13% of us being happy with Congress, congressmen won 90% of their races in 2012. In fact, we haven't had a reelection rate in the House lower than 80% since at the very least 1964 and likely since longer.

Look, the House of Representatives is supposed to be the Peoples' House. Our founders designed it to be the most direct way for us to influence policy. They made elections frequent for the very purpose of giving us the ability to constantly show the government our displeasure with it by throwing the bums out.


And this is perfectly legal. There was a short time when the Supreme Court tried to say, "for real guys don't do it anymore, it's not nice" (not a direct quote) when gerrymandering was done on racial grounds, so that's good? Oh, but wait, they never established a good standard for what constitutes racial gerrymandering and in fact, in Hunt v. Cromartie, the courts allowed a race based district because it wasn't purely racist, it also had "political implications" (direct quote). Well duh … since every single category of voter tends to favor one party over another (that's why they aren't all ties) any use of any category to gerrymander a district results in political implications. If the court's reasoning in that case is not circular, I don't understand wheels.

There's been a reversal of even this minor limitation. When the Court overturned part of the 1964 Civil Rights Act it removed the power of courts to review voter registration laws and gerrymandering plans in States that were historically really quite racist. Those limited protections … gone. (Correction: Thanks to xyzzy12345 for noting this nuance. "The supreme court did not overturn preclearance [requiring states to get approval of a district map from a federal judge]. They only overturned the rule that a history of racism in 1964 does not mean a state must preclear forever more. Any federal judge can put a state or county back on the pre-clear list and many places are still obliged to preclear.")


Technically, it may not be an absolute power (Davis v. Bandemer - 1986) but no map has been found too gerrymandered by the court system.

And words alone won't solve this. Heck, Florida passed a law against gerrymandering with a State amendment reading "No … district shall be drawn with the intent to favor … a political party." But what is favor anyway?


Okay. Well that. That's clearly gerrymandered. Sigh. Good job Florida's 5th district.


Is it Getting Worse? Why yes … yes it is.

See … gerrymandering requires predicting how people will vote. Something, I'm sure people were bad at doing in the 18th, 19th, and most of the 20th century. This led to the use of labels as horrible proxies for how people vote, like race and well …. race mostly. Then the 2000s saw an explosion in data to crunch. Suddenly you had soccer moms and NASCAR dads, and Nate Silver. Hell, I volunteered for a campaign in 2008 and we knew the phone numbers of the people that voted in the 2004 election as well as their addresses and even the likelihood of them voting again.


We live in a world of big data. A world where Target can tell your parents you're pregnant before you do (that totally happened). By 2020 parties will be able to use, not just census data, but all our social media and consumer information to draw the most exquisitely detailed gerrymandered maps ever. Ever complain about a politician on twitter? Donated to a party? Gone to graduate school? So get ready, cause the parties in power in 2020, will be in power … until 2030 at the least when they will be able to draw the districts again … and … okay so pretty much they'll be in power forever.

Okay, I Want to Solve It. How About a Bipartisan Council?

NO! Those that watched the video above all the way through know what's wrong with this idea. When bipartisan councils set up districts they set them up to maintain the status quo, not get an accurate representation of the people in the their districts. In fact, anytime you let people be in charge of drawing districts there will be some level of corruption. Even independent counsels of experts are not free from lobbying, bribery, and just plain old incompetence. We need to limit the amount of manipulation people can accomplish when drawing the district, that's why I think my proposal might work.


So Let's Implement Some Proportional Representation in this House … Like Parliament

Um … you know we live in a place I sometimes refer to as "We Hate Europe … and Communism." I mentioned this in the intro, but look … there is no freaking way that we are going to create a system by which our elected officials are elected in anything other than a one person, one vote, one representative type deal. Sorry. Wish it weren't so, but so it is. And in a harsh … Ron Paul will never be taken seriously because he's short and looks old kind of way; it's unfair, even cruel, but totally true.


Moreover, politicians will never, and I mean never support reform that changes the way we vote. These schemes tend to create many more than two parties and no matter what, the power of the parties as they are today is more than it would be if a permanent third, fourth, and fifth party became a reality.

There are some truly awesome ideas for changing how we vote. My favorite is the scoring method. It's like the Olympics; you rate the candidates on a scale from 0-10 based on how much you like them. The highest average wins. This is an awesome way to vote.


We could even try the Alternative Voting System where we rank candidates by who we like more. Then, if no person gets a majority of "first place votes", we eliminate the least popular candidate and add their voters' second place votes to the other candidates' totals (rinse repeat as necessary).

We even covered some of these ideas in io9 because San Francisco was running with the concept.


Ranked Choice Voting: Does a mathematical algorithm make for better elections?

These types of voting systems remove the Spoiler Effect, which if you need a primer is why Bill Clinton won in 1992 when Ross Perot was a thing. A majority of the U.S. was on the right, but the right got split by Perot. It's also why all Democrats want the Tea Party to start its own party, and why Tea Partiers, no matter how much they claim not to be Republicans, will never do that.


I love these ideas, but it's politically unfeasible for the country as a whole. Changing how we vote is harder than changing to the metric system. No matter how much it makes sense, it won't ever happen. 'Merica!

(Yes, I know I'm basically saying SF is not part of the U.S. for political purposes, am I really that wrong?)


Okay … That Was Harsh. So, Should we take Humans Out of the Equation Entirely?

HA! I get it. That's funny. No, but seriously, this is a cool alternative and there are a lot of options to choose from.


The Shortest Split Line Method which in simple terms is just to take an area and draw the shortest line possible to get an equal number of people in the allotted number of districts. If this sounds complicated, it is. Or at least it doesn't make intuitive sense, particularly when you get to drawing it. I'll let the cartoon animal people explain it.

The Minimized Distance Method which tries to minimize the aggregate distance of each constituent from the center of the district. So pressure toward circular districts.


The OPRA Method which makes the shortest perimeter between districts to create two districts of equal size. Seems complicated, but it's probably very pretty.

The Moment of Inertia Method which … sigh … okay here I go … draws districts such that populations are within 2% of the average district size and the sum of the squares of distance between each census tract weighted by population size and the districts centroid is minimized. I feel light-headed.


The Diminishing Halves Method recursively divides the state in half by a line, perpendicular to the statistical best-fit line describing the regions census tracts. Yup. That made sense to a lot of people that are not me.

These methods are attempts to use computer models to create the perfect district. There is a 50+ year history of suggesting computer models can create perfect districts without human interference … depending on your definition of perfect.


And this the problem with all of the hands-off solutions. They define a perfect district and then move on. It's almost always some level of compactness, equality of population, contiguity, and other metrics with various definitions depending on the person that runs the model. The last two methods took into consideration historical maps, some take into account trajectory of population increases and decrease in the past or expected over the next decade. Not to mention that some are biased toward rural districts and others toward urban districts with all the political bias that entails.

So … what if they're wrong? What if we don't always agree on what the perfect district is? Can we just change the algorithm later? And if so, how easy is it to change? How easy is it to manipulate the algorithm while no one is paying attention? Remember, like Net Neutrality, people fall asleep when thinking about gerrymandering because it's boring. If you're still reading this, you're just aces in my book.


Besides, are we really willing to let a machine draw our maps? To put the job in a black box and assume it knows best? The answer is probably no. Politicians already have a hard time handing over their redistricting powers to bipartisan committees, or worse, independent committees, but giving it over to a machine, whose algorithm is difficult to understand?

Well You're Pessimistic, What's Your Great Plan Then?

Okay. Are you ready? An equation that … WAIT hold on, don't leave me yet, I'm going somewhere with this. Look the problem with people is that they're corrupt. The problem with algorithms is that they're hard to understand and politicians won't support plans that drastically diminish their power.


So what about a compromise? Be as corrupt as you want, within the confines of not making us look stupid.

What's my definition of stupid? Buzzfeed makes a list about it and leaves off Illinois's 4th district (that image above) because there are that many to choose from.


So let them gerrymander I say, but don't let them cross the stupid line. What simple algorithm could be enacted to limit theses Machiavellians? Something easy to implement, devoid of judgment, and completely understandable? My proposal?


Simple IMHO.

The squared circumference of a district divided by its area cannot exceed 40.

Circumference^2 / Area <= 40.

Boom. Done. Just for reference we'll call that 40 the district's complexity. A perfect square district has a complexity of 16. A circle has a complexity of 12.566 or 4pi. This allows for some manipulation but not to the point of stupid manipulation. That 40 is a bit arbitrary and I'm willing to debate it in the comments section.


Here are a few simple drawings to help get the idea across.


So How Does this Play Out Right Now?

Good question. I ran the ratio against some raw data I found. Wyoming and North Dakota each having one congressional district in the state come away with complexity scores of about 16.9 and 20.0 respectively. The least complex district for a state with more than one district was Arizona's 5th with a complexity of 22.6. Good on ya Arizona! (They have a five person bipartisan commission set up.)


Other than that it's pretty much as expected. The average complexity of a district is about 78.6 with a standard deviation of 56. If that seems high, that's because it is. That picture of the Illinois 4th district … scored a 208.6 and it's not even close to the worst offender. Ladies and gents, the worst district known to the U.S. is … drum roll … North Carolina's 12th district with a whopping complexity of 479.5 a full 100 higher than the second worst.


It may not look as silly, but the high complexity comes from the incredible detail the map drawer went into to get just the right neighborhood mix. Some points are no longer than a highway lane. It's the district that got approved in the Hunt v. Cromartie case I mentioned above and was created to give African-Americans a representative. (Never mind that it also gave the opposing party simple majorities in a lot of other districts that otherwise would have been more competitive. Paternalistic crap theory got constitutional approval from Sandra Day O'Connor.) Again … both parties do this, it just happens to be much more obvious when you make a district with the clear intent to segregate an entire minority class.


The average district looks like California's 53rd district (Complexity = 78).


And a complexity of 40 gets you something like Utah's 2nd district.


I'm more than willing to share my data set if you guys want. It took some time to come up with it but it's based on the Mode blog analysis found here.

You Came Up With This? ... Really?

Well … yes in the sense that I thought of it one day without prompting, but no in the sense that I was not the first person ever to do so. So really, no, but I'm still proud of "discovering" how to create infinite perfect right triangles when I was a kid cause I have that kind of an ego. I've learned quite a bit about redistricting after I thought of this idea. There are a ton of great thinkers on this subject that have used this simple principle as a basis for a lot of analysis. In fact, this ratio is the crux of a lot of indexing for which districts are the most gerrymandered. This site here summarizes a number of these analyses. Here is another that is also fun and mathematical.


But, as far as I can tell, no one has proposed to use a ratio like this as a means to limit gerrymandering (save for their inclusion in the more complex "hands-off" redistricting ideas I mentioned above). And there are some valid reasons why not.


First, this ignores geographic boundaries. One of the reasons it's currently difficult for indexers to use ratios of area to circumference is that circumference data is hard to find. A lot of districts run along rivers and other natural landmarks that are nigh impossible to measure outside of fractal math.

Second, it ignores some of the qualities that people look for when trying to come up with a "fair" district.


As to the former point, I don't think it's an issue in practice. It would not be hard for people just to make straight or simple curving lines around rivers that set a quantifiable boundary, and map drawers would be incented to minimize the length of that line to lower the complexity number. Plus, the idea that water cannot be a part of a district is merely convention. It's not like if a lake or part of the pacific is included in the technically defined district suddenly previously unknown hyper-intelligent fish will demand the right to vote (though if this happens, I think that's a win for everyone). Besides, some districts already gerrymander the ocean.

The later point, I addressed above. Fairness is in the eye of the beholder. Part of the beauty and simplicity of this plan is that actual people can get together and debate what a fair district looks like … so long as it doesn't end up looking stupid.


Okay. So If I Buy Into This, What Can I Do?

Get to twitter ye activists! Okay, but seriously now, this will take some effort. First step is to spread the idea. We don't need a constitutional amendment, though passing one would be awesome, what we need is as many state legislatures as possible to get behind the idea before 2020. [Side-note: did you know that only the federal legislature is referred to as Congress? I didn't.]


The pros:

1) It's simple to understand.

2) It doesn't force politicians to give up all their power.

3) It's entirely objective.

4) It helps both parties.

5) It helps all Americans.

Send them this article with the drawings so they understand it. Look I'm not saying you should spam people, but totally spam politicians. They only move when they truly believe that their constituency cares about something, and they measure constituency support by the amount of mail, email, and voicemail they get.


State legislators are the best sources because they are the ones that have to pass legislation to this effect, but if you don't want to google them, here is a site to find your federal representative. They will put pressure on the state legislature for you if they get enough people that care.

But I'm not from the U.S., I Just like Reading a Lot

First off, Gold Star! If you read this far and aren't even a U.S. citizen kudos. You too have an incentive to care about our broken political system. For some reason our political débâcles directly affects the entire world; probably something to do with all our guns and butter.


Welp, thanks to the magic of the internet you too can contact our congressmen just as easily as we can. Thanks to the Citizens United case you can even donate anonymously in our elections, which for this purpose I'm happy about.

I'd suggest making a really effort in those States that are the most egregious.

1) Maryland

2) Hawaii - [Edit: Yes, I know this is a weird outlier but its high on the list from the indexers I linked to above and running the numbers ia pain]


3) Louisiana

4) North Carolina

5) West Virginia

Conclusions – TL/DR

This is a proposal to fix gerrymandering by acknowledging what is politically feasible. We may have a very difficult time completely removing people from the process (and we may not want to), but it is possible to limit the extent to which they can manipulate those maps based on the simple formula:


Circumference^2 / Area <= 40

I'd love to get feedback on this idea, but more than anything I'd love for us to make a difference in the way we elect our representatives. Right now, Congress only has a 13% approval rating but a 90% reelection rate. We've seen Congress stagnate and continue to stagnate for the last 10 years. Gerrymandering will only get worse as information about our voting preferences becomes easier and easier to crunch. We need to fix this issue and we can do it. What do you say?

Share This Story