9 ms·
Gerrymandering and a cure - shortest splitline algorithm
- georgemcbay 14y agoIf only the immense amount of political problems we have could be solved through simple, rational technical solutions! What a fantastic world we'd live in right now! A bit more on-topic is that I feel some issues like gerrymandering wouldn't be such a big problem if more root-level problems were addressed, like implementing instant run-off voting to help weaken the two-party system in the first place.
- ianb 14y agoI don't see how instant run-off voting would change much of anything, unless perhaps you got rid of the endorsement process at the same time so that people of the same party could more easily run against each other.
- weeksie 14y agoIt removes the fear of throwing away a vote on a third party. Look at the Australian government for example, there are a ton of third parties represented.
- JoshTriplett 14y agoI agree with you that voting systems represent a root-level problem (and in particular that first-past-the-post causes the two-party system, per Duverger's Law), but IRV is pretty much the only voting system worse than first-past-the-post. It allows people to vote for a third party, but completely ignores their preferences among other parties until eliminating that third party; thus, it does a good job of convincing people to vote third-party, but actively works against the preferences of people who do so.
- taejo 14y ago> completely ignores their preferences among other parties until eliminating that third party I'm afraid I don't follow. Eventually, the third party either wins or is eliminated, and then the preference between the two main parties is effected. Am I missing something?
- JoshTriplett 14y agoWith IRV, you express a ranked preference between candidates, but IRV ignores all preferences other than your first choice until your first choice gets eliminated. Simplest example: suppose you have three parties, A, B, and C. A and B represent the two primary parties, and C represents a third party, which tends to draw more voters from B than from A. C voters would prefer B over A, as the lesser of two evils. (Think either C=Green B=Democrat or C=Libertarian B=Republican.) With first-past-the-post, most C voters will vote B to make sure A doesn't win. IRV wants to make it reasonable to vote third-party by allowing C voters to vote CBA. However, this has at least two major problems. First, with votes like CBA, IRV ignores the preference for B over A until C gets eliminated. So, as long as the third-party C has fewer votes than B and A, the third-party votes get ignored as expected. But if the third-party actually gets enough votes, which IRV claims to encourage, then B will get eliminated first, followed by C, then A wins. So, C voters ended up hurting their preferred primary party by voting third party, exactly the situation IRV claims to avoid. Second, IRV ignores compromise candidates: if both A and B voters would prefer C over the opposing party, IRV will still eliminate C because not enough people had it as their first choice. Both of these problems occur because IRV ignores preferences beyond your first choice until your first choice gets eliminated. (That leaves aside all sorts of other crazy problems, such as monotonicity failures: sometimes, ranking a candidate higher can cause that candidate to lose when they'd otherwise win.) Take a look at some of Wikipedia's excellent articles on voting systems for more details on some of the criteria used to evaluate voting systems. IRV fails quite a few of the important ones.
- ccleve 14y agoNice sentiment, but the implementation is inept. The courts have determined that existing boundaries, including county and village boundaries, matter. Physical features matter. Existing "communities of interest" matter. Population equality matters. And, sadly, race matters a lot. The only sort-of mathematical awareness that the courts have concerns contiguity and compactness. Contiguity is easy. Compactness has a million definitions, and the courts haven't settled on a meaningful one. Come up with a compelling, mathematically-sound definition of compactness and you might have something. Also, the author ought get at least a basic familiarity with the literature on political map-drawing. If he had it, he would know that maps are drawn on census blocks and that houses are never "split".
- cduan 14y agoI wonder if this algorithm would still present biases given the fact that large cities tend to vote differently from less densely populated areas. My hypothesis is that this algorithm would favor city voters over rural voters, for the following reasons: 1. Successfully gerrymandering in favor of one party generally involves stuffing the other party's voters into one district, while splitting your party's voters into multiple districts. 2. The proposed algorithm favors straight-line divisions through regions such that about half the population of the region is on each side of the division. 3. Thus, chances are, the straight lines will run through the middles of cities, cutting cities into multiple districts. I'd be interested to see someone test this out--it's likely that my hypothesis is wrong, but it also seems pretty likely to me that there will be at least some urban/rural bias, one way or another.
- 3JPLW 14y agoMy favorite example is Illinois' 4th Congressional district[1]. But: there can be positive outcomes to such gerrymandering. Chicago is very strongly majority democratic. Splitting the city with straight lines would not change this. All of the congressional districts that include the city are already held by democratic incumbents. But this funny shaped district connects the two large latino neighborhoods: predominantly Puerto Rican in the North and Mexican in the South. And as such, Luis Gutiérrez has represented the district since 1993, when the district was defined as such. He was the first Latino to be elected to the House from the midwest. More information on the general 4th district wikipedia page[3]. 1. https://upload.wikimedia.org/wikipedia/commons/a/a0/IL04_109.gif https://upload.wikimedia.org/wikipedia/commons/a/a0/IL04_109... 2. https://en.wikipedia.org/wiki/Luis_Guti%C3%A9rrez https://en.wikipedia.org/wiki/Luis_Guti%C3%A9rrez 3. https://en.wikipedia.org/wiki/Illinois%27s_4th_congressional_district https://en.wikipedia.org/wiki/Illinois%27s_4th_congressional...
- ccleve 14y agoJust curious: who are you? I'm also a Chicago software developer political type. And I was involved in a lawsuit challenging the constitutionality of the 4th CD in the 90s.
- frozenport 14y agoIts not immediately clear disproportional representation is a positive outcome.
- mjn 14y agoIt's closer to proportional rather than disproportional representation in this case. Latinos make up about 30% of Chicago's population, but they're spread out so they wouldn't form a majority of any single compact district, so wouldn't be able, as a community (if indeed it's a coherent community), to send a representative to Washington to represent their views. However, if you connect demographically similar areas, then the community gets representation closer to their actual share of the population. Of course, we could just use proportional representation outright and achieve a similar result, or elect candidates from multi-member districts and a slate of candidates (e.g. put all of Chicago in one district, and elect the top 4 candidates). But that's a more radical reform that's less likely to happen.
- jacques_chester 14y agoIn Australia electoral boundaries set by the Electoral Commissions. Officers of those commissions are by law required to be politically neutral. For example, because I have been a member of a political party, I am ineligible to work for them. There's a modest amount of argy-bargy every time the boundaries are redrawn, but in general seats are contiguous and have shapes that follow the population distribution.
- ianb 14y agoIn the case of a lot of states with one major city, the algorithm basically just splits that main city, for instance New Mexico: http://rangevoting.org/Splitline2009/nm.png http://rangevoting.org/Splitline2009/nm.png – when the city is big enough to have a couple districts of its own it's not as bad, like: http://rangevoting.org/Splitline2009/mn.png http://rangevoting.org/Splitline2009/mn.png I'm kind of unimpressed. There are mechanisms to remove partisanship and political gerrymandering without making the whole thing into an algorithm.
- kenthorvath 14y agoWhat's wrong with algorithms?
- brownbat 14y agoSplitlines are very likely to cut through urban areas. Since the urban/rural distinction is meaningful in American politics, this has interesting consequences. I'm not sure "arbitrary" always gets you "unbiased," or "unbiased" always gets you to "fair." Fairness requires tallying up the numbers of disenfranchised in the new method, comparing it to the old. Splitline will exclude some urban voters from the political process for the foreseeable future, and some rural voters. It's not obvious which group loses more (it's highly contingent on the population distribution). But it's worth tallying up, just to see. If disenfranchised is really our concern, why not just try to minimize it more directly? Let's say someone is defined as "disenfranchised" if they identify with a party with no chance of victory, or if they contribute overflow votes that will never be necessary for their party to win the district. If you're in a minority party that gets less than 40% of the vote, you're disenfranchised. If your party has over 60% of the vote, then everyone beyond that 60% is disenfranchised. (We can quibble over the definition, or generalize it for more parties, but let's just assume we have some operational definition approaching "disenfranchised," without any definition, we couldn't evaluate the success/failure of splitline anyway.) Now just draw all the constitutional district maps, and pick the one that disenfranchise the fewest number of voters possible. Ok, that may be computationally infeasible. So instead, just have each party draw it's own recommended map, and then pick the one of those offered that disenfranchises fewer voters (using data from the last election, maybe). [1] A man-made definition is exactly what splitline recommends against, but even imperfect but vaguely reasonable definitions of "disenfranchisement" in this method will tend to be more "fair" (in terms of reducing disenfranchisement) than purely arbitrary dividing lines.
- eli 14y agoThis would have the practical effect of reducing minority representatives. Some people think this is OK, but it's worth at least mentioning.
- w1ntermute 14y agoThat could be easily solved by adding some proportionally elected representatives.
- eli 14y agoWouldn't that then result in less geographic diversity or representatives who don't represent just your district? Or am I misunderstanding?
- w1ntermute 14y agoThat would result in some of the representatives being elected on a national basis. So any voting bloc that makes up 5% of the population nationwide, but doesn't have a majority anywhere in the country, would still be able to get representation in Congress - they would have 5% of the proportionally elected seats. This would also be a big step towards eliminating the two-party situation we're currently stuck with and greatly increase political competitiveness. If you had 100 proportionally elected representatives, for example, any party with 1% national support could get one representative in the House. This would allow for 3rd parties to get a political foothold which they could then use (if they become popular enough) to eventually unseat the Democrats or Republicans in non-proportional seats as well.
- gizmo686 14y agoWhile we are at it, why not move to a delegative system. Every voter can directly vote for or against any given bill. They can also 'give' their vote to any other voter, who can vote on their behalf or pass the vote to yet another voter. At any point in time, you can change who has your vote, or overrule them on a given issue. (And the person who you gave your vote to can do the same if he delegated it to a third person, and so on down the line.) This has the advantage that if there is a bill that you care about, you can be sure that your vote goes the way you want. And otherwise, you can give your vote to the politician whose views most align with your own, and politicians have power directly proportional to their support.
- mikeash 14y agoGerrymandering is only a symptom, and cannot be cured without tackling the underlying disease. That disease is a political culture in which politicians will do whatever it takes to win, regardless of popular sentiment. No, I have no idea how to fix that.
- gizmo686 14y agoSet up the system so that the politician's best chance at winning comes from doing as good a job as possible. This means providing as few oppurtunities as possible to game the system.
- nathanstitt 14y agoThese type of articles annoy me. Not the article necessarily, but the thinking that goes into it. The solution presented is technically sound and would absolutely solve the problem of crazy gerrymandered districts. The possibility of something like this actually being implemented: zero. Does anyone seriously think that congress would allow their re-election prospects to be controlled by a software program? It sounds like they are attempting a ballot initiative route, but good luck explaining this to a skeptical public while facing negative campaigning that makes it sound like 'computerized death panels' all over again. I see the same thinking all the time - a new developer shows up all bright eyed and tries to talk everyone into switching to his favorite 'Framework/Language X' without even considering considering if it's at all feasible. I'm not saying that things can't be improved. But if you're proposing a change you have to be smart enough to consider it's chances and propose ways to overcome the inevitable pushback.
- graeme 14y agoNot to mention that it ignores the motives for Gerrymandering. Politicians surely know that Gerrymandered districts are 'bad', if a logical electoral system is your priority. Politicians Gerrmander because they see an advantage in it, and they don't have compunctions against it. I suspect good, non-Gerrymandered districts could easily be generated using traditional means as well. But the officials in charge have signalled they aren't interested.
- danso 14y agoThere is nothing easy about drawing districts...what fundamental rule is there that says it is better for my household to be grouped with the politically-polar-opposite area south of me rather than the more politically similar group east of me? Is it bad if each district contains demographically homogenous groups that vote as a bloc? How else to introduce diversity than by organizing districts differently than traditional neighborhood boundaries? Obviously, there are some who will attempt to game the system? But one candidate's gerrymander is another's fair apportionment...to dismiss it all as a big conspiracy controlled by indefeatable evil interests is to practice the worst kind of intellectual ambivalence.
- 14y ago
- danso 14y agoI'm sorry but this is the kind of "solution" that gives analytical thinking a bad name. Ignoring the political will to implement this, this solution's biggest problem is its willful ignorance of physical boundaries. A four lane highway can be as stark a dividing line between groups as miles of physical space. This article simply dismisses this reality by saying "well politicians never consistently considered those boundaries either, so there" Ironically, this ostensibly logical argument ends up going further into anti-intellectual territory than the politicians it derides. The thinkers behind this can't think of any other way that math might be used to accommodate or weight the effect of these different boundaries? Really? Or would the resulting boundaries be so aesthetically unappealing that its just better to go for the elegance of straight lines? That is among the stupidest rationales I've read in the whole gerrymandering debate. The whole post just smacks of incredible naïveté. No wonder engineers and scientists have a bad reputation in representative politics.
- gizmo686 14y agoPutting aside the problems of districted voting, is this naive attempt really that bad? Why is it a problem if I am in a different voting district than my next door neighboor, we both still go to vote, then have our votes evaluated by the system resulting in a winner. The goal is to make said system as fair as possible (while maintaing the concept of districts). I do not see why physical boundries are relevent at all. As cduan pointed out, this algorithm might have a bias favoring urban or rural voters, which is a problem. I have an even simpler algorithm that I think solves all problems of bias. Say you want a state to have N districts. Create N district labels (probably the numbers 1 through N). Now assign each voter in the state to a randomly selected district. For each district, take all but the largest contious regions and re-asign each voter within them. Repeat until all districts are continuous. Or we could move to a popular vote.
- mjn 14y agoAs far as the relevance of physical geography, I do see some benefits for the original idea of representatives actually representing an area in Washington. That's more likely to happen if boundaries are drawn along somewhat coherent lines that delineate cities or neighborhoods. On the other hand, those concerns are clearly not the main thing that goes into drawing today's gerrymandered boundaries anyway, so maybe losing that notion for geometrically drawn boundaries isn't a large price to pay.
- btilly 14y agoA random note for those who might be inclined to think that gerrymandering does not have a significant political impact. In the 2012 elections, Republicans carried Congress by 233 to 200 seats. Yet across all Congressional elections they lost the popular vote by about 1.35 million. (http://skeptics.stackexchange.com/questions/14315/did-the-democrats-win-the-popular-vote-in-the-2012-us-congressional-elections http://skeptics.stackexchange.com/questions/14315/did-the-de... is where I ran across this.) The cause of the discrepancy is largely gerrymandering. Unfortunately lots of states have laws about trying to draw boundaries that put ethnic groups together, etc in non-trivial ways. There are algorithms that can produce maps satisfying those restrictions, but they are more complicated. And, for obvious reasons, they often have the goal is producing such maps with a definite electoral bias...
- danso 14y agoSo is the 1.35 million number just the absolute difference between all votes cast nationally? Doesn't there have to be so e adjustment for districts in which many D voters elected an easy winner but in which excess votes don't really do much since it is an easy win? And of course, there's other explanations besides gerrymandering...it could be that the RNC and its financial allies spent more savvily in tight races and pulled through. The problem of money is of course its own discussion
- chimeracoder 14y ago> And of course, there's other explanations besides gerrymandering...it could be that the RNC and its financial allies spent more savvily in tight races and pulled through. Yes, this issue is incredibly complicated to analyze - for example, in the House, contentious bills like the ACA are oftentimes passed with ~220 votes (usually ~217 defines a majority). That doesn't mean that only 220 people would have been been willing to support the bill; just that the winning party didn't need to waste any resources whipping another ten votes in excess of what was needed to secure a victory. Just looking at the final vote tally is misleading, because if the threshold for passage were higher, it's very likely that the actual vote tallies would miraculously adjust accordingly.
- Dove 14y agoI'm not apologizing for gerrymandering in the least, but . . . http://catb.org/jargon/html/koans.html#id3141241 http://catb.org/jargon/html/koans.html#id3141241 . . . just because you didn't put bias into a system doesn't mean it isn't biased.
- alwaysdoit 14y agoDistricts should be deprecated altogether. Instead we should elect representatives via proportional representation. When we we had approximately 30,000 citizens per representative there was a chance at your representative actually knowing you, but at this point we are more likely to have our views represented by a third party.
- vidarh 14y agoNorway is below your 30,000 citizen per representative number, and frankly even then, while there certainly is a chance, not many people do know their representative even there. In Norway, some regional connection is arranged by grouping seats into multi-seatconstituencies by region. Then more proportional results are obtained by taking any votes "leftover" for a party after assigning whole seats and assigning seats from a non-geographic pool with some caveats. It satisfies those who want a regional link of sorts, and doesn't skew the proportionality too much.
- alwaysdoit 14y agoFor the US, I think that electing representatives by proportional representation within states would work similarly and would work pretty well with our existing constitutional framework.
- mjn 14y agoSome interesting mathematical background, on whether formal convexity measures can work as a gerrymandering metric: http://mathdl.maa.org/images/upload_library/22/Polya/Hodge2011.pdf http://mathdl.maa.org/images/upload_library/22/Polya/Hodge20...
- aneth4 14y agoLet me challenge some assumptions about gerrymandering. The most common misconception of gerrymandering is the assumption that close races are required for the voters to have a choice. While in current practice this may be true, the reason is due to other factors in our voting habits and electoral structure, not inherent in gerrymandering. The solution is not to get rid of gerrymandering, but to make other more fruitful changes to our electoral habits and systems. Gerrymandering is neither good nor bad. Gerrymandering works by making certain districts overwhelming in a single party's favor, leaving few competitive districts. While this may sound like it undermines voter choice, that is not the case. A district dominated by one party does in fact have a choice in their primaries. In fact voters there have a better choice than in competitive districts. Voters in a single party district get to focus on whether their representative precisely reflects their views rather than whether their party can win in the general election. Take for example a 90% Republican district. We know that this district will be representative of a swath of Republican voters - that's a good thing. Those voters can now focus on the more subtle leanings of their candidates during the primaries. Voters and primary candidates can openly debate about social conservatism and gun control, rather than what the small percentage of "swing" voters care about. Essentially, party factions become the choice, exactly what this country needs to evolve two parties that do not do a good job of representing the country. This is how I foresee democracy emerging in China, and is common from time to time, though not necessarily implemented to the voters advantage, in frequently defacto single party states, such as Mexico, Japan, and Canada. Pushing the impact of our representative system into more subtle debates in a primary election is only a bad thing because our primary system has some fixable issues. Fewer voters participate in primary elections, and there are not enough primary challengers. Making primaries more competitive and getting voters more involved in the primaries of the gerrymandered districts would help, and this leverages the positive aspects of gerrymandering rather than wasting effort trying to ban the practice. In some ways the issue should be self-correcting. The more guaranteed a district is for a given party, the more candidates should emerge in the primary, the more competitive the primary should be, and the more voters should care. This is because winning the primary is nearly as good as winning the general election. Perhaps a better solution to competitive primaries is something similar to what is being tried in California - non-partisan primaries. These shift the more subtle intraparty distinctions in single party districts to the more widely attended general election where they should be. In the above-mentioned Republican district, the district would likely have two Republican candidates in the general election. Such a system does both a good job of representing the constituents and of providing voter choice. The impact is similar to what one gets in a proportional representation or many-party system. The reverse of gerrymandering - engineering all highly competitive districts - also has major disadvantages, the largest being that every district has a choice between two distinct parties who could have radically different views. The winner of the election represents a slim majority, is chosen by a very small percentage of "swing voters," and small changes in "swing voter" sentiment can lead to major changes in representation. While that may have the advantage of throwing the bums out quickly, it can also lead candidates to appeal to their bases and push for voter turnout rather than having substantive distinctions. Sounds familiar. Conclusion? Stop being outraged about gerrymandering and practice some political Aikido. Push for active primaries and support legislation for non-partisan primaries. These actions don't require Supreme Court intervention or Constitutional amendments, are practical, and may have a better result than banning gerrymandering.
- drcube 14y agoWhy do we have to use geography at all? Can't we apportion representatives alphabetically, by last name? Or numerically, by SSN? I don't really see what geographical representation buys us, except gerrymandering and pork.
- freehunter 14y agoIt gives us neighborhood-specific representation (or at least should). This neighborhood may have different needs and wants than the neighborhood next to it, and both should be represented by a politician who understands those needs.
- 5teev 14y agoThis article reminded me of a Flash game I played with several years ago: http://www.redistrictinggame.org/ http://www.redistrictinggame.org/ If you work through the levels, you'll see that voting districts sometimes must have highly irregular shapes to ensure maximum representation of the various groups of constituents. You'll definitely see cases where "objective" boundaries leave some groups completely unrepresented. Gerrymandering definitely happens, and it's a big problem, but it's not a simple one.