Gerrymandering and a cure for it – the shortest splitline algorithm | Hacker News Reader

Gerrymandering and a cure for it – the shortest splitline algorithm

rangevoting.org

160 pointsclukic175 comments

Isn't gerrymandering a much more complex problem than that makes it out to be? I've always thought the problem was that it's actually difficult to even come up with a definition of what fair is. For instance, do you want each district to be as representative of the population as possible? Or do you want the house of representatives to be as representative of the population as possible? I suspect that you can't have both. For instance, if you have a state that is 60/40, then do you want each district to be 60/40? Or if it has ten representatives/districts, do you want 6 reps of one party and 4 of the other? What is districting even supposed to aim for? I don't think there is one reasonable answer to this question, which is why I think any attempt to answer this question for all states is doomed to fail.

The only way I can think of to get both is to increase the number of representatives. The House of Representatives was originally supposed to scale with the size of the population, and they eventually put a stop to that. If they had continued, then we'd have a house with a makeup that matches national polling much more closely than it does now, and gerrymandering wouldn't be as much of a problem. Isn't it true that as you add more representatives, gerrymandering distortions lessen?

> Isn't gerrymandering a much more complex problem than that makes it out to be?

No, its vastly simpler: the problem with gerrymandering is that there is no good way to draw single-member winner-take-all districts, and the solution is just don't do that. Multimember districts with a system that provides proportional representation within each district (e.g., five-member districts with STV) solve the problem pretty much completely.

Algorithmic line-drawing of single-member winner-take-all district pretty much guarantees poorly-representative districts, but without anyone being able to control how and where they are bad (well, except that there are quite a lot of algorithms that can be chosen, so the political choice of which to choose will be influenced by the effect the particular algorithm would have on the particular jurisdiction adopting it.)

> Algorithmic line-drawing of single-member winner-take-all district pretty much guarantees poorly-representative districts

A) The status quo is massive Gerrymandering, so something like the Shortest Splitline Algorithm would be a massive improvement.

B) The point is not to make quasi-proportional-representation with districts. The point is to make a legislature whose laws are representative. To that end, it does not matter if the constituents of a given district happen to have a lot in common.

No line drawing method on single member districts does that, and the best way to get laws which represents e people effectively is by assuring better proportionality. This had, in fact, been studied quite a bit, see, e.g., Arend Lijphart's Patterns of Democracy. Mucking around with variations on how our by whom lines are drawn for single-member, plurality or majority runoff districts and thinking that that had some how of meaningful improvement is self-delusion. It doesn't work, it can't work, we know why it can't work, and we know what does work, all from actual experience in real representative democracies.

Playing around with how district lives are drawn isn't democratic reform, it's democracy theater, a means of diverting attention without changing anything meaningful.

"Perfect enemy of the good" fallacy, if that's a thing.

Gerrymandering can add ~30% of representatives for the party controlling it, so changing it to a party method does improve things a lot, even if it is arguably very far from perfect.

> "Perfect enemy of the good" fallacy, if that's a thing.

Its only a thing when the "perfect" is not a real alternative.

Systems that produce more proportional representation than any means of line drawing in single-member winner-take-all districts are real alternatives that are well-established in the real world.

Well, proportional representation is mathematically and theoretically possible in the US, but culturally and politically as likely as adopting Chinese as official language. That won't change for decades at least.

But changing how districts are drawn actually does happen in this country. California degerrymanderified their process a few years ago.

Which optimization criteria and algorithm to use is also a political decision, and the people making that decision are the same as the ones responsible for massive gerrymandering. It's actually easier to defend perverse results[1] if they're the consequence of a redistricting from scratch by the "fair" algorithm you've chosen rather than arbitrarily switching parts of cities from one countryside district to another in a way which looks suspiciously like you've been paying attention to their demographics.

[1]like the Olson criterion, which when optimised by a computer program results in district boundaries apparently likely to return mostly Republicans in California...

Its worth noting that almost any party-blind algorithm for drawing district lives is going to overrepresent Republicans, because Democratic support is very concentrated in large cities, resulting in most party-blind algorithms creating few super majority Democratic districts and many bare majority Republican districts. (Few and many both being in relation to the proportion of each party in the populace.)

In the rural US, any district outside an urban area is NOT a 'bare majority republican district'; they're all large majority republican districts. Rural America is devoutly republican, outside recently formed border districts (formed since 1980) that are largely immigrant.

Most gerrymandering doesn't bias rural districts; it biases suburban districts that are marginal majorities (usu republican) only after redrawing the lines. Any automatic line redrawing algorithm will defeat this, since it takes a lot of perverse effort to draw boundaries that are sufficiently biased to serve Machiavellian pols.

The Mixed-Member Proportional system proposes to keep the advantages of FPTP and PR by electing most members FPTP and adding a few PR seats in the same body so that gerrymandering is ineffective.

You just lay out about 75-80% of your districts with traditional geographic community lines so each area gets a representative for its local interests responsible to local voters. Then voters vote for their local rep and also for a preferred party. You count the reps elected from districts by party and the remaining PR reps are allocated to fill in the discrepancy between that result and the party preference votes.

New Zealand, Mexico, and Germany use some variation of MMP for their legislatures, so it's working in prosperous stable first world countries already.

re Germany: "Working". It took over 60 years to fix the overhang seat allocation to not give big parties more power than they should have.

Is the US system better in that regard?

They don't even have more than two parties that matter, so… definitely no.

Agreed.

But even when the USA sees the light on proportional representation, there will continue to be winner takes all elections. Executive positions (president, governor, mayor), senators, judges, ballot issues, etc.

In those cases, approval voting is the correct answer. Easy to explain, easy to tabulate, most mathematically fair, hardest to game.

Per Duvenger's Law, approval voting would probably also help replace the current two party system with more choices.

On the other hand, you can't gerrymander Senate seats. (it's not representative now, but it's not changing either)

North and South Dakota are two states instead of one so that Republicans could have an extra two senators. Hawai'i and Alaska were admitted to the Union together so that neither party would gain an advantage in the Senate from the addition.

So yes, you can gerrymander Senate seats. It's just harder and less reliable than House seats.

If you're willing to change the definition of the word gerrymander, sure.

I wonder is this one reason some in the US Establishment are opposed to Peuto Rico being a state? Since it would probably be 2 Democrats?

There is no question about it. Same with the District of Columbia and various other places that should arguably already be states. They'd be likely to provide Democratic Senators, so there is a gigantic roadblock to keep this from happening.

Likewise, the promise of 2 Democratic senators is the main reason that Democrats are pushing for DC statehood rather than the more logical outcome, which would be retrocession of the city of Washington to Maryland.

It's more difficult to make an argument against Puerto Rican statehood, but DC is expressly not a state, and should either remain a federal district with no voting representation or be part of Maryland (as stated elsewhere).

I'm totally on board with starting over with new state boundaries.

Unlike the House, which represents the people, the Senate represents the states. That's why each state has 2 senators regardless of its size and population: I think the idea was to prevent oppression by large states.

The Senate representing the states was abolished a century ago when the Seventeenth Amendment was adopted. Today senators are chosen by popular election the same as congressmen, they just have wildly disproportionate electorates with some elected by half a million and some by tens of millions.

So splitline-partition the country.

The intent of the Senate was to preserve the independence and influence of the states. If you splitline-partition the country, you are eliminating the means of achieving this goal, and it would be simpler to abolish the Senate. It can be argued that the current system of directly electing Senators has already made the institution unrepresentative of the interests of the state governments, but your proposal would further alienate Senators from their home states.

Screw the states. Most of them only exist by historical coincidence anyway, rather than genuine cultural commonality.

I mean, hell, at this point, a lot of "state lines" are either way too far in (resulting in close-by cities within the same broad metro-area having no common government in the Northeast) or way too far out (resulting in close-by cities in the Bay Area having no common government that doesn't also serve rural Central California).

The point of drawing constituency lines is to unite people who have common interests while giving mutual autonomy to people with competing interests. The current states in the USA do none of that: they basically just systematize the shapes of former colonial land-holdings.

There is no constitutional provision that would ever allow abolishing the Senate. The Constitution of the USA prohibits any amendment that could change the disproportionality of the Senate.

An amendment can alter the text of any portion(s) of the constitution. While I agree that abolishing the Senate would be a radical change, I do not know of any restrictions which would bar it.[1] Even if this were not possible, a constitutional convention could be held to make the change.

[1] https://en.wikipedia.org/wiki/Article_Five_of_the_United_Sta...

Technically it says that "no state, without its consent, shall be deprived of its equal suffrage in the Senate" so the Senate could presumably be abolished by constitutional amendment if all states agreed. (I wonder about replacing the Senate with another legislative house and giving the existing Senate largely ceremonial responsibilities without formally abolishing it -- would that count as depriving the states of equal suffrage in the Senate?)

Presumably any amendment that would abolish or sufficiently alter the Senate would simply remove that article as well.

The phrase "the Constitution prohibits this amendment" doesn't make any sense because an amendment to the Constitution cannot by definition be unconstitutional.

https://en.wikipedia.org/wiki/Entrenched_clause

Perhaps that was meant to say, it would make the Constitution internally contradict itself?

It prohibits an amendment to eliminate the equal representation of States in the Senate, but it does not prohibit either of:

1. An amendment eliminating all actual powers of the Senate, either with our workout creating some new body to which those powers are transferred, or

2. An amendment eliminating the restriction on amendments eliminating the equal representation of States in the Senate.

I'm actually really intensely curious how one would design a voting system for people to choose what subentities to belong to. How would you populate the ballot? Would you run a clustering algorithm on the votes?

What if New Jersey and Michigan just decide they want to be as they are, but excluding Camden and Detroit respectively?

Or would you do something whimsical like draw lines based on football/baseball teams and cleave states like CT in half?

> But even when the USA sees the light on proportional representation, there will continue to be winner takes all elections. Executive positions (president, governor, mayor), senators, judges, ballot issues, etc.

Maybe, maybe not. Most of those don't fundamentally have to be single-winner elected things, e.g., executive positions don't have to be single-winner systems; multiseat executives do exist, after all -- even many US states have multiseat executives, but do those by single-winner election with seats allocated by portfolio (see, e.g., California's "Constitutional officers".)

> In those cases, approval voting is the correct answer.

Approval voting is, IMO, generally a poor choice for most elections (its good for elections in groups where disapproval or approval is tied to some substantive commitment -- e.g., if you are voting on where to go to for lunch, and "approve" on an option means "I will go if this option is chosen".) It doesn't make a lot of sense for most other purposes (because "approval" doesn't mean the same thing across different ballots -- this is actually true, though to a far lesser degree, of systems that require a forced ranking ballot, which is why I prefer ranked voting systems that can use unforced rankings) -- for ballot props that potentially might conflict, a system that merges features of majority and approval makes some sense (that is, it takes a majority to pass, but in the case of conflict the larger majority trumps the smaller one) and is, in fact, already used in some places (e.g., its what California already uses for ballot props.)

> Easy to explain, easy to tabulate, most mathematically fair, hardest to game.

IRV isn't much worse on any of those, and is if anything more familiar (and if you are using STV for multiseat elections, IRV is identical, so even though its a little harder to explain on its own, if you are using STV elsewhere, IRV for single-seat elections is pretty much free in terms of complexity, either of explanation or administration.)

IRV isn't much worse on any of those...

I was an active IRV proponent. Then I figured out they require computerized tabulation (for non trivial elections, such as my jurisdiction) and are infeasible to audit.

I kind of like approval voting but there's one big problem.

How to we enact it?

I also see a small problem.

Party representatives stand outside of polling places and present voters with lists of party endorsed candidates. People who currently vote solely on the basis of party will just approve only candidates who are on the party provided list.

The two major parties control virtually every legislative body in the country and the current system serves them best.

Knee-jerk answers like "Vote the bums out" isn't realistic. I'm thinking back to when the people of Washington DC voted back into office as their mayor, a man who was videotaped smoking crack cocaine. Even when it should be a simple matter, party politics trumps common sense.

> How to we enact it?

(And this applies to any electoral reform that there is resistance to by incumbents, not just approval voting, which I don't particular prefer.)

In states with citizen initiative processes that bypass incumbent office holders, by that method. Elsewhere, by building support in the electorate and then running candidates on a platform of doing that, and keeping organized in the electorate to hold those candidates accountable.

Low hanging fruit.

Join your local party and sneak it into your platform.

Use it whenever it makes sense. I used approval voting for feature planning and triaging bugs.

Our study group uses it to choose dinner.

Etc.

"But even when the USA sees the light on proportional representation"

"Sees the light"? Proportional representation is far from a panacea. It tends to lead to a fragmented government (which is usually bad). It also privileges party over individual candidates (which can be good or bad, but those who dislike partisanship would certainly disapprove). There are other factors at play as well.

Most (in)famously, the Weimar Republic used PR. That didn't end well.

> Proportional representation is far from a panacea.

Among modern, established democracies, proportionality of representation is pretty closely linked to public satisfaction with government, and public participation with government. Its as close to a panacea for effective democracy as you could have -- which isn't surprising, proportionality of representation is a pretty direct measure of the extent to which a government is democratic.

> It tends to lead to a fragmented government (which is usually bad).

Any government in a system with a diversity of viewpoints is either going to be minoritarian or coalition based; winner take all tends to majority-of-a-majority -- which can be minoritarian -- and, whether it is that or not, tends also to involve non-overt coalition building in structuring big-tent parties. PR systems tend to be more likely to be actually majoritarian, and tend to feature explicit -- and thus more accountable -- coalition building. It doesn't lead to "fragmented" government by any meaningful measures.

Itssometimes portrayed as leading to less stable government, because in parliamentary systems its associated with more rapid turnover of government administrations/cabinets, but by another measure, it leads to more stable government, as while there is more frequent turnover of cabinets, there is more continuity between successive governments/cabinets, and more continuity in governments/cabinets over time.

> It also privileges party over individual candidates (which can be good or bad, but those who dislike partisanship would certainly disapprove).

Certain methods of achieving proportionality (e.g., party list systems) may do so; candidate-centered election methods with rules that produce proportional results (like single transferrable vote in small multimember districts) do not. In fact, such systems would do the opposite compared to the status quo system in the US, emphasizing individual candidates over party by offering more competition, even in general elections, between individual candidates even in districts with a clear majority for a single party.

> Most (in)famously, the Weimar Republic used PR. That didn't end well.

Constitutional provisions for Presidential rule-by-decree had a lot to do with that, PR -- or even the particular form of PR used by the Weimar Republic -- well, if you want to argue that that was responsible for something, then go ahead and make that argument.

"Among modern, established democracies, proportionality of representation is pretty closely linked to public satisfaction with government"

When one of these "modern, established democracies" manages to maintain a stable government (or even survive) for as long as the United States has, perhaps that statement could be made.

https://en.wikipedia.org/wiki/Proportional_representation#Li...

Other than perhaps the Scandinavian countries, that's not exactly a list that makes one optimistic about stability.

Edit: "Constitutional provisions for Presidential rule-by-decree had a lot to do with that, PR -- or even the particular form of PR used by the Weimar Republic -- well, if you want to argue that that was responsible for something, then go ahead and make that argument."

See my response to the other poster below. The PR system (and the myriad of squabbling parties produced by it) led to there being no effective government for two subsequent parliamentary elections.

Of the top twenty most developed countries in the world (ranked by HDI), over half (13) have some form of proportional representation. They aren't all Scandinavian countries, though predictably many are. Which of them would you not consider a stable government?

    Y    Norway 
    Y    Australia 
    Y    Switzerland 
    Y    Netherlands 
         United States
    Y    Germany 
    Y    New Zealand 
         Canada 
         Singapore 
    Y    Denmark 
    Y    Ireland 
    Y    Sweden 
    Y    Iceland 
         United Kingdom
    Y    Hong Kong 
         South Korea 
         Japan 
    Y    Liechtenstein
    Y    Israel 
         France

Correlation isn't causation. The best conclusion you can draw from your list is that neither PR or FPTP inherently hampers development.

> Which of them would you not consider a stable government?

Whoa, that's some strawman. The standard we're looking to prove is "proportionality of representation is pretty closely linked to public satisfaction with government".

> Which of them would you not consider a stable government?

Whoa, that's some strawman.

No. The person who was being replied to specifically wrote about stability.

>When one of these "modern, established democracies" manages to maintain a stable government (or even survive) for as long as the United States has

"Has" up to what date? Is there some set length of time they have to surpass? Or are you suggesting the US can't learn anything from other countries until after it collapses and is outlived? Seems like a moot point by then.

> When one of these "modern, established democracies" manages to maintain a stable government (or even survive) for as long as the United States has

That's a lovely artificial requirement you've made there. I assume this means that you think the US system of government is inferior to the UK system of government, since longevity is apparently the only requirement?

And apparently having an extremely bloody secession war counts as 'stability'? Or being at war with your neighbours for most of the first half of your life? The US lost 50% more people fighting itself than when fighting the Axis in WWII a century later.

"That's a lovely artificial requirement you've made there."

Wanting a system of government with proven stability is not a "straw man".

"I assume this means that you think the US system of government is inferior to the UK system of government"

The UK doesn't have PR either. And no, it means nothing of the sort. My argument is that PR tends to be unstable, not that the US system is the only possible stable system.

No, the UK doesn't have PR, but it's a different system to the US. From what you're arguing, the US would be better off switching to their system. If you want to do it properly, you should probably import some Germanic hereditary figurehead - perhaps Schwarzenegger could start the US royal line?

Still, I like you completely sidestepped the question of how 'stability' matches up with having a civil war that was one of the bloodiest in history. You're not actually interested in finding the best form of government; you're only interested in defending your own.

"No, the UK doesn't have PR, but it's a different system to the US."

Nice attempt to shift the goalposts. Not going to work.

Stability isn't the only goal, but it's an important one.

"Still, I like you completely sidestepped the question of how 'stability' matches up with having a civil war that was one of the bloodiest in history.

Britain had a nasty civil war also.

"You're not actually interested in finding the best form of government; "

No, because I don't believe that there is a "best" form of government. Attempting to immanentize the eschaton usually (always) winds up with a few million people dead.

I'm interested in avoiding forms of government that are demonstrably bad, based on past performance.

But it breaks the personal link with the electorate with party placemen/insiders getting the top positions on the list.

Pr also gives to much power fringe party's eg Germany totally abandoning nukes to keep the 1 or 2 greens onside to prop up Angela's government.

What the Greens want is completely irrelevant to Merkel's government and has been for as long as she has been Bundeskanzlerin.

Abandoning nuclear energy is a widely accepted move by the majority of the population and across political parties. If anything the people wanting to keep nuclear energy, are in a minority. A minority that is not willing to risk actually advocating for nuclear energy.

You know how much the consumer price of electricity has gone up in Germany?

And I bet until last week the German "consumer" loved diesel powered cars (and the tax benefits) and ignored the down sides of targeting CO2 vs NOX

The parent's comment is accruate on the only important point here -- Germans don't like nuclear power.

The phase out of nuclear power isn't the result of a coalition trying to appease a radical fringe component. It's broadly reflective of public will in Germany.

http://www.world-nuclear.org/info/Country-Profiles/Countries...

The consumer price of electricity has gone up but that's a complex problem with many reasons behind it. Abandoning nuclear energy is not the only reason and probably one of the least important ones.

Even if it were a reason, reversing again on this topic and not abandoning nuclear power would have no impact on prices by itself. New nuclear plants would have to be built which would take decades, even if you could find locations where such a plant would be tolerated.

Breaking the link between the electorate and individual candidates/officeholders is true of party list methods which aren't the only way of achieving PR. STV in small multimember districts is just as candidate centered as FPTP methods (arguably moreso, in that there remains candidate choice within the same path preferences at the general elections.)

> proportionality of representation is pretty closely linked to public satisfaction with government, and public participation with government.

Is the causal relationship really that proportional representation drives political representation and satisfaction?

Democracies with a high level of political participation may be likelier to adopt experiments like proportional representation...and then public satisfaction is driven by the high level and effectiveness of political participation.

Italy and Israel are two countries that have had very proportional systems in recent time frames.

Having lived in Italy, I am not that impressed by the politics there, and think that proportional representation with party lists is part of the problem.

I think STV is pretty interesting, but producing one winner is no bad thing either. Of course, as per Arrow, nothing is perfect.

How on earth can it be that yours is the only reference to Arrow?

For those who don't know what that is:

https://en.wikipedia.org/wiki/Arrow%27s_impossibility_theore...

Having lived in Israel, its politics are often poisoned by those of the United States, but are otherwise much, much healthier than those of the United States.

It tends to lead to a fragmented government (which is usually bad).

Already the case.

It also privileges party over individual candidates

Distinction without a difference. Even if the focus is on the candidates, they're all firmly partisan with little variation.

Most (in)famously, the Weimar Republic used PR. That didn't end well.

http://tvtropes.org/pmwiki/pmwiki.php/Main/HitlerAteSugar

"http://tvtropes.org/pmwiki/pmwiki.php/Main/HitlerAteSugar"

No, that is not the point. Hitler eating sugar had nothing to do with his government. Proportional representation did. That's how he came to power. Look it up.

https://en.wikipedia.org/wiki/Adolf_Hitler#Appointment_as_ch...

"Hindenburg reluctantly agreed to appoint Hitler as chancellor after two further parliamentary elections—in July and November 1932—had not resulted in the formation of a majority government."

They had large number of splinter parties, all of whom hated each other's guts, and which were unable to form an effective government. The Nazis were not a majority (no one was), but were a large, well-disciplined bloc. The rest is, as they say, history.

If one reads about Weimar politics, one striking thing is how basically nobody wanted a democracy. The Communists didn't, the Nazis didn't; Von Papen, Chancellor second before Hitler, of the "Centre" Party tried to get the monarchy re-instated, and ruled Prussia effectively as a dictator. Ending democracy was basically the one thing a majority of Weimar politicians could agree on.

A plurality of German voters voted for the Nazis. If Weimar Germany had the US or UK system of government, that plurality of voters might have given the Nazis an outright majority of seats, just as the UK Tories were recently reelected with only a plurality of the popular vote.

What alternative electoral system would have kept the Nazis out of power in spite of their huge lead in vote share in three successive elections?

The Nazis weren't too concerned with legalities

I am not sure what to make of your first point, as it is not clear that we (HN) have a common understanding of what is meant by "fragmented". Is it a government which does not pass laws or spend money on specific programs which different groups oppose for varying reasons, or would that be a unified opposition? Would fragmented government pass a great many laws and spending programs with no unified goal, to appease disparate interest groups, or would this be a coalition?

>Distinction without a difference. Even if the focus is on the candidates, they're all firmly partisan with little variation.

Does it matter whether the politicians are partisans unified in party goals? It seems that in the USA, politicians frequently go against their own parties, relative to other countries. Is partisanship in favour of non-party aligned interest groups part of what you perceive as a problem? Please describe a non-partisan politician, party, or system.

Lastly, I think we can all agree that a single anecdote does not make the case for or against the PR system, though it does show that PR is not a sufficient condition for stable and everlasting democracy.

The rigid party discipline of the Republican party (in Congress) is starting to look like a temporary aberration. It was caused by a period of unusually strong party control over campaign finances. For a couple of decades there, Congressional leaders were able to strongly prevent Congressmen from being re-elected if they bucked the party.

Ironically, I suspect SuperPACs are contributing to the decay of Republican cohesion -- the party has lost its strong control over re-election. At this point the only thing Republicans in Congress can agree upon is that they dislike anything they suspect a Democrat -- especially the President -- of liking.

As for Democrats, Will Roger's famous quote is something like "I am not a member of any organized political party. I am a Democrat."

> Most (in)famously, the Weimar Republic used PR. That didn't end well.

Just look at the fixes the German Federal Republic put into place in 1949. That turns out to work quite well, allowing party plurality while still maintaining effective coalitions.

The European Union, Ireland and Scottish local elections elections use a PR voting system.

Germany and New Zealand use MMR in some of their elections.

> It tends to lead to a fragmented government (which is usually bad)

The stalemate of the past two terms in the US doesn't say much for winner-takes-all.

> It also privileges party over individual candidates

Only depending on the proportional rep type. But again, in the US, there are only two parties. Very rarely you might glimpse an independent, but they mean nothing in the overall political landscape. There is no functional way for another party to rise in the current political system in the US - so party is inherently privileged over individuals.

> Most (in)famously, the Weimar Republic used PR. That didn't end well.

Ah, you're just trolling.

Or more sophisticated analysis of the pros and cons of PR.

http://scorevoting.net/PropRep.html

"PR gave you Nazis!" isn't a sophisticated analysis.

Edit: chock full of nazis, apparently: https://en.wikipedia.org/wiki/Proportional_representation#Li...

"The stalemate of the past two terms in the US doesn't say much for winner-takes-all."

You think it's bad with two parties who can't agree on anything? Try twenty parties who can't agree on anything.

"Ah, you're just trolling."

No, I certainly am not trolling. The multitude of parties in the Weimar Republic, which could not form an effective government for two entire election cycles, were directly responsible for Hitler's rise to power.

> Try twenty parties who can't agree on anything.

What, multipolar politics, like modern Germany or Sweden or Australia? Wealthy, socially progressive countries with high standards of living?

You're just spreading FUD. PR was one of a great many factors involved in the rise of Hitler, but keep in mind that the US system also let a president get elected that didn't win the popular vote, who then went on to destroy a foreign country to drum up domestic political support. Literally hundreds of thousands of people died in Iraq to make Bush more popular in the US, despite Iraq having nothing to do with threats to the US. But it would be stupid to say that the US electoral system was directly responsible for the Iraqi bloodbath.

And as I say in the other comment, there are a shitload of countries using PR that aren't fascists.

"What, multipolar politics, like modern Germany"

Modern Germany does not have PR. For good reason.

" PR was one of a great many factors involved in the rise of Hitler, but keep in mind that the US system also let a president get elected that didn't win the popular vote"

The U.S. President has never been elected by popular vote. Again, for good reason. Feature, not bug.

Modern Germany hasPR via a mixed member proportional (MMP) system -- which, IIRC, was a German innovation when introduced; it doesn't have straight party-list PR, but then the more proportional system for the US proposed upthread isn't party-list PR, either. Proportionality is a (really, continuous valued, not binary) outcome, there are a number of systems for achieving it to a greater or lesser extent.

I see no appeal in proportional representation. The idea that I can only vote for a party, and that they get to choose which people actually hold those seats... it is repulsive and sickening.

Any belief you have that the present system works any differently is a fiction.

You aren't describing PR, you are describing party list systems, which are one method of achieving PR. Single Transferrable Vote in multimember districts also achieved proportionality of representation, but involves directly voting for candidates. MMP mixes elements of FPTP and party list proportional. And there are a bunch of other mechanisms for achieving PR.

I don't think the point is to make the result fair, I think the point is to make the process fair. The US ideal has always been about equality opportunity rather than equal outcome. By agreeing on an algorithm for drawing lines, you remove any bias from the process, which is probably as good as we're going to get.

I do think the algorithm presented here is too simplistic. It's likely to create situations where people have to travel long distances to their polling places and, if anything, we should be optimizing for voter turn out. It could, perhaps, work if cities were never split between districts, but even then people who live just over the border may have to travel quite a ways to vote.

I wonder whether we're arguing over the best bad idea rather than something simpler which would yield a better result. Is there really a benefit to geographically grouping voters? I'd argue that you're likely to get better representation from grouping voters by age range rather than locality. We've consistently seen policies that benefit older demographics over younger demographics...perhaps if there was a candidate elected only by 20-somethings, that demographic would actually be represented for once.

With a quick visual survey of their results, it seems this is a particular weakness of this algorithm. It appears to me that it's tending to split large cities into chunks and tie each chunk in with a huge outlying area.

Not only does that inconvenience polling places, but I think it does impose some socio-political outcome. I'm not sure whether it's good or bad, but the decision to not have cities voting as a block, and instead to include slices of them with their suburbs and outlying rural areas, certainly impacts the way representation will be determined.

True. Rural residents will often have completely different issues than city dwellers. To include them in a slice of the city pie, is probably going to entirely disenfranchise them. Depends on the numbers.

Why is voter turn-out the relevant metric? You are implicitly assuming that increasing voter turn-out is a worthy goal in and of itself, and I am not sure why. Is it important to you that each voter feel involved, or that they are responsible for their representatives, or do you believe that an increase in the number of voters would improve policies and outcomes? I would support an attempt to increase voter turn-out if it would improve policies and outcomes, but there is little evidence of this; in fact, the current evidence shows that current voters are better informed and more intelligent than eligible non-voters.[1] Before you accuse me of being an anti-democratic autocrat or something of that sort, please keep in mind that I am not advocating the abolishment of elections, that I used to agree with you (but was persuaded to change my mind), and have voted at every election I am eligible for since reaching the age of majority.

[1] https://en.wikipedia.org/wiki/The_Myth_of_the_Rational_Voter

The ethical case for democracy rests not on whether democracy leads to optimal policies and outcomes (which is ridiculous), but on the degree to which it includes individuals in the politics of the state exercise of power over them. The state necessarily has power over individuals, and the ethical basis for the use of that power declines to the extent that the state excludes people from participation in political decisions. It erodes the consent of the governed. The core injustice of things like poll taxes and grandfather clauses is not that they led directly to suboptimal policies, but that they disenfranchised black voters, which is what led to policies that the white supremacists considered optimal.

Voter turn-out indicates that information about platforms is passing from politicians to voters, and information about preferences is passing from voters to politicians. Low voter turn-out has tended to co-occur with the managerial turn of politics, in which no party or candidate offers any substantial choice to even the most enthusiastically civic subset of the population.

Essentially, low voter turnout is a correlated symptom of de-facto oligarchy.

I'm not sure why you have the assumption of one polling place per district, or even the assumption of having polling places at all. Washington and Oregon have both transitioned to 100% vote by mail.

Not really. The congress works the way it does because it's designed to represent a union of nominally sovereign states. Not matching national polling is a feature -- otherwise, the interests of NY, Boston, Chicago, LA, Atlanta would completely and utterly steamroll everything. The nation doesn't work if nobody gives a shit about Kentucky.

Gerrymandering is a different problem made worse by the Supreme Court's decision that every elected body with the exception of the Senate requires proportional representation.

Since districts are drawn by state legislatures, and state legislatures are only driven by districts cut up by population, there's no incentive do to anything but preserve/expand the numbers for your party. In other words, you don't get the "adult supervision" role that the US Senate delivers (for better or for worse), and get the quality of governance that you see in the US House -- that is drivel.

> Isn't gerrymandering a much more complex problem than that makes it out to be? I've always thought the problem was that it's actually difficult to even come up with a definition of what fair is.

That's perhaps true, and I observe what you say to be correct, but the lack of an acknowledged fair solution means that the existing solutions are all (party) political.

When Party A is in charge, they gerrymander in one direction, and when Party B is in charge, they then gerrymander in exactly the other (or perhaps a convenient tangent). The purpose of gerrymandering is to unfairly extend the period that you're in power.

Having an ostensibly party-neutral solution that all parties can agree to is a useful solution.

It's not immediately clear that this solution will really be fair, but it's not immediately clear that it's unfair, and at least it removes the politicking over boundaries.

If even a few regions start using this solution with the recognition that it is party-neutral, then parties in power elsewhere will start to be pressured into doing the same, to avoid appearing undemocratic.

It's a one-step improvement on what happens now.

This is very true--trying to create a single outcomes-based ideal of what's fair is both difficult and subject to the same strategizing that caused gerrymandering in the first place. Instead of choosing what "fair" is and maximizing it, the proposal takes a different approach: by using an extremely simple geometric rule, the algorithm seeks to avoid choices that would be susceptible to influence for political gain. Those are the choices that tend to result in gerrymandering, which by locking down power is expressly designed to be the opposite of fair to future voters. So instead of the difficult task of directly maximizing fairness, the proposal aims to minimize a great source of unfairness. I'd be curious to know how well it would work in practice.

That site already acknowledges that there is enormous complexity in the issue.

http://scorevoting.net/TheorDistrict.html

You don't so much need an algorithm that's fair, but one that's seen as unbiased.

Gerrymandering is simply incumbency protection.

Fairness is simply maximal competitiveness.

It's not clear to me that maximum competitiveness is "fair". Suppose you do in fact have a state that is perfectly divided into a red district and a blue district, but instead of drawing a representative from each side, you split it on the orthogonal side to create maximum competitiveness. Then, by essentially random chance, you might end up with two blue reps or two red reps when fair representation would probably be one red and one blue.

>Then, by essentially random chance

Or instead, red and blue will actually start listening to their constituents because if they do then they could get two and if they don't they could get zero.

This is a fair criticism. I agree that they would be more sensitive to constituent concerns, but isn't it still perverse if half the population has no voice in government because the other side was tactically better? Also, it is still possible to discipline an egregiously bad rep even though they're technically in a "safe" district, so it's not like you've totally lost control.

If you did that, you would tend to elect purple candidates, rather than the bright red and sky-blue ones which gerrymandered districts tend to produce: see what caused Boehner to resign.

I really don't know enough to have an opinion on this matter, but I also don't see the substance of your argument. If you flip a fair coin twice, you might get two heads. That doesn't mean the coin isn't fair; fairness is a property gleaned of repeating the trial many times.

So over many elections, the aggregate result would be fair.

That's true, but every time an individual result is unfair... it's still unfair. If I live on the red side and there are two blue reps, I have literally no representation despite comprising 50% of the population. The fact that I sometimes get to sock it to the blue side with two red reps is unfair to them. This isn't a case of turnabout is fair play, as the promise of representative government is that everyone gets a voice and decisions are generally made by consensus, compromise, and occasionally a smack down vote.

Let me put it another way, which system would you prefer to live in?

It sounds like you are confusing "everyone gets a vote" with "everyone gets a result." If you're the only red person in a room of blues, there's no way it's going to be fair unless you can convince others to turn, or you run away. A voting system is not meant to make everything possible, but to make it so that the one red in the room has an out: get more votes. (Because without something, you've got a very volatile situation.) You have a voice, it's just not codified in your vote unless you're a near majority.

The point of voting isn't to get fair representation. You can't have fair representation in this sense due to the Pigeon-Hole principle. You can have it in a long term, systemic sense.

    every time an individual result is unfair... it's still unfair

Fairness is not a property of individual results. It's a property of a statistical system incorporating many data points.

On what timescale? If district I'm talking about went 100 years red then 100 years blue but retained the same characteristics, that would be aggregate fairness, but clearly unfair. In your case of one red person in a sea of blue, it's obvious that they would be at a disadvantage. The situation I'm specifically talking about is one where red and blue are roughly equal. Does it make sense to randomly remove one side's voice?

Define "maximal competitiveness" in a 60-40 state with ten districts. Ten 60-40 districts? Eight 51-49 districts, and two 96-4 districts? Either could be argued as incumbency protection.

Flip a coin.

Duverger's Law suggests that a two party system is the logical result of winner takes all elections, because aspirant will attempt to win with the smallest coalition.

In other words, fair redistricting will reduce polarization.

The goal isn't pretty lines.

The goal is maximal competitiveness. Which boosts voter participation. Which gives legitimacy to the results.

https://en.wikipedia.org/wiki/California_Citizens_Redistrict...

"While the long-term results will bear out over time, independent studies by the Public Policy Institute of California, the National Journal, and Ballotpedia have shown that California now has some of the most competitive districts in the nation, creating opportunities for new elected officials.[12][13][14]"

Independent redistricting commissions don't work.

http://scorevoting.net/GerryComm.html

Apparently they don't work in the US.

That link says that they work in Canada, and they work in Australia too.

Without an underlying reason identified, this conclusion is somewhat unsatisfying.

The link also gives unsatisfying reasons for why it says they don't work, e.g. incumbents keep winning elections.

I haven't studied the matter really closely, but from what little I do know of the issue, it makes me wonder--are Arizona, Washington, and NJ examples of the hyper-gerrymandered states that citizen redistricting is supposed to prevent?

Could it be that, in those states, they're already starting out from a pretty equitable district layout, such that when the citizenry gets to redraw the layout, it doesn't actually matter much?

Or, put another way, how would a citizen redistricting panel work in Florida or North Carolina or Texas?

Gerrymandering gets attention because it is malicious misrepresentation, but unintentional misrepresentation is just as bad. Voting is a flawed system that will always give some voters more voice than others.

- The people who voted for the loser will not be represented (one of the reasons minorities are underrepresented).

- The need of local representation is illusory. Make a list of all the political issues you care about. How many of them are local?

- The effect of a bi-cameral legislature means that in the Senate, voters in Wyoming have over 60x the voting power of voters in California.

- Voting itself is irrational under a cost/benefit analysis. So by definition, the least rational citizens are picking candidates.

- Name recognition is huge for elections, meaning money will always effect elections.

- Candidates suffer self-selection bias (as well as others). To run, you need to be good at public speaking (fear of public speaking is one of the most common fear in America), look a certain way (no tattoos or piercings allowed), and have a lust for power.

Representation is a math problem. The only solution is random sampling. Replace Congress with 1 house made up of 1,000 randomly sampled citizens and we will have true representation.

> Gerrymandering gets attention because it is malicious misrepresentation, but unintentional misrepresentation is just as bad.

From an ethics point of view, the intent makes gerrymandering worse than unintentional misrepresentation. If you are not moved by ethical considerations, there is still the purely practical matter that the intent makes it much harder to fix.

You'd get a good sample, but a poor representative. If a group of people is attempting to choose a person to advocate their interests, they'll choose the person best at advocacy, who is likely atypical in a variety of ways.

Because typically people are bad at public speaking, rhetoric, bargaining, and other means of advocating their interests.

Not everyone has to address Congress. As long as each valid viewpoint has at least one advocate, I trust people could figure out the best option. Less bargaining means fewer riders and pork, which is a good thing. I'm fine if only the best ~10% of Congress wrote/debated laws and the rest simply voted.

Also, remember we are comparing against the current Congress, which only represents the wealthiest among us. Pretty low bar.

Gerrymandering applies as much to local school, county, and state districts as it does federal districts. At the school and county and state level, all issues are local. In fact, most legislation that has a large tangible effect on your kids, your home, and your job occurs locally, not federally.

The Federal government makes decisions for the entire nation. State-specific issues should not make their way into Congress.

State issues are likewise at the state level. There is no reason to make sure two counties are represented, since they won't get separate laws/taxes.

All that matters is that the people being affected by the laws are equally represented. That cannot happen with districting.

This is an elegant solution with a lot of good properties: Equal population balancing, politically unbiased, reasonably understandable by the layman, typically geographically compact districts. But where's the discussion of the desirable and undesirable properties in a districting algorithm and how this algorithm achieves or doesn't achieve these properties?

specialist already mentioned maximal competetiveness, ROFISH mentioned clustering due to natural features. Is regional or ethnic representation important, or is it an undesirable bias? How about travel distance (either for voters or campaigners)? I'm sure you guys can think of others you'd like to see or to avoid. Feel free to suggest some, I'm curious.

There is a little more information at http://rangevoting.org/SplitLR.html but did I miss the real weighing of the tradeoffs and advocacy for this particular method above all others?

Edit: Found a discussion of the theoretical issues at http://rangevoting.org/TheorDistrict.html which touches on some, but not all of the points raised.

This is no more a cure for gerrymandering than Condorcet voting is a cure for the various ills of the two-party system. Until you have a plan to implement this algorithm in the current political system, you're just solving sudokus and crosswords.

The problem with gerrymandering isn't that people aren't capable of drawing simple lines to split states into districts.

Oh, but this algorithms removes all the choice. So there's nothing up anyone's sleeves.

It just moves the problem.

Instead of arguing about the borders, you have arguments about which clustering algorithm to use, and what the weights to the inputs should be.

A simplistic algorithm like shortest splitline will give impractical results in many cases. For sure you're going to put Eagleton and Pawnee in the same district.

Unless you have a mathematical definition of districts that lead to good government, a purely mathematical solution won't be possible.

> Instead of arguing about the borders, you have arguments about which clustering algorithm to use, and what the weights to the inputs should be.

Yes, that's true. But with any luck the weights will have less knobs to turn, ie less information, than drawing lines.

> Unless you have a mathematical definition of districts that lead to good government, a purely mathematical solution won't be possible.

We just need to avoid the worst government.

Or perhaps the Americans can just look around the world to see how other countries are drawing their voting districts, and compare what works and what doesn't. (Shouldn't federalisms make that kind of experimenting work inside the US alone, too?)

In any case, there's a more interesting simple technocratic fix to try: count ballot papers that tick more than one candidate as a vote for each of the ticked candidates.

This way you can vote for the lesser evil and who you actually want.

(See https://en.wikipedia.org/wiki/Approval_voting)

Here's a better algorithmic non-gerrymandered districting, solving for compactness: http://bdistricting.com/2010/

Gerrymandering is an intentional act carried out by politicians, not one of happenstance. It's not something that can be cured with algorithms.

I take it the proposal is that there should be a law requiring that such-and-such an algorithm be used. That would take away the politicians' power to gerrymander.

(Of course it would also need to be voted for by the same politicians who presumably won by gerrymandering. Good luck with that.)

The Redistricting Game[1] is an interesting approach to teaching about how gerrymandering works.

[1]: http://redistrictinggame.org/game/launchgame.php

I'm concerned with the stability of the district borders. Because of the way this algorithm seems to like to slice through cities, it seems like minor changes in neighborhood density of a city is likely to significantly change the angle of the bisecting line, thus shifting large areas of the outlying region from one district to another.

This is bad for two reasons.

First, it seems a big weakness of democracy is voter ignorance. Throwing people repeatedly from one district to another, so they don't have time to learn the issues relevant to their district and the record of their representatives, will exacerbate the problem.

Second, the system can be gamed. It looks to me like approving or denying the construction of a large apartment complex near the center of a city can be used as a tool to push lines one direction or another. So approving that big apartment building in the city will increase population in that region, tending to tighten the angle made around it, thus freeing some voters from that district and pushing them into a neighboring one. Indeed, since the algorithm is recursive, this could have big follow-on effects subsequent iterations.

One problem with this: it makes sense that district lines could be along natural (rivers, lakes, mountains) or political (towns, county) lines. Trusting a simple line algorithm may bisect a town down the middle.

Yep. Consider Eastern California: the Sierra Nevada runs north to south creating a natural barrier. There's a small sequence of cities east of the barrier. During the Spring and Fall campaign season, it can be literally impossible to cross that barrier: you must go a long way around or fly.

The suggested algorithm would unite cities on both sides of the divide. It would be very difficult for an underfunded candidate to campaign on both sides.

The problem with taking into account natural or political lines is that there are so many to choose from.

You don't want to leave anything to choice. That's where gerrymandering comes in.

A while back, I was friends with someone who was a forensic accountant and former state senator. She talked about gerrymandering quite a bit. She'd say "you don't choose your representatives, they choose you."

She said both parties cooperated in drawing safe seats for everybody, and claimed that most of the people in our state congress were corrupt. Anyone who didn't play along with the graft would find their district redrawn out from under them, if the timing made that possible.

This actually happened to another friend who was a former state house member. The two of them were instrumental in revealing the corrupt activities of state House Speaker Jim Black, who ended up serving time in prison.

I imagine Jim Black was a sacrifice as well - he didn't play along with somebody, so he got tossed to the wolves. And the public thinks "Well, that's been dealt with" and the rest of the corrupt remain safe from further scrutiny. For a while.

I think this is fascinating even if it will never happen. It's important to keep in mind the nominal and actual result of human drawn district boundaries ("gerrymandering" is a pejorative).

In theory, human drawn district boundaries can create districts which are more homogenous so that, for example, a state with a few representatives and a few defined geographic or political divisions can divide things up in a way that makes sense. For example, it would seem odd for Las Vegas to be split into multiple districts in vast barren Nevada (it isn't -- LV is the smallest district).

What happens in practice is: "Contrary to one popular misconception about the practice, the point of gerrymandering isn't to draw yourself a collection of overwhelmingly safe seats. Rather, it's to give your opponents a small number of safe seats, while drawing yourself a larger number of seats that are not quite as safe, but that you can expect to win comfortably." http://www.washingtonpost.com/news/wonkblog/wp/2014/05/15/am...

Contrary to one popular misconception about the practice, the point of gerrymandering isn't to draw yourself a collection of overwhelmingly safe seats. Rather, it's to give your opponents a small number of safe seats, while drawing yourself a larger number of seats that are not quite as safe, but that you can expect to win comfortably.

That almost sounds like a tautology.

What's tautological about it?

Here's the same sentiment expressed in image form: http://i.imgur.com/bRPcQ7O.jpg

In the original statement, the latter is just a slightly more detailed strategy for achieving the former.

I think the ideal strategy in gerrymandering is not immediately obvious (although it may technically be a tautology, it is not obviously tautological). If you asked people which of the two following strategies is ideal, many would give the wrong answer:

1. Begin by drawing the districts such that your opponents win almost 100% of the votes in as many districts as possible.

2. Begin by drawing the districts such that you win almost 100% of the votes in as many districts as possible.

Upon analysis the mathematics of the situation is obvious, but one's gut reaction may be that winning elections handily for yourself seems obviously good.

Not at all. The first thing people think of when they first learn the idea of gerrymandering is "draw districts that my party will never lose". In fact, the opposite is true. You want to concentrate the opposing party in a small number of districts, which then become overwhelmingly safe. But the seeming benefit of overwhelmingly safe seats is outweighed by the benefit of having a small advantage in a larger number of seats.

Splitline doesnt honor community or geographic boundaries. I'd suggest a modification that honors zipcode boundaries. The USPS has figured out basic units already.

Fun fact: there is actually no such thing as ZIP code boundaries.

ZIP codes are actually based on delivery routes: your ZIP code is the post office your mail carrier leaves from. ZIP codes will not actually respect community boundaries: if a town lies between two post offices, it is frequently more efficient for mail to be delivered by carriers from multiple post offices.

Moreover, not every address in the US receives home delivery. If the post office does not have a delivery route going past your home address, you can instead get a post office box from whichever post office you find most convenient. What happens to packages addressed to those street addresses is ... complex.

The "ZIP code maps" you have probably seen are actually creations of third-parties. If you ask very nicely, the Post Office will give you a list of which street addresses belong to which ZIP codes. A couple of different third party map-makers will then try to create a map based on these lists. They are not always useful or accurate.

>Moreover, not every address in the US receives home delivery. If the post office does not have a delivery route going past your home address, you can instead get a post office box from whichever post office you find most convenient. What happens to packages addressed to those street addresses is ... complex.

Emphasis mine. Would you care to expand? I'm interested in complex things.

In general, the US Post Office will attempt to deliver any unambiguously addressed parcel.

Some of the things that can happen if you address a parcel to a non-delivery address include:

  * delivery to a PO Box owned by the addressee
  * delivery to the physically closest post office to that address
  * delivery to the routing facility serving that address

This is especially a problem when a confused business demands street addresses instead of PO box numbers. Your parcel may end up somewhere you don't know about.

Street addresses in the US primarily exist for the use of emergency services, to ensure that it is easy for an ambulance or fire truck to find every house. They aren't even the primary key on land in most of the country; usually, the county or state will use a separate lot number which is independent of the road system to identity land for tax or other purposes. Prior to the 9-1-1 emergency system, the post office just used "rural route # / box #" to identify rural addresses.

Interesting. Thanks for sharing.

Ok I wondered about this. Our local village has no mail delivery; everybody in town has to have a PO box. I thought Federal mail delivery was a given thing, like a right. How can the town just decide to blow it off?

Home delivery was never quite universal. Rural Free Delivery (https://en.wikipedia.org/wiki/Rural_Free_Delivery) wasn't even a thing until the 1890s. Hilariously, with budget cuts, there has been more of a tendency to keep RFD while cutting home delivery in some of the smaller towns and villages, since it is easier for people in the village to get to the post office.

> I thought Federal mail delivery was a given thing, like a right.

Its not. Establishing post office and post roads is a Constitutional power of the federal government, but providing postal service to every building (or even just every dwelling, or, for that matter, providing any postal service at all) is not an obligation of the Federal government, and there is no entitlement to postal service.

The algorithm uses US Census data, so perhaps using county boundaries would be ideal? I'm not american, but I imagine both census and zipline boundaries as honouring geographic and community boundaries.

> The algorithm uses US Census data, so perhaps using county boundaries would be ideal?

Counties are too big to be atomic units in any algorithm that assigns districts that would be meet the requirements for reasonably equal size districts.

> I'm not american, but I imagine both census and zipline boundaries as honouring geographic and community boundaries.

ZIP codes don't do either particularly well, and census blocks might approximately do one or the other (not always the same for different boundaries) but they can't consistently do both, since the two can conflict.

County boundaries in much of the US are like state boundaries: people drew a bunch of straight lines through a map, with only the slightest regard for natural barriers, to create areas of pleasing size. It is quite common to have large population centers crossing county boundaries, or have natural barriers in the middle of a county with population centers on either side.

Wouldn't just counting total votes rather than winning regions be the simplest fix?

(Honest question, not USAmerican)

In this context, each region sends one member to the House of Representatives in Washington. So it is a "count total votes" process within the region, but there's no way to aggregate between regions (without fundamentally changing our political system).

It is pretty... but vastly oversimplifies the problem. If we are ever going to get something for real it needs to take into account real problems. Talking about things like actual operation of elections, not producing high numbers of different ballots because of splitting of districts.

For something to really work it is going to need to take into account things like existing political, and probably some physical, boundaries.

I'd love to get rid of gerrymandering. I live in a massively gerrymandered district where the representative can very safely ignore the area where I live.

This algorithm appears to have some interesting results though. Check out Maryland: http://rangevoting.org/Splitline2009/md.png

(O/T) And I live in the town where Gerry was from (his name is pronounced with a hard-G by the way, like Gail, not Jerry).

We have a drinking club named after him!

https://en.wikipedia.org/wiki/Elbridge_Gerry

That seems to be a very oversimple approach. Cities are the most obvious problem. You kind of want a population center to have a single congressmen representing it. Mathematically, you want districts to be fairly compact. There have a been several Operations Research papers that try to do things in a more practical way.

Also, a greedy approach (split in half, repeat) seems weird to me.

Technology can be used to help with this problem. Take a look at the maps from CommonCensus:

http://commoncensus.org/maps.php#local_maps

First people "vote" for the boundaries of their neighborhood and then congressional districts are created using these neighborhoods as atomic.

The author of this site seems to think that in the ideal world, districts would be decided up with clear straight lines.

This is not the case. Districts should represent distinct groups of people. Ideally, with well drawn districts, political representation will be accurate to the popular vote.

I'd recommend you all read the following article if you haven't yet.

I think if you wanted to solve districts algorithmically, you would need a neural network.

http://www.washingtonpost.com/news/wonkblog/wp/2015/03/01/th...

Similar to this other algo:

http://bdistricting.com/2010/

Although I don't necessarily disagree with this concept, I am not sure it is well refined. I noticed, looking at a certain city, that even with this shortest splitline the city was still rather split by more or less ethnically demographic lines.

Another more obvious solution is to get rid of congressional districts and then apportion the seats down the list of highest vote getters to lowest vote getters until there are no more seats (across the entire state).

https://www.youtube.com/watch?v=kUS9uvYyn3A

this explains it quite well.

Gerrymandering has more to do with Simpson's paradox. Look it up :)

Dude... thank you

This describes it quite well.

Doesn't gerrymandering mostly carve out a lot of districts for black congressmen who otherwise wouldn't have districts? I think that's kinda what it boils down to in America. There are conservative areas adjacent to and mixed with majority black areas and they want different representation.

I'm not sure how common it is, but the linked site actually has an interesting discussion of this phenomenon, which Wikipedia calls "affirmative racial gerrymandering"[0], and how it would interact with the proposed scheme: http://scorevoting.net/TheorDistrict.html#Minority

[0] https://en.wikipedia.org/wiki/Gerrymandering_in_the_United_S...

Um, no. Here's one example: http://talkingpointsmemo.com/muckraker/janet-adkins-corinne-...

This was more true in the era of overt white supremacy, when majority-white electorates would refuse to vote for black candidates, and thus black candidates could only win elections in strongly majority-black districts. This has changed from the 1990s, which black candidates started winning even in majority-white districts.

However, this also led political marginalization as well. To tilt a purple state Republican, you gerrymander districts so that you have more slightly-Republican districts than strongly-Democratic one. Concentrating black Democratic voters in one district is a strategy both to guarantee the presence of a few black representatives as well as an overall Republican majority.

No one ever thinks of it that way.

gerrymandering is good--for the electorate. why?

Because gerrymandering increases homogeneity. A gerrymandered district is more homogeneous--the voters are more alike than they were before. In general.

Homogeneity in general increases unity. A more homogeneous district means the voters in general share more common interests. If the voters share more common interests then it is easier for them to elect and hold accountable a politician who can represent their common interests.

Of course if you want the corporations to have more control over the government, and you want the people to have less control, then gerrymandering is indeed bad.

Texas's 23rd stretched from Austin to the Mexico border. You think this has "more common interests"?

https://upload.wikimedia.org/wikipedia/commons/2/24/TravisCo...

Regardless of desirability, gerrymandering does not by definition increase homogeneity. It's a tool that is used for political gain, whatever that may be.

Nope. Gerrymandering makes districts more or less homogenous in a way that benefits one party.

To tilt a balanced state Republican, you concentrate Democratic votes into a few districts where Republicans will never be competitive, then create a larger number of districts where Republicans can consistently win with razor-thin margins. You'll have one homogenous district that is 70% Democratic, and seven Republican districts that are 53% Republican.

gerry mandering is a symptom of democracy not the problem.

democracy is the problem, or rather it is a form of self governance which presents many dillemas.

historically , we know the bigger and more mature a democracy gets , the more it ages into socially defunct patterns of corruption , lowest common denominator, the decay of ethics, and embedded social passivity.

the solution to 'gerrymandering' is the end of our current democratic system, not some 2 bit claim that an algorithm is a solution to a symptom of a bigger problem .

And replace it with what?

Duh. Something better.*

*"better" is left as an exercise for the reader. I have a better solution but the margins of this textbox don't provide enough room for it.

Reply on news.ycombinator.com