Re:This is the best explanation of gerrymandering you will e...
by waqf on Wednesday February 08, 2017 @05:53PM
(#13602985)
If people are even thinking about gerrymandering, then it's apparent that the system of having one representative elected by each district no longer serves its supposed purpose of making sure each "community" is represented.
Re:This is the best explanation of gerrymandering you will e...
by Lazare on Wednesday February 08, 2017 @06:12PM
(#13603104)
Yes, and yes.
It has some real advantages (it's nice being able to vote for someone that actually agrees with you AND not have your vote be wasted), but my experience in New Zealand is that:
1. It enhances the power of parties and thus increases partisanship
2. It allows small parties to play kingmaker. In the NZ case, the perennial kingmaker party is corrupt and virulently nativist.
If, hypothetically, you were concerned about the partisanship, corruption, and nativist policies of current American politics, PR may not be the panacea you hope. I think it was quite unlikely for the American system to yield Trump as president (we just got very unlucky) but it would be quite likely for a PR system to yield Trump as a kingmaker with a vastly outsize influence on the policy of the resultant government. That's not obviously better, on balance.
Re:This is the best explanation of gerrymandering you will e...
by dragonwriter on Wednesday February 08, 2017 @06:42PM
(#13603302)
The kingmaker effect happens in the US system with factions within major parties, it's just less visible to people who don't deeply study politics, because the factions are less visible than actual parties in a multiparty system.
Enhancing the power of parties is a function of systems that establish proportionality by party list (either wholly or by MMP) more than it is directly of proportionality (proportionality vs. a duopoly-reinforcing system encourages party cohesion, because it reduces the cost of people splitting off to form their own party because they dissent from existing parties, but that decreases rather than increasing the power of parties, in that the party majority or leadership -- outside of party list systems -- has less that they can threaten wayward members with to keep them in line.)
Re:This is the best explanation of gerrymandering you will e...
by waqf on Wednesday February 08, 2017 @06:32PM
(#13603245)
Thanks, I was expecting a rebuttal based on experience with PR but this is a particularly interesting one.
Actually I'd be interested in seeing some kind of delegative democracy system like Liquid Feedback. It has some features in common with PR (mostly the emphasis on having to negotiate a coalition given the voting results), but I'm hoping it wouldn't share the problems you outlined?
Re:This is the best explanation of gerrymandering you will e...
by dragonwriter on Wednesday February 08, 2017 @06:50PM
(#13603355)
Arguably, it's what failed to happen; Hitler ended up leading a minority government with the largest party (well, tied with the Communists), no one acted as kingmaker with a small party putting some other coalition over the top extracting concessions for that boost.
There was kingmaking of a more personal kind going on behind the scenes, but not of the minor-party-exercises outsized influence kind.
Re:This is the best explanation of gerrymandering you will e...
by dragonwriter on Wednesday February 08, 2017 @06:46PM
(#13603326)
The first is a problem with party list proportional representation, but there are constituency-and-candidate-centered means of achieving proportionality (pragmatically, less perfect partisan proportionality than party list, but arguably better direct representation and possibly equal practical proportionality of substantive views) such as STV in multimember districts.
The second is just making visible a problem that exists even in the US system with intraparty factions.
Re:This is the best explanation of gerrymandering you will e...
by fictioncircle on Wednesday February 08, 2017 @06:35PM
(#13603263)
In the US, the party list would likely be chosen by party primaries (i.e. voters) rather than the party. That would largely negate the first problem.
Similarly, for the second, the US has two very strong political parties and this would really only change House of Representatives (one portion of one branch of the federal government) so the kingmaking power would be extremely limited as the vast majority of the Senate & Presidency would still end up with the stronger parties.
Re:This is the best explanation of gerrymandering you will e...
by AnthonyMouse on Wednesday February 08, 2017 @07:26PM
(#13603552)
> If people are even thinking about gerrymandering, then it's apparent that the system of having one representative elected by each district no longer serves its supposed purpose of making sure each "community" is represented.
The gerrymandering problem isn't created by individual districts, it's created by first past the post voting.
Suppose you replace this with range voting or approval voting. Now it doesn't matter where you draw the district lines, because there can be twelve viable candidates and no matter where you draw the lines, each district will choose the candidate who best represents the district. A candidate can run, and win, on a platform which is specific to that individual district without having to align with an existing party's platform because winning is no longer contingent on the backing of a major party.
Which means you can't draw the lines to favor your own party because all it will do is cause some new party/candidate to appear that better represents whatever arbitrary district you created than you do.
And then you get to keep all the advantages of individual districts and avoid all the disadvantages of PR.
Re:This is the best explanation of gerrymandering you will e...
by dragonwriter on Wednesday February 08, 2017 @07:42PM
(#13603620)
> The gerrymandering problem isn't created by individual districts
Yes it is.
> it's created by first past the post voting.
FPTP creates other problems, but not the gerrymandering problem. OTOH, systems that support parties may mitigate some effects of gerrymandering.
> Now it doesn't matter where you draw the district lines, because there can be twelve viable candidates and no matter where you draw the lines, each district will choose the candidate who best represents the district.
The problem with gerrymandering is that single member districts let you choose what kind of candidates will be elected by how you draw draw district lines, and by doing so you can assure that some interests are over or underrepresented in the legislature compared to their prevalence on the population. You've not stated a way in which range or approval voting change that (and you won't find one, because they don't.)
Re:This is the best explanation of gerrymandering you will e...
by AnthonyMouse on Wednesday February 08, 2017 @07:50PM
(#13603657)
> You've not stated a way in which range or approval voting change that (and you won't find one, because they don't.)
Sure they do. If you concentrate all the people who care about a particular issue into one district, and you make candidates not affiliated with a major party viable, then that district can elect a representative who cares about that issue a lot and will be willing to trade issues the district cares less about in order to get it, without having to worry what the party leadership wants.
And how is PR supposed to solve this problem? It only replaces districts with parties, and then you're stuck choosing from a small handful of parties whose members all stick to the party line and you're forced to choose one that doesn't represent you on some subset of issues.
The only better way to solve that problem is direct democracy, which would bring in a whole different set of problems.
Re:This is the best explanation of gerrymandering you will e...
by ClayShentrup on Thursday February 09, 2017 @02:49AM
(#13605360)
PR mitigates Gerrymandering because it reduces the amount of "discardable" surplus votes.
Imagine you have 100 single-winner districts that are each roughly 50/50 Dem/GOP, with the slight edge breaking 50/50, so that your legislature has 50 reps from each party.
Now we redistrict to pack a bunch of Dems from the GOP-majority districts into one super-Dem district, shaving down their margins in numerous districts. We could conceivable get 99 GOP reps even though the population is 50/50 Dem/GOP. That super-Dem district "discards" or "wastes" 50% of the votes of the Dems inside it, because they're surplus votes in excess of what they need to win the district.
Now imagine you switch to 20 5-winner districts elected with PR. The Dems could win all five seats with conceivably 83.3% of the vote (5 candidates getting 16.6% each means they get all the votes). So the most you can possibly wast is 16.7% of the votes instead of 50%.
> It only replaces districts with parties
No. Most PR systems still have districts. Exceptions like Israel, where the whole country is one big district, tend to be the poster children of PR gone wrong. Plus there are PR systems where you vote for candidates, not parties. Single Transferable Vote in Australia's Senate is an example. You just rank the candidates (although many voters blindly follow the guidance of their preferred party).
However I agree that even the single-winner form would reduce the negative impact of Gerrymandering (also, getting Score Voting is a much more consequential/impactful change than eliminating Gerrymandering). The reason it mitigates Gerrymandering is that it tends to reduce polarization and two-party dynamics, electing broadly appealing centrist winners. If you try the "packing" strategy with Score Voting, you tend to "move the ideological center", so a district that's already going Democrat doesn't become a /dev/null for surplus Dem votes that you want to effectively burn. If you pile on more Democrats, you can just keep pushing the center more Democratic. This is a theoretical argument mind you, but I think a reasonable one.
Clay Shentrup
Co-founder, The Center for Election Science
Re:This is the best explanation of gerrymandering you will e...
by AnthonyMouse on Saturday February 11, 2017 @04:19AM
(#13621877)
> Imagine you have 100 single-winner districts that are each roughly 50/50 Dem/GOP, with the slight edge breaking 50/50, so that your legislature has 50 reps from each party.
Imagine you have 100 single-winner districts with range voting so that there aren't two parties, or any strong parties at all.
Then the platform needed to win a district is district-specific, because a candidate with a district-specific platform beats any party's general platform in that district. Even if the winning candidate is a member of a party, the candidate still has to do what the district wants over what the party wants or they'll lose the seat.
The winning candidate in single-district range voting is literally the one that gets the highest average rating within the district, so it optimizes for no surplus at all.
If there was any then an independent candidate could show up with a platform that captures it by aligning more closely with the district and thereby win the seat.
The only way to do better than that is to do the same thing but with smaller districts, at the cost of having more members of the legislature (and in the limit it becomes direct democracy).
> No. Most PR systems still have districts.
I don't mean there are no districts, I mean that in PR instead of being stuck with a district that might not represent you in some ways, you're stuck with a party that might not represent you in some ways.
And PR STV doesn't fix that. Anything that increases the number of constituents per representative inherently can't, because that is what causes the problem to begin with. Diluting each vote by 20 times but giving each constituent 20 times as many representatives doesn't cancel out.
If a state upstream from your district is dumping poison in the river, no candidate could win that district without addressing it. But with any kind of PR, it's possible for every representative to ignore it because the affected voters split between twenty representatives, none of which need 1/20th of their votes enough to address the problem.
And RRV looks like it would create a lot of strange incentives, like giving a lower score to candidates you strongly like if you think they're likely to win regardless, except that if everyone thinks that way then that candidate loses.
Re:This is the best explanation of gerrymandering you will e...
by e28eta on Wednesday February 08, 2017 @05:48PM
(#13602951)
Obviously the title is hyperbole, but in order to be the "best explanation" I'd like to see some of the reasonable arguments that are used to justify "non-compact, non-fair" districts.
Re:This is the best explanation of gerrymandering you will e...
by tunesmith on Wednesday February 08, 2017 @08:15PM
(#13603782)
Even the word "fair" implies the belief that a state's district breakdown should track with the state's population breakdown. This is not necessarily axiomatic. If you want competitive districts with a smaller incumbency advantage, then maybe you don't want that. IOW it depends on how you define "fair".
Re:This is the best explanation of gerrymandering you will e...
by favorited on Wednesday February 08, 2017 @06:22PM
(#13603165)
Fun fact: "gerrymander" is a portmanteau of the name "Gerry" and "salamander." Massachusetts Governor Elbridge Gerry signed the state's initial electoral districts into law (which his party had drawn in their favor), and one of the districts looked like a salamander.
Re:This is the best explanation of gerrymandering you will e...
by tunesmith on Wednesday February 08, 2017 @08:19PM
(#13603802)
Can PA-7 be deemed indefensible for reasons other than how it looks? Not a leading question, I wonder if there are statistical data about it. You could argue that if you drew proportional representation as districts (basically: pointillism) it'd make PA-7 look wonderful in comparison.
Re:This is the best explanation of gerrymandering you will e...
by caf on Wednesday February 08, 2017 @10:27PM
(#13604425)
It is abundantly clear that that district has been drawn in the way it has for nakedly political reasons, and not to group together contiguous communities of interest.
You could argue that if you drew proportional representation as districts (basically: pointillism) it'd make PA-7 look wonderful in comparison.
If you argued that, you'd be wrong. Proportional representation districts should group together communities of interest in exactly the same way as single-member districts - the only difference is that PR districts are represented by more than one representative.
Re:This is the best explanation of gerrymandering you will e...
by tunesmith on Thursday February 09, 2017 @12:26AM
(#13604894)
Sure, if you just completely redefine my point. But if you take a PR "district", like a state, with multiple representatives, and imagine them a a collection of single-representative districts, they're about as far from contiguous as you get.
I'm not arguing PA-7 isn't a problem. But I don't think "looks" is a good standard. What's better is something like the standards being argued in the current Wisconsin case that I think might still be going to the Supreme Court.
Re:This is the best explanation of gerrymandering you will e...
by mymex1 on Wednesday February 08, 2017 @06:32PM
(#13603246)
It won't solve the problem of "how" to redistrict, but as far as finding out if gerrymandering IS happening it seems like there are good ideas out there such as the "efficiency gap" used in a recent Wisconsin case.
Re:This is the best explanation of gerrymandering you will e...
by my_first_acct on Wednesday February 08, 2017 @07:16PM
(#13603496)
Even "compact, fair" districts can lead to outcomes that one party, at least, would find unfair.
For instance, in the US, Democratic voters tend to be clustered in cities, so urban congressional districts (even "compact, fair" ones) vote overwhelmingly for Democrats. This means that there a lot of "wasted" Democratic votes contributing to landslide victories in urban districts, instead of bringing about Democratic victories elsewhere.
Even if the Republicans have only a narrow majority in the suburban and rural districts, and are an overall minority of voters in the state, they are able to win the majority of congressional districts.
Re:This is the best explanation of gerrymandering you will e...
by dragonwriter on Wednesday February 08, 2017 @07:27PM
(#13603556)
Right. Any method of drawing single-member FPTP districts will, unless voters of different persuasions are perfectly uniformly distributed, will:
(1) Produce disproportionate advantage for some party, and/or
(2) Produce some number of safe seats where the general election is a forgone conclusion and the mechanism that selects candidates decides the winner in practice.
You can select methods to choose which of these effects predominate, and which party benefits from either disproportionate representation or safe seats (there's a certain tension -- you get disproportionate representation for your party by giving the other party super safe seats), but you can't avoid them unless you throw out FPTP, and really single-member independent districts entirely.
Re:This is the best explanation of gerrymandering you will e...
by tunesmith on Wednesday February 08, 2017 @08:13PM
(#13603773)
There isn't really yet an operational definition of gerrymandering. Such that, if x conditions exist, it's gerrymandering, and if they don't, it's not. You can't do it without pointing at results and going, "Well that's clearly unfair."
I believe there is a recent legal case that is still in the courts where they are trying to arrive at a legal standard. Something about the percentage of unrepresented voters in a state's average district. So there might be progress on that point.
But even politicians that complain against gerrymandering are inconsistent. Some politicians complain about "safe" districts not being competitive enough, and also complain that house elections don't mirror the popular vote breakdown enough. Those are actually on opposite sides of the loose tradeoff.
As you move towards your house results mirroring the popular vote (like if today's House were 48.2% Democratic and 46.1% Republican), then you move towards less competitive districts (except for primaries), like the "perfect representation" example in the article. Because, in order to guarantee that your state's district breakdown will mirror your state's vote breakdown, you basically would want all Republican-voting people in their districts, and all Democratic-voting people in their own.
So, what do you want? Do you want a House that is overall representative of the national popular vote? Do you want districts where the maximum number of citizens are represented? Do you want competitive districts? Do you want compact, contiguous districts? Do you want districts that respect traditional neighborhood boundaries, district history, and geography? Because they all imply different districting strategies. And saying "let's just let a computer split-line it!" means explicitly rejecting some of these principles - you can't get past making choices about tradeoffs, because it is inherently messy.
Re:This is the best explanation of gerrymandering you will e...
by justinzollars on Wednesday February 08, 2017 @07:11PM
(#13603474)
Gerrymandering results in more competitive districts. So it will be a fact a life because every party wants more.
From the point of view of actual Representation it does not make sense. For example, my former Congresswomen was personally heartbroken to lose the town of Bowling Green, in Ohio. Her new district isn't really representative of any logical organization of people, so its hard to be responsive to the needs of the people.
I think my preference is that a computer cut it in a way that considers the geography and land topology and types of communities whether cities or towns - with the goal of being as square as possible.
Re:This is the best explanation of gerrymandering you will e...
by dragonwriter on Wednesday February 08, 2017 @07:31PM
(#13603568)
> Gerrymandering results in more competitive districts.
Only when the goal of gerrymandering is to maximize competitive districts, but it's usually not; it's usually to maximize safe districts and districts in which one party has an electoral advantage. (Also, frequently, it's designed to create particular benefits or problems for particular incumbents.)
Re:This is the best explanation of gerrymandering you will e...
by pasquinelli on Wednesday February 08, 2017 @07:18PM
(#13603510)
or just keep the same process for drawing district boundaries but take away the notion of a district to higher elections. that's the source of the bad incentives that lead to gerrymandering.
Re:This is the best explanation of gerrymandering you will e...
by hibikir on Wednesday February 08, 2017 @07:24PM
(#13603540)
While I am no friend of gerrymandering, arguing that there is a geographic ideal is disingenuous: Voters clump up, and thus any automated mechanism that just looks at geography will be skewed, and, in the US today, Very often skewed towards republicans compared to the ratio of voters in a state.
There is no such thing as a fair way to assign representatives without making some judgement call on what fair means. If you prefer short split line, that's because of your very personal preferences, which might have something to do with good looking maps, but they absolutely have nothing to do with making sure that groups of people are epresented in a way that will match proportional representation. Shortest split line will still give you a state that is 50/50 in a senate vote, but hands you a congress that is a 7/3 split.
Re:This is the best explanation of gerrymandering you will e...
by specialist on Wednesday February 08, 2017 @07:29PM
(#13603561)
Geometric esthetics aside, I'm eagerly awaiting the Supreme's ruling on voter efficiency (wasted votes). I anticipate it'll moot (supplant) most of the arguments about compactness, shapes, etc.
In that case, there's no reason not to switch to some more fairly representative system, for which there are various possibilities but PR systems are the best known (https://en.wikipedia.org/wiki/Proportional_representation).
I spent a number of years in Italy, and several come to mind that I saw there:
* Representatives are beholden to the party, not to some local area.
* Smaller parties can play 'kingmaker'.
It has some real advantages (it's nice being able to vote for someone that actually agrees with you AND not have your vote be wasted), but my experience in New Zealand is that:
1. It enhances the power of parties and thus increases partisanship
2. It allows small parties to play kingmaker. In the NZ case, the perennial kingmaker party is corrupt and virulently nativist.
If, hypothetically, you were concerned about the partisanship, corruption, and nativist policies of current American politics, PR may not be the panacea you hope. I think it was quite unlikely for the American system to yield Trump as president (we just got very unlucky) but it would be quite likely for a PR system to yield Trump as a kingmaker with a vastly outsize influence on the policy of the resultant government. That's not obviously better, on balance.
Enhancing the power of parties is a function of systems that establish proportionality by party list (either wholly or by MMP) more than it is directly of proportionality (proportionality vs. a duopoly-reinforcing system encourages party cohesion, because it reduces the cost of people splitting off to form their own party because they dissent from existing parties, but that decreases rather than increasing the power of parties, in that the party majority or leadership -- outside of party list systems -- has less that they can threaten wayward members with to keep them in line.)
Actually I'd be interested in seeing some kind of delegative democracy system like Liquid Feedback. It has some features in common with PR (mostly the emphasis on having to negotiate a coalition given the voting results), but I'm hoping it wouldn't share the problems you outlined?
Hmm, now you mention it, didn't something like that happen to the Weimar Republic?
There was kingmaking of a more personal kind going on behind the scenes, but not of the minor-party-exercises outsized influence kind.
The second is just making visible a problem that exists even in the US system with intraparty factions.
Similarly, for the second, the US has two very strong political parties and this would really only change House of Representatives (one portion of one branch of the federal government) so the kingmaking power would be extremely limited as the vast majority of the Senate & Presidency would still end up with the stronger parties.
* https://en.wikipedia.org/wiki/Mixed-member_proportional_repr...
The gerrymandering problem isn't created by individual districts, it's created by first past the post voting.
Suppose you replace this with range voting or approval voting. Now it doesn't matter where you draw the district lines, because there can be twelve viable candidates and no matter where you draw the lines, each district will choose the candidate who best represents the district. A candidate can run, and win, on a platform which is specific to that individual district without having to align with an existing party's platform because winning is no longer contingent on the backing of a major party.
Which means you can't draw the lines to favor your own party because all it will do is cause some new party/candidate to appear that better represents whatever arbitrary district you created than you do.
And then you get to keep all the advantages of individual districts and avoid all the disadvantages of PR.
Yes it is.
> it's created by first past the post voting.
FPTP creates other problems, but not the gerrymandering problem. OTOH, systems that support parties may mitigate some effects of gerrymandering.
> Now it doesn't matter where you draw the district lines, because there can be twelve viable candidates and no matter where you draw the lines, each district will choose the candidate who best represents the district.
The problem with gerrymandering is that single member districts let you choose what kind of candidates will be elected by how you draw draw district lines, and by doing so you can assure that some interests are over or underrepresented in the legislature compared to their prevalence on the population. You've not stated a way in which range or approval voting change that (and you won't find one, because they don't.)
Sure they do. If you concentrate all the people who care about a particular issue into one district, and you make candidates not affiliated with a major party viable, then that district can elect a representative who cares about that issue a lot and will be willing to trade issues the district cares less about in order to get it, without having to worry what the party leadership wants.
And how is PR supposed to solve this problem? It only replaces districts with parties, and then you're stuck choosing from a small handful of parties whose members all stick to the party line and you're forced to choose one that doesn't represent you on some subset of issues.
The only better way to solve that problem is direct democracy, which would bring in a whole different set of problems.
Imagine you have 100 single-winner districts that are each roughly 50/50 Dem/GOP, with the slight edge breaking 50/50, so that your legislature has 50 reps from each party.
Now we redistrict to pack a bunch of Dems from the GOP-majority districts into one super-Dem district, shaving down their margins in numerous districts. We could conceivable get 99 GOP reps even though the population is 50/50 Dem/GOP. That super-Dem district "discards" or "wastes" 50% of the votes of the Dems inside it, because they're surplus votes in excess of what they need to win the district.
Now imagine you switch to 20 5-winner districts elected with PR. The Dems could win all five seats with conceivably 83.3% of the vote (5 candidates getting 16.6% each means they get all the votes). So the most you can possibly wast is 16.7% of the votes instead of 50%.
> It only replaces districts with parties
No. Most PR systems still have districts. Exceptions like Israel, where the whole country is one big district, tend to be the poster children of PR gone wrong. Plus there are PR systems where you vote for candidates, not parties. Single Transferable Vote in Australia's Senate is an example. You just rank the candidates (although many voters blindly follow the guidance of their preferred party).
Another cure to Gerrymandering is algorithmic redistricting, which I'm supportive of. https://www.youtube.com/watch?v=kUS9uvYyn3A
As for Score Voting (aka Range Voting) there is a proportional form of it. http://ScoreVoting.net/RRV.html
However I agree that even the single-winner form would reduce the negative impact of Gerrymandering (also, getting Score Voting is a much more consequential/impactful change than eliminating Gerrymandering). The reason it mitigates Gerrymandering is that it tends to reduce polarization and two-party dynamics, electing broadly appealing centrist winners. If you try the "packing" strategy with Score Voting, you tend to "move the ideological center", so a district that's already going Democrat doesn't become a /dev/null for surplus Dem votes that you want to effectively burn. If you pile on more Democrats, you can just keep pushing the center more Democratic. This is a theoretical argument mind you, but I think a reasonable one.
Clay Shentrup Co-founder, The Center for Election Science
Imagine you have 100 single-winner districts with range voting so that there aren't two parties, or any strong parties at all.
Then the platform needed to win a district is district-specific, because a candidate with a district-specific platform beats any party's general platform in that district. Even if the winning candidate is a member of a party, the candidate still has to do what the district wants over what the party wants or they'll lose the seat.
The winning candidate in single-district range voting is literally the one that gets the highest average rating within the district, so it optimizes for no surplus at all.
If there was any then an independent candidate could show up with a platform that captures it by aligning more closely with the district and thereby win the seat.
The only way to do better than that is to do the same thing but with smaller districts, at the cost of having more members of the legislature (and in the limit it becomes direct democracy).
> No. Most PR systems still have districts.
I don't mean there are no districts, I mean that in PR instead of being stuck with a district that might not represent you in some ways, you're stuck with a party that might not represent you in some ways.
And PR STV doesn't fix that. Anything that increases the number of constituents per representative inherently can't, because that is what causes the problem to begin with. Diluting each vote by 20 times but giving each constituent 20 times as many representatives doesn't cancel out.
If a state upstream from your district is dumping poison in the river, no candidate could win that district without addressing it. But with any kind of PR, it's possible for every representative to ignore it because the affected voters split between twenty representatives, none of which need 1/20th of their votes enough to address the problem.
And RRV looks like it would create a lot of strange incentives, like giving a lower score to candidates you strongly like if you think they're likely to win regardless, except that if everyone thinks that way then that candidate loses.
Maybe that's because I remember reading this article (TBF, it is linked), particularly the section about "community of interest": https://www.washingtonpost.com/news/wonk/wp/2014/06/03/this-...
Essentially the graphic shows how compactness can be unfair, but doesn't tell us what reasonable explanations can be used for non-compact districts.
You could argue that if you drew proportional representation as districts (basically: pointillism) it'd make PA-7 look wonderful in comparison.
If you argued that, you'd be wrong. Proportional representation districts should group together communities of interest in exactly the same way as single-member districts - the only difference is that PR districts are represented by more than one representative.
I'm not arguing PA-7 isn't a problem. But I don't think "looks" is a good standard. What's better is something like the standards being argued in the current Wisconsin case that I think might still be going to the Supreme Court.
https://www.nytimes.com/2016/11/21/us/wisconsin-redistrictin...
For instance, in the US, Democratic voters tend to be clustered in cities, so urban congressional districts (even "compact, fair" ones) vote overwhelmingly for Democrats. This means that there a lot of "wasted" Democratic votes contributing to landslide victories in urban districts, instead of bringing about Democratic victories elsewhere.
Even if the Republicans have only a narrow majority in the suburban and rural districts, and are an overall minority of voters in the state, they are able to win the majority of congressional districts.
See this 2014 article (sorry, behind NYTimes paywall) for more details: https://www.nytimes.com/2014/01/26/opinion/sunday/its-the-ge...
(1) Produce disproportionate advantage for some party, and/or
(2) Produce some number of safe seats where the general election is a forgone conclusion and the mechanism that selects candidates decides the winner in practice.
You can select methods to choose which of these effects predominate, and which party benefits from either disproportionate representation or safe seats (there's a certain tension -- you get disproportionate representation for your party by giving the other party super safe seats), but you can't avoid them unless you throw out FPTP, and really single-member independent districts entirely.
I believe there is a recent legal case that is still in the courts where they are trying to arrive at a legal standard. Something about the percentage of unrepresented voters in a state's average district. So there might be progress on that point.
But even politicians that complain against gerrymandering are inconsistent. Some politicians complain about "safe" districts not being competitive enough, and also complain that house elections don't mirror the popular vote breakdown enough. Those are actually on opposite sides of the loose tradeoff.
As you move towards your house results mirroring the popular vote (like if today's House were 48.2% Democratic and 46.1% Republican), then you move towards less competitive districts (except for primaries), like the "perfect representation" example in the article. Because, in order to guarantee that your state's district breakdown will mirror your state's vote breakdown, you basically would want all Republican-voting people in their districts, and all Democratic-voting people in their own.
So, what do you want? Do you want a House that is overall representative of the national popular vote? Do you want districts where the maximum number of citizens are represented? Do you want competitive districts? Do you want compact, contiguous districts? Do you want districts that respect traditional neighborhood boundaries, district history, and geography? Because they all imply different districting strategies. And saying "let's just let a computer split-line it!" means explicitly rejecting some of these principles - you can't get past making choices about tradeoffs, because it is inherently messy.
From the point of view of actual Representation it does not make sense. For example, my former Congresswomen was personally heartbroken to lose the town of Bowling Green, in Ohio. Her new district isn't really representative of any logical organization of people, so its hard to be responsive to the needs of the people.
I think my preference is that a computer cut it in a way that considers the geography and land topology and types of communities whether cities or towns - with the goal of being as square as possible.
Only when the goal of gerrymandering is to maximize competitive districts, but it's usually not; it's usually to maximize safe districts and districts in which one party has an electoral advantage. (Also, frequently, it's designed to create particular benefits or problems for particular incumbents.)
[1] http://scorevoting.net/SplitLR.html
There is no such thing as a fair way to assign representatives without making some judgement call on what fair means. If you prefer short split line, that's because of your very personal preferences, which might have something to do with good looking maps, but they absolutely have nothing to do with making sure that groups of people are epresented in a way that will match proportional representation. Shortest split line will still give you a state that is 50/50 in a senate vote, but hands you a congress that is a 7/3 split.