The problem with the efficiency gap is that relies on voter patterns, but voter patterns are affected by gerrymandering.
So if you know that your vote is unlikely to affect the outcome, then you won't vote, which distorts the efficiency gap metric. In other words, it incentivizes districts that are so partisan that they discourage the opposition to vote.
In contrast, the compactness metric [0] only relies on census data and ignores voting patterns, party membership, or other metrics. Though this has downsides, at the very least it avoids the feedback loop of the efficiency gap.
I think this may be going for the perfect by sacrificing the good. There is nothing inherently wrong about relying on voter patterns. It's only been wrong to use voter patterns to gerrymander.
One could require that a redistricting lead to a layout that averaged 0% efficiency gap for all elections since the last redistricting effort. From that point forward, voter patterns and distribution would change, but that is exactly the reason we periodically redistrict anyway. There would still be some distortions but it would probably be a marked improvement from the status quo.
Redistricting without looking at voter demographics would lead to greater efficiency gaps. It's true that it would favor one party or another from state to state, but 50 states (or ~400 districts) is not enough to expect it would cancel out. And there are plenty of other good reasons to look at voter patterns - cohesive neighborhoods, historic geographic personality of districts, etc. It's more complicated than just aiming at contiguousness itself.
There are factors beyond compactness which merit consideration, the problem being that neither has been well-defined in a quantitative sense.
States usually point to two kinds of "communities of interest" with regard to the redistricting process: racial minorities and citizens connected by some lower political boundary ie. city limits.
A better re-districting algorithm would have to take those factors into consideration.
Let's suppose we want to introduce other factors. We now have to ask:
* who gets to choose the factors?
* when and how can they be changed?
* how do we measure them?
Let's suppose we go with the "community of interest" approach. As you pick and choose particular communities of interest to include in the algorithm, it changes the resulting district. This means that the districts resulting from such an algorithm can be adjusted by adjusting the types of communities. Thus, a party with great influence over the process will petition to have communities of interest that it knows will produce districts favorable to it. This will work for a time, until the opposition party realizes that they too can influence the districts by petitioning for particular communities of interest.
Then, when the opposition party comes into power, they will also try to get those communities changed. No factor is immune to this -- even the influence of racial categories on districts can be chosen and manipulated (and both parties ARE doing that, including the "good" party).
So, while the algorithms could be "better" than the pure compactness approach in an ideal world, it would only work if we could 100% prevent the adjustment of definitions and weights of communities of interest. And given the nature of politics, I think that will be very unlikely.
What if instead of worrying about whether the districts are fair, we worried about making it so that the fairness of the districts doesn't matter?
If we used mixed member proportional representation, the districts could be draw using any method and we could still be sure that all voters would be represented fairly.
The biggest problem with relying on compactness alone is that there are many maps that are equally compact, but have different impact on voter efficiency.
So at the end of the day, you'll end up having to look at efficiency anyway to verify that a compact map is 'fair'.
The problem with the efficiency gap as a metric is that, even when districts are drawn by algorithms blind to partisan affiliation, (considering only geometric compactness, for instance,) often an efficiency gap persists.
"The challenge for House Democrats is that their while their voter coalition has proved large enough to carry the popular vote in five of the past six presidential elections, it remains intensely concentrated, mostly in urban areas. That concentration provides a systematic advantage to Republicans in the struggle for the House—even before considering the GOP's edge in control of redistricting after the 2010 census." (https://www.theatlantic.com/politics/archive/2015/01/the-gop...)
Gerrymandering is bad, but fair and equal representation is an individual right, not something to which political parties are entitled.
The problem with the efficiency gap as a metric is that, even when districts are drawn by algorithms blind to partisan affiliation, (considering only geometric compactness, for instance,) often an efficiency gap persists.
I think that just illustrates that no rule will meet all definitions of fair.
The proposal in the article hand waves it away by suggesting that the efficiency gap be used as a limiting test on proposed districts rather than drawing districts so that it is minimized.
>Gerrymandering is bad, but fair and equal representation is an individual right, not something to which political parties are entitled.
Individuals vote for parties or candidates of their choosing. Regions do not have any right to representation. One person, one vote means we reallocate seats to urban areas which are currently deprived of representatives for their voters.
Political parties(at least in the US) are themselves a form of gerrymandering. The issues for which the 2 main parties stand for seem almost arbitrary except that one party is "of the future" and the other party is "of the past". Yeah there's some philosophical basis for the positions but they're often post-hoc to serve certain special interests.
[edit]
I've frankly become slightly sympathetic to non-democratic forms of government because I feel political parties inherently lead towards forms of treachery. They will be bought off or will try to invite outsiders as a means to grab onto power. My ideal political system these days is a meritocratic lottery system.
In ancient Athenian democracy, sortition was therefore the
traditional and primary method for appointing political
officials, and its use was regarded as a principal
characteristic of true democracy.
There's also the question of who is eligible for the lottery and who ultimately controls that determination. That form has a strong incentive to pervert the lottery process, both by choosing who is eligible and by perverting the draw itself. The latter is probably easier to secure than the former.
And even philosophically, I tend to doubt there's any widespread agreement on what makes a good leader to begin with. If anything, we disagree quite bitterly regarding that.
The Venetian Republic had an interesting solution to that. The initial lottery was just to pick an electoral college, who would would then vote (using approval voting) on another randomly-selected pool of candidates. This was repeated for several rounds! See http://www.counter-currents.com/2013/06/lot-in-venice/
That seems like a pretty clever way to prevent anyone from stacking the deck in advance. Although it was ultimately an oligarchic system, since only the rich ruling class were ever considered for office at any stage.
A good leader isn't, in a democracy. It's supposed to be rule by the people, not rule by whatever dictator can get past whatever bar we set. This is why debating where to put the bar and how to draw the line is doomed to fail - people will tend to support whatever system is more likely to return the dictator they themselves will prefer - which again often seems to be the candidate that will damage your opposition the most. Even the meme of "the lesser of two evils" seems to talk past this point entirely, because a dictator we must certainly have!
We need to get back to a much less powerful presidency, one we'd be comfortable leaving to Trump for a bit, but more importantly, one that he probably wouldn't be attracted to in the first place.
(Since we're harking back to ancient definitions of things: during the roman republic, that is, pre-Julius Caesar, dictator was a title given to a single person only in short periods of time, to deal with a crisis. During peacetime, power was with a large senate and two consuls, the latter only elected for a year. This a short and incorrect summary of the history, but TL;DR the pre-Caesar Romans would probably have considered the US president a dictator, regardless of third world cleptocrats hugging the contemporary definition)
Where I'm from there's two political blocks that are made out of 3-4 parties each.
This means that it's very easy to first pick the general side, left or right, and then you can throw your actual vote on the party whose profile you like the most.
For example, both blocks have a party that is profiling itself on environmental concerns, so you can vote "green left" or "green right" if you want. Also, there's no first-past-the-post nonsense, each electoral district has several mandates, and there's a few spare mandates that are used to make sure that the final parliamentary mandates pretty accurately reflect total popular vote of each party, while still maintaining regional representation. This means every vote counts, and voting for a party that gets 5% of the vote is perfectly fine, they'll get 5% of the seats. As it should be.
I really don't get the whole winner-takes-all mentality of the American system. I think it would be an incredible boost for democracy if there were more parties. If you could have an actual Tea party, one Christian Values party, and one fiscally conservative party, instead of those being factions inside the Republicans, fighting for control, and whichever faction "wins" gets to set the agenda of the entire thing.
> political parties inherently lead towards forms of treachery. They will be bought off or will try to invite outsiders as a means to grab onto power
'Democracy is the worst possible option, except for all the others.' All the least corrupt, most free, most properous countries in the world are democracies. Can you name a system that has worked better?
> Second, we compute the efficiency gap for congressional and state house plans between 1972 and 2012. Over this period as a whole, the typical plan was fairly balanced and neither party enjoyed a systematic advantage. But in recent years — and peaking in the 2012 election — plans have exhibited steadily larger and more pro-Republican gaps.
I think the way we normally talk about gerrymandering as a tool that one party uses to gain an electoral advantage over another party is kind of misleading and may lead us to measure success by the wrong metric.
Another way to thing about gerrymandering is not as a Republican vs Democrat thing, but an elected officials versus their constituents thing.
To use the US House of Representatives as an example, House members of both parties want the same thing, which is to get re-elected. Furthermore, they want to be re-elected by such a wide margin that their job is secure and they don't have to spend a lot of time and effort campaigning.
The voters, on the other hand, want their elected representatives to be accountable, and so they want them to be afraid of possibly losing the next election. Voters want their preferred party/candidate to win, but it's also in their best interests that elections are close enough that the implicit threat of withholding their vote is enough to motivate their representative to do their job well (whatever that means to the voter).
When party A is able to re-arrange their districts so that all their candidates have safe seats, they usually cram a bunch of party B's supporters into districts that also give a certain (smaller) number of safe seats to party B.
If you think of this as a zero-sum game between party A and party B, then party A wins a certain number of seats and party B loses an equal number.
If you think of this as a game between the voters and their Representatives, then the Representatives (of both parties, except for the handful in party B that lose their seats) won because fewer of them have to work to get votes and the voters lost because they have less influence than they did before.
The efficiency gap seems like a good metric to measure whether party A or party B is winning, but it's not obvious to me that it's a good metric for measuring how safe the seats in a given state have become.
The efficiency gap metric would say that 50% of the votes were wasted in a 2-way near-tie for a not-safe seat, but also that 50% of the votes were wasted in an election for a safe seat if one candidate got 100% and the other got 0%.
I'm not sure how best to measure seat safeness in a way that can distinguish an unbalanced result due to gerrymandering versus an unbalanced result due to one candidate running a better campaign than his/her opponent. We can look at margin-of-victory statistics in many races in many districts and make generalizations about how there are more safe districts now than some time in the past, but it's harder to say conclusively (in a way that would persuade a judge) that one particular district is gerrymandered.
So if you know that your vote is unlikely to affect the outcome, then you won't vote, which distorts the efficiency gap metric. In other words, it incentivizes districts that are so partisan that they discourage the opposition to vote.
In contrast, the compactness metric [0] only relies on census data and ignores voting patterns, party membership, or other metrics. Though this has downsides, at the very least it avoids the feedback loop of the efficiency gap.
[0] https://www.washingtonpost.com/news/wonk/wp/2014/06/03/this-...