Dumb question about redistricting:

[MinneapolisMatt]

Why can't we just slap a piece of graph paper over the entire country and call it a day?

[dsc]

1. districts have to be equal in population

and then there are some legitimate things to take consideration of. Majority minority districts so that minorities can be represented. Keeping communities of interest together. Not splitting up cities so they have no power. It can be complicated.

[napoleon_in_rags]

2. Easily done:

Draw a vertical line down the middle of the state so half the population is on each side. for each division, draw a horizontal line so 1/4 of the population is on each side, etc.

There is no moral argument whatsoever for the gerrymandering that is putting people who the popular electorate doesn't support into power.

[dsc]

5. then you should eliminate districts

Even if we had simply used that method in a state like PA, you still would have had a bad result. Unless you purposely gerrymander to split Philly into way more districts than your method would do, you are going to have a lot of districts where Democrats will win say 75 to 25 with some close but narrow GOP districts elsewhere. In some states, like Iowa, Maine, New Hampshire, and some others, your method would likely work out OK, but in states with large cities it wouldn't. Now if you had proportional vote where each party had an ordered list of people and you as a voter just voted for the party, that would work. In NC, we got 51% of the vote with 13 seats total, we won 4. We should have gotten 7 to the other sides 6. Eliminating districts would produce that result. Ones drawn under your system might well have produced a 5 to 8 split against us, maybe a 6 to 7 split against us, with a small likelihood of a 7 to 6 split in our favor. My guess is we would have had a 5 to 8 split. We would win one down east seat, two triangle seats, one charlotte seat, and one triad seat for sure. We might get one more east seat or maybe a triangle/triad seat or a slim chance of both. I can't see a way for us to get another seat in a map drawn by your method.

[napoleon_in_rags]

6. I heard Maddow putting forth the numbers.

She explained the cost (in votes) per person seated in congress, and the Dem cost was way higher than Republican cost, which is to say it took more voters to elect Dems to congress than to elect Republicans. This tells me all I need to know: A neutral system like I am advocating should get rid of such biases.

[dsc]

7. No it wouldn't

at least not in states that have cities which are the size of one or more congressional districts. Philadelphia voted 85% for Obama and about 14% for Romney. Which means that this would be probably around an 85 to 15 Dem area. http://www.phillyelectionresults.com/

In 2010, Philadelphia had around 1.5 million people. That would be a bit over two seats. Unless you draw several districts all of which cut into Philly, which directly violates what you said you want to do. We end up wasting votes. That is why your method can't work in a state like PA. Yes, it would be less unfair than the current districts, but your method would likely lead to at best an 8 to 10 division against us or a 7 to 11 division against us in PA the current division is 5 to 13 in favor of the GOP. A fair one would have been either 9 to 9 or 10 to 8 our favor depending upon just how much we beat them by.

[napoleon_in_rags]

9. Are you going to make me do math while drinking my saturday night beers?

Here are the presidential results for PA:

Obama votes: 2,990,274 = ov
Romney vote: 2,680,434 = rv

Here are the Senator results:

Casey (D): 3,021,364
Smith (R): 2,509,132

Now, the first question is, how do we compute the probability of an individual voting for Obama? Its
ov/(ov+rv) = 0.527319340018918
Easy. Now the more complex question: Given an arbitrary partition of voters of size n, how do we compute the probability that k voters of partition size n voted for Obama?

Let binomial(n, k) be the function mathematicians call "choose". Let * be "times" and ^ be "to the power of". Then:

binomial(n, k) * p^k * (1-p)^(n-k) = odds that k voters out of an arbitrary partition of size n voted for Obama.

Now how do we compute the odds that a majority of voters from an arbitrary partition of size n voted for Obama? Let sigma(n, i=min) be the notation mathematicians call sigma notation. then. probability of obama win (pow):

pow = sigma(n, i=n/2) : binomial(n, i) * p^i * (1-p)^(n-i)

equivalent of the discrete integral of the asymmetrical binomial distribution of an Obama win, from the sufficient votes needed for an Obama win and up.

Now lets compute it for some various sizes of n: (odds of Obama win in first position, Romney win in second)

n = 20: (0.680919164663950, 0.319080835336050)
n = 40 (0.693499502364917, 0.306500497635083)
n = 250 (0.823468231377216, 0.176531768622784)
n = 1000 (0.960895921815634, 0.0391040781843661)

Anyway, so what does all this mean? It means that when Obama won the popular vote by the amount that he did, the odds of any arbitrary partition of voters going for Obama increases as the size of the partition grows. Which means it takes a REMARKABLE FEAT OF GERRYMANDERING to deliver wins against the popular vote in most circumstances, and there is nothing to fear in my scheme, or any other arbitrary scheme of dividing popular votes for whoever won the popular vote. As Obama did, and as the PA senate Dem did by an even larger margin. Its only very precisely planned schemes which can stop the popular vote from winning.

http://www.huffingtonpost.com/2012/11/08/house-candidates-votes_n_2096978.html

[dsc]

15. you are assuming where you live has nothing to do with how you vote

and that is just not true. People in Philly had an 85% chance of voting for Obama, people outside of Philly had way less than 85% of voting for Obama. The binomial probability function can only be used if the probability doesn't change from place to place, but it clearly does. If you don't split Philly up, which I don't think your method would do, then we would have wasted votes. To a lesser extent Pittsburgh presents a similar problem. It would likely give you one district with around 70/30. Again, those votes are wasted. You can't get around it.

[napoleon_in_rags]

17. Yeah, today I realized the problem with what I wrote.

The beers made me miss that.

The core message is still true: Any arbitrary partitions of voters - (by which I mean randomly chosen) are incredibly likely to go with the popular vote as the number of the voters in each partition grows. Obviously once a partition is a big as the whole state, the probability of results matching popular vote is 1. But yes, geographic area has an effect too, it skews results on its own. That makes the math a bit more complicated when you talk about drawing a geographic map.

I believe the results of the maps should match the popular vote. And I believe there are probably arbitrary ways to draw them which make that very likely, but unfortunately it would be a fair amount of work in order to prove that, which I don't have the time to do. I am willing to say that gerrymandering which intentionally causes election results contrary to the popular vote is anti-democratic.

[dsc]

16. one last example using NC

NC has three Congressional districts that are down east, two that are far west, five districts that come from the triangle and triad, about 1.5 Charlotte districts, and about 1.5 from suburbs of Charlotte and rural areas between Charlotte, the triad, and the triangle. Your method really wouldn't alter that. We would almost certainly lose the two western districts. In the east best case is 1D, 1 DINO, 1R. Likely would be 1D, 2R. We should win at least 3 out of 5 of the triangle/triad seats. 4 isn't out of the question. The Charlotte seat should be a win, the 50/50 seat a true swing, the other 2 losses. That gets us either 6 or 7 depending on the swing.

[DRoseDARs]

8. Go here and please learn just how not easy redistricting is.

http://www.redistrictinggame.org/

[napoleon_in_rags]

18. Oh I know there's a lot to it.

I don't know the ins and outs of redistricting itself, but just the overall mathematical idea: You have one probability distribution, and then you're dividing it - quanitizing it - into a discrete set of others, with discrete win/lose answers. That makes all kinds of wacky things happen which I won't go into here, suffice to say one fun example is the Young's double slit experiment, in which the universe redistricts itself: You have a wave based probability distribution (the underlying distribution) interfering with itself as it goes through two slits. But when you observe one of the slits to see where the particle is going through, the universe makes each particle (quantum) act as a district, and it either must go entirely through one slit, (Romney slit) or the other (Obama slit)... When previously it went through both. (just as each district has people who support each candidate, it only becomes an entity when you declare it to be and take its majority as its value.) So yeah, that's from quantum mechanics. So there's plenty of complexity in this neighborhood of thing, more than enough for me thank you!

[DRoseDARs]

22. *facepalm* OK, no, enough. Continually spouting math and quantum physics shows you don't understand.

Mein gott. Just click the link, play the game, learn just what's involved with redistricting. Seriously, it won't bite you, it doesn't take all that long and you'll get a good grasp on the subject.

[napoleon_in_rags]

23. With all due respect, this situation is very much about math.

The quantum mechanics comment was cute more than useful, but my core point stands.

You link to this site, which brings up all the concerns in drawing districts: things like ensuring minority representation, and all this. But what we've got, what's been reported on Maddow is a situation where the it costs more Dem votes to get somebody elected than Republican votes. We've got situations where actual winners are different from those popularly elected quite commonly. What the math does is it tells us that this situation is remarkably improbable. That means gerrymandering is basically subverting democracy in these contests. What I'm saying is that democracy is priority one in drawing districts. The elected officials should be the winners of the popular vote a huge majority of the time. Priority one. Anything which distracts from that fact honestly feels pretty sleazy to me.

[DRoseDARs]

24. You're missing the forest... for the bees. The only math that matters is ensuring each district has

...as close to equal numbers of constituents in a given state between the number of House Representatives that that state is apportioned after the Census. The Electoral College is a separate though related matter and should frankly be done away with. However, I'm getting the sense you did not do the 5th mission, where it tried to demonstrate how the proposed Tanner Bill would forbid political affiliation (incumbents would scream about that one) or ethnic enclaves (certain neighborhoods do in fact have higher populations of any given ethnic group) from consideration when drawing districts, also requiring said districts to be as compact as possible (eliminating gerrymandering) and having as close to equal numbers of constituents as possible (equal representation per House Rep. in keeping with the spirit of democracy). That mission was all about dividing the population of a state into four equal, compact parts. You have no idea where the Democrats or Republicans live, no idea where Blacks and Whites live, nothing. Just raw population densities. A bipartisan committee would vote on the plan and then a court would decide whether it was acceptable. The committee could turn it down, but if the court finds it keeps to the letter of the law under the Tanner Bill, the plan would be accepted over the committee's objection. The governor of that state doesn't even get a say in the matter.

Small 'd' democratic reforms to redistricting would do away with a lot of the problems with the Electoral College, but obviously not as much as getting rid of it altogether. However, we can't do one and not the other if we're to fix the mess of both.

[JVS]

10. population is not uniformly distributed throughout any state

your method won't work

[napoleon_in_rags]

11. Very enlightening, Sherlock.

What I said is:

"Draw a vertical line down the middle of the state so half the population is on each side"

Not

"Draw a vertical line down the middle of the state so half the area is on each side"

If you'll look, you'll see that there is always some vertical line for any state that puts half the population on each side. Same with horizontal lines or any sub-partitions.

[JVS]

13. "Draw a vertical line down the middle..."

middle

That's where you went wrong. What you meant to write was "Draw a vertical line such that..." But that isn't what you did write.

[napoleon_in_rags]

14. middle: refering to the midpoint of the population density, not the area.

You understand what I was saying. Let's not squabble.

[jeff47]

19. States aren't uniformly populated.

Colorado: the vast majority of the state's population lives in Denver and Colorado Springs. The rest of the state is fairly empty. Your lines mean one district gets Denver, one district gets Colorado Springs, and the remaining 5 districts have very little of the population left.

New York: About 1/2 of the population lives in New York City - 8M of 19M in the state. You are proposing that one district hold half of the state's population.

North Carolina: Again, the population is concentrated into cities, most notably Charlotte and Raleigh, followed by Ashville, Winstom-Salem and Wilmington. The rest of the state is must less populated.

There is no moral argument whatsoever for the gerrymandering that is putting people who the popular electorate doesn't support into power

Well, you'd be doing just that. A whole lot of rural Republicans elected to represent "empty" parts of the country with a small number of people elected to represent the majority of the population...which is the exact opposite of what the House is supposed to be.

[DianaForRussFeingold]

3. Huh...I think that's a very good question

I don't understand this issue and asking questions is the only way to learn. Thank you..

[MinneapolisMatt]

4. Thanks!

It seems like there has got to be a simple(r) and less-nefarious way to do this hahaha!

[Egalitarian Thug]

12. We certainly could, that would be the rational and obvious solution (although it wouldn't be the

nation, it would be each state), but you are making the mistake of assuming that a solution is desired by those that have the power to implement any solutions to any problems. It took them a lot of time and work to create these easily surmounted problems and they're not about to allow some smart-ass to solve them with simple and rational solutions.

[krispos42]

21. We probably shouldn't even have districts.

Each party, in the primaries, assembles a slate of candidates in a particular order. I'll use Connecticut as an example; we have 5 congresscritters.


So in the primaries, the top 5 candidates for each party goes on the slate in order of popularity. In the general election, the voters vote for the party, not the individuals (although I would presume they are listed on the ballot).

If the a party wins between 10 and 30% of the vote, the top candidate on the list goes to Washington. Between 30 and 50 percent, the top 2 go. Between 50 and 70%, the top 3 go. Between 70 and 90%, 4 go. Over 90%, all 5 go.

It would get us a smattering of minor-party candidates as well, especially in populated states. For example, California has 53 congresscritters, so the brackets would be only a couple of percentage points wide.