mancalablog

posts to days old
of topics
with text
post

a more perfect union

m (28 Jan 2010 9:36): I like Gail Collins.
m (28 Jan 2010 9:39): and wordle!
m (28 Jan 2010 9:51): nice pdf from the House Committee on Energy and Commerce on the differences between the House and the Senate HCR bills.
Paul (17 Dec 2010 16:50): Does anyone get the sneaky suspicion that Obama actually works better with Republicans than with Democrats? Is it easier to negotiate with opposition than with allies?
Paul (4 Sep 2011 6:09): When computers talk to each other conversationally
m (4 Sep 2011 7:39): I see you've got your Flash Plugin working!
Paul (4 Sep 2011 8:34): Yes, Chrome for linux "does flash" out of the box. Not quite sure how it works...
Paul (7 Sep 2011 5:35): I'm so "in the know": I linked the cleverbot-cleverbot chat before it made it to xkcd
G (13 Jan 2012 14:01): ...Texas abortion law that requires pregnant women to listen to the fetal heartbeat... :(
In before "Texas abortion law that requires pregnant women to shed at least 10 tears," "Texas abortion law that requires physicians to play a sound clip of babies screaming..."
Paul (6 Nov 2012 21:51): So... looks like the Presidency and Senate are in good shape. But what's up with the house? Looks like it's all fucked up by gerrymandering. How can we quantify how unfair the house results are?

* Maybe, find the total popular vote for Democratic house candidates versus Republican house candidates, and compare it with the percentage of seats won...

* Maybe, plot a histogram of the margin of victory in each house district. An unfair result has democrats winning by wide margins and republicans winning by narrow margins, with a gap in the middle (ie, not many democrats winning by narrow margins)...

Are any of these feasible? Is anyone else asking this?
m (7 Nov 2012 19:33): (from a FiveThirtyEight interview last year with David Wasserman about redistricting)
DW: All districts have to be contiguous. Compactness is a different matter all together. Mathematicians and geometrists have attempted for years to come up with a uniform way to calculate compactness. But there is really no one-size-fits-all solution to evaluating whether a map is compact or not. If there were, courts would probably be drawn to it.

Don't you think there is, though? Wouldn't it be relatively straightforward to construct some simple geometric rules for splitting states up? Maybe the rules themselves could be subject to gerrymandering, though. Hm.
m (19 Nov 2012 10:34): Wow! Talk about gerrymandering: "Democrats led Republicans by 56 million to 55 million votes nationally" [cumulative over all House races]
Paul (20 Nov 2012 5:37): yeah. Perfect gerrymanderers can keep 1/2 + epsilon of the seats as long as they keep 1/4 + delta of the voters. Specifically, put 1/2 - delta of the voters into seats that are 100% democrat, and the other 1/2 + delta of the popular vote into seats that are ((1/4+delta)/(1/2+delta))% republican. So what lower bounds can we put on delta? (If delta is 1/4, we're back to normal democracy.) Delta is mildly lower-bounded by state borders, somewhat more lower bounded by district compactness requirements, and substantially lower bounded by states like California where a nonpartisan commission draws the districts...
Paul (20 Nov 2012 5:48): oh no! some of my epsilons and deltas are wrong, and I'm late for work. correction forthcoming!!!
Paul (20 Nov 2012 19:54): Ok, here it is. In order for republicans to win 1/2 + epsilon of the seats, they only need to get 1/4 + delta of the votes, where delta > (epsilon/2), provided they gerrymander the districts. Specifically, draw 1/2 - epsilon of the districts to have 100% democrat votes, and draw the remaining 1/2 + epsilon of the districts to evenly distribute the remaining votes which are ((1/4+delta) / (1/2+epsilon)) > 50% republican.


arrange it so that 1/2 - epsilon of the votes are 100% democratic for 1/2 - epsilon of the seats, and 1/2 + epsilon of the votes are evenly distributed (. of yeah. Perfect gerrymanderers can keep 1/2 + epsilon of the seats as long as they keep 1/4 + delta of the voters. Specifically, put 1/2 - delta of the voters into seats that are 100% democrat, and the other 1/2 + delta of the popular vote into seats that are ((1/4+delta)/(1/2+delta))% republican. So what lower bounds can we put on delta? (If delta is 1/4, we're back to normal democracy.) Delta is mildly lower-bounded by state borders, somewhat more lower bounded by district compactness requirements, and substantially lower bounded by states like California where a nonpartisan commission draws the districts...voters are selected to yeah. Perfect gerrymanderers can keep 1/2 + epsilon of the seats as long as they keep 1/4 + delta of the voters. Specifically, put 1/2 - delta of the voters into seats that are 100% democrat, and the other 1/2 + delta of the popular vote into seats that are ((1/4+delta)/(1/2+delta))% republican. So what lower bounds can we put on delta? (If delta is 1/4, we're back to normal democracy.) Delta is mildly lower-bounded by state borders, somewhat more lower bounded by district compactness requirements, and substantially lower bounded by states like California where a nonpartisan commission draws the districts...
Paul (20 Nov 2012 19:55): And just ignore the garbage after the first paragraph. It was late and I was tired.
G (28 Nov 2012 8:29): I think that there is probably not any good rule for mechanically determining if a district guideline is reasonable or not. Simply because you could still have compact regions that are very clearly designed to favor your side, but also because maybe having a long district on a coastline seems like it might not be so bad, and stuff like that. Or I guess what I meant is that compactness, while good, doesn't really prevent abuse.
m (15 Dec 2012 16:39): No wonder gerrymandering makes you grumpy, Dex. Is/was that your district?

What about just the requirement that districts be convex? It's certainly simple, and though Grant's right that it can (will) still be abused, at least we'll avoid ludicrously, blatantly abusive shapes like this one.
G (16 Dec 2012 14:35): The grid nature of streets makes a convex requirement untenable.
m (16 Dec 2012 22:22): Good point, but there's no reason I can think of that district boundaries would need to align with streets.
m (16 Dec 2012 22:24): But the initial point still stands: they'd just end up being long skinny rectangles of cheating instead of wiggly cheater-fractals.
m (30 Dec 2012 17:41): This article makes me think that they're doing it wrong: 242 districts in which the result was more than 20 percentage points off of the presidential result (117 of those going Dem, 125 for Republicans).
m (30 Dec 2012 17:49): but here's a thought:
Meanwhile, the differences between the parties have become so strong, and so sharply split across geographic lines, that voters may see their choice of where to live as partly reflecting a political decision. This type of voter self-sorting may contribute more to the increased polarization of Congressional districts than redistricting itself.
G (31 Dec 2012 23:49): Requiring convex districts means that either you'll need to make perfectly rectangular districts in perfect grid cities, or you have to deal with deciding which houses or which rooms are on which side of which lines. The stakes are lower for every decision, but there are a lot more decisions that have to be made, this way.

If we have weird district lines that don't align by streets even now, then I guess I think we oughta do that!

I think the blue-skies perfect-world solution would be to have non-partisan groups doing the redistricting. Maybe you have to lock them in a room with no internet until the lines are all drawn, and they can't check how the territories voted?
G (31 Dec 2012 23:51): Also we have states with concavities along their borders. Their state-borders. So you'd have to at least add exceptions for that. So take that!
Paul (1 Jan 2013 21:16): Center for Range Voting always has something interesting to say. Did you realize that the whole concept of specially-drawn districts is specified only by law, not by constitution? "In some sense the ultimate proportional representation system is simply not to have any votes and voters at all; the legislature is simply selected randomly!"
m (2 Jan 2013 10:37): [continuing from paul's quote]: "This practice was introduced in Ancient Athens where a machine was invented for generating the randomness."
m (2 Jan 2013 20:44): here's a new one (to me)
m (2 Jan 2013 20:53): The very first proposed amendment (also new to me) actually tries to do this (districts max out at 50,000 residents) and fails due to a scrivener's error(?!)
But failed amendments are still live(?!) so if it had been properly worded we would be able to pick it up for ratification today (I think)
G (5 Mar 2013 23:28): In order to form a more perfect union, it is imperative that we change the national anthem to R. Kelly's 2003 hit "Ignition (Remix)."
G (28 Apr 2013 14:46): America’s blacks voted at a higher rate than other minority groups in 2012 and by most measures surpassed the white turnout for the first time, reflecting a deeply polarized presidential election in which blacks strongly supported Barack Obama while many whites stayed home.
G (1 May 2013 22:14): ...
m (5 Sep 2015 12:00): Bachelor party
LAN party
Party all the time
m (7 Sep 2015 7:36): Good party guys.
G (7 Sep 2015 13:59): That's my line!
Thanks very much, I had a great time.
m (18 Sep 2015 13:29): trolling mancala for any good old quotes that might be nice to break out tomorrow
z (19 Sep 2015 23:55): Congrats G!
Paul (12 Oct 2016 16:06): My deep reflection on visiting Waffle House in Atlanta: If the North won the civil war, how come only the South got the Waffle House chain?
m (13 Oct 2016 9:04): Lasting grudge against Sherman for burning down all initial franchise locations?
m (13 Oct 2016 9:04): Sic semper tyrannis?
R (1 Mar 2017 16:32): Barack Obama Was Kicked Off Of A White House Tour For Shouting, ‘I Already Know That,’ After Everything The Tour Guide Said
R (1 Mar 2017 16:35): So, so dumb, but it got me.
sam (2 Mar 2017 15:54): picture says it all