Finally: I have thought a lot about competitive districts and how they should be drawn, but I couldn't really eke a math paper out about it, and I'm not sure what else to do when you've thought about something for a while.
I know! You write a post on Unfogged.
I really like multi-member districts as an end-run around this kind of nonsense, but it's one of those things that will never happen (makes voting "too complicated", either makes districts larger or requires increasing the size of the House, people are too attached to having a specific individual as "their representative").
While I'd like to see much better representation of parties in our system, it feels weird to make that a core standard for judging implementation of a system crafted around individuals. Of course I'll take what I can get.
I'm trying to write something to improve awareness/understanding of how instant runoff (ranked choice) works here in the Bay and how it could be improved, but while I like it at the local level, for state and national I'm more and more inclined toward mixed-member / additional member system.
Pivoting way too fast to my current hobbyhorse but I claim OT.
I'm struck as I analyze ranked-choice ballot images how many ballots end up effectively discarded ("exhausted") because you can only rank 1-3, not all candidates, no matter how many are running.
Oakland, 2010 mayor (11 candidates): 11.3% of ballots exhausted
Oakland, 2014 mayor (16 candidates, about 7 plausibly viable): 23.5%
SF Supervisor District 10, 2010 (22 candidates, incredibly scattered vote): 46.8%
Those are extremes - the median percent exhausted (among the elections where choices need to be reallocated, e.g. no initial majority winner) is more like 8%. But the fact that you can get results like this at all seems like a drawback approaching the seriousness of the spoiler effect.
Wait, those are elections that are ranked choice but that only allow ranking three candidate? I haven't seen that before - all the ranked-choice voting I've done (Cambridge, MA) lets you rank all the candidates, even when there are 20 of them.
5: Yes, in exactly four cities around here (SF, Oakland, Berkeley, San Leandro). I think the idea was to make it feel quicker and easier ("pick your top three choices"), consistent with the rebranding to Instant Runoff.
Possibly also for cost reasons? The way our ballots presently work, all the candidates have to be listed three times so you can make your #1 pick from list 1, #2 pick from list 2, etc. Ranking everyone would take a lot more paper.
I can sort of see the "quicker and easier" appeal - ranking 11 candidates? That's a tall order. Picking three? That's much better.
Ideally, you'd want to be able to pick your three favorites and mark with a stigma those who you most loathe. Then the mushy middle would go unranked.
Ranking 20 candidates seems ridiculous. Is there any voter who can articulate a difference between nos. 8, 12, and 17?
The implementation is a grid ballot - one line per candidate, one row per rank, like this.
I agree with 9 and would like to suggest that making ballots more complicated is a big potential advantage for a whichever party might be more less vulnerable to issues with voter turnout.
Whoa, paper? We've got fancy voting machines that do all the work for you.
10 does work for just letting you rank however much you feel like. Are you allowed to rank your least favorites last and leave the middle unsorted?
It seems like in Australia you can (or used to be able to?) say "I allocate my #2 and lower choices according to the preferred order registered by X party". I like the idea of being able to defer in this way.
10: Our balloted - this doesn't show local offices, but they're formatted the same way, in that eight options take up a third of a page, and no grids available.
What about sizing/shaping districts by dividing a state into small squares/regular polygons and then assembling them to meet some sort of objective (even population size, economic diversity)? If people didn't like the arrangement, they could re-randomize a certain number of times.
You can't fix this without gaining control of the state legislatures. Why not just do that, but instead of fixing it, gerrymander the other way?
13: I imagine so, but that raises another of the issues, whether you're okay with lower-information / less engaged voters having less of an impact on the outcome than the ones who take the trouble to rank everyone, and don't mess up in the process. In the 2014 Oakland mayoral race, of the 9,608 ballots exhausted, 1,670 had only registered one choice out of the allowed three, and 3,118 had only two. Was this because they genuinely didn't care about any candidates other than the ones they picked, or because they were too tired to pick a third, or mistakenly thought registering only one would indicate stronger support? (A large proportion of ballots indicate the same candidate three times, and that proportion varies by candidate.)
16 is often the way mathematicians think of these things. Then each atomic level polygon can be represented as a vertex in a graph, and graphs are easy for computers to chew on.
whether you're okay with lower-information / less engaged voters having less of an impact on the outcome than the ones who take the trouble to rank everyone, and don't mess up in the process.
I'm picturing myself in a school board ballot or something where I'm uninformed - would I be ok just ranking two candidates, putting my last choice last, and then leaving it to more informed voters? Yes, in that context it seems like a fair decision I've made.
That is to say, it seems fair on an individual level. Whether it aggregates to something problematic, I'm not sure.
I'm not sure what else to do when you've thought about something for a while.
Write it up and either post and self-advertise, or submit to an editor who has an audience? Vox, maybe the Atlantic? Failing positive response, or if you'd prefer to avoid the hassle, maybe Medium and self-promotion/community promotion?
Your idea is clear and should get a wider audience I think.
Substantively, you don't consider tactical voting/vote exchange, it seems relevant. I saw a few references to tactical vote exchanging in the recent UK parliamentary votes, and laments that the platforms for doing this were unfair to lib dems.
Here is one site
Your idea is clear and should get a wider audience I think.
Thanks! This is mostly re-hashing what other people have done, though. I thought out a more technical way to draw districts to optimize competitivity, but there's no "proof" since optimization is subjective - just a short algorithm.
Also wildlife biologists.
OR! Get everyone to vote in their little polygons and then assemble the areas after the election based on how similar the votes were! You could even add a ranked list of issues if you want more dimensions. And areas don't have to touch. Gerrymander everything.
Perfect. I will def get this platform implemented.
I kind of agree with 20. If I voted for only one person, it'd be because (a) all the other people are tied for last place; (b) I don't care about any of the other people; or (c) I don't know anything about the other people. In the latter two cases, more informed people should make the decision, but in the first, it should be counted differently.
22.last. And yet the LibDems gained 4 seats (50% increase) in spite of a reduced overall vote. What would they have considered fair?
I don't know about the substance of the complaint. Just noting that it's a new dynamic that can affect voting outcome. Definitely polling data and staggered timing of primaries matters in the US.
Does anyone know of good reviews of whether "tactical voting" encouragement/data or vote exchange websites had an effect in the recent election?
Anyway, this isn't a math problem. Somebody needs to run out into the annoying parts of this and similar states, tell obvious lies to racists, and then go to the legislature or Congress and vote Democrat.
I'll even provide the first lie.
"I'm so happy to be here in X standing before real Americans."
I'm pretty convinced now that the problem is intractable and multi-member districts are the only partial solution. But even if we increased the size of the House to my preferred level (~600 reps), 3-member districts are 1.5 million people each and 5-member districts are 2.5 million. Those are big districts.
Adding: You probably want 5-member districts. 3-member districts will give 2-1 splits almost every time. 4-member districts give a lot of 2-2 ties. 5-member districts give a lot of 3-2 splits, but it is still possible for 4-1 splits or minor parties to win if they can manage 17% of the vote.
31. You might like to look at the experience of Ireland, which has a mix of 3, 4 and 5 seat districts, which are adjusted very frequently (six times in the last 20 years). But they still end up with several minor parties and independent members.
Like, multi-member districts where the top X candidates get seats? I think the UK had those historically, phased out over time with the last abolished in 1918. Not sure how they perceived them to work/not work.
I know Japan had them for a long time, where it provided incentives for minority parties to split ad infinitum. After they moved in 1994 to a combination of single-member plurality districts and PR, opposition parties started merging on the quick.
Along the lines of your polygons, I've often had the vague idea of using a similar process to redefine state boundaries.
1. Divide the country into small polygons, influenced by natural terrain boundaries, shooting for ~100000 people per polygon.
2. Let the polygons decide whether to join a state with their neighboring polygons.
3. When a group of polygons representing at least 0.5% of the US population decide to unite, then they can officially be a state.
4. Groups of polygons below 0.5% remain territories until they decide to unite.
Obviously it's unworkable.
If the plan is to toss us in with Idaho, the answer is 'no thanks.'
You don't have to put ranch dressing on your steak.
I understand your hesitation. But as a map guy, I would totally group the mountainous parts of Montana with Idaho. Or at least the parts west of the Divide.
3. When a group of polygons representing at least 0.5% of the US population decide to unite, then they can officially be a state.
My first thought was that this would result in ~200 states because everybody would want to maximize power in the Senate. My second thought is that 0.5% is ~1.6M people, NYC is 8+M. Either NYC becomes 5 states or large urban populations remain underrepresented in the Senate.
This is interesting, there's an element of randomness in the distribution of ranked ballots under the "Cincinnati rules." See "Transferring the surplus. Basically, all #1 votes are counted first, and any candidate with more than 1+ [(total votes)/(# open seats +1)] has the number of ballots above that number transferred to the #2 choice on the transferred ballot. If a ballot doesn't rank a #2 (or higher numbers in subsequent rounds) then it isn't transferred and a ballot that did have a preference is transferred instead. The selection of which ballots are transferred is based on two random factors, the ordering of precincts and an integer n which says transfer every nth surplus ballot.
I wonder if the random nature of that has ever been shown to matter- that a narrowly losing candidate can show with a different set of random choices they would have won- and whether that's grounds to redo the election. Probably not on the latter since many places say draw straws/flip coins in the event of a tie so randomness is accepted as a means of resolving close elections.
39. NY boroughs are counties, why not states?
39. NY boroughs are counties, why not states?
It would be weird for the Mayor of NYC to out-rank 5 governors and 10 Senators (or, to be more serious, more budgets and services are controlled at the state level and coordinating service across 5 states seems like a pain -- though perhaps a solvable problem).
38 So it was from March 3, 1863 to May 28, 1864. Or you can go back to March 2, 1853, if, instead of being ruled from Lewiston, you want us governed by Olympia.
You probably want 5-member districts. 3-member districts will give 2-1 splits almost every time. 4-member districts give a lot of 2-2 ties. 5-member districts give a lot of 3-2 splits, but it is still possible for 4-1 splits or minor parties to win if they can manage 17% of the vote.
I'm really amused that the logic in this quote is vaguely similar to my logic when I was ribbing Urple about expanding the gender-binary to five competitive levels here:
And are you picturing tiers that make it clear that we're not just cloaking traditional categories - like maybe 5 tiers of testosterone? 3 tiers would get read as M/F/Intersex. 4 tiers would get read as Big M, Little M, Little F, Big F.
IT MUST ALWAYS BE FIVE!
37 sounds intriguing. I'm going to have to try that. Maybe not *on* the steak, on the side, with the ketchup. I still haven't tried that, either, but I'll do it next time I eat out.
you want us governed by Olympia
Olympus.
43 I've mentioned before that I have a friend who tells a story of bribery (with fake gold nuggets) and other DC skulduggery involved in getting us split off in 1864.
39
I can't change the representation in the Senate, though I suppose I am already imagining a pony. But the point was that people would take into account the tradeoffs. There are benefits of scale for states and finding the balance between those benefits and the disadvantage in the Senate.
Thought experiment: if all states were allowed to unilaterally divide under the current Constitutional system, how many of them would? Would NYC divide into 5 states to take advantage of the extra Senate represenation? Would CA or TX?
Texas wouldn't. We are so heavily invested in our silhouette, I think people here think it's the shape of God. The shape is an obsession.
Looking at cities, it seems that those tradeoffs change with time. Cities chose to merge all the time in the pre-war period, including Brooklyn with NYC, but that all seemed to stop and so you have western cities with weird enclaves that refuse to merge (LA) or abnormally small land area for their size (SF).
49
That's what I mean. And TX and CA are the two whose splitting makes the most sense.
But as a map guy...
Interesting. As a map guy myself what kind of map guy if you don't mind my asking
Oh, sorry, didn't mean to mislead. Amateur only. But I find maps fascinating and I've sometimes thought about trying to learn how to use fancy GIS stuff. I could look at maps and draw fake borders for hours.
Cool, no worries. I came to maps a bit late myself but...I won't say much more here right now.
The shape is an obsession.
On the veldt, states used a large panhandle to signify reproductive fitness.
I'm a map guy too.
So Moby are you basically saying Texas got cucked when it gave up the Oklahoma panhandle?
It totally irritates me that Texans call the northwest square of Texas the "panhandle". Have they ever seen a pan? Here is a Texas-shaped skillet and they did not put the handle at the panhandle.
F, Charley, and Barry are actually the map guy from Dora.
They call the western part of Nebraska the panhandle too.
Anyway, Texas's need to make ridiculous claims about being messed with and being bigger than other states (except Alaska) obviously stems from their status as the donor in a metaphorical dong-transplant to Oklahoma. Yes.
12: Yeah, paper. In California, we vote by connecting the arrows on a cardstock ballot (like the sample ballot that Minivet linked) which then gets fed into a machine that automatically counts it, thus ensuring that there is a paper audit trail which matches what the voter designated in event of a recount. Works pretty well, and is highly tamper-resistant, unlike the fancy voting machines that either (1) produce a paper record that the voter has to (but usually won't) check manually, or (2) don't produce a paper audit trail at all, leaving you with no effective way to audit a suspicious result. The only real problem is the high number of absentee/vote by mail ballots that turn out to not be automatically readable due to coffee stains or whatever.
but there's no "proof" since optimization is subjective - just a short algorithm.
So, it's not a math paper. Maybe it's some other kind of paper.
(Does it use the word "Voronoi"? On the basis of thirty seconds of thinking about it, I feel like anti-gerrymandering algorithms could use the word "Voronoi.")
I can't change the representation in the Senate, though I suppose I am already imagining a pony.
As long as we're talking ponies, let's abolish the Senate.
62: 12 was a joke. My whole state is a joke.
I should go see San Antonio and Austin before global warming turns that part of the state in the Sahara 2.0.
I've seen Dallas and Houston. They kind of sucked, IIRC.
(Does it use the word "Voronoi"? On the basis of thirty seconds of thinking about it, I feel like anti-gerrymandering algorithms could use the word "Voronoi.")
It doesn't, but Voronoi cells did come up in a brief conversation I had about this with another math person.
Also I'm scared to write non-math papers. I need a collaborator.
Outside of math, they say "accomplice".
Inside of math, you have to use $ $.
65. That's what Santa Anna said.
Without looking it up, I think Oklahoma has the one true panhandle, and Texas uses the name to cover their shame. Idaho and Nebraska play along.
Maryland is okay on the handle, but the rest of the pan could use some work.
Idaho has a perfectly panhandle. It's just that they already dropped the eggs on the floor.
I stand by the Florida panhandle. It's in profile.
To Florida, it's a panhandle. To America, it's a taint.
||
Is there any consensus these days how quickly to contact someone following a first date?
I had a nice one last night - someone from my grad program I hadn't seen in years, so more common ground than average. Hard to tell if my feelings were reciprocated, I always spend too much time talking about myself, but I live in hope.
|>
I don't hold with dating people who went to graduate school. They're way too emotionally invested in the life of the mind to be useful.
I have no idea what the rules are, but if you wanted to communicate enthusiasm, a reasonable hour in the morning the next day seems as if it'd work. For trying to communicate a precisely calibrated balance between enthusiasm and complete self-sufficiency, I can't help you.
Trying not to appear emotionally needy just makes you look emotionally needy.
My other dream plan for states is to use Voronoi using cities as loci and travel time instead of physical distance as criterion. IOW, states are defined by whatever large city is the shortest travel time away.
The only hard and fast rule is to wait until she's 18 before you send pictures of your genitals if you're married to somebody working for a presidential candidate.
I wouldn't describe that sort of rule as either hard or fast.
76: My rule is always text a good morning the next day. Let her know you enjoyed hanging out, wish her a good day. That's it. She'll either respond with something that invites further texting or she won't. Neither one means anything.
My approach to dating is not to try to play it cool and barely interested. If I'm interested I tend to just say it. I can't do smooth, so I try to do unambiguous.
I can't do smooth, so I try to do unambiguous.
Keep that in mind in case someone ever says "I like my men like I like my peanut butter."
I wonder if the random nature of that has ever been shown to matter- that a narrowly losing candidate can show with a different set of random choices they would have won- and whether that's grounds to redo the election. Probably not on the latter since many places say draw straws/flip coins in the event of a tie so randomness is accepted as a means of resolving close elections.
I don't see why it would matter to say that if people voted differently someone else would have won. However, it would look worse for the system if you could tie a change in outcome in the opposite direction from a change in votes. For example, could you construct an alternative scenario where a small number of supporters of losing candidate X either don't vote, or rank X further down, but with the result that candidate X actually wins. (Keyword here for further reading is "monotonicity".)
Southeast Alaska is sometimes called a panhandle, but not often.
The problem with most panhandles isn't that they aren't handles, but that the remainders aren't pans. OK being the only exception, I guess. I could buy pothandle for FL, WV, MD maybe.
I'm learning about the Salisbury Convention in the U.K. I guess the House of Lords used to favor bills covered in flour, gravy, and mushrooms.
On topic, because of complicated voting scheme.
I don't see why it would matter to say that if people voted differently someone else would have won.
That's not what I'm describing. Same sets of ballots cast leads to different outcomes by chance at the counting stage. Let's simplify and say 2 open seats, 3 candidates, 10 voters. So require 4 first place votes to win a seat which is next integer greater than votes/(seats+1)- not sure how they handle fractions there, I think it's just the next highest whole number even though it says 1+[votes/(seats+1)].
Anyway, assuming 4, say candidate A gets 6 first place votes, candidates B and C each get two. So four of candidate A's ballots are removed, and two redistributed to second choice. Which four and which two are randomly chosen. So if three of A's voters had B second and three had C second, depending on which two get redistributed to second choice you can imagine all three scenarios where either B wins, C wins, or B and C tie, and which scenario obtains is based on a randomly chosen factor.
Oh, I see. No multimember districts in the Bay where I've been analyzing.
||
So I googled "icloud app specific password" and set thought I had set it up but I still can't get iCloud mail working with my Android phone again. This has been my main email since the first .mac days. Any idea what's going on? I can't find current answers in any forums or help pages.
|>
Reiterated thanks for the advice above. Very satisfactory initial response received.
97 Glad to hear that.
Now if I can just get my email working again I send her a quick note to vouch for you.
Mixed-member proportional representation seems like a great system to me. You vote for an individual candidate in your (single-member) district, and also a party (on a national list). The winning district candidates win seats, and then a bunch of extra seats are allocated on the basis of the party-list vote shares, so that the total seats are as close to proportional as possible (subject to various things like minimum party threshold, etc.).
Downsides: it would require either a large (and probably variable) expansion of the House, in order to be able to guarantee proportionality, or extreme disparities in district size between e.g. Wyoming and big-state districts (although Wyoming's already almost 30% smaller than average). Neither seems like a big deal to me.
Combine that with replacing the senate with a randomly-selected deliberative assembly, and we're good to go.
Of course, nothing will ever change and America is doomed. Oh well.
97: Be careful. Women who respond satisfactorily can be kind of clingy.
100. It works fine in Germany, which is like the US insofar as it's a federal country whose component states have strong historical identities. It would take some getting used to, though, especially the variability implicit in "overhang seats".
You would have to increase the size of the House considerably. Districts are already unwieldy in many places; halving the number of constituency Representatives to allow for the Additional members would make it insane, plus what do you do about the States that have a single member at large?
104: I think it or something quite close to it also applies in trapnel's current country of residence.