Re: The Parable of the Polygons

1

Is this an analogy?


Posted by: Moby Hick | Link to this comment | 12- 8-14 9:37 PM
horizontal rule
2

D'oh, it's a parabola.


Posted by: JP Stormcrow | Link to this comment | 12- 8-14 9:39 PM
horizontal rule
3

I had a programming assignment freshman year that was basically this--a matrix of colors with probabilities of bias and a function describing how to reshuffle. Iterating over it and watching it evolve, it was amazing watching how even with low probabilities, people are total dicks.


Posted by: dalriata | Link to this comment | 12- 8-14 9:39 PM
horizontal rule
4

There's a very high probability that people are total dicks.


Posted by: Moby Hick | Link to this comment | 12- 8-14 9:40 PM
horizontal rule
5

2: I think there's a bifurcation point where segregation doesn't occur, but it's surprisingly low.


Posted by: dalriata | Link to this comment | 12- 8-14 9:41 PM
horizontal rule
6

What if you have more than two races and lots that aren't perfectly square?


Posted by: Moby Hick | Link to this comment | 12- 8-14 9:44 PM
horizontal rule
7

6: She sounds pretty hot, bro.


Posted by: Flippanter | Link to this comment | 12- 8-14 9:49 PM
horizontal rule
8

I did it with somewhere around six or ten colors, and with a big enough grid that the squareness wasn't as apparent (although I guess that still matters on the micro level). You'd still get segregation, but the neighborhoods would be smaller and more interestingly shaped. That's probably paramterizable, but I recall running into the limit of what Matlab could do on the lab machine I was working on.


Posted by: dalriata | Link to this comment | 12- 8-14 9:50 PM
horizontal rule
9

Thomas Schelling's Micromotives and Macrobehavior is short and pretty readable on this kind of toy model.


Posted by: essear | Link to this comment | 12- 8-14 10:00 PM
horizontal rule
10

I already read one of his books (The Strategy of Conflict) and I think I shouldn't read another until I've read at least one book from everybody else who has written one.


Posted by: Moby Hick | Link to this comment | 12- 8-14 10:02 PM
horizontal rule
11

Vi is the best. It's great to see how much success she's having doing awesome interesting things.


Posted by: Unfoggetarian: "Pause endlessly, then go in" | Link to this comment | 12- 8-14 10:05 PM
horizontal rule
12

I don't even see shapes.


Posted by: fake accent | Link to this comment | 12- 8-14 10:24 PM
horizontal rule
13

My neighborhood is 85% Latino, mixed new immigrant and longtime U.S. born, but my street is pretty diverse. There's a white normy couple two doors up who I don't know except to say hi to and a bohemian pair five doors up who we're friendly with (I saw the woman perform in a puppet show years ago and we have a bunch of friends in common). When we were looking at the house, the Mexican next-door neighbor came out to tell us that there were a number of Chinese people on the block, as if to say, don't worry, it's not all Mexicans.


Posted by: k-sky | Link to this comment | 12- 8-14 10:29 PM
horizontal rule
14

I'm torn about this sort of thing. On the one hand, the Schelling model is absolutely beautiful as a mathematical parable. On the other hand, as an explanation of how actually-segregated societies like the US got to be that way, it's a piece of ludicrous white-washing, and I use the word advisedly. (To be fair, I think from reading Schelling that he realized it had nothing to do with what had really happened.) On the third hand, maybe "demand diversity near you" is more likely to have a beneficial effect than "undo the legacy of centuries of violent oppression".
(Mostly I'm kicking myself for not assigning the model as a final project in my programming class.)


Posted by: Cosma Shalizi | Link to this comment | 12- 8-14 11:06 PM
horizontal rule
15

normy and bohemian are a good example of measuring `like us' on more than one axis. Which could lead to micro-segmentation (hello PRISM) or to being able to look around and think `I have something in common with each of my neighbors, we'll be OK.'

(But Love and I had the wit to win; we drew a Venn diagram and confused the issue entirely.)


Posted by: clew | Link to this comment | 12- 8-14 11:15 PM
horizontal rule
16

The parallel to race is obvious, but this can also model the liberal/conservative split.


Posted by: Spike | Link to this comment | 12- 9-14 4:29 AM
horizontal rule
17

14. This. Segregation doesn't happen only because of individual homeowner actions; it's collective via all sorts of social actions that encourage or even mandate it. Covenants and red-lining, lending practices, and so on.

The model is too simplistic, anyway. People don't only move because they are unhappy with the racial makeup of their neighborhood. They change jobs, move to larger or smaller houses/apartments as their incomes and life-needs change, etc., even if they are happy or "meh" about where they are currently living.

If I had more spare time I'd write that simulation and see what happens. It might not be any "better" but it would be less of a (cough) whitewash.

Also, FWIW I liked "The Strategy of Conflict." Not that l've re-read since the asteroid struck my fellow dinosaurs.


Posted by: DaveLMA | Link to this comment | 12- 9-14 8:01 AM
horizontal rule
18

14: I would go further. The Schelling model is completely worthless, and the fact that it has a certain amount of elegance makes it worse because it lets people write pseudo-serious thumbsuckers about how mathematics proves that segregation isn't anybody's fault. I can only assume from first principles that it featured in a cover story in the old TNR.


Posted by: Walt Someguy | Link to this comment | 12- 9-14 8:06 AM
horizontal rule
19

I disagree that it's worthless. I mean, I agree with the white-washing, but it's a good illustration of how individual good intentions alone isn't sufficient to solve societal problems.


Posted by: heebie-geebie | Link to this comment | 12- 9-14 8:10 AM
horizontal rule
20

people write pseudo-serious thumbsuckers about how mathematics proves that segregation isn't anybody's fault

I absolutely agree that that this happens, but c'mon. It's not the model's fault. And my own personal experience with the model has been it being used to demonstrate that even if people individually prefer some diversity, the interactions of stuff can wind up resulting in segregation anyway, so you have to be careful not just to have spotless individual thoughts and aspirations but also to think carefully about your place in a larger society. And how to adjust your personal behavior in view of (say) the whole neighborhood, rather than just your immediate vicinity. Or develop a super strong preference for diversity or something.

My dad made me read about it when I was in high school, at which age it was a very interesting and effective provoker of thought.


Posted by: E. Messily | Link to this comment | 12- 9-14 8:15 AM
horizontal rule
21

This kind of reminds me of the cafeteria in middle school, at a school that was 40/30/20/10 Black/White/Asian/Hispanic. The groups more or less got along, but lunch tables were pretty highly segregated. That's an example that iterates at a far higher frequency than housing, and which is less effected by outside policies.


Posted by: Spike | Link to this comment | 12- 9-14 9:22 AM
horizontal rule
22

19. But it artificially restricts the range of cases where their good intentions come into play. That's part of the problem, other than 14's takedown.

21. I know it's just an anecdote but I wonder if the social class segregation was important layered on top of racial, where you say "pretty highly segregated." My memory of middle school is that there were racial divisions, but also "class" divisions, such as jocks (of all races) sitting together, and nerds sitting together, and for that matter, boys sitting together and girls sitting together.


Posted by: DaveLMA | Link to this comment | 12- 9-14 9:37 AM
horizontal rule
23

It's not the model's fault.

No, but the model didn't somehow turn animate and put itself up online with a subtitle about "the shape of society".


Posted by: Minivet | Link to this comment | 12- 9-14 9:53 AM
horizontal rule
24

My memory of middle school is that there were racial divisions, but also "class" divisions, such as jocks (of all races) sitting together, and nerds sitting together, and for that matter, boys sitting together and girls sitting together.

We didn't get a choice about seating; everyone sat down at one long table per class (as in, year, not as in upper/middle/working). It never occurred to me before how much grief this saved us all.


Posted by: ajay | Link to this comment | 12- 9-14 10:15 AM
horizontal rule
25

There was a meme a while ago about how privilege is like playing a video game on an easy setting. It's be interesting to try to make a semi-realistic family wealth building game with different settings. A sim inheritance. Start in 1800 and make decisions to try to maximize wealth in 2000.


Posted by: Unfoggetarian: "Pause endlessly, then go in." (9) | Link to this comment | 12- 9-14 10:22 AM
horizontal rule
26

I played something a bit like that in text-only format on a BBC Micro in primary school. It was called "Making Ends Meet" and simulated the life of a working-class family in 1900 Britain. You had to make decisions about how much to save, whether to contribute to the co-op etc, and every now and again it would randomly chuck a family health emergency or something at you and either you would spend down your entire savings, or you'd die of scarlet fever, or both.

No doubt it was really part of the covert pro-NHS indoctrination scheme.


Posted by: ajay | Link to this comment | 12- 9-14 10:34 AM
horizontal rule
27

24. The Middle Ages just haven't ended there yet, have they?

25. I have a friend whose ancestors owned about five blocks of Manhattan. If only they hadn't sold it.


Posted by: DaveLMA | Link to this comment | 12- 9-14 11:00 AM
horizontal rule
28

26: It's like Oregon Trail, but more boring.


Posted by: Moby Hick | Link to this comment | 12- 9-14 11:05 AM
horizontal rule
29

Right. We in the self-reliant US were taught that whatever health problems bedeviled you, you could always go out and shoot bears.


Posted by: Eggplant | Link to this comment | 12- 9-14 11:07 AM
horizontal rule
30

endorsing 19 & 20; a lot of people seem to think that removing outright pernicious laws and having most people mean well is all it will take for society to fix itself up. But -- adding 25 -- if our little shapes had wealth-endowments, and wealth was slightly spatially contagious (as it is! Propitty Valoos! Don't marry tha for funny, but... ), one could see how much effect resilient segregation had on wealth-balancing.

Which might be just an illustration of what Coates and historians have been telling us over and over and over, but apparently it does need to be told over and over and over.


Posted by: clew | Link to this comment | 12- 9-14 1:18 PM
horizontal rule
31

I think the (pardon the phrase) tipping point for me, on the Schelling model, was reading Easley and Kleinberg's Networks, Crowds, and Markets (which is a good book!), and seeing them not just explain the model (fine), but offer racial segregation in Chicago (!) as an empirical illustration. In doing so, they were following a paper by some economists which actually tried to fit the model to data from Chicago 1940--1960, naturally without so much as mentioning things like redlining, official policy, threats of violence, etc., etc. The fact that it didn't occur to Easley and Kleinberg to remember recent history and realize that the economists' paper was crazy points, I think, to the way the model misleads even smart and well-informed people, and, through them, many others.


Posted by: Cosma Shalizi | Link to this comment | 12- 9-14 2:24 PM
horizontal rule
32

the model misleads even smart and well-informed people

I'd really rather just blame the people. And I don't think I would call people "smart and well-informed" if they can't figure out the difference between a mathematical model and real-life Chicago. (Or remember history, and that economists are crazy.)

I feel like this happens on a pretty regular basis. Somebody comes up with some sort of model or analogy or hypothetical scenario that makes some sort of useful point or other, and then somebody else (probably an economist!) comes along and uses it for nefarious purposes, and then everyone's all "BURN DOWN THE MODELS! BAN ALL THE ANALOGIES!" That's a silly response. Let's just ban poor interpretations and misleading, racist misuses of those things instead.


Posted by: E. Messily | Link to this comment | 12- 9-14 2:33 PM
horizontal rule
33

I like models.


Posted by: Moby Hick | Link to this comment | 12- 9-14 2:38 PM
horizontal rule
34

23: No, but and the model didn't somehow turn animate and put itself up online with a subtitle about "the shape of society".

Exactly.


Posted by: E. Messily | Link to this comment | 12- 9-14 2:48 PM
horizontal rule
35

32: Right, and I think the occasional demonization of formal Game Theory is another example.


Posted by: JP Stormcrow | Link to this comment | 12-10-14 5:54 AM
horizontal rule
36

25: There was a meme a while ago about how privilege is like playing a video game on an easy setting.

For a completely abstract version of this, I offer the the DIVE variants that I've mentioned here. The one really is "kinder and gentler" in that the random draws are morel likely to be useful in the one version compared to the other, but during game play it seems the same--you're just "luckier" in the one. The typical score is at least 2x in the kinder, gentler one. So my evil experiment is to give one half of a motivated group of folks one version and half the other version and see the myths of competence that develop.


Posted by: JP Stormcrow | Link to this comment | 12-10-14 6:10 AM
horizontal rule