Dr. Schelling's neighborhood
July 9, 2006 1:37 PM   Subscribe

Dr. Schelling's neighborhood. Is segregation the holdover of a racist past or an inevitable result of simple mathematical processes? After you've read the theory, try it for yourself here, here & here. Dr. Thomas Schelling won the 2005 Nobel Prize in Economics for developing these ideas, but not everybody agrees that he deserved to.
posted by scalefree (31 comments total) 9 users marked this as a favorite
 
mathematical segregation looks much different than racial segregation.
posted by j-urb at 2:12 PM on July 9, 2006


ooops.
posted by j-urb at 2:14 PM on July 9, 2006


Really good post. Here's some more about how his theories were used in vietnam (which is why some folks protested his Nobel). I assume everyone else is talking about headbutts and soccer. Philistines.
posted by Sparx at 2:14 PM on July 9, 2006


Very interesting.
posted by 517 at 2:19 PM on July 9, 2006


Some of Albert Einstein's theories were used in war too.

Truth is not inherently good or evil, and cannot be negotiated with, or protested against, or spun, and rarely can it be changed through petitions. Schelling's models establish that, given certain desires and certain conditions, automatons will behave in certain ways. This is clearly relevant to human behavior.

I for one would rather be governed according to what I and others actually are doing, rather than according to what some naif thinks that people are doing, according to an emotionally appealing religious/political ideology.

Ideologies, and other emotional desires, are for setting goals. For example, "We want people to be healthy" is an ideological statement. Given this desire, the way to achieve it is to be humble before the facts, and examine under what conditions people are and are not healthy, and promote the conditions that achieve the goal.
posted by aeschenkarnos at 2:59 PM on July 9, 2006 [1 favorite]


Excellent post!
posted by Spacelegoman at 3:28 PM on July 9, 2006




Thanks. I was running through some old bookmarks & came across the Atlantic article. Had to dig around to find the full text of it & then grabbed some simulation software to go along with it. I think it has significant implications for online communities & how they can tend towards negative patterns like extremism & mediocrity.
posted by scalefree at 4:00 PM on July 9, 2006


One thing that struck me about the "racism" model was this passage:

In the simulation I've just described, each agent seeks only two neighbors of its own color. That is, these "people" would all be perfectly happy in an integrated neighborhood, half red, half blue. If they were real, they might well swear that they valued diversity.

First, if every agent seeks to have at least two neighbors of its own color, it's not quite looking for a 50-50 neighborhood. When you include the agent's own color, the split is more on the lines of 40-60. And secondly, it's (mildly) specious to say that these agents "value diversity". Rather, they tolerate it; in the model, no agent actively seeks diverse surroundings.

But such nitpicking is not intended to detract from the interest of the results. Game theory is just amazing.
posted by matematichica at 4:51 PM on July 9, 2006


in the model, no agent actively seeks diverse surroundings.

matematichica, I was thinking the same thing when I read that, as I recently moved to a neighborhood where I am in the distinct minority, but prefer it that way. In fact, I would feel very uncomfortable living somewhere where everyone was the same and would leave if the neighborhood tipped that way.

While it is amazing that the results end up mimicking real life, is it possible that the rules for some simulations are chosen based on assumptions (that every individual wants to be with a certain number of their own kind) that are not necessarily true? Or that may hold true at one time but change at some point?

Or am I focusing on something not important?
posted by maggiemaggie at 5:39 PM on July 9, 2006


It only works when people have a desire to stick to 'their own kind', so racial segregation would only happen if people viewed being around other members of their race as a positive, rather then a neutral.

You don't see people segregated by hair color, or blood type, or any other genetic trait even when those traits have more effect on the genome? So why would skin color be different, unless there was an underlying sense of 'otherness' caused by skin color.

Without that underlying sense of otherness, no amount of mathematical mumbo jumbo would cause segregation.

(by the way, his mathematical models are very naïve, he's basically talking about simple thermodynamics applied to society. If we add 'heat' to cause things to re-randomize slightly and continuously that 'heat' can over come the 'desire' and keep things un-segregated, even if there were a slight desire to keep things segregated)
posted by delmoi at 5:59 PM on July 9, 2006


It's a shame all the agents in the programs online follow the same rules... if we changed it so, say, 10% of the agents needed two neighbors of different ethnicity, I wonder if racially diverse enclaves would spring up, sort of like cool bohemian areas with homogenous suburbs around them.
posted by Spacelegoman at 6:15 PM on July 9, 2006


Also, sort of off-topic but from one of the links:

This is a cool model of birds flocking.
posted by Spacelegoman at 6:17 PM on July 9, 2006


Some of Albert Einstein's theories were used in war too.

Not really, and Einstein wasn't sitting around shooting the shit with Truman trying to figure out exactly how much plutonium to use in order to make as many Japanese die from radiation poisoning as possible

I for one would rather be governed according to what I and others actually are doing, rather than according to what some naif thinks that people are doing, according to an emotionally appealing religious/political ideology.

Which is why the Vietnam War was such as smashing success. People behave in somewhat predictable ways, but what they certainly do not do is try to minimize their own suffering, especially when they're thrusting for revenge for Snelling's analytical terrorism.
posted by delmoi at 6:18 PM on July 9, 2006


not that this isn't interesting, but it doesn't take away the "racist" part - it's not an "inevitable result of mathematics" unless everyone starts out with a desire for a minimum racial representation. Why would they desire that? Do you seek a neighborhood where "at least two of my neighbors have brown hair"? Or "at least two of my neighbors are left-handed"?

The fact is that race is still an important cultural marker and source of identity in many cases, and most people are going to notice the racial composition of a neighborhood. It doesn't have to be that way - my martial arts school is very diverse, and I think part of the reason is that when new students come to check it out, they really only notice the technique & ability, and race just doesn't register as much (also, there is that "martial" structure so that there's less worry about "not fitting in"...). But when people look for a home, they seek a place where they feel like they belong - things like age, marriage status, parental status, income level, type of profession, etc, are all aspects that overtly come into play. Things like race, religion, culture, politics, are less explicitly sought, but people talk about the "feel" of a neighborhood, and they look for people like themselves.

I don't really see how we can claim it's not racist/segregationalist, although you can make the argument that it's an acceptable "separate but equal" scenario. but it's not "just math". It's math according to certain preferences of human beings. If everyone sought 'at least two blue and at least two red', you'd have a different outcome.
posted by mdn at 6:25 PM on July 9, 2006


not that this isn't interesting, but it doesn't take away the "racist" part - it's not an "inevitable result of mathematics" unless everyone starts out with a desire for a minimum racial representation.

Sure, but I think it's a stretch to claim that someone who doesn't want to stand out in his or her neighborhood is a racist. At the very least it's a different sort of racism than that which wants to keep "those people" out. I live in a neighborhood which is purported, by its residents at least, to be one of the most diverse in the US, and nobody seems to mind. But I experience a definite hostility when I venture into parts of Chicago where there are no residents of my ethnicity [1]. I don't think it's unreasonable of people to avoid that hostility by living in places where their ethnic group and socioeconomic class have some significant representation.

1. I experience the same thing in predominantly white, English-speaking neighborhoods where the residents are of a wildly different background than myself, which leads me to think that it's overly reductionist to make this about race.
posted by IshmaelGraves at 7:16 PM on July 9, 2006


Racist segregation is exclusionary. The dominant social group denies the submissive group access to its territory. It's discriminatory & overt & the submissive group has no choice in the matter. But even in the absense of that, self-segregation happens anyway without a conscious explicit rule in place. It's an emergent behavior without explicit boundaries established & enforced in advance of the result. That's what I meant by "just math", the emergent nature of it. But yes, a different set of preferences would result in a different set of boundaries, community sizes & shapes.
posted by scalefree at 7:27 PM on July 9, 2006


Sure, but I think it's a stretch to claim that someone who doesn't want to stand out in his or her neighborhood is a racist.

first, do you stand out as the only red head in the neighborhood? It's indicative of our racial presumptions and fears that we worry about standing out to begin with. I would have joined my martial arts school even if I had been the only white person there, because the program was what I wanted, and I liked the master. I did not worry about or even really notice the racial constitution of the student body. In that context it wasn't an important factor.

second, the preference is specifically for a minimum of group A but total indifference to group B. That means they are fine with total segregation and overtly opposed to total immersion. They will tolerate some mixing, but they are not seeking it at all. Hardly surprising what the result is...

this has been said for ages: the race problem in the modern age is a subtle kind of racism that most people do not consider themselves part of - but most of us are, just in the way we are uncomfortable when too many x are in the room. You can say it's not really racism, but that's exactly what the 'separate but equal' claim was. Everyone can ride the public transport, but it's not unreasonable not to want to "stand out" in your section of the bus, right?
posted by mdn at 8:22 PM on July 9, 2006


Very interesting FPP, scalefree.

I love applied chaos theory; I think it can be argued successfully that Schelling's models can be seen as simple special cases of more generally random, possibly repeating ones. His cases are interesting because they appear to self-organize from apparently near randomness, according to simple principles, but the observations tend be focused largely, if not entirely, on initial conditions, evolutionary schemes and rules, and end states where something can be said to have "happened." Perhaps many, many more systems of similar design could have been run where things didn't "evolve" to some clear cut end, but just meandered around awhile, and then fizzled out. It's hubris to claim you "understand" something, if you're just good at noticing interesting end states for problems with apparently simple bounds, and particularly if you tend to end "runs" when you have an apparently stable end situation to wrap up in neat theories.

That said, one other important difference these artificial societies have with real ones is that in real societies, we can expect alteration of policy and outcome if the society senses that actions aren't producing desired outcomes. Having a bad war? Takes a while for public sentiment to shift, but eventually, the lack of desirable results generates enough political and economic pressure to bring war to a halt, even though the society halting it could have, conceivably, continued to prosecute the conflict.
posted by paulsc at 8:34 PM on July 9, 2006


Neither an argument for or against Schelling's theory, but:

mdn : "first, do you stand out as the only red head in the neighborhood?"

Jeremy Decker ('the Redheaded Woodpecker') stood out as the only red head in my neighborhood. Or, at least, the only red head in my neighborhood who rode the bus.
posted by Bugbread at 8:50 PM on July 9, 2006


first, do you stand out as the only red head in the neighborhood?

No, because red-headedness is not culturally marked. If I felt that everyone else was sneering and shouting epithets at me because I was a redhead, then yes, I'd feel that I stood out and I'd probably think about moving.

It's indicative of our racial presumptions and fears that we worry about standing out to begin with.

Yes, it is. But again, I think this is about class more than race per se; it's due to the historical (and, of course, continuing) disenfranchisement of black people and of recent immigrants that class and race tend to fall along similar lines. (And I think you'd have a very difficult time finding any human society in which people didn't worry at all about standing out.)

second, the preference is specifically for a minimum of group A but total indifference to group B. That means they are fine with total segregation and overtly opposed to total immersion. They will tolerate some mixing, but they are not seeking it at all. Hardly surprising what the result is...

No, it isn't surprising, but I don't think this is an inherently immoral attitude. I have a hard time buying the idea that Joe Random has an ethical duty to maximize the diversity of his surroundings. I think that a tolerance for other kinds of people is, realistically, the best that you can hope for on a large scale in an inherently tribal species like humans.

this has been said for ages: the race problem in the modern age is a subtle kind of racism that most people do not consider themselves part of - but most of us are, just in the way we are uncomfortable when too many x are in the room. You can say it's not really racism, but that's exactly what the 'separate but equal' claim was. Everyone can ride the public transport, but it's not unreasonable not to want to "stand out" in your section of the bus, right?

You seem to be arguing that the difference is one of degree, not kind, but I think there's a definite qualitative distinction to be drawn between "I'm going to move here where there are other X-type people who share my basic cultural outlook" and "Y-type people must not be allowed to move here because they don't share my basic cultural outlook." Then again, perhaps you're right and it's a slippery slope. I think we'd both agree that in an ideal world there would be no racially-homogenous neighborhoods. I don't think this is an unrealistic goal, and in fact it seems inevitable that race and class will grow less closely correlated over time. I think it's less likely that we'll ever live in a society in which there is no segregation by class, though I think people would enjoy it more than they realize; the most interesting places I've lived have been those where you could find the local politicians and developers drinking beer with the local homeless guys at the local dive.
posted by IshmaelGraves at 9:14 PM on July 9, 2006


For the mathematically inclined, a paper by Epstein (PDF) that goes into a couple of the civil violence models. These models seem slightly more complex than the segregation/racism models we've been discussing. More light reading for a Sunday night.
posted by matematichica at 10:01 PM on July 9, 2006


mdn:

> first, do you stand out as the only red head in the neighborhood?

If you had lived in a time and place where Irish immigrants were culturally marked, you bet you would have, carrot top. That passed, this will too.


> this has been said for ages: the race problem in the modern age is a subtle kind of racism
> that most people do not consider themselves part of - but most of us are, just in the way
> we are uncomfortable when too many x are in the room.

That's not remotely pernicious enough to deserve the epithet "racism." That's just ordinary life.

People are concerned with their own social standing--rightly and realistically, because of everything else it affects including how much they earn, who they can marry, and how long they're likely to live. Inevitably people ask themselves "Does it raise or lower my own social standing to live next to this person X?" Let me know when you get people to stop worrying about their social standing, because I've got several other miracles I'd like you to pass.

posted by jfuller at 7:45 AM on July 10, 2006


Pardon me, but this does seem like a dead obvious result: that agents seeking self-similar agents will aggregate. How does this study reveal anything about racism? It's just a mathematical trick.

I think it's useful to talk about 'racism' as immoral at a societal scale, and 'prejudice' as immoral at a personal scale, and this study just confuses prejudice for racism. Is this the state of economics in 2002? Where is the economic theory? This is about as scientific as reading tea leaves.


I have a hard time buying the idea that Joe Random has an ethical duty to maximize the diversity of his surroundings.

The logic is that you can't be motionless on a moving train, and 'we' must take 'affirmative action' as a society to fight the legacy of racism, if 'we' think racism is bad. In the United States, none of us alive wrote the laws or policies that gave more power to "white" indentured servants, or settled on the compromise that chattel slaves would be counted as less than people for voting purposes, or fought in the Civil War or against Reconstruction. But we all have to live with the baggage. Ignoring history doesn't make it go away.

So if you don't take some action, you are acting immorally by saying you are fine with a racist system.

Peter Singer and others might have some equation for how much action, but it's beyond me at the moment.

I think our larger economic system is part of the problem, because neither central-planning nor market-planning systems are set up to give people information to weigh, quantitatively or qualitatively, the advantages and disadvantages of valuing anti-racist action versus other types of positive social action.

Market planning systems are actually pretty anti-social in general. There are a lot of problems; this book sums a history of the economic debate along with a solution (that i haven't read yet, and maybe i'll understand it in a few years) and this book looks at the problems from a philosophy of science perspective.

the language you are using--[to] "maximise [racial] diversity"--implies that we are limited to thinking that no other move towards racial diversity--like a 70% move--would be ethical, and we're stuck with choosing only 0 or 100%. I think that we need to consider that a 70% move could be ethical, but there's not really any way to determine what 70% would look like, bcos neoclassical economics doesn't take valuing (racial) diversity into account, and so we're stuck using vague or overbearing political solutions and weird economic patches, along with changing our social habits.

I think most of us do, in fact, value (racial) diversity, and so it is frustrating that we can't make individual economic decisions that reflect this.

Whether or not the economic science is this deficient on purpose is something we on metafilter will not agree upon, but that doesn't mean that we can't recognize past failures and move forward with good vision.
posted by eustatic at 8:25 AM on July 10, 2006


I just read this again and i realize I sound like an asshole. i should have read this guy's original work, i'm sorry.
posted by eustatic at 9:17 AM on July 10, 2006


eustatic: Pardon me, but this does seem like a dead obvious result: that agents seeking self-similar agents will aggregate. How does this study reveal anything about racism? It's just a mathematical trick.

I rather agree. I read this article, publishing the results of the civil relations models, and again, the author interprets pretty reasonable mathematical results (take away the "cops" and eventually people kill each other!) as though they were surprising and insightful comments on human behavior. There's more complexity to the models, and so maybe they're a decent first order model for people, but I kind of got the impression that the guy set up models that were designed to have outbreaks of violence, and then played them up.

Some of his findings: unrestricted random movement of agents results in occasional outbreaks of violence (you essentially get a critical mass of agents), a sudden drop in perceived "regime legitimacy" of a society causes a surge in violence not seen in a slow erosion of legitimacy.*

*I think this might have to do with the edge phenomena you get in approximating a step function with fourier series, but I'd be curious to see what actual applied math people have to say.
posted by matematichica at 9:33 AM on July 10, 2006


Let me know when you get people to stop worrying about their social standing, because I've got several other miracles I'd like you to pass.


Of course you can't get people to stop worrying about their social standing (barring evolutionary change), humans are social creatures; but you can enlarge people's ideas about what social standing is to produce a society that values diversity and understanding more than discrimination and ignorance (and oppression).

The humanities do an ok job of exploring what "social standing" can be; i just wish economics would listen to the humanities and try to work with them, and incorporate more of what people actually value into their models so that we have to tools to make those decisions more precisely and effectively.
posted by eustatic at 11:04 AM on July 10, 2006


*[to][the]
posted by eustatic at 11:06 AM on July 10, 2006


I wrote this to the nytimes guy

This simulation is weak and silly, even if you are only looking at "race" as a factor of two types. It only lets you pick how intolerant you are of other people, as opposed to how much you like them. An unprejudiced person would pick "zero." a slightly prejudiced person would pick "2." an aryan nation / NOI mofo would pick "4."

In a racist society (like ours), the underprivileged, prejudiced
actors might be "-2" or "-4," ("moving out of the ghetto"), but there is not even that choice in this model, and I imagine it would be inordinately simple to change the program to effect this change.

And what if my comfort level is determined by the diversity of types in my neighborhood? What if I seek novelty and balance? Then I do not exist in this game. this model assumes everyone is un-curious, prejudiced and intolerant. prejudiced and intolerant people self-segregate. you do not need a computer simulation to figure this out, and this is not news!

Worse, this is grist for hating on mathematics and simulations, which are beautiful things being made useless and ugly.

That everyone is intolerant may not be a bad assumption in a racist society; but it's far from interesting. One is made to wonder at the more interesting motives of such an uninteresting piece of work so prominently displayed.

posted by eustatic at 11:20 AM on July 10, 2006 [1 favorite]


eustatic wins!
posted by Mental Wimp at 4:43 PM on July 10, 2006


Thanks eustatic! I'd be curious to see a reponse if the nytimes guy gets back to you.

The worst part about this whole deal seems to be that it's more than just the reporter making excessive generalizations based on simple mathmatical models; some of it seems to already be in the interpretations by the economists/social scientists/what-have-you doing the research. I get the impression that they're trying really hard to make what's still a very basic sort of investigative technique seem extra relevent.
posted by matematichica at 6:04 PM on July 10, 2006


« Older C'mon, you knew it was going to happen...   |   Deadball Newer »


This thread has been archived and is closed to new comments