6 ms·

I watched once a lecture on Coursera about Schelling's model of segregation, which I found extremely insightful. If you see maps like that with strong segregat

by WA 13y ago

I watched once a lecture on Coursera about Schelling's model of segregation, which I found extremely insightful. If you see maps like that with strong segregation, one of the first attempts to explain it is "racism". However, if you divide your neighborhood in 9 blocks such that you are in the center, you have 8 neighbors. Schelling's model now shows that even in cities where most people say:"I'd like to live next to only 35-40% of people who are similar to me (whatever that means exactly, could be ethnical background, skin color, income, ...) and I'd accept 60% of neighbors who are different from me", you would still get a segregation of 80-90% in the city. So, the reason for strong segregation is not necessarily that people are racists. People can be very tolerant, but if you try to fulfill the minimum requirement (40% like me), you end up with strong segregation patterns just because there's no other way to fulfill the minimum requirement for most people. Note: Numbers aren't exact, but pretty close. Edit: Link to lecture: https://class.coursera.org/modelthinking-003/lecture/16 https://class.coursera.org/modelthinking-003/lecture/16

acchow 13y ago
Any chance you have a direct link to the lecture without having to register for a course?
- WA 13y ago
  This is it: https://class.coursera.org/modelthinking-003/lecture/16 https://class.coursera.org/modelthinking-003/lecture/16
twic 13y ago
Reminds me of a similar thing in developmental biology, where if you have two populations of cells, each having a slighly higher affinity for other cells of its own type, then they will very efficiently sort themselves into separate compartments. This is thought to be a pretty important mechanism in the formation and maintenance of structures in the body. In case anyone's interested, here is a nice (if now rather old, although to be honest scientific writing was so much better back in the '90s) review on the whole business: http://www.sciencedirect.com/science/article/pii/S0092867400812970 http://www.sciencedirect.com/science/article/pii/S0092867400... In which we find a paragraph describing this process: "The selector genes do more than specify the pattern and the structures that the compartments will eventually make—they also specify, indirectly, a surface property. This property has been termed cell affinity, meaning that cells that share the same affinity, owing to the same binary code of selector genes, will intermingle during growth. There are a number of different experiments that lead to this conclusion, but perhaps the simplest is the observation that when the selector gene engrailed is removed, in vivo, from a posterior clone of cells in the wing, those cells gain anterior affinity: they now sort out from posterior cells and, if they are in contact with anterior cells, will sort into and mingle with them. Cells from neighboring compartments will have different affinities and tend to minimize their mutual contact, so that where the two compartments abut, there is a relatively straight line across which the cells do not stray."
- chrischen 13y ago
  Is true in physics as well. That's how stars are formed... Gravity is extremely weak, and so the affinity for one atom and another is extremely small, but after millions of years they still cluster together so hard sparks fly.
- jtfairbank 13y ago
  Why do you say that scientific writing was better in the '90s? I'm just curious, not challenging your claim.
  - twic 13y ago
    Back when i used to read a lot of papers, i found that i enjoyed the ones from the 90s more. I think the further you go back in time, the more honest and open scientists were in their writing; as time has gone on and academia has become more crowded and competitive, the more papers read like marketing. Authors writing in the 80s would happily admit that there were things they didn't know or hadn't proven; to do that now would be a sign of weakness. I remember a paper from the 80s, the first one reporting the localisation of some cytoskeletal protein, vinculin or talin or something, where the authors present a set of microscope images showing that the protein is found exclusively at the ends of stress fibres (the main structural members of many kinds of cell). Great finding! And then they show a picture of a cell where it isn't, and is instead assembled into lots of little rings. The caption says that they have no idea what kind of cell it is, or why it's doing that. I just can't imagine that happening today. In this case, this is a review paper, so that doesn't directly apply. But i think the ethos still pervades the writing; the authors are writing to explain what's known, not to advance some hypothesis or demonstrate how much they know.
ryusage 13y ago
I saw that lecture also. Wish I'd had more time to go further with the Modeling class, it was really cool. Anyway, the equally interesting alternative to what you explained is what happens when everybody actually is super racist. If every person's requirement is that they refuse to live near anyone that is different, then you actually get a weird pattern where things are as mixed as they could be. The neighborhoods never stabilize, because there's always going to be border areas where each group can't help living next to a different group. Actually, it just occurred to me that the model might not be accurate at that point, though, because people can't always move away like the model implies. Instead of people moving around chaotically, I wonder if the real result is violence intended to force the others to move. EDIT: changed 'counterpoint' to 'alternative'
- samatman 13y ago
  http://en.wikipedia.org/wiki/1967_Detroit_riot http://en.wikipedia.org/wiki/1967_Detroit_riot http://www.flickr.com/photos/walkingsf/5560480146/in/set-72157626354149574 http://www.flickr.com/photos/walkingsf/5560480146/in/set-721... http://en.wikipedia.org/wiki/Baltimore_riot_of_1968 http://en.wikipedia.org/wiki/Baltimore_riot_of_1968 http://www.flickr.com/photos/walkingsf/5559896701/in/set-72157626354149574 http://www.flickr.com/photos/walkingsf/5559896701/in/set-721...
SiVal 13y ago
Yes, this is a "map of America's racial distribution". Calling it a "map of America's racial segregation" means someone is pushing a political agenda. Calling it a map of racial integration would be just as accurate but probably didn't support the right narrative. As a map of distribution, you would expect significant clustering from randomness alone. We've all seen the "which of these dot patterns is actually the random one?" tests, where few people guess the correct answer: the one with all the clusters. If you take that clustered map as a seed, then add any affinities, you get positive feedback, increasing the clustering. Those affinities don't have to be racial selection but can just be in the general form of, "I make decisions based on what I hear, and I hear more from people I know than from people I don't know." So, you hear from your cousin that there is a new apartment building opening across the street from him. You hear from several members of your church that X is the best school district. A factory in your town closes and the laid off employees discuss their options and decide to go check out a company in the next town that is hiring (with lots of reasonably-priced apartments nearby). Just getting more information from people you know than people you don't know could be sufficient for Schelling's model, but add to that the tendency to get more information from people from your background (whether you actually know them or not), because it's likely to be more relevant to you, and you magnify the effect significantly. An urban liberal and his soccer-loving Mexican gardener don't have to dislike each other for the former to read more NY Times and the latter to watch more Univision, and the information about where the opportunities are as presented in those two media outlets are not the same. This doesn't mean that there is no race-based component in deciding where to live. It just means that, as the Schelling model demonstrates, you can't tell how much there is by looking at how much clustering there is on a map.
- paganel 13y ago
  > So, you hear from your cousin that there is a new apartment building opening across the street from him That's the thing,in a truly color-blind society your cousin could be "black" while you could be "white", but this seems not to be the default setting in the States. If it matters I have "light" skin (after my father) while my brother has "brown" skin (after my mother). I hadn't notice that until recently when my brother visited me (I live in an Eastern European capital city) and there were a couple of people in the tram who looked at me, and at us, differently compared to when I was just by myself. I only realized the probable reasons (my brother has the skin-color of a "gipsy") later on.
gcb0 13y ago
> (40% like me) yeah... you know, the definition of racism is when you even take those categories as a differentiation attribute in people. So, it's racism after all. If you'd say singles vs big families. rich vs poor. etc. it would be personal preference. The things you listed, racism.
- newnewnew 13y ago
  If that's racism, then it is a lot less virulent of a thing than we normally think of. Almost everybody in the world has some sort of affinity for their own cultural group, practices and norms. Maybe we should develop another term for active inter-racial hatred?
  - gcb0 13y ago
    But it's still racism :) you are thinking of hate-crime.
erikpukinskis 13y ago
> I'd accept 60% of neighbors who are different from me How is that not racism? Why do you care if your neighbor is a different race than you? Isn't it more relevant whether they're someone you'd enjoy living near?

by WA 13y ago

acchow 13y ago

Any chance you have a direct link to the lecture without having to register for a course?

WA 13y ago
This is it: https://class.coursera.org/modelthinking-003/lecture/16 https://class.coursera.org/modelthinking-003/lecture/16

twic 13y ago

Reminds me of a similar thing in developmental biology, where if you have two populations of cells, each having a slighly higher affinity for other cells of its own type, then they will very efficiently sort themselves into separate compartments. This is thought to be a pretty important mechanism in the formation and maintenance of structures in the body. In case anyone's interested, here is a nice (if now rather old, although to be honest scientific writing was so much better back in the '90s) review on the whole business: http://www.sciencedirect.com/science/article/pii/S0092867400812970 http://www.sciencedirect.com/science/article/pii/S0092867400... In which we find a paragraph describing this process: "The selector genes do more than specify the pattern and the structures that the compartments will eventually make—they also specify, indirectly, a surface property. This property has been termed cell affinity, meaning that cells that share the same affinity, owing to the same binary code of selector genes, will intermingle during growth. There are a number of different experiments that lead to this conclusion, but perhaps the simplest is the observation that when the selector gene engrailed is removed, in vivo, from a posterior clone of cells in the wing, those cells gain anterior affinity: they now sort out from posterior cells and, if they are in contact with anterior cells, will sort into and mingle with them. Cells from neighboring compartments will have different affinities and tend to minimize their mutual contact, so that where the two compartments abut, there is a relatively straight line across which the cells do not stray."

chrischen 13y ago
Is true in physics as well. That's how stars are formed... Gravity is extremely weak, and so the affinity for one atom and another is extremely small, but after millions of years they still cluster together so hard sparks fly.
jtfairbank 13y ago
Why do you say that scientific writing was better in the '90s? I'm just curious, not challenging your claim.
- twic 13y ago
  Back when i used to read a lot of papers, i found that i enjoyed the ones from the 90s more. I think the further you go back in time, the more honest and open scientists were in their writing; as time has gone on and academia has become more crowded and competitive, the more papers read like marketing. Authors writing in the 80s would happily admit that there were things they didn't know or hadn't proven; to do that now would be a sign of weakness. I remember a paper from the 80s, the first one reporting the localisation of some cytoskeletal protein, vinculin or talin or something, where the authors present a set of microscope images showing that the protein is found exclusively at the ends of stress fibres (the main structural members of many kinds of cell). Great finding! And then they show a picture of a cell where it isn't, and is instead assembled into lots of little rings. The caption says that they have no idea what kind of cell it is, or why it's doing that. I just can't imagine that happening today. In this case, this is a review paper, so that doesn't directly apply. But i think the ethos still pervades the writing; the authors are writing to explain what's known, not to advance some hypothesis or demonstrate how much they know.

ryusage 13y ago

I saw that lecture also. Wish I'd had more time to go further with the Modeling class, it was really cool. Anyway, the equally interesting alternative to what you explained is what happens when everybody actually is super racist. If every person's requirement is that they refuse to live near anyone that is different, then you actually get a weird pattern where things are as mixed as they could be. The neighborhoods never stabilize, because there's always going to be border areas where each group can't help living next to a different group. Actually, it just occurred to me that the model might not be accurate at that point, though, because people can't always move away like the model implies. Instead of people moving around chaotically, I wonder if the real result is violence intended to force the others to move. EDIT: changed 'counterpoint' to 'alternative'

samatman 13y ago
http://en.wikipedia.org/wiki/1967_Detroit_riot http://en.wikipedia.org/wiki/1967_Detroit_riot http://www.flickr.com/photos/walkingsf/5560480146/in/set-72157626354149574 http://www.flickr.com/photos/walkingsf/5560480146/in/set-721... http://en.wikipedia.org/wiki/Baltimore_riot_of_1968 http://en.wikipedia.org/wiki/Baltimore_riot_of_1968 http://www.flickr.com/photos/walkingsf/5559896701/in/set-72157626354149574 http://www.flickr.com/photos/walkingsf/5559896701/in/set-721...

SiVal 13y ago

Yes, this is a "map of America's racial distribution". Calling it a "map of America's racial segregation" means someone is pushing a political agenda. Calling it a map of racial integration would be just as accurate but probably didn't support the right narrative. As a map of distribution, you would expect significant clustering from randomness alone. We've all seen the "which of these dot patterns is actually the random one?" tests, where few people guess the correct answer: the one with all the clusters. If you take that clustered map as a seed, then add any affinities, you get positive feedback, increasing the clustering. Those affinities don't have to be racial selection but can just be in the general form of, "I make decisions based on what I hear, and I hear more from people I know than from people I don't know." So, you hear from your cousin that there is a new apartment building opening across the street from him. You hear from several members of your church that X is the best school district. A factory in your town closes and the laid off employees discuss their options and decide to go check out a company in the next town that is hiring (with lots of reasonably-priced apartments nearby). Just getting more information from people you know than people you don't know could be sufficient for Schelling's model, but add to that the tendency to get more information from people from your background (whether you actually know them or not), because it's likely to be more relevant to you, and you magnify the effect significantly. An urban liberal and his soccer-loving Mexican gardener don't have to dislike each other for the former to read more NY Times and the latter to watch more Univision, and the information about where the opportunities are as presented in those two media outlets are not the same. This doesn't mean that there is no race-based component in deciding where to live. It just means that, as the Schelling model demonstrates, you can't tell how much there is by looking at how much clustering there is on a map.

paganel 13y ago
> So, you hear from your cousin that there is a new apartment building opening across the street from him That's the thing,in a truly color-blind society your cousin could be "black" while you could be "white", but this seems not to be the default setting in the States. If it matters I have "light" skin (after my father) while my brother has "brown" skin (after my mother). I hadn't notice that until recently when my brother visited me (I live in an Eastern European capital city) and there were a couple of people in the tram who looked at me, and at us, differently compared to when I was just by myself. I only realized the probable reasons (my brother has the skin-color of a "gipsy") later on.

gcb0 13y ago

> (40% like me) yeah... you know, the definition of racism is when you even take those categories as a differentiation attribute in people. So, it's racism after all. If you'd say singles vs big families. rich vs poor. etc. it would be personal preference. The things you listed, racism.

newnewnew 13y ago
If that's racism, then it is a lot less virulent of a thing than we normally think of. Almost everybody in the world has some sort of affinity for their own cultural group, practices and norms. Maybe we should develop another term for active inter-racial hatred?
- gcb0 13y ago
  But it's still racism :) you are thinking of hate-crime.

erikpukinskis 13y ago

> I'd accept 60% of neighbors who are different from me How is that not racism? Why do you care if your neighbor is a different race than you? Isn't it more relevant whether they're someone you'd enjoy living near?