Segregation from Slight Individual Bias
In a fun, interactive demonstration, authors Victoria Hart and Nicky Case demonstrate how even small individual biases can lead to large collective biases. Their simple demonstration allows us to explore the behavior of two distinct groups, triangles and squares, when individual bias is present in each member of the population. Members of each group are laid out on a square grid and are given a “bias” that determines the percentage of their neighbors that must be the same type to prevent a shape from “moving” to a different spot. The grid is then measured for “segregation” based on the work of Nobel prize winning game theorist Thomas C. Shelling. Running the simulation it is easy to see the impact of even slight bias between shapes. At just 33% the grid
Running the simulation it is easy to see the impact even slight bias between shapes has. At just 33% the grid becomes 50% segregated, at 45% it goes up to 77%, and at 55% it can go as high as 90%. Incredibly, after the grid becomes segregated, lowering the bias and running the simulation again does not lower the segregation score.
These insights are extremely relevant in today’s society as the United States faces widespread protests over race issues and the United Kingdom continues to struggle following the vote to exit the European Union; a decision made in part due to issues concerning open borders and immigration policy. It is extremely interesting to see how fast segregation can multiply even in Victoria Hart and Nicky Case’s simple simulation and the issues raised are even more poignant in our world’s diverse society.
I think that game theory’s applications in creating a segregation score are very interesting but it is the local dynamic between a shape its neighbors that I think is more relevant to our discussions in class. Shapes of the same type can be said to have strong ties to one another and those of different types can be said to have weak ties. It makes sense that a shape surrounded by shapes of the same type is the most stable configuration due to the high number of strong ties. As a result, as the shapes are prompted to move by an unacceptable number of weak ties, they will start to congregate in clusters that make it hard for the other shape to move nearby. These clusters are inherently stable and thus once formed are unlikely to fall apart. As a result, they can either stay the same size or grow.
This clustering of strong ties to create stable groups of shapes leads to the high segregation. It seems that the level of “bias” set, simply works to either increase or decrease the size of these clusters. Clearly, if our society is to continue its cosmopolitan trend individual biases, however small, must be addressed.
Interactive Simulation: http://ncase.me/polygons/
Segregation Measure: http://www.stat.berkeley.edu/~aldous/157/Papers/Schelling_Seg_Models.pdf