Some lecture notes/ scribble on Model Thinking by Scott E. Page. Lecture “Segregation and Peer Effects”, Part 2: “Schelling’s Segregation Model”:
Schelling built a model and executed it in order to analyse the empirical phenomenon of segregation:
Problem: people tend to live in the neighbourhood of people similar to them self. How can this be explained?
Approach: every person has a threshold t saying: “I want to have at least a percentage of t people similar to me my neighbourhood”. Schelling allocated the thresholds and then ran the model.
Result: roughly speaking, already a ‘low’ threshold lead to a ‘high’ degree of segregation.
|=: So Schelling found a surprising effect by running a model. Now, from the point of computational thinking I’d like to ask:
does running a model provide understanding?
Personally, I don’t think so: execution to me is more like a discovery, an expedition into the ‘jungle of knowledge’, that reveals some effects that then can be analysed. Thus ‘running’ in the above case provides ‘just’ an effect. The nice thing is, that we can rely on this result, ‘it does the job’. Nevertheless, we cannot tell, why it does so. Thus, imho ‘running’ is a way to obtain results, bypassing understanding.
Seems to be a typical example of what epistemologists call emergence, i.e. macro-level effect from micro level-causes, here as ’emergence by execution’.