social reading

the small world problem

The small world problem has been in the news recently, due to the publication of Duncan Watt's re-creation of Stanley Milgram's original experiment.

This week we'll be reading Milgram's original paper, Watt's recent paper, and Granovetter's comments on the latter. There is useful supporting information about the Dodds et al paper.

Here are some questions to think about:

• What effect did the information Dodds et al provided about the targets have on the search path? What might happen with different information? What would happen if the same target was given, but with different clues?

• Clearly, people are motivated to match the data they are given about the target when they pick their next-in-chain person. What other considerations do you think go into this decision?

• How random are the targets (in both the original and recent studies)? In the Dodds et al supplementary data, look at the range of incomplete chains. What do you think accounts for this? Look at the authors' discussion (in the main paper) of the professor who had many completed chains. Do you agree with their assessment?

• Milgram notes that in the past, there were many disconnected sets of people. Are we really all connected now? Is the degree of connectedness (ie, the number of people it would take to connect 2 random people) changing rapidly? In all populations?

• Granovetter suggests combining this sort of experiement with real world network data. Can you think of a way to do this?

• Both Dodds et al and Granovetter say that the results can be skewed by motivation. If so, how can the experiment be redesigned to motivate the participants to continue the chains as best as they can?

• Milgram wrote about "sociometric stars", which would be similar to the "hubs" of which Dodds et al say they saw little. What do you think caused the discrepancy?

• What do you think this sort of experiment really shows?