I watched Dan Gilbert’s interview on Prediction & Psychology, to help inform my final project which will be on the predictive systems of Online Dating. Gilbert raised some interesting psychological points that link to the data that is used in these matching algorithms used by dating sites. He mentioned that people think about the future differently than how they think about the past since the past is constrained by data, and as such when we think about the past, we are building a simulation of what happened, and what our preferences are. So, when it comes to inputting data into these algorithms, it is important to note that the closeness to what reality actually is (in terms of how past experiences shape preference) is more of a simulation rather than exact accuracy.
He also raises the interesting point that the easier question we ask ourselves is “what will happen?” rather than the more challenging, yet more important “will I like it or not?”. Matching algorithms are ones that aim to reach a stable outcome (Gale & Shapley, 1962) where there is no incentive to deviate given a set of pairings. These outcomes, however, just because they are stable does not mean on the individual level that they are utility maximizing. So, it is interesting to map the notion of WHAT the future is vs will I LIKE the future to how these matching algorithms work.
Gale, D., & Shapley, L. S. (1962). College admissions and the stability of marriage. American Mathematical Monthly, 69(1), 9-15. Retrieved from https://www.eecs.harvard.edu/cs286r/courses/fall09/papers/galeshapley.pdf