Small cities face greater impact from automation by Frank, Sun, Cebrian, Youn, and Rahwan

  • Frank et al. analyzed the relative impact of automation on employment and find that cities with fewer than 100,000 people face the greatest risks.
  • This is a good reminder that most social processes are spatially correlated. In the words of Tobler: “Everything is related to everything else, but near things are more related than distant things.”
  • To get the policy response right, we’ll have to take these spatial correlations into account.

Counterfactual Fairness by Kusner, Loftus, Russell, and Silva

  • Focusing on causality helps us quantify fairness and prevent bias in machine learning.
  • Dropping biased predictors isn’t good enough. We have to understand the causal structure of the bias and account for it our analysis. The counterfactual approach Kusner et al. propose looks like a promising way to do this.

Your Research Garden by Chris Parnin

  • Parnin’s garden metaphor helps me deal with the churn of disparate projects.
## make gapminder2


ggplot(gapminder, aes(x = pop, y = continent)) +
  geom_density_ridges2(color = "white", size = .2, alpha = .5) +
  coord_polar() +
  theme_void() +
  theme(plot.background = element_rect(fill = "black"))