Scott Alexander examines Google Correlate, finding it produces both obvious and nonsensical correlations, and explores its potential uses and limitations. Longer summary

Scott Alexander explores Google Correlate, a tool that finds search terms correlated with specific datasets. He finds that while some correlations are obvious (like 'skiing' with cold temperatures), others are nonsensical, especially for diseases. He tests various datasets, including parasite load and family ties, finding odd correlations that often seem to be confounded by factors like race or geography. Scott concludes that Google Correlate isn't very useful for revealing deep connections, but might serve as a sanity check for confounding factors in studies or as a cheap method for creating datasets. He ends with a humorous correlation between searches for an anti-parasitic medication and curiosity about Lil' Wayne's sexual orientation. Shorter summary