Science Quote by John Tukey

Tukey’s line reads like a tidy aphorism, but it’s really a warning shot at the complacency of “the answer” culture. Numerical summaries collapse a dataset into a few sanctioned facts: mean, variance, a p-value with a ceremonial asterisk. That compression is useful, but it also rewards what you already think you’re looking for. “Expected values” isn’t just a technical term here; it’s a psychological jab. When you reduce data to a number, you tend to confirm the story you brought with you.

Graphical summaries do something ruder. A plot refuses to behave like a verdict. It makes room for the awkward particulars: a second cluster that suggests two populations, a fat tail that wrecks normal assumptions, an outlier that might be an error or the entire point. “Unexpected values” is Tukey’s shorthand for discovery - the part of analysis that doesn’t feel like bookkeeping. A scatterplot isn’t merely decoration; it’s an instrument for surprise.

The context matters: Tukey helped build the modern toolkit of exploratory data analysis in an era when computing made it easy to produce more numbers than insight. His subtext is anti-ritual. Don’t outsource thinking to summary statistics, and don’t mistake elegance for truth. Numbers are great at reporting what happens on average; graphics are great at revealing where “average” is a lie. In a culture that loves dashboards and single-score rankings, Tukey is still arguing for curiosity over closure.