Calling Bullshit: The Art of Skepticism in a Data-Driven World: Summary & Key Insights

The modern information landscape is unlike anything humans have faced before. The printing press democratized knowledge, but it could not have prepared us for the algorithmic firehose of social media. Every minute, millions of posts, memes, and data visualizations flow through platforms optimized not for truth but for engagement. In this environment, bullshit doesn’t just survive — it thrives.

I want you to imagine the internet as an ecosystem. In any ecosystem, what thrives depends on its environment. Here, attention is the nutrient, and anything flashy, emotional, or conveniently simple grows fast. Algorithms reinforce this, amplifying what people click on, not what holds up to scrutiny. When a misleading headline or a fake statistic spreads, it does so not because people are stupid, but because the system rewards content that provokes over content that informs.

This is the paradox we face: more data does not mean more understanding. Abundance breeds confusion unless paired with discernment. Our attention becomes fragmented, our trust wandering from one half-true narrative to the next. The only antidote is awareness — the conscious realization that we are creatures of limited cognitive capacity living in an environment that exploits those limits. It’s not enough to know facts; we must also understand the social and technological frameworks that shape which facts reach us.

Numbers carry a false aura of authority. A claim backed by data feels objective, immune to argument. But numbers can lie — not because math itself deceives, but because humans manipulate how data are collected, framed, and presented. We see this daily: a study showing that teenagers who play video games score lower in school, or a headline about unemployment rates improving, neglecting to mention that calculations changed along the way.

One common tactic is cherry-picking: selecting only the data that support a conclusion. Another is using misleading averages. The mean, median, and mode each tell different stories, yet communicators often choose whichever paints the rosiest picture. There are countless ways to make statistics say what one wishes — by truncating axes on graphs, by using relative percentages that conceal actual magnitudes, or by presenting small datasets as definitive.

When I analyze such examples with my students, I encourage them to reverse engineer the claim. Ask: how was this number produced? What was excluded? What assumptions hide in those percentages? True data literacy is not about memorizing formulas — it’s about recognizing that every statistic is an argument, and every argument deserves scrutiny. Numbers earn trust only when their context is transparent.