Library:Calling Bullshit
| Calling Bullshit: The Art of Skepticism in a Data-Driven World | |
|---|---|
| Author | Carl T. Bergstrom & Jevin D. West |
| First published | 2020 |
| ISBN | 978-0-525-50918-9 |
| Role | Analysis |
| Domains | Science & Environment Technology & Media |
| Tactics | Statistical & Evidentiary Distortion |
Calling Bullshit: The Art of Skepticism in a Data-Driven World is a 2020 book by Carl T. Bergstrom & Jevin D. West.
Summary
Carl T. Bergstrom, an evolutionary biologist, and Jevin D. West, an information scientist, both at the University of Washington, turned a popular university course into this book. The course, launched in 2017 under the same name, drew national attention and wide adoption at other universities. The book is its most complete form: a practical manual for recognizing and refuting misleading claims in an era when data is everywhere.
Old-school and new-school bullshit
The authors distinguish two kinds. Old-school bullshit relies on words: flattering rhetoric, weasel phrases and confident assertion. New-school bullshit wears the costume of science and numbers: statistics, charts, algorithms and published studies. Its power comes from the fact that most people feel unqualified to argue with a number, which lets bad reasoning hide behind quantitative authority.
The asymmetry problem
The book takes seriously what is often called Brandolini's law, or the bullshit asymmetry principle: the energy needed to refute bullshit is far greater than the energy needed to produce it. A false claim can be produced in seconds; debunking it can take days of research. Much of the book is about making refutation efficient.
What it teaches
Chapters work through the main machinery of data-driven deception: confusing correlation with causation; selection bias and samples that cannot support the conclusions drawn from them; misleading data visualizations, including truncated axes and decorative charts that distort proportions; and the hype around big data and machine learning, including a study that claimed an algorithm could detect criminality from facial photographs. A chapter on scientific publishing explains how predatory journals, p-hacking and press-release exaggeration let flawed findings enter the public record.
Spotting and refuting
The practical sections offer heuristics anyone can use: question the source, beware claims that seem too good or too bad to be true, think in orders of magnitude, and look for the unfair comparison. On refutation, the authors advise using analogy and counterexample, redrawing a misleading graph correctly, and correcting without humiliating, since people rarely change their minds when they feel attacked.
Relevance to Bullshitopedia
This is the closest thing the field has to a working manual for editors. Its toolkit maps directly onto the Statistical & Evidentiary Distortion tactic, and its insistence that refutation be clear, fair and evidence-based is a standard Bullshitopedia entries should meet.