Baloney Detection Kit by Carl Sagan

After reading Sagan's book, I needed a reminder of his Baloney Detection Kit — today more than yesterday. This is my TL;DR.

Baloney Detection Kit by Carl Sagan
Carl Sagan in 1981. (Credit: Tony Korody/Sygma)
“Like all tools, the baloney detection kit can be misused, applied out of context, or even employed as a rote alternative to thinking. But applied judiciously, it can make all the difference in the world — not least in evaluating our own arguments before we present them to others.”
— Carl Sagan (The Demon-Haunted World, 1995)

Tools 🧰

For skeptical thinking

  • Independent confirmation — facts need outside verification
  • Substantive debate — invite all knowledgeable perspectives
  • No argument from authority — experts err; there are no authorities, only experts
  • Multiple hypotheses — generate all possible explanations, then test to eliminate
  • No attachment — your hypothesis is a waystation, not a destination
  • Quantify — vague + qualitative = open to anything; measure to discriminate
  • Chain of argument — every link must hold, including the premise
  • Occam’s Razor — when two hypotheses fit equally, choose the simpler
  • Falsifiability — if it can’t be disproved in principle, it’s not worth much
  • Controlled experiments — design out bias; run double-blind when needed
  • Separate variables — one change at a time

Fallacies ⚠️

Which are common

  • Ad hominem — attacking the person, not the argument
  • Appeal to authority — unverifiable claims from the powerful
  • Adverse consequences — “it must be true or else…”
  • Appeal to ignorance — absence of proof is not proof of absence
  • Special pleading — invented exceptions to rescue a failing argument
  • Begging the question — assuming the answer in the premise
  • Observational selection — counting the hits, forgetting the misses
  • Small numbers — drawing big conclusions from tiny samples
  • Statistics misread — e.g. confusing median with average
  • Inconsistency — applying standards selectively
  • Non sequitur — conclusion doesn’t follow from premise
  • Post hoc — “it happened after, therefore caused by”
  • Meaningless question — false paradox with no real referent
  • False dichotomy — ignoring the vast middle ground
  • Short-term vs. long-term — urgency used to dismiss larger stakes
  • Slippery slope — extreme outcome presented as inevitable
  • Correlation ≠ causation — co-occurrence is not mechanism
  • Straw man — caricature a position to make it easier to attack
  • Suppressed evidence — half-truths, selective disclosure
  • Weasel words — euphemism as camouflage for meaning