Bullshit is speech that aims to persuade without caring whether it is true. It may be false, true, or unfalsifiable. The key feature is indifference to accuracy. Spotting it is a skill, and there are times when even careful thinkers cannot be sure. Here is a practical way to navigate both situations.

Fast triage in seven questions

1. What exactly is the claim in one sentence with a measurable outcome
2. By when does it happen or apply
3. What would count as disproof
4. What is the base rate for similar claims in the real world
5. What evidence is offered and can I check at least one source
6. How does it work in plain language
7. Who benefits if I believe this

If you cannot answer at least four of these, treat the claim as unverified and keep your decisions conservative.

Common red flags

• Vague verbs and big adjectives with no numbers
• No time frame or threshold for success
• Unfalsifiable promises that “cannot be disproven”
• Cherry picked anecdotes without a denominator
• Moving goalposts when evidence arrives
• Gish gallop with many shallow points that are hard to check quickly
• Jargon that does not translate to a working example
• Appeals to secret sources or unnamed experts
• All upside with no cost, risk, or tradeoff
• Personal attacks in place of data

Reliable green flags

• A clear claim that could be wrong
• Numbers with units, error bars, or ranges
• Time bounded predictions you can later score
• Methods and data you can inspect
• Replication by independent parties
• Known limits, caveats, and failure modes
• Conflicts of interest disclosed
• A precommitment to update or retract if new evidence arrives

Why you sometimes cannot tell

• Domain complexity that exceeds your current knowledge
• Information asymmetry where the other side knows facts you cannot access
• Noisy or delayed feedback that hides outcomes for months or years
• Adversarial optimization where someone is trying to game your checks
• Strong social incentives that punish dissent
• Cognitive load, fatigue, or emotional arousal that narrows attention
• Novel events with no base rate to anchor judgment
• Sparse or low quality data that makes inference unstable

In these cases, the goal is not certainty. The goal is damage control and learning while you wait for better signals.

What to do when unsure

• Price the uncertainty. Decide what level of risk is acceptable and size your bet accordingly.
• Prefer reversible moves over irreversible ones. Try before you commit.
• Run a cheap test. Create a small experiment that would generate one strong signal.
• Ask for skin in the game. Tie claims to consequences for the claimant.
• Set tripwires. Define a clear stop rule that triggers if certain facts appear.
• Time box your research. Stop after a fixed interval and decide based on what you have.
• Document your reasons. Future you can audit where you were right or wrong.

Two minute protocol

1. Rewrite the claim as a forecast with a date, a number, and a pass or fail condition.
2. List one piece of disconfirming evidence that would change your mind.
3. Search for that evidence first.
4. If still uncertain, make the smallest useful move and schedule a review date.

Worked example

Claim. “Supplement X boosts energy by 30 percent.”

• Measurable. What metric defines energy, for example daily step count or time to complete a standard task
• Time bound. For the next 14 days
• Disproof. Average change less than 10 percent compared with last month under similar sleep and work conditions
• Base rate. Most over the counter supplements show small or no effects in controlled tests
• Evidence. Check for randomized trials, sample sizes, and funding sources
• Mechanism. How does it affect physiology, in terms you can explain to a teenager
• Incentives. Who sells it, who measured it, and who stands to gain

Decision. Run a two week N of 1 trial with a prewritten stop rule and record keeping. Spend only what you are willing to lose.

Rules of thumb

• Extraordinary claims require extraordinary evidence
• Boring explanations beat exciting ones most of the time
• Never accept a claim that offers zero risk and infinite reward
• If nothing could change their mind, talk about decisions not about truth
• Always know what would change your mind

Bottom line

Define the claim, set a time frame, and ask how it could be wrong. When you cannot tell, downshift into small tests, reversible moves, and clear stop rules. Certainty is rare. Good decisions come from disciplined doubt paired with measured action.