Critical Thinking

When was the last time you changed your mind about something important?

Most information is noise. Most arguments have holes. Critical thinking protects you from bad decisions dressed as good ones.

Source Check

Before accepting any claim, run five questions:

Question	What It Catches
Who wrote this?	Authority bias — credentials don't guarantee truth
What do they gain if I believe it?	Incentive bias — follow the money
What's the funding source?	Institutional bias — who pays shapes what's said
Who disagrees, and why?	Selection bias — one side of the story
When was this produced?	Recency bias — context changes everything

If you can't answer at least three, you don't know enough to have an opinion.

Evidence Type	Strength	Watch For
Anecdote	Weakest — one story proves nothing	Vivid stories feel more true than they are
Correlation	Moderate — things move together	Doesn't mean one causes the other
Controlled experiment	Strong — isolated variables	Small samples, unreplicated results
Meta-analysis	Strongest — patterns across studies	Publication bias, heterogeneous methods

Ask: What's the sample size? Is it reproducible? Would the opposite finding have been published?

Common failures that sound right but aren't:

Fallacy	Pattern	Example
Ad hominem	Attack the person, not the argument	"They're biased, so ignore the data"
False dichotomy	Only two options when more exist	"You're either with us or against us"
Appeal to authority	Credentials substitute for evidence	"The expert said so"
Survivorship bias	Only see winners, ignore losers	"Dropouts built billion-dollar companies"
Sunk cost	Past investment justifies continuing	"We've already spent too much to stop"
Straw man	Weaken the argument then defeat it	"They want no rules at all"

The test: Does the conclusion follow from the premises? What must be true for this to work? What's assumed but unstated?

Your own thinking fails predictably:

Confirmation bias — You seek evidence that supports what you already believe. Fix: actively search for disconfirmation.
Anchoring — The first number you hear shapes all subsequent estimates. Fix: generate your own estimate before looking at others'.
Availability — Recent or vivid events seem more likely. Fix: check base rates — what usually happens?
Dunning-Kruger — Incompetence prevents you from recognising incompetence. Fix: seek feedback from people better than you.

The diagnostic question: What would change my mind? If the answer is "nothing," that's faith, not thinking.

Before committing to a position:

Steel-man the opposition — State their strongest possible argument, not the weakest version
Ask "compared to what?" — Every option has an alternative. Evaluate relative to that.
Pre-mortem — Imagine this decision failed. What went wrong? Work backward.
Falsifiability test — How would you know if you're wrong? If you can't answer, it's not a testable claim.

Cynicism. Dismissing everything. Analysis paralysis. Mistaking skepticism for wisdom. The point is better decisions, not never deciding.

Archetype	Thinking Style
Realist	Grounds claims in evidence — what's actually true?
Philosopher	Questions assumptions — what's assumed but unstated?
Engineer	Tests claims against systems — does this hold under load?