Part 3-B: Examine the Problem | Inspect the Pieces

Title Card

Stop flattening useful differences before you know what kind of variable you are looking at.

Two things can belong to the same category without being equivalent in magnitude, consequence, or importance.

When the variable changes by degree, preserve the degree instead of collapsing everything into yes/no.

Sometimes moving farther along a spectrum crosses a boundary where the practical consequences change.

If two cases differ along more than one meaningful dimension, stop forcing them onto a single ruler and add another axis.

Notice recurring configurations without turning a pattern into an essence or assuming every member contains every feature.

Sometimes you can confidently rank cases without pretending you know the exact distance between them.

Figure out what a piece of information actually bears on before letting it vote on something else.

Wanting a conclusion, or hating it, can distort your map. It cannot edit the territory just to reward or disappoint you.

Pointers bear on the proposition. Indicators tell you something about the conditions around the reasoning and where scrutiny may be useful.

Before deciding whether evidence is good or bad, name the proposition it is supposed to affect. Ask: evidence of what?

A pointer can legitimately raise, lower, or leave unchanged the probability of the claim being tested. A pointer is not a verdict.

Catch recurring cases where a real fact is made to vote on a proposition it does not actually support.

Some facts matter because they tell you where to look harder, not because they settle the external claim.

The same fact can be neutral relative to one question and informative relative to another. Pointers belong to evidence-question relationships, not permanently to facts.

Pull apart concepts that got welded together and inspect what crossed the seam.

Association is ordinary biological machinery, not necessarily intentionally bad reasoning. Your subconscious did not ask permission before building half this shit.

A false or overgeneralized association may have had a perfectly understandable origin in the environment that taught it.

Separate A, B, and the relationship between them instead of reasoning from one fused object.

Two things can travel together repeatedly without becoming the same thing. Shared history does not create an equals sign.

Emotional, moral, or social charge can bleed from one concept into another until the second concept inherits properties that properly belonged to the first.

A stored association can manufacture an automatic pointer before you have inspected whether the current event actually supports the conclusion.