Growing Your Critical Thinking

Critical thinking is reasoning from first principles to the real cause, past the first plausible answer and backed by data, never pattern-matched. It is the attribute the hiring loop is built to detect, because it is the one that cannot be faked under pressure and the one every other spoke leans on. Product sense without it picks the wrong problem confidently. Craft without it builds the wrong thing well.

This page is how you move yourself up one rung on it. It is a mirror, not a scorecard: find the row just above where you sit, and put your reps there. The reasoning tools themselves live in Structured Problem Solving, Thinking Like an Engineer, and Diagnosis.

Where you are now#

Find the line that sounds like how you actually reason, not how you wish you did.

Level	The tell
Intern	You stop at the first plausible cause and repeat the number you were handed.
Junior	You ask “why” one level down to a contributing factor, and you check a claim against data.
Mid	You reach the system-level cause and back every claim with real numbers.
Senior	You reframe when the real cause sits higher up, and you know which number would change the call.
Lead	You coach it in review and set what “checked against data” has to mean.

What moves you up#

The whole spoke is one discipline applied harder: do not accept the first answer that would explain what you see, and do not accept a claim you have not measured.

Ask why until you hit something you can change. “Checkout is slow” is a symptom. “Checkout is slow because the address lookup makes one call per keystroke” is a cause. The first stops you at the screen; the second tells you what to fix. This is the issue tree instinct: split down until the branch is actionable.
Separate the cause from the first thing you noticed. They are usually different, and pattern-matching quietly swaps one for the other. The bug you can see is often three layers below the one you should fix.
Back the claim with a number, and know which number would change your mind. A claim you cannot attach a measurement to is a hunch with confidence. If no possible number would change your call, you are defending, not reasoning.
Distrust the number you were handed. Data misleads more often than it lies: a rate with no denominator, an average hiding a bimodal split, a count over the wrong window. See Misleading Data.

The habits that keep you stuck#

Everyone pattern-matches under time pressure; it is fast and usually right, which is exactly why the wrong times are expensive. Catch yourself in these:

Bad: “Signups dropped, it is probably the new pricing page, let us roll it back.” First plausible cause, no measurement, action taken.

Good: “Signups dropped 12% week on week. The pricing page is one candidate, but the drop started Tuesday and the page shipped Thursday, so it is not the cause. The Tuesday change was the email verification step. Let me check its completion rate before touching anything.”

The tells that you are pattern-matching:

The first cause becomes the fix with nothing ruled out in between.
A number gets repeated, never checked, because it came from someone senior or a dashboard that looked official.
“It is probably X” where the “probably” is doing all the work and no data follows it.
The reframe never happens: you keep debugging the branch you started on even as it stops paying out.

The reps that level it up#

Move a rung by doing the next row’s reasoning one step before it feels comfortable, then close the gap in review:

Re-derive instead of recalling. When you catch yourself reaching for “we always do it this way,” stop and rebuild the reasoning from the actual constraint. Half the time the old answer no longer fits.
Before you touch code or take an action, write the one number that would prove you wrong, then go get it. This is the single highest-leverage rep on the spoke.
Take a claim you believe and argue the other side hard enough to break it. What survives is reasoning; what does not was a preference. Do this on your own conclusions, not just other people’s.
Ask to be challenged in review. Have someone make you re-derive a decision from scratch and slip in a wrong premise. Catching it is the muscle; missing it tells you exactly where to work.

The one question#

For any conclusion you are about to act on, ask: what number would change my mind, and have I actually looked at it?

If nothing could change your mind, it is a belief, not a finding. If something could but you have not looked, you are guessing with a straight face.