Think for Yourself

Normal thinking produces normal results. Abnormal results need abnormal inputs, and the inputs are the beliefs you already run on.

Most people inherit their beliefs and call the result judgment. I treat judgment as a built object, assembled from many fields and tested before it gets trusted.

Discipline and drive sit on top of a worldview. Those qualities matter, and none of them picks a direction. When the worldview is borrowed, discipline moves you faster along a road somebody else picked.

Mastery and rule-changing run on the same input: a way of seeing that the field does not already share. A borrowed way of seeing caps how far the effort carries.

Where your beliefs came from

You did not pick most of what you believe. Family handed you the first set, school handed you the next, and culture supplied the rest before you had a vote.

Borrowed beliefs are cheap to hold. Testing one costs time and standing, so the testing stops early, and the questions stop with it.

What remains is a box built from the inside. The walls are answers you never checked, and from in there the walls look like the edge of the world.

The box is comfortable and cheap to maintain. Nothing inside it argues with anything else inside it, so it feels like clarity.

A borrowed worldview still works. It obeys the room, and it returns the results the room already gets.

Certainty is a function of your viewpoint

From where you stand, the earth looks flat. The reading is honest, the instrument is your own eye, and for most of human history nothing better was available.

The earth kept its shape the whole time. Height changed the answer.

Truth, as anybody holds it, depends on position, information, and the year. Absolute truth needs a person who has seen every angle and knows everything there is to know. Nobody holds that job.

What a field calls a fact today is the best account available today. New instruments arrive, and the account moves.

History is a long record of confident people being wrong, and the recent pages carry their share.

That leaves a working rule. Hold the belief that survived a repeated test, and hold it loosely, because the view can still widen.

Updating is the price of accuracy. You pay it in public, and the alternative is running an old answer while the world runs on a newer one.

The rules of a field are somebody's old answer

Every rule in your field is an old answer that worked. Somebody watched what happened, wrote an explanation that held, and enough people agreed to stop looking.

That work was real. It was done with the information available then, and the explanation carried as far as the information reached. The rules are the accepted facts of a time, and the time keeps moving.

Most people obey the rules because obeying was trained early. Challenging one is also a claim about your own ability, and few people want to make that claim out loud.

A master starts in the same room with the same training. The difference arrives later, when the rules stop explaining what the master keeps seeing.

The glitches come first. One result does not fit, then a case the theory skips, and the questions start there.

Breaking a rule to look brave produces noise. A master breaks a rule because it stopped explaining the world, takes a different path, and other people notice when the path works.

The new path becomes the norm, then the status quo. Then somebody else finds its glitches, and the cycle runs again.

Idiosyncratic thinking is the habit of asking the question nobody in the room has asked. Unusual questions are the only reliable source of unusual answers. Question the unquestioned, because that is the ground nobody has picked over.

Assume the claim is wrong until it is shown

Information arrives all day from news, podcasts, coworkers, and family. Filter it, or a worldview gets built out of whatever showed up.

The default filter is the source. “It was on the news” and “everyone says it” both report status, and status gets treated as proof.

A crowd holding a belief is a fact about the crowd.

The working posture inverts that default. Assume a claim is wrong until somebody shows that it is right.

The challenge that does most of the work is short. “How do you know?” The follow-up is shorter. Show me the data.

Good information arrives as an argument, and an argument has parts you can pull apart. Four moves do the rest.

• Strip the identities and name the objects. Two people become A and B, and you map what moved between them.
• Separate the claim from the premises and the conclusion, then check whether the premises carry the conclusion.
• Name who benefits if you believe it, and who pays. That check alone dissolves a lot of persuasion.
• Ask why until the real question appears, then answer that one instead.

Stripping the identities is the hardest of the four. You read the situation as a third party rather than a participant. You give each thing one name and keep it, and you rebuild events in the order they happened.

When four words point at one thing, group them and keep one. People argue about the words long after the thing stopped being unclear.

The map holds the objects and the influence between them. Influence runs one way, both ways, or in a loop. People narrate in loops and in feelings, and emotions are real without being events. They matter, and they belong in a different column from what happened.

Premises follow words like since and because. Conclusions follow therefore and thus. Once the parts sit apart, most arguments come apart at the joint between the reasons and the conclusion.

Then rebuild the argument from the other side. The strongest counterargument you can write is the real test of what you hold.

Two question families pay more than the rest: where this breaks, and when it fails. One more earns its keep. Ask what a safe test on this costs. Put the same questions to the people involved, and then to yourself.

This is why I often do not answer the first question somebody asks me. I ask why they are asking, then why that matters. By the third or fourth layer the real problem is on the table, and the original question was the wrong one.

Reduce the thing to parts you can recombine

Reasoning by analogy is the cheap method. “It has always been done this way” is a report about the past, and it keeps your options inside the past.

Analogy is easy because somebody else already did the assembly. You inherit the finished object and the name that came with it.

A first principle is something you know to be true that you cannot derive from anything simpler. Working from first principles costs more attention: break the thing into parts, then rebuild from the parts.

Put a motorboat towing a skier, a tank, and a bicycle in a row. Analogy sees three finished machines. Decomposition sees a motor, skis, treads, handlebars, gears, and a seat. One arrangement of those parts is a snowmobile.

Finished forms hide their parts. The parts carry the options.

Decomposition is half the work. Recombination is the other half, and a pile of parts with no new arrangement pays nothing.

After a while this changes what you look at in somebody else's work. The pieces are usually available to everybody, and the advantage sits in how they were configured and sequenced.

The experiment is the only judge

The experiment settles it. If somebody says a thing does not work, and you run it and it works, then it works.

Claims are weaker than results. A claim costs nothing to make, and a result costs whatever the test costs.

The method is observation, measurement, and a prediction that can fail. The loop is short: observe, predict, test, then keep or reject the hypothesis.

A test reads clean only when one thing moved. The one-variable rule from the note on cleaning up your mind applies to beliefs the same way. Each turn of the loop buys one change and returns one result you can read.

One occurrence is not enough. A method that worked once worked once, and a method that solved the same problem many times with the same steps is portable.

Portable is the word that matters. A proof of concept is what the second run buys you. You can hand a replicated method to somebody else and watch it run without you, and a habit does not survive that handoff.

A belief you adopt today gets applied for years. The effect horizon from the note on second-order consequences is the reason one clean run does not earn your trust.

The same loop runs on a price, and on the problem you picked to solve. It returns iterations, and each iteration is one correction.

Every new model changes the ones already there

You already run a set of models. Most were installed without your consent, and the ones you cannot name are the ones in control.

Naming them is the first move. A model you can name is one you can test and remove, and the unnamed ones get a free pass forever.

School installed models and work installed more. Some still earn their place and some distort everything you look at, and you decide which is which.

One discipline held long enough becomes a lens you cannot take off. The lawyer sees contracts everywhere and the accountant sees cost everywhere, and both are narrow in the same direction.

Physics has a name for the ambition: a unified field theory, one account that covers every force. Your version covers the fields you work in, and it never gets finished.

A worldview drawn from many fields reads one situation from several angles at once. That comes from running two fields against one problem, and trivia collection never gets there.

The angles pay only when you act from the wider view. A wider view that changes none of your moves is a hobby.

Do not try to hold all of it at once. The note on deciding cuts how many choices reach you at all, and the same load rule applies to models. Name the two or three you run every week, and let the rest wait.

Adding a model looks like software integration. Everything connects to everything, and a change in one place shows up in another. Add one, run it against what you already hold, and repair the contradictions before you add the next.

Your set is yours, and its shape is part of the advantage. Copying somebody else's list returns somebody else's results.

A worldview you can inspect is one you run. An unexamined one runs you.

Write your worldview down

Open a file tonight and write down what you currently believe about your work. Plain sentences, in the order the beliefs come out.

Most of what you hold collapses when you try to put it in one written sentence. That collapse is free information, and it arrives before the belief costs you anything.

Then improve the file. Cross out what a wider view already voided. Run the survivors against a test that changes one thing, and keep the file open for years.

Most people never find out what they think. Write the file and keep improving it. The beliefs that survive the rewriting are the ones you can carry through the stretch where nothing shows.