I.34 — Logic Isn’t One Thing: The Switchboard, the Smuggling Routes, and the Razor

The Decision Papers — Logic Isn’t One Thing: The Switchboard, the Smuggling Routes, and the Razor. Banner with several different bridges crossing a misty ravine, each a separate route, in watercolor on aged paper.

“The object of reasoning is to find out, from the consideration of what we already know, something else which we do not know.”

— Charles Sanders Peirce, “The Fixation of Belief” (1877)

Someone in the meeting says the logic is clear, and the room nods. Everyone feels they see the same solid thing. They are not. The person who said it was reasoning from a rule. The person nodding was reasoning from a pattern she had seen before. And a third, quietly unconvinced, was reasoning from a story that seemed to fit. Three different machines, all called logic, all humming in the same room, and no one notices they are not the same machine.

That is the trap hiding inside the word logic. We treat it as one clean tool, when it is really a switchboard of different reasoning modes. Each one runs on its own kind of fuel. Each one is strong over a certain range and useless outside it. And each one fails in its own particular way. Call it the Monolith Fallacy: believing logic is a single thing, when it is a handful of very different things wearing the same name. The job here is not yet to pick the right mode for a task, which is a skill of its own. It is the simpler step that comes first: to see that there is more than one mode at all, so you can stop saying the logic is clear and start asking which logic, and how much weight it holds.

Start with the three main kinds. Deduction runs from a rule to a certain answer. All men are mortal, Socrates is a man, so Socrates will die. When the rule really holds, the answer is guaranteed. But that is also the trap: it is easy to treat a passing trend as if it were an iron rule. Software has no marginal cost, someone reasons, as if it were geometry, and the whole plan collapses the day that trend turns out to be a fashion in disguise.

Induction runs from the past to the future. The sun rose every day so far, so it will rise tomorrow. It gives you good odds, not certainty. Its trap is that you assume tomorrow looks like yesterday. It works beautifully for a thousand calm days, then a market turns and the pattern that never failed fails all at once.

Abduction runs from a clue to the best explanation. The grass is wet, so it probably rained. This is the logic of diagnosis and strategy, and its trap is that you quietly decide the most likely story is the only story, when the grass could as easily have met a sprinkler. Abduction hands you a suspect, not a verdict. It has earned a small test, not a full commitment.

Two more slip into the room dressed as logic when they are really something looser. Analogy borrows understanding from one thing to explain another. It is wonderful for getting your bearings and treacherous as proof, because it assumes the two work the same way underneath, and when they do not, the argument drops through the floor. And heuristics are the rules of thumb you carry around. Quality always wins. Buy the dip. They worked somewhere once, so you trust them everywhere, and the trap is that a rule which was gold in one setting can bankrupt you in another.

Here is why this matters so much. When the modes blur together, we reach a confident, unanimous conclusion while standing on exactly the wrong kind of reasoning, and never feel the ground shift. We used a pattern where we needed a rule. We used a story where we needed a test. And because it all felt like logic, no one caught the swap. The worst decisions were not illogical. They were perfectly logical, in the wrong mode. Worse, when the reasoning feels shaky, people rarely stop. They reach for complexity instead, piling on variables until the model looks impressive enough to trust. That is complexity theatre, and it does the opposite of what it promises. Every extra assumption is one more thing that has to be true, so the more elaborate the model gets, the more fragile it becomes.

So what is the fix? It is almost embarrassingly small. Before anyone says therefore on a decision that matters, name which kind of logic is carrying the argument, and how much confidence that kind has earned.

You might think this sounds pedantic, that a room forced to label every sentence would seize up and never decide anything. And you would be right, if the rule applied to everything. It does not. Most of what gets said in a meeting is routine, and you leave it alone. You save the move for the moment real weight is about to rest on a single argument. There, naming the mode takes about five seconds, and it stops the room from spending a pattern as if it were a law, or a good story as if it were a proven fact, right before the money moves. The whole skill is noticing which arguments are heavy enough to deserve the question.

In the room, it comes down to one short declaration. Before a conclusion that matters, say which mode you are in and what it buys you. Deduction, so this holds only as long as the rule holds. Induction, so this is a good bet that assumes things stay stable. Abduction, so this is the leading explanation, not a verdict, and it earns a pilot. Analogy, so this is pointing us in a direction, not proving we have arrived. That one sentence costs nothing and changes everything, because it sets the confidence level before the conclusion lands, instead of letting a guess borrow the authority of a proof. And when the model starts swelling with detail, run the razor over it: ask whether each new assumption buys real accuracy or just props up a weak one, and cut whatever earns nothing. Every part of an argument should pay its way.

So the next time the room goes quiet and someone announces that the logic is clear, do not nod with everyone else. Ask the small, awkward question first. Which logic? And how much weight has it earned? Because a room that never asks will keep mistaking a confident guess for a settled fact, and it will keep being surprised when the guess gives way.

Decision Rule — The Mode Card

Before anyone says “therefore” on a consequential claim, name the kind of reasoning in use and the confidence it earns:

  • Name the mode and its warrant. Deduction is conditional on the rule. Induction is probabilistic and assumes a stable regime. Abduction is a leading explanation awaiting a test. Analogy is orienting, not confirming.
  • The Occam gate. When the model grows complex, ask whether each added assumption buys real precision or just props up a weak premise, and cut what buys nothing.
  • Save the declaration for arguments carrying real weight. Naming the mode is a five-second act that stops a pattern posing as a law, or a story posing as a proof.

One essay a week

The Decision Papers examines how institutions actually decide — and where decisions fail before anyone sees them fail. New essays arrive by email, free.

Leave a Reply