This one is personal: Philosophy 201, Formal Logic, is the single best course I took in college for my technical career, more than anything in my actual major. Everything below is the version of that class I wish someone had handed me before I needed it.

Breaking Down a Problem

Every stuck problem starts life as a single bundled sentence: "this plan won't work," "the team isn't aligned," "the model is underperforming." Treated as one claim, it is unarguable, you either accept it or reject it wholesale. But almost no real claim is atomic. It is a conjunction, several independent propositions wearing one sentence as a coat, and a conjunction only fails if any one part of it is false.

The first move is just unbundling: writing out what would each, separately, have to be true. Hover each claim below, every one has its own test.

"This plan won't work" breaks into:
Claim A The budget is insufficient for the scope. Falsified by a line-item estimate that comes in under budget.
Claim B The timeline ignores a known dependency. Falsified by showing the dependency is already resolved.
Claim C The team lacks a skill the plan assumes. Falsified by naming who on the team already has it.

People argue the bundle as though it stands or falls as a block. Once it is unbundled, most "disagreements" turn out to be agreement on two of three claims and a real, narrow dispute on the third. The conversation stops being about the plan and starts being about the budget line, which is a conversation that can actually end.

Finding the Load-Bearing Premise

Once a claim is isolated, ask what it depends on. Every conclusion sits at the end of a chain. Trace it back far enough and you do not hit more logic, you hit an axiom: a value judgment or an empirical claim that is not derived from anything further back. That is the actual foundation, and it is usually smaller and more specific than the argument built on top of it.

P1: the vendor's uptime SLA is 99.9%
P2: a rate below 99.9% breaks our checkout flow
modus ponens
we can build checkout on this vendor

Nobody disputes the inference here, it is valid. What they dispute, if anything, is P1 or P2: whether the SLA number is real, or whether the actual break-point is 99.9% and not 99.5%. Chasing the conclusion in circles is what happens when this step gets skipped. Chasing the premise ends the conversation, because a premise is the kind of thing you can actually check.

Understanding a Person

The same machinery reads a person, not just a plan, with one adjustment. With a document, every premise is on the page. With a person, you only ever see the conclusion: what they decided, what they said, what they did. So that is what gets argued with. Someone reaches a conclusion you find alarming, and the instinct is to attack the conclusion directly, to explain, again, more slowly, why it's wrong. That almost never works, because the conclusion was never the problem. It followed correctly from a premise you have not found yet.

The premises are invisible, so the job is abductive: working backward from a conclusion to the smallest, most coherent set of premises that would make it the reasonable output of a valid inference. Find that premise and put it on the table, and the conversation moves to a place it can actually be resolved.

The Principle of Charity
Before concluding someone reasoned badly, check whether you correctly reconstructed what they started from. Most of what reads as a broken inference is actually a valid inference from a premise you don't hold, or didn't know they held. Given their premise, their conclusion is the rational one. It only looks irrational from where you're standing, on a different premise.

"Why would anyone think that" almost always has an answer once the missing premise turns up. Sometimes, once you find it and discuss it directly, you find you actually agree, and the conclusions realign on their own. And sometimes you surface the premise and still don't share it. That is fine too. You do not have to end up agreeing to end up understanding. Knowing that someone values speed over certainty, or the group over the individual, or the letter of a rule over its intent, is enough to see that they are being consistent and rational from where they stand, not stubborn or slow. The disagreement moves from "they don't get it" to "we start from different places," which is a much smaller and more honest place to be stuck.

The same read applies when a person seems to contradict themselves, arguing for caution here, for boldness there, on cases that look identical from outside. That usually is not confusion either. It is evidence of two premises held at once, both real, that have not yet been tested against each other. That seam says more about what someone actually values than either position alone does.

Where They Meet

A person stuck on a problem is both objects at once: someone reasoning correctly from premises that have not yet been checked against the problem's actual structure. The friction is almost never "their logic is broken" or "the plan is bad." It is that the premises doing the work, on both sides, have not been said out loud yet, so both sides keep restating conclusions at each other instead.

Problem
Person
shared premises:
where agreement is
actually possible

Decomposition, traced premises, and named contradictions all do the same thing: they move the argument off the conclusion, which is unarguable, and onto the premise, which is not. Sometimes that premise turns out to be shared, and the conclusions fall in line behind it. Sometimes it does not, and what you get instead is the clearer, calmer kind of disagreement: not "you are wrong," but "I see why you'd land there, from where you're standing."

01
List the atomic claims
Don't argue the conjunction as a block. Unbundle the sentence into the claims that actually have to hold.
02
Trace each claim to its premise
Follow the chain back to the one thing it can't do without, usually smaller and more checkable than the conclusion.
03
Ask what would falsify it
The shape of a claim's negation tells you what is actually being claimed.
04
Argue the premise, not the conclusion
When a person seems wrong, reconstruct the premise behind their conclusion and discuss that, out loud, before assuming the inference itself is broken.
05
Different premises are an okay place to land
If you surface the premise and still disagree, that's fine. You now know they are being rational from their premise, not wrong or irrational, and that is worth more than winning the argument.
Further Reading
  • Aristotle, Prior Analytics — the founding text on the syllogism and premise-conclusion structure underlying the claim-unbundling method above.

Bridge the Gap. Empower Their Decisions.

Teaching a team to reason clearly, together, is a humanities skill as much as a technical one. Tech CoLab's games build exactly that muscle across technical and non-technical audiences alike.

Read: Improving Spock's EQ → Read: People, Process, Tools →