Your words are making your decisions for you.

In 1889, an Ethiopian emperor and the Kingdom of Italy signed the same treaty in two languages. The two versions agreed on everything except one word. That one word decided whether his country would stay free.

His name was Menelik II. The document was the Treaty of Wuchale, comprising twenty articles meant to settle matters of friendship and trade between Ethiopia and a rising European power. Nineteen of them caused no trouble. The trouble lay in Article 17: it existed in two languages that did not say the same thing.

The Amharic version, the one Menelik read, said that Ethiopia could use Italy as its channel to other European governments. Could. A convenience, his to take or leave. The Italian version, the one Rome filed away, said that Ethiopia would conduct all of its foreign affairs through Italy. Would. An obligation. The whole distance between the two texts was the distance between may and must, one word, carrying the sovereignty of a nation on its back.

Italy knew exactly what it was holding. Count Pietro Antonelli, the envoy who negotiated for Rome, had put his name to both texts, and the Italian one he produced carried an obligation the Amharic one left as a choice. Italy told the governments of Europe that, on the strength of its version of Article 17, Ethiopia had become an Italian protectorate, and for years afterward it called the gap between the two texts an honest slip of translation. A whole country, claimed, in the space of one verb.

Then Menelik found out what he had supposedly agreed to.

The word comes first

Hold the emperor there a moment, because his situation exposes the thing most of us get backwards every day.

We treat a word as a label, a sticker we attach to a decision after the decision is already made. In that picture, the deciding comes first, and the naming comes second, so the word is downstream, cosmetic, safe to swap and safe to ignore.

The picture is upside down. The word comes first. It is the instruction you follow before you have noticed you are following it. May and must were two different arrangements wearing one article number, and everyone who acted on them acted differently. Rome moved to swallow a country. Menelik moved to keep one.

You can watch the same mechanism in any room where being wrong carries a cost. A doctor looks at a fever of 103 and picks a word for it. Call it the illness, and the work is to push the number down. Call it a symptom, and the work is to hunt for the cause. Same fever, and the word walks the doctor toward two different patients. NASA lost a spacecraft on its way to Mars because one team wrote a number in pounds and another read it in newtons, and the orbiter was gone before anyone caught the mismatch. In every field where a mistake leaves a body or a crater, people are obsessive about their terms, because the word is the first link in a chain that ends in a consequence.

They guard the vocabulary because words fucking matter.

Business is the one room that forgot. It has talked itself into being the exception, the place where words are only labels, and you can swap them around a conference table because nothing physical depends on the swap. So the sharpest person at the table waves a distinction away as semantics and feels sophisticated doing it. In a cockpit or an operating theatre, the same move would get them escorted out of the building.

Sometimes it really is just semantics

Here I have to be fair, because the lazy version of this argument is a lie. Most of the time, the person who says “that’s just semantics” is right. Most word fights are exactly that. Two terms, one meaning, no difference in anything anyone does. Calling those fights what they are is a service, and the people who do it should be thanked.

So the claim is not that words always matter. Plenty of words are interchangeable, and treating every one of them as sacred is its own kind of paralysis.

The whole game is telling the two apart, and one question does it. A distinction is merely semantic when the two words send you to the same action. If you would do the same thing either way, drop it and move on. If the two words send you somewhere different, the words are carrying a decision, and that decision is the real fight.

That is the test. The question is always the same. Do the two words make you behave differently? Article 17 failed that test in the most expensive way a sentence can. May and must sent two nations toward two different centuries.

Why the smartest person in the room is the worst at this

Notice that nobody in the Wuchale story was fooled by a word. The Italians chose theirs with care, because they understood what a word could carry. Menelik grasped the stakes the instant he saw the discrepancy. The people in that story all took the word seriously, the Italians most of all. The danger of “it’s just semantics” belongs to a different room, the one where smart people have decided words are beneath them.

A man named Chris Argyris spent decades at Harvard working out why the most capable people in an organization are so often its worst learners. His answer is one of the most useful and least flattering ideas in management.

Argyris drew a line between what you say you are doing, which he called your espoused theory, and what your behaviour shows you are doing, your theory-in-use. He found that people vary enormously in what they say and barely at all in what they do, and that the gap between the two is invisible to the person standing inside it. It is widest in the people most certain they have no gap.

Why are smart people the worst at closing it? Because they have rarely had to. They have spent their lives being right; being right is their advantage and their identity, and they have almost no practice at the one move learning asks for, which is staying inside the feeling of being wrong long enough to look at it. So when something threatens to expose a gap, they defend instead of investigating. Argyris called these defensive routines, and he saw that they are most polished in the most accomplished people. He had a name for it I cannot improve on. Skilled incompetence. The talent for looking like you are reasoning while you are protecting yourself.

He also split learning into two kinds. Single-loop learning fixes the answer and leaves the assumptions that produced it alone. A thermostat sees the room is cold and turns up the heat. Double-loop learning turns on the assumptions themselves and asks whether the thermostat is set right at all. Rewriting your website because the market is confused is single-loop. Asking whether you have earned the position you keep trying to broadcast is double-loop. Re-examining the words you use is double-loop learning by definition, because your words are your assumptions in everyday clothes.

Seen this way, the reflex stops looking like rigour. “It’s just semantics” is a defensive routine. It lets a smart person wave off a distinction while leaving their own assumptions untouched. It feels like rising above pedantry, and what it does is pull the door to real learning quietly shut, with the one person most able to walk through it standing on the wrong side.

Roger Martin, whom Thinkers50 once named the most influential management thinker alive, built his whole definition of strategy on Argyris’s distinction. For Martin, your strategy is whatever your actions reveal it to be, and the words on the slide are only a claim about it. A man of his standing does not anchor a life’s work on something cosmetic.

The same word that started a war, on a smaller battlefield

Bring it down from emperors to the version that drains companies a quarter at a time.

There are two things people in business fuse into one. There is the concept you come to own in a customer’s mind, and there is the way you describe yourself.

They feel like the same activity, and they are worlds apart. You author your description, and you can rewrite it on a Tuesday. You earn the concept slowly, through decisions that cost you something, and the market keeps the final say. Confusing the two is confusing your résumé with your reputation, the gap between what you say about yourself and what the room says about you once you have left it.

That blur is where the money goes. If you believe describing yourself and earning a perception are the same act, you believe you can fix a perception problem by editing the description. You cannot, and here is the clean proof. If a company decided how it was seen, the gap between how it wants to be seen and how it is actually seen could not exist. That gap is the most ordinary thing in business. The market decides, and the gap is the receipt.

It is also why announcing what you are tends to backfire. The instant a claim is spotted as a claim, people discount it, because everyone walks around with a fine detector for being sold to. A claim is cheap. Anyone can say they are innovative or trusted or the leader. A costly decision carries a weight a sentence never will, which is the whole reason it persuades. You do the expensive thing only a serious company would do, and you let the room reach the conclusion on its own, because a conclusion people arrive at themselves is the only kind they defend.

The bill comes due downstream

The dismissal would be harmless if it were free. It hands the invoice to someone else, later.

Picture the team that has decided a perception problem is a messaging problem, because the word they reached for told them so. They leave the product alone, and the price, and the question of who they should refuse to serve. They rewrite the homepage. They run the workshop. They argue about adjectives while the cause sits untouched, and the word is what sent them there.

History keeps the receipts on this too. General Motors spent the better part of a decade and enormous budgets telling people the Oldsmobile was not their father’s car. The ads were polished, the brand died anyway. Roger Martin, who argues that a brand is built through how a company competes, used Oldsmobile as his example of a name the product underneath had stopped backing. The problem lived in the car, and no sentence about the car could fix the car. Quibi raised close to one and three-quarters billion dollars, hired famous founders, and bought a Super Bowl ad. The marketing was flawless, and the company was gone in roughly six months, because communication can do everything communication is able to do and still cannot manufacture a position the thing underneath has not earned.

What the emperor did

Back to Menelik, holding a treaty that said one thing to him and another to Rome.

He could have let the word stand. The pressure to do exactly that was enormous, and the power behind the Italian reading was real. His wife, the Empress Taytu, was the first to press him to throw the treaty back at them. He refused the Italian reading. He told Europe he had signed up for a convenience, and that no country of his would be handed over on the strength of a verb Rome had quietly changed. When Italy held to its version, the disagreement over a single word escalated into war.

It ended on the first of March, 1896, near a town called Adwa. Menelik’s army met the Italian force in the field and broke it. Ethiopia stayed free. Adwa became the most famous defeat a European army suffered in the whole scramble for Africa, and it kept Ethiopia independent while its neighbours were being carved into colonies. A new treaty was signed that autumn, and this time the word came out the way Menelik had always read it.

A single verb started a war. A battle decided which reading of the verb the world would honour. At no point was any of it just semantics, and everyone with something real at stake knew it.

So the next time you hear yourself say “that’s just semantics,” stop for one second and run the test. Name the action each word would send you to take, out loud and specific. If both words point you at the same place, you were right; let it go with my blessing. If they point you somewhere different, you have caught yourself, because the words were carrying a decision you did not want to look at.

A wrong answer is cheap. It gets caught and corrected. The expensive sentence is the one that ends the conversation a single beat before the learning would have started, said with total confidence, by the person in the room who needed it least.