Every cat is made of glass. Fluffy is a cat. Therefore Fluffy is made of glass.
Ask a seven-year-old whether that is a good argument and you will usually get an objection to the first line, delivered with some energy. Cats are not made of glass. The child is right, and the child has also answered a different question from the one that was asked.
The question was whether the argument holds together — whether, supposing the first two lines were true, the third would have to be true as well. It would. There is no way for every cat to be made of glass, and for Fluffy to be a cat, and for Fluffy not to be made of glass. The argument is airtight. It also happens to start from something false, which is why its conclusion is false, and those are two separate facts about it.
That separation is the first genuinely difficult idea in logic, and we teach it before almost anything else.
Two questions, not one
An argument invites two questions and they come apart cleanly.
Does the conclusion follow from the premises? That is validity. It is a question about the relation between the parts, and you can answer it without leaving the page.
Are the premises true? That is a question about the world. Answering it may require looking something up, or asking someone, or years of study, and nothing about the argument's arrangement will help you.
An argument that passes both is sound, and soundness is what you actually want. But you cannot get there by treating the two questions as one, and almost everyone does treat them as one until they are shown otherwise. The natural response to an argument is a single verdict: good, or not. Splitting that verdict in half is the whole skill.
Why the nonsense helps
Which brings us back to the glass cats.
If you teach validity using true premises, the exercise quietly collapses. Every animal has a heart; a dog is an animal; therefore a dog has a heart. A learner nods along, and there is no way to tell whether they followed the argument or simply recognized that dogs have hearts. Their knowledge of the world does the work and the reasoning never has to.
Absurd premises take that shortcut away. Nobody has an opinion about the glass content of cats. The only thing available to think with is the shape, and a learner who gets the answer right has read the shape, because there was nothing else there to read. This is also why the exercises get funnier as they get harder, and that is not decoration — it is the same principle applied further.
The moment we are aiming for is small and specific: a learner says the argument is good, and then, unprompted, that the conclusion is false anyway. Once someone can hold those two thoughts at once, they have the distinction. Everything above it is elaboration.
What it buys, years later
The reason this comes so early is that a great deal is built directly on top of it.
You cannot study a formal fallacy without it. Affirming the consequent is invalid, full stop, and seeing that requires the ability to ask whether the conclusion follows while setting aside whether you agree with it. A learner who has not separated the questions will judge the argument by whether the conclusion sounds right, which is exactly the error the fallacy exploits.
You cannot locate a disagreement without it either, and this is the part that matters outside a classroom. Two people arguing usually agree on more than they realize. Sometimes they accept the same facts and disagree about what follows from them. Sometimes they reason identically from different starting points. Those are different disagreements, and they are resolved in completely different ways — one by argument, the other by evidence. Someone who cannot tell them apart will spend an hour producing facts at a person who never disputed the facts.
By the upper tiers, the distinction has become a habit of reading. Faced with an argument you find objectionable, the first question is not whether you object. It is: does it follow? A great many arguments you dislike are valid, and their premises are where your actual disagreement lives. Finding that out is not a concession. It is the only route to an objection that survives a reply.
Starting where it is easy
None of this requires a technical vocabulary. Tier I never uses the word "valid." It asks whether the story hangs together, and it uses cats made of glass to ask.
The words arrive later, once there is something for them to name — which is the order the ideas came in historically, and, we think, the order they are easiest to receive in. A learner who has spent a year noticing that arguments can hold together while being wrong is not learning a definition when they meet the term. They are learning the name of something they already have.