Analogies are a loan against understanding

A good analogy is the fastest teaching device there is. Someone says “a database index is like the index at the back of a book” and a concept that would have taken four paragraphs lands in six words. You feel the click. You can use it immediately.

It also has to be paid back, and most courses never come back to settle up.

What an analogy actually does

An analogy borrows structure. You already have a working model of something — books, water, post offices, restaurants — and the analogy maps the new thing onto that existing model so you can reason about it before you understand it.

That is a genuinely brilliant move, and the emphasis belongs on before. You are not being given understanding. You are being given a usable approximation on credit, against understanding you will build later.

The debt comes due when you need to do something the analogy does not cover. And every analogy has a boundary, because if it mapped perfectly it would not be an analogy, it would be the thing.

The boundary is invisible from the inside

This is the whole problem in one sentence: the analogy does not tell you where it stops working.

Electricity is like water in a pipe. Voltage is pressure, current is flow, a resistor is a narrow section. This is excellent, it gets you through a great deal of basic circuit reasoning, and it quietly falls apart the moment you meet inductance or anything that behaves like a wave rather than a fluid.

A promise is like an IOU you can redeem later. Fine, until you need to know that the work started when the promise was created rather than when you asked for the value, at which point the IOU model produces confidently wrong predictions about what runs when.

DNA is a blueprint. This one is actively misleading and is taught to essentially everybody. A blueprint is a scaled plan of the finished object, read by someone who builds it. DNA is nothing like that — there is no plan of the organism anywhere in it and no reader standing outside. It is closer to a recipe, and really not much like that either.

In each case the failure has the same signature: you reason past the boundary, you get an answer, and it feels exactly as confident as the answers you got inside the boundary. The analogy has no mechanism for saying “not here”. Nothing about being wrong feels different from being right.

Which is why unpaid analogies compound

Now watch what happens across a whole course that leans on them.

Chapter two gives you the water pipe. Chapter five explains a new idea in terms of the water pipe. Chapter eight explains a third idea in terms of the fifth. You are now three levels deep in a metaphor that was approximate on its first use, and every layer has inherited the original boundary while adding one of its own.

The reader who finishes that course can produce fluent-sounding explanations of the entire subject and cannot answer a question that was not anticipated. Worse, they cannot tell — because the fluency is real. They really can explain it. They explain it in terms of pipes.

The tell, if you want to check yourself: if you cannot say what the thing actually does without reaching for the comparison, you are holding the loan and not the asset.

How analogy should be used

Not never. It is too useful for that. But rationed, and repaid:

• Once, at the moment an idea is genuinely unreachable otherwise. Not as the default register for a whole document, and not as the opener of every section.
• Immediately dismantled. The sentence after a good analogy should be the one that says where it breaks. “This is a useful picture, and it stops working as soon as X.” That single sentence converts a loan into a tool, because now the boundary is marked.
• Replaced. By the end, the mechanism should be doing the work and the analogy should be a memory aid rather than the load-bearing structure. If the course ends with the reader still standing on the metaphor, it did not finish the job.

The thing that beats an analogy

There is an alternative that is almost always better, and it is not more rigour.

An analogy compares the thing to something else. A concrete example is an instance of the thing itself, in a domain the reader already knows. Not “a queue is like a line at a coffee shop” but “here is the actual queue in the system you work on every day, and here is what happens when it fills up.”

The example has no boundary problem, because there is no mapping to break. It is the real mechanism, observed in familiar surroundings. Everything you learn from it is true of the thing, and every intuition it builds is one you can keep.

So why do courses reach for analogy instead? Because an example from your world requires knowing what your world is.

A course written once for everybody cannot use the system you work on, the subject you already studied, or the problem that made you go looking in the first place. It has no idea. So it must reach for something it can assume everybody has — coffee shops, libraries, post offices, water — and that requirement for universality is exactly what forces it into comparison rather than instance.

The analogies are not a stylistic weakness. They are what you get when the writer is not allowed to know who is reading.