Half of this year's learning has an expiry date

You will spend some evenings this year learning things. Nobody tells you, while you are doing it, which of them will still be true when you need them.

Here is a way to find out that costs nothing. Go and look at the technical videos people were making a few years ago, and see which ones are still being watched.

The walkthroughs are quiet now. Very few people are following a step-by-step guide for a version that stopped shipping. But the ones that explain a choice, and what each side of it costs you, are still collecting views this month. Same year of upload. Often the same channels. Half of that work survived and half did not, and the half that survived was not the harder half.

What actually separates them

Everything you learn sits on one of two footings.

The first is something true about a product. This setting lives in this menu. This flag does this thing. That knowledge arrives with a version number attached, and an expiry date nobody shows you at the time.

The second is something true about the problem. Adding machines gives you more of them to fail, and more of them to keep in step with each other. Buying one bigger machine gives you a single place for everything to stop at once. That knowledge has no version number, because problems do not ship releases.

From the inside, while you are learning them, the two feel exactly alike. They cost the same evening. They give the same small satisfaction at the end of it. And only one of them is still standing in four years.

Why the second one lasts

The obvious answer is that it is deeper, and the obvious answer is wrong. Plenty of hard, specific tool knowledge has gone every bit as cold as the easy kind.

Think about what has to happen in the world for something true about a product to stop being true. Almost nothing. Someone renames a flag over a lunchtime. What you learned did not become false, exactly. It stopped pointing at anything, because the thing it named has moved and did not tell you.

Now think about what has to happen for a trade-off to stop holding. The problem itself would have to change shape. Coordinating work across machines would have to stop costing anything. Failure would have to stop being something you design around. That is an enormous thing to ask of the world, and the world mostly does not do it.

So it was never about depth. It is about what the knowledge was fastened to. One layer is fastened to decisions somebody else makes on their own schedule. The other is fastened to the shape of the work, and the work is not on anybody's release calendar.

The question, and where it earns its keep

Before real time goes into anything: is this true about a product, or true about the problem underneath it?

You are not banning the product half. You need it to be useful on an ordinary Tuesday, and you will keep learning it for as long as you do this job. You are deciding the split on purpose, rather than letting whatever has the best tutorial decide it for you.

It works the same way when you are the one writing things down. The next time you hand a document to somebody who will open it in two years, write down what the choice cost you. Not the steps, because the steps will already be wrong by the time they read them. What you gave up when you picked this, and what you would need to see before picking differently.

Try this against whatever your own team keeps on a shelf. Which files are people still opening after a couple of years? It is rarely the ordered how-to. It is the small number of pages where somebody was honest about what the team agreed to give up. Everything underneath those pages gets replaced, more than once, and the reasoning keeps working anyway.

And the shape is bigger than learning. A procedure that is already wrong. A guide that aged out. A design document nobody opens. In each case what decayed was the scaffolding, and the reasoning inside it was fine. The technical symptom is not necessarily the actual problem.

The sort takes about ten minutes on paper and roughly forever in your head, which is why there is a page for it. It has the list on one side and, underneath, the three questions worth answering at the foot of anything you hand over. Run whatever you were going to start tonight through it first:

https://dareomotosho.com/resources/field-kit