I used to begin learning by looking for the best explanation. This feels sensible. It is also a good way to borrow someone else’s certainty.

Once the answer is on the page, every step towards it starts to look inevitable. I recognise the argument, think “of course”, and mistake that feeling for understanding. Close the book a day later and very little remains.

So I am trying a small reversal: guess first.

Before I read the chapter, watch the video or ask a model, I write down what I think will happen and why. The guess does not need to be polished, or even plausible. Its job is to leave evidence of my mind before the answer edits it.

Before checking a chess engine, choose the move. Before reaching for the equation, sketch which way the object will move. Before learning why something failed, predict where the failure began. The guess gives the explanation something to collide with.

When I am wrong, the difference is useful. It shows the exact place where my picture of the world and the world parted company. When I am right, I still have to explain why, then change one condition and see whether the explanation survives.

Two paths from something to learn: reading the answer first fades within a day, while a guess that collides with the answer ends in a correction that sticks.
Both paths reach the same answer. Only one records what it changed.

This is more uncomfortable than reading first. A prediction creates the possibility of being visibly wrong, even when I am the only person who will see it. But that small commitment is what turns the answer from information into a correction.

Afterwards I can study, redraw the idea in my own words, try it, break it and return to it later. Those steps matter, but the guess gives them an address. Without it, I mostly learn that a good explanation sounds good.

I do not want to finish a chapter knowing that the author made sense. I want to know what I believed before it, where that belief broke and what replaced it. The answer is more useful after a guess.