On puzzles

A puzzle is a problem that promises it has an answer. That promise is the whole trick.

Faced with an ordinary problem, most of the effort goes into deciding whether it is worth solving at all, whether it is solvable, and whether you are the person who should be solving it. A puzzle removes all three questions. Someone built this. It has a solution. You are allowed to try. What is left is the pure activity of thinking, with none of the anxiety that usually surrounds it.

That is why I like them, and why I think they are underrated as training. The skills a good puzzle exercises, holding several constraints in mind at once, noticing when an assumption is doing more work than it should, knowing when to stop pushing and start looking, are the same skills that hard real problems need. Puzzles just let you practise them without the stakes.

The best ones have a particular shape. They look impossible for a while, then one observation makes everything fall into place, and afterwards you cannot imagine not having seen it. That moment is what I am chasing in everything I build.