Not all friction is a failure of design. Some of it is the design.
The consent screen I built for a research survey last year is a good example. Ethics required the participant to actually read the disclosure — not skim it, not scroll past it. So the confirm button stayed disabled until the user reached the bottom of the text. One extra step. Measurable drop-off. Intentional.
The instinct in interface design is to remove friction wherever it appears. That instinct is usually right. But there’s a class of interactions where friction is doing real work: confirming a destructive action, committing to a goal, agreeing to terms that actually matter. In these moments, the tap that costs the user two seconds is also the tap that makes the interaction mean something.
The risk is confusing necessary friction with lazy design. A button that’s hard to find is not friction — it’s a bug. A confirmation step before deleting something that can’t be recovered is friction earning its keep.
The question worth asking: what does removing this friction make easier? Sometimes the answer is “the thing we want the user to do.” Sometimes the answer is “doing it by accident.”
Friction that protects against accidents has a different character than friction that just slows things down. The first kind is architecture. The second kind is a problem.
Building with friction means knowing which one you’re adding — and being willing to defend it when someone asks why it’s there.