A nudge (from Thaler and Sunstein's book of the same name) is any small design choice that predictably shapes behavior without restricting anyone's options or changing their economic incentives — the philosophy behind it is sometimes called libertarian paternalism: guide the easy path toward a good outcome, without blocking any other path.
Whatever option is pre-selected gets chosen disproportionately often, simply because changing it takes more effort than accepting it. Organ donation opt-out countries have dramatically higher donation rates than opt-in countries for exactly this reason — the outcome-relevant choice barely changed, only which option required effort.
Some of the most effective nudges don't ask for a decision at all — they build the desired outcome into a step the user is already taking for another reason. A retirement contribution automatically increasing a little with each raise (unless opted out) achieves the savings goal as a side effect of a raise the user already wanted.
| Question | Nudge | Dark pattern |
|---|---|---|
| Is the alternative option still genuinely easy to find and choose? | Yes | No — hidden, tiny, or deliberately confusing |
| Does it benefit the user, or only the business? | The user, primarily | The business, at the user's expense |
| Would the user feel tricked upon noticing it? | No | Yes |
| Is it reversible without real difficulty? | Yes | Often deliberately hard to undo |
This is the boundary that matters most in this whole track
A confirm-shaming cancel button ("No thanks, I don't want to save money"), a subscription that's one click to start and eight steps to cancel, or a pre-checked box adding a paid add-on at checkout are all dark patterns, not nudges — they fail the table above on every row. Recognizing this line matters as much as knowing the techniques themselves.