What is a closure in JavaScript, and when is it useful or surprising?

I use closures deliberately for small private state and dependency capture, but I keep ownership visible. A long-lived listener can retain a much larger object graph than expected through its lexical environment, so cleanup still matters. In React, stale-closure bugs are another form of the same mechanism: a callback belongs to one render and sees that render's values unless dependencies or state-update patterns reconnect it to current state. When capture intent becomes difficult to see, I prefer an explicit state object or reducer whose lifetime and transitions are easier to inspect.