The Unexpected Examination
The puzzle
A teacher tells the class on Friday: "There will be a test one day next week, Monday to Friday, and on the morning of the test you will not know it is coming."
A pupil argues it is impossible. It cannot be Friday, since by Thursday evening it would be the only day left and so expected. Then it cannot be Thursday either, since Friday is already ruled out. By the same step it cannot be any day. The class relaxes - and the test arrives on Wednesday, entirely unexpected. Where is the error?
Stuck?
One at a time, smallest first. Open only as many as you need — the whole point of the puzzle is the part you work out yourself.
Hint 1
The argument assumes the teacher's announcement is true in order to prove it cannot be carried out. Is that stable?
Hint 2
Ask what the pupils actually know on Tuesday evening, given that they have just proved no test can happen.
Show the answer Hide the answer
The elimination argument undermines its own premise. Having concluded that no test can be a surprise, the pupils expect no test at all - so any day becomes a surprise, and the Friday step they started from no longer holds.
How to get there
The Friday step is sound on its own: if Thursday passes with no test, Friday is forced and therefore not a surprise. The trouble is the next step, which treats "Friday is impossible" as something the pupils reliably know while they are still reasoning about Thursday.
They only know it because they are assuming the announcement will be fulfilled. But the conclusion they reach is that it cannot be fulfilled, which destroys the assumption that each step relied on. The chain of reasoning is self-defeating rather than valid: it uses a premise to derive the premise's falsity.
And the outcome proves it. By Tuesday evening the class has convinced itself there will be no test, so Wednesday's test genuinely is a surprise, and the teacher's announcement comes out true. The paradox is usually classed with the self-referential ones: the announcement is about the pupils' own state of knowledge, and reasoning about it changes the thing being reasoned about.
The technique worth keeping
When an argument derives a contradiction, check whether it used its own conclusion as a premise along the way. Statements about what someone will know are especially prone to this, because the act of reasoning alters the knowledge the statement describes.