A tiny, unusual little wordplay game, scarcely IF. World-model is irrelevant, and the parser only accepts one type of command. You are given words, and have to supply a rhyme for each word. Your score is how many acceptable rhymes you supply within 60 seconds. The framing is that this is a psychotherapy exercise; you're being asked whether you feel emotional state X, and you respond with, 'no, I'm feeling Y'. Y does not always have to be an emotion, since the game recognises a wide range of rhyming words, some of which aren't even adjectives.
On the one hand, gameplay is simple and easy, so much so that it's tempting to think of this as a toy rather than a game; getting a good score might take a few tries, but mechanically it's not difficult. The game tells you the puzzle's parameters straight off. On the other, the framing creates a weird uneasiness about the exercise. A fair amount has been written about how the psychotherapy premise of Eliza tended to make players take it more seriously. Similarly, WDHYF has something of the tension of a word-association game about it; it's a silly exercise, but there's the sense that you're being judged on it. This is particularly true because the act in question is laying claim to emotions, which is something people generally invest a lot in. Laying claim to an emotion can be a vulnerable act of self-expression, and it's also a commitment of sorts, a reinforcement; we don't look kindly on people who fake emotions, and we don't want to be seen as someone whose emotions change rapidly for no reason ('emotionally unstable' is a euphemism for 'crazy'). Saying 'I am happy' or 'I am sad' costs you something.
But because of the constraints you're usually forced to lay claim to a rapid, little-considered series of random, mismatched, nonsensical emotions. So there's a sense -- slight, but disquieting -- of brainwashing-exercise about this: repetition, illogic, pressure, self-obliteration.