Creative metaphor is a phenomenon that stretches and bends the conventions of semantic description, often to humorous and poetic extremes. The computational modeling of metaphor thus requires a knowledge representation that is just as stretchable and semantically accommodating. We present here a flexible knowledge representation for metaphor interpretation and generation, called Talking Points, and describe how talking points can be acquired on a large scale from WordNet (Fellbaum, 1998) and from the web. We show how talking points can be fluidly connected to form a slipnet, and demonstrate that talking points provide an especially concise representation for concepts in general.