If we wanted to bring a HAL-like being into existence, what are the millions of things that should be used to prime HAL's knowledge pump? How should they be represented inside the machine so that it can use them efficiently to deduce further conclusions when needed, just as we would? Who will do the actual entering of all that data? Assuming it's done by a large group of individuals, how would they keep from diverging or contradicting each other? A review of the development and implemtation of the CYC program shows us how, through applications such as natural language understanding, checking and integrating information in spreadsheets and data bases, and finding relevant information in image libraries and on the World Wide Web, the key point is that if you have the necessary common-sense knowledge, you can make the necessary inferences quickly and easily; if you lack it, you can't solve the problems. Ever.