People can write if-then rules by hand, or a computer can learn from data. Machine learning is a kind of statistical inference that finds patterns in data, and it needs a great deal of training data, usually supplied by people. We compare both ways on Sky Sorter and measure which works better.