What do Boston drivers have in common with migrating geese, flocking cranes, flashing fireflies, and swarming ants? All of them are acting out a "natural algorithm." Unlike their software engineered cousins, these algorithms have been optimized over millions of years through natural selection. They do not come neatly encoded in readable form, yet scientists have become experts at modeling them. To call these dynamical systems "algorithms" is to suggest a new way of studying them, which goes beyond numerical simulations and relies on analytical tools from theoretical computer science. Can "algorithms science" do for complex systems what classical math has done for physics? I cannot answer this question but I can try to convince you why asking it is not completely ridiculous.