History of programs and nature

Given the idea of using programs as a basis for describing nature, one can go back in history and find at least a few rough precursors of this idea. Around 100 AD, for example, following earlier Greek thinking, Lucretius made the somewhat vague suggestion that the universe might consist of atoms assembled according to grammatical rules like letters and words in human language. From the Pythagoreans around 500 BC through Ptolemy around 150 AD to the early work of Johannes Kepler around 1595 there was the notion that the planets might follow definite geometrical rules like the elements of a mechanical clock. But following the work of Isaac Newton in the late 1600s it increasingly came to be believed that systems could only meaningfully be described by the mathematical equations they satisfy, and not by any explicit mechanism or rules. The failure of the concept of ether and the rise of quantum mechanics in the early 1900s strengthened this view to the point where at least in physics mechanistic explanations of any kind became largely disreputable. (Starting in the 1800s systems based on very simple rules were nevertheless used in studies of genetics and heredity.) With the advent of electronics and computers in the 1940s and 1950s, models like neural networks and cellular automata began to be introduced, primarily in biology (see pages 876 and 1099). But in essentially all cases they were viewed just as approximations to models based on traditional mathematical equations. In the 1960s and 1970s there arose in the early computer hacker community the general idea that the universe might somehow operate like a program. But attempts to engineer explicit features of our universe using constructs from practical programming were unsuccessful, and the idea largely fell into disrepute (see page 1026). Nevertheless, starting in the 1970s many programs were written to simulate all sorts of scientific and technological systems, and often these programs in effect defined the models used. But in almost all cases the elements of the models were firmly based on traditional mathematical equations, and the programs themselves were highly complex, and not much like the simple programs I discuss in this book. (See also pages 363 and 992.)