I designed the Minivac 601, a digital computer trainer that taught business people how computers work. The Scientific Development Corp sold it from 1961, bringing computer technology education beyond engineers.
With John McCarthy I co-edited Automata Studies, published in 1956. Its categories followed the subject headings of my 1953 paper "Computers and Automata", framing early research on the science of intelligent machines.
I applied my mathematics to investing. Shannon's demon holds equal parts cash and a stock and rebalances regularly to profit from random price jitter. A 1986 Barron's report found my return beat 1,025 of 1,026 mutual funds.
In 1950 I designed and built Theseus, a learning machine: a mechanical mouse moving through a maze, tracked by an electromechanical relay circuit beneath the surface. It was one of the first experiments in artificial intelligence research.
At Bell Labs in World War II I worked on fire-control systems. I invented signal-flow graphs in 1942 and co-authored a 1945 essay treating data smoothing as separating signal from noise — engineering that heralded the Information Age.
In a paper completed in 1949 and published in March 1950, I estimated the game-tree complexity of chess at roughly 10^120. The figure, now the Shannon number, is still regarded as accurate in computer science research.
In 1949 I presented "Programming a Computer for Playing Chess", one of the first articles on computer chess. I described position scoring and move selection, laying foundations for future chess programs and AI research.
In 1956, as information theory spread fast, I wrote the one-page editorial "The Bandwagon" for the IRE Transactions on Information Theory, urging researchers to keep a scientific attitude.
I compressed Kerckhoffs' principle into five words: "The enemy knows the system". Now called Shannon's maxim, it is a design rule for cryptography and security engineering.
In classified Bell Labs research, published in 1949, I proved the one-time pad unbreakable and showed every unbreakable cipher must share its properties. It put cryptography on a mathematical, scientific footing.
