Comment Re:Proved conclusively? (Score 1) 269
I don't see how you can prove something conclusively in silico, you put in what you know and you get a distillation of it out. How can you discover* completely new physics when the computer can only start with a potentially incorrect/inaccurate theory and make deterministic calculations based on that input? I mean, you can't get out more than you put in, can you?
Actually, yes, you can get out more than you put in. These guys made the machine extrapolate laws of physics without any knowledge of physics or geometry.
They used a genetic algorithm to explain the measurements of a pendulum sways, and in the process the computer "invented"/"learned about" things like adding, substracting, multiplying, dividing, some algebra, conservation of momentum and Newton's second law.