[@Bishop]

why is it idiotic?

Sure, it might fail. The first few (hundred) times it will probably come up with inaccurate results, but what's the harm in seeing how good it can get at such? For research purposes, y'know?

There's all sorts of factors out there that can mess up results, but that's where it gets interesting in our opinion. How much data does a system like that need to correct for such factors?