Re: [silk] Building intelligent machines with casual reasoning

2018-08-23 Thread Charles Haynes
You mention Bayesian statistics as a thing like Pearls causality maths that's too complex.for most people and so hasn't caught on. I'd argue the exact opposite. Bayesian statistics ARE complicated but the first time I saw them my reaction was Oh My God this is going to change everything about how

Re: [silk] Building intelligent machines with casual reasoning

2018-08-23 Thread Charles Haynes
Thinking out loud, it seems to me that for most things Bayesian logic supercedes Pearls causality. The probability of something given a prior, versus the probability without the prior is, in some sense, the degree to which the prior "causes" the result. The beauty is that you can usefully reason