[Aron]

I think if you have good natural instincts about artificial intelligence you realize that Hinton really is the god in this space. If you don't, you get distracted by LeCunn and Schmidhuber and whatever lesser minds. The capsule theory here is maybe not the best implementation, but the intuition behind it is still leading the way. The undifferentiated mass of neurons is not how evolution has solved our problems. 3d geometry is intrinsic to the low level design. It shouldn't be learned. It should be assumed.

I reject your generic devotion to process. The real leadership and the process are far different. The process of peer review might do a good job of rejecting bad ideas, but it does a lousy job of accepting revolutionary ideas. I bet you don't understand the difference.

[aoeusnth1]

While I agree wholeheartedly that peer review is fundamentally a risk-averse, conservative process, I bristled when I read the statement that LeCun as a "lesser mind." That's quite rude, and uncalled for.