Personally I've always found Larry's talks very poor. They seem to wander all over the place, full of buzzwords and much of what he claims doesn't seem to have any connection to what he actually did.<p>Huffman encoding? Then why are <i>so many</i> of the defaults wrong? Nearly any perl programmer would tell you that your programs should start with "use strict", so wouldn't Huffman encoding dictate that "strict" would need to be turned <i>off</i> if you <i>didn't</i> want it? And choosing "special return value" to signal failure can't possibly be shorter than using exceptions, it can only be <i>as good</i> if you don't handle any errors.<p>He's always reminded me of someone trying to be "deep" after smoking a joint.