The amount of Claude Shannon worship in YC is astounding. Shannon basically lied (possibly to himself too) about AI for about 15 years from 1950 on.
It appears as if Shannon created a maze-solving robot mouse with memory, and also published the basic idea for computer chess around that time. Both ideas seem fundamentally sound.
And even if he had somehow been wrong on something at that stage in his life, he had basically created the science of communication single-handedly long before. Plus some minor work in cryptography.
Dollar cost averaging doesn't work because it makes an assumption that doesn't hold in the real world. It assumes that stocks trade in a range, and revert to a "true" price over time. If there were a true price, then you would in fact buy more when the price was low and less when it was high and DCA would work.
But stock prices look more like a random walk, and they display no tendency to revert to a mean.
Here's an article I just found: http://www.crossingwallstreet.com/archives/2010/11/dollar-co...
They don't use it.