pg provided no data, he provided a sketchy anecdote and made an unsupported claim.
The idea of offering more rewards is that people who otherwise go into other fields (which are more rewarding, socially and financially), would then be attracted programming. A percentage of those people will be great programmers.
Grouping great programmers together gives them a tiny, tiny, tiny chance at large rewards. It's a gigantic risk. People aren't stupid. Most go to paths with less risk. The market has corrected itself. It attracts people in a manner appropriate to its risk/reward equation.
The way to reduce that risk is to provide more guaranteed rewards. Which means more compensation. Which means transferring more of that unused capital sitting around in VC accounts to the great programmers. Which means the limiting factor is the capital put into the system.
Absent some big payday from a startup, which is a roll of the dice not in their favor, programmer compensation is low relative to other, much less risky career paths. If the field truly wants to attract the best and brightest, it's got to offer a lot more rewards.
How much of a reward are "great programmers" 71-100 that company needed in pg's example going to get, anyway? How huge a success does the company have to be for someone that far down the line to come out way ahead? You have to offer those taking such a position a lot of up front compensation, or you won't get them.
As far as "great programmers" being intrinsically motivated, that's only true of people who are currently in the field, as the extrinsic motivation isn't really there, is it? So how do we know extrinsic doesn't work? We're not trying it.
It's ludicrous to think we have all the great programmers we could have. There are any number of people who are capable. Any number of highly intelligent, incredibly driven people who could fit the bill. They just go into other fields with better odds of superior rewards.
Want more great programmers, offer better rewards. They will come.