I think the biggest driver of the "10x programmer" concept, the thing that validates it, is the related concept of the "negative net productivity programmer."

Some programmers, in some environments, produce less than they cost to supervise and for others to fix their mistakes.

Some programmers, in some environments, look productive because they do the 20% of the work that gets you 80% of the way there. They start a whole lot of projects which almost work, then it takes a really good programmer a lot of time to repair the relationship with the customer.

At this point in time, however, I see the productivity of the organization as the real issue. Even a 10x or 100x programmer is only going to be that way if they are in a zero-bullshit environment. It takes one bad co-worker or bad boss to turn somebody like that to an 0.1x programmer. A lot of it is choosing your environment.