The "10x productivity" idea is somewhat silly anyhow - sure, some people are quite productive, but mostly if one guy is 10x another, the other guy just sucks.
It is more like 10000x rather than 10x.
First off, we humans are just barely intelligent enough to write nontrivial computer programs in the first place. I believe strongly if humans were on average ten percent less intelligent we would still be stuck on the towers of hanoi and bubblesorts. Of course, what "intelligence" means in the context of computer programming is highly idiosyncratic and not very well understood. Finding those people with the right cognitive toolkit for solving a particular set of coding problems can make a huge difference in the success or failure of a company. Or at least a product.
Second, I've lost count of the number of times where I've ran into someone (or some enormous team of people) who labored on some project for years and then ran into some dude (nearly always a dude, sorry ladies) who solved exactly the same problem in a few weeks. Or less. For an infamous example consider the history of the Xanadu project versus the early http servers and Mosaic. Yes, yes, I know Xanadu was trying to do something totally different than the WWW did. Of course, WWW worked and Xanadu never did.
Third, a lot of the real value add is the "Aha!" type of insights that translate large, intractable problems into easily solvable ones (see again Xanadu/WWW). The people who can provide those kinds of insights are rare and precious. The whole purpose of the enterprise is to solve problems, and if you aren't finding and taking vicious shortcuts to solve those problems you aren't doing your job.
In the end I absolutely agree in the sense that writing software is largely a creative enterprise, not an engineering one. And the people who write software need to be managed and compensated as creative people are, not like they are replaceable parts. Because that is the other point where software authoring is so different from other enterprises -- coders are often extremely specialized, to the point where managing coders, especially talented, productive coders, becomes largely a problem of matching tasks to appropriate talent.