I think it's sort of funny when people think 10x engineers don't exist. For most of us, it's pretty obvious because we've probably worked with one. (If you haven't: you might want to rethink your job choices :-))

If you think about it, there's essentially no physical or finite limit to how much you can get done in a day with software (outside of, like, typing speed), so it boils down to how fast and comprehensively you can think. So you might think, well, nobody is 10x smarter than anyone else, and that's probably true, but you could certainly imagine someone that has learned 10x more and has 10x more experience and given the right context can apply that effectively.

I strongly agree with the no-jerks part. I think, short term, you can be a jerk and still be effective, but long term if you're unpleasant you'll develop blind spots because people don't want to deal with you. Once you have significant blind spots, I suspect it's very hard to maintain the 10x pace. There is of course, a strong counter example here, which would be someone like linus torvalds (who is both extremely effective, and, by his own admission, kind of a jerk), but I think while exceptions like that exist they're not the common case.