This is an eternally frustrating topic. The 10x effect is very real, not mysterious ... and also somehow very overhyped.
The original research in the 1970s didn't claim some programmers were 10x average, just that some were 10x some others. That is directly observable in any shop with more than a half dozen people.
There are "10x typists" too. I type 65 words a minute, sad as a typist but fast for an engineer. Our admin types something crazy like 170 wpm. People who never have a typing class can be 15-20 wpm. 170/17 = 10x. I'm sure a real typist would say 17 wpm people "can't type." From that person's perspective it's true.
The reason people doubt it is that developers tend to group at skill levels. It's rare to have a 10x difference in a single shop or on a single team. But it's very common for the top dev at a top company to be 10x the worst dev at another company for any given task.
What most people forget is that the 10x figure only measured those who completed the task. For any given task, there are definitely 0x devs and devs with negative velocity.
I tend to always see it within a single shop at least. Whether it's 10x or more like 3x is debatable but "significantly more productive" is usually obvious.
I'm typing way faster when I'm typing code than when I'm writing text, because I know most of my IDE's keyboard shortcuts and hence can type 5 lines of code with very limited key strokes.
The original research in the 1970s didn't claim some programmers were 10x average, just that some were 10x some others. That is directly observable in any shop with more than a half dozen people.
There are "10x typists" too. I type 65 words a minute, sad as a typist but fast for an engineer. Our admin types something crazy like 170 wpm. People who never have a typing class can be 15-20 wpm. 170/17 = 10x. I'm sure a real typist would say 17 wpm people "can't type." From that person's perspective it's true.