The original research that 10x came from used 1x for "worst", not "average". Their scale was 1x - 10x. IIRC it was also "within a given team" or some sort of

The original research that 10x came from used 1x for "worst", not "average". Their scale was 1x - 10x. IIRC it was also "within a given team" or some sort of similar scope, not over all developers, which would explain why they didn't see that 28x.