Those studies just gave already filtered people a small project and only measured the students who actually completed the project. So, problems with those stud

Those studies just gave already filtered people a small project and only measured the students who actually completed the project. So, problems with those studies which would imply that the difference is bigger in the real world: 1. Filtering means that the population was more homogeneous than real world populations. 2. Talent matters more the larger project you work on as a very big part of being fast is designing good abstractions to simplify the problem. 3. Not counting people who didn't complete the project means that we remove the long tail of worst developers. If you do include the worst developers productivity which was 0 then it isn't 10x, it is infinite x.