Author here. The argument, compressed: headcount beat talent for fifty years because one great engineer could only type so fast. Brooks' surgical team (one surgeon, nine support roles) was the right structure and nobody could justify nine salaries that produced nothing a CFO could see. Agents made the nine cheap, but they only replaced the hands, not the pushback, the memory or the pipeline. Which is also why the rockstar engineer shouldn't come back, and probably will anyway. The rest of the piece is about which seats go and what happens to the people who used to arrange them.
Happy to be told where it's wrong.