I'd be really interested in some empirical foundation for the described "plateau" effect. Are you aware of any?
You may also want to look into the concept of deliberate practice ("prolonged efforts to improve performance while negotiating motivational and external constraints", see also my related post).
We're a place where coders share, stay up-to-date and grow their careers.
We strive for transparency and don't collect excess data.