One thing that people are very confused about is the idea that post-doctoral training is only about training. It is also about selection: identifying the most talented and accomplished scientists to give a shot at scientific independence.
It is delusional to think that post-doctoral training “trains” scientists to become PIs in the same way that plumber apprentices are trained to become plumbers: if you just slog through the training and keep your head down, you will become a decent plumber. The more accurate analogy is to minor league baseball: yeah it is necessary training to learn how to play ball in the major leagues, but it is also a selection mechanism to identify those players who have a decent shot at success in the majors.
This is exactly why the idea that two-year post-docs–like in the old days–are more than sufficient to “train” scientists to run their own labs is a fucking joke. It may be sufficient to “train” scientists, but it is insufficient under the vast majority of circumstances to implement the selection function of post-doctoral training. There are *many* more scientists seeking PI positions nowadays, and thus the selection function of post-doctoral training becomes more important and more stringent.
Longer post-docs should be welcomed by those aspiring to PI positions, as it provides a much fairer opportunity to prove one’s mettle. Many post-docs start slowly for a variety of reasons, and so just because you don’t have much to show after two years, doesn’t say much about your potential. But if you haven’t achieved much after 5+ years as a post-doc, it is reasonable to conclude that it is not just a matter of bad luck, bad mentors, or anything other than a simple–and unfortunate–lack of the skills and talents required to be a PI.