Good point. But really, all advice from older persons to younger persons carries a strong element of survivorship bias. The data are biased but RCT studies are not practical so it's reasonable to use what we have (recognizing the limitations of our data, as you pointed out).

I don’t believe survivorship bias applies... yet. Even though I do have a bachelors degree (not computer science) I barely graduated high school. Even still I am doing well as a senior developer having surpassed many of my peers. Yet many of the strongest developers I have worked with personally are also self taught. I believe that once I take my peers into consideration there is a frequent reproducible pattern.

It becomes survivorship bias when I become more specific and speak to what works versus what does not, particularly in contrast to common faulty social expectations.