I don't see it that way. Anecdotes are not data, certainly, but the article really only uses those anecdotes to disprove the premise that education == success. Anecdotes are more than sufficient to accomplish that. Just one success story would have been enough.
He does go on to question if a degree could even be a liability, but I would suggest that the wording acknowledges that he doesn't have sufficient data to form any hard conclusions. I didn't take it as an argument, just a tangent on something that would be interesting to measure in the future that relates to the rest of the story.
Sure, anecdotes do prove the point that it's possible to have success without education. But there have always been examples here and there; even outside of tech, you could become the next Keith Richards. It's just not very likely.
I think it's possible things are shifting (at least in tech, probably not as much in other areas) so that college degrees aren't as big a benefit, but a handful of extreme successes aren't where you'd find evidence of that.
Mind you, success (depending on how you define it) is not very likely, regardless of your background. 95% of the population in North America earn less than $80K per year. A decent income at the high end, but not what most people consider successful. Looking closer at the numbers, the top 1% is comprised of 27% without a degree and 24% with a non-graduate degree. That seems indicate to me that there is no real benefit to having a "plain old" degree with regards to success at all.
The remaining 49% have a post-graduate degree. This indicates that there may be some merit to pursuing post-graduate studies. However, we already know that most JD and MD graduates generally make 1% incomes due to market protections associated with that education, so it is not clear if any old post-graduate degree would also put you at the same level. Unfortunately, the data that I have access to does not break it down any further.
The numbers most certainly demonstrate a correlation with education and achievement in work, but the variability between people is so wide the numbers are basically meaningless with respect to any common interpretations. So far, I have not found any compelling argument for a college education in this context (there are a million other good reasons, of course), outside of areas where legal protections exist. If you dream is to be a doctor then a college education is a must, obviously, but it is not so clear cut for the common man.
He does go on to question if a degree could even be a liability, but I would suggest that the wording acknowledges that he doesn't have sufficient data to form any hard conclusions. I didn't take it as an argument, just a tangent on something that would be interesting to measure in the future that relates to the rest of the story.