Well if you've ever taken a university class where they 'take you through' the topic, which is to say along the path that discoveries were made and then updated, you realize that they learned stuff originally that was wrong, and really smart people went off on really wild tangents that were a waste of time (sometimes provably). So going through a solid college program on engineering should 'move you through' the mistakes to get you close to state of the art, with an understanding of why those were mistakes, and why things are done the way they are now, and how to evaluate your progress with some tools like complexity analysis and to figure out how to break what seems like an impossible problem into achievable segments.

Good college students learn to ask a lot of questions and to learn from the answers.

That being said, people who don't 'go to school' but have a 'learning lifestyle' (which is to say always curious, always reading, always tinkering with a variety of ideas) can get just as good as any university student can. The challenge for them is to convince an employer that they are in fact well educated on the topics and tools they will be expected to use to do their job. If they start their own company, well they don't really have to convince anyone since the results speak for themselves.

"...They can build you a simple website for a thousand dollars but they can’t build the next Google and they won’t create the technologies of the future."

well, i always chuckle when people use this (admittedly worn out ex) for the their argument, but folks like Bill Gates, anyone, eh? ...well, guess it's not like he created "Google"? alright, sarcasm aside, I don't think there's any doubt about the importance of a college degree to get a broad perspective on understanding computational what-have-yous, esp compared to learning just a language or two, but the author overstates his point a bit -- of course exceptions are plenty where people can succeed w/o degrees, though i'd just imagine less likely.

Weren't they mostly maths/physics/electronics graduates, though? Besides, university attendance was much less commonplace in those days, so it's not a particularly fair comparison.

Before universities were teaching CS, they were teaching mathematical logic, linear algebra, calculus, linguistics, and electrical engineering.

A general CS degree just distills the good parts out of those fields and mixes in courses teaching the foundations of computer and systems architecture. (In general anyway)