Not exactly a valid analogy, because his start date isn't "the beginning of human history", rather it's the development of the linked list data structure, which is the first point at which humanity possessed sufficient knowledge and tools to figure out the cycle-detection algorithm.
Given that we didn't have the knowledge or tools necessary to make a compiler hundreds/thousands/millions of years ago, you'd have to bump up your start date considerably.
Sure, I stretched that too far. But hopefully the point is made. Then should we expect no students (or anyone) to be able to build a compiler (or os kernel) in any amount of time given the internet and whatnot? While I may be articulating this poorly, I think my fundamental issue with his point is solid.
Given that we didn't have the knowledge or tools necessary to make a compiler hundreds/thousands/millions of years ago, you'd have to bump up your start date considerably.