More silly, macho, "real programmer" nonsense. It's like saying that the best way to learn calculus is to pick up a calculus book in first grade and read it over and over until it all becomes clear. Yeah, that's the express lane to mastery.
If you want to master something, sequentially master its prerequisites and methodically work your way up. If you do that, then a lot of the ideas of the masters will have begun occurring to you before you even read their works, and you will be the sort of person they were writing for.
Trouble with this is that prerequisites differ. I taught myself multivariable calculus out of a differential geometry textbook that I lacked roughly an entire degree's worth of "prerequisites" to, in order to pass a bog-standard calculus course, because I found the ostensibly "easier" approach advocated by the undergraduate calc book utterly incomprehensible.
This is why you look at the related works section first. If you don't know any of them, go find those articles and do your favorite recursive look up of those articles until you find documents that you can grok.
At that point you work forward, annotating, high lighting, and answering a few questions like:
If you want to master something, sequentially master its prerequisites and methodically work your way up. If you do that, then a lot of the ideas of the masters will have begun occurring to you before you even read their works, and you will be the sort of person they were writing for.