A standard CS undergraduate curriculum teaches core subjects like computer architecture and networking. With the bevy of freely available CS preprints on such topics on sites like arXiv.org would it be possible for a highly motivated individual to learn the theory just from preprints? I’m wondering the same for the field of math, economics, or any other technical field. I haven’t read that many papers but my intuition is that the background section often attempts (sometimes effectively, othertimes not so much) provide the context and vocabulary needed to understand the core theoritical aspects of the paper.<p>As an example to flesh out my question, for say the topic computer architecture - can we stitch together enough preprints (given the corpus of whats available on arXiv - not sure if they have all the important CS papers) so that if they go through them they could learn everything theoritical that is learned in an undergraduate or graduate computer architecture course.
No, not at all. Even a standard CS curriculum doesn't teach enough in the required classes to give enough context to fully understand the preprints in the two subfields you used as an example. An important part of academic research is understanding the topic of a paper within the context of other current and previous work. It's very challenging to start at the end and work backwards when you don't have a solid grounding in the fundamentals.
The problem is that many preprints are hit of a very high quality.<p>You'd be better off obtaining basic literature from books and then grabbing research papers off scihub.
As a general rule, no. A lot of the preprints assume a detailed (grad level) understanding of the problem and they explain just enough to frame the problem in order to introduce their solution. I wager someone going down this path would get frustrated quite quickly.