[karatekidd32v]

I mostly agree with this sentiment - with the added note that there is significant value in 'going deep' on a small subset of those papers. In my opinion, best bang for the buck there is to focus on the well known and impactful papers in that domain. I think there are big benefits to really digging into what makes a particular solution work, and how the authors really 'prove out' the full idea in the paper.

[actually_a_dog]

Right. That's why there are 3 steps to the process. The more interesting the paper is, the deeper you go with it.