Neat! I’ve never figured out the trick of getting value from a math paper but maybe this will help!
When reading a paper, orient your eyes towards the future whenever possible, like reading several words ahead in a game of TypeRacer. Scan the paper at a high level to understand the big picture, then read all the theorem and lemma statements to see how they fit together, and only then decide which weeds to get into. Only check a difficult computation once you already know what its payoff will be.
The way I read the papers that I can get something out of (which are all experimental papers that collect givens from the physical world) is: (1) read the abstract to figure out if downloading the PDF is worth the effort, (2) look at the figures and if the figures tell a story that might be important if true, then (3) double check by reading the conclusions to see if the authors know the story they are themselves trying to justify and then (4) read the methods to figure out if the experiment had subtle bullshit hiding in it that grossly invalidates the central idea, and finally (5) read the methods carefully to figure out how to copy their techniques.
In the olden days (maybe still true, I stopped bothering to try to verify this) computer science papers were totally garbage for the most part, because they never included the code or a link to the code, which is the only part of them that would help with step four or five “for really reals” and so all of academic CS was basically just a bunch of “real science” larping :-(
I’m interested in trying your technique on a few math papers. Do you know of any classics that would be good to practice on?
Here are two recentish papers I really enjoyed reading, which I think are fairly reasonable to approach. Some of the serious technical details might be out of reach.
Neat! I’ve never figured out the trick of getting value from a math paper but maybe this will help!
The way I read the papers that I can get something out of (which are all experimental papers that collect givens from the physical world) is: (1) read the abstract to figure out if downloading the PDF is worth the effort, (2) look at the figures and if the figures tell a story that might be important if true, then (3) double check by reading the conclusions to see if the authors know the story they are themselves trying to justify and then (4) read the methods to figure out if the experiment had subtle bullshit hiding in it that grossly invalidates the central idea, and finally (5) read the methods carefully to figure out how to copy their techniques.
In the olden days (maybe still true, I stopped bothering to try to verify this) computer science papers were totally garbage for the most part, because they never included the code or a link to the code, which is the only part of them that would help with step four or five “for really reals” and so all of academic CS was basically just a bunch of “real science” larping :-(
I’m interested in trying your technique on a few math papers. Do you know of any classics that would be good to practice on?
Here are two recentish papers I really enjoyed reading, which I think are fairly reasonable to approach. Some of the serious technical details might be out of reach.
https://arxiv.org/abs/1909.03562
http://www.cs.tau.ac.il/~nogaa/PDFS/induniv1.pdf