A paper is an argument

Most students read a paper like a textbook: they try to learn every fact in it. Researchers read it like a lawyer's case: they ask what is being claimed, and whether the evidence is good enough.

Three parts of every paper

1. The claim. What the authors say is now true or possible.
2. The evidence. Experiments, proofs, data, figures and tables that support the claim.
3. The limits. Where the claim stops being true, or has not been tested.

Every section of a paper serves one of these. The introduction sets up the claim. Methods and results give the evidence. The discussion should state the limits.

Why this changes how you read

When you know the claim, you can ask a sharp question of every page: does this make the claim stronger? A figure that does not help the claim is either support for a smaller point, or a sign of padding. A missing experiment is easy to notice once you know what the claim needs.

ExampleA paper claims "our method makes image classifiers more robust to noise". You now know exactly what to look for: tests with noisy images, a fair comparison with other methods, and more than one kind of noise. If the paper only tests one kind of noise, the claim is bigger than the evidence.

The gap between claim and evidence

The most useful thing you can find in a paper is the gap between what it claims and what it shows. Small gaps are normal. Big gaps are where weak papers hide, and where new research questions come from.

RememberRead every paper as an argument: claim, evidence, limits. Then ask whether the evidence is enough for the claim.