A very honest, slightly lazy, but highly effective survival guide for drowning in literature review. Learn how to stop fighting with bad authors and start actually reviewing their science.
I had a minor realization the other day. It’s definitely not a world-changing discovery, but it’s honestly changed how I handle literature review, so I figured I’d share it.
As a PhD student, a good chunk of my life is spent reading and reviewing papers. And let’s be brutally honest: some papers are just terrible to read.
It’s not necessarily that the science is bad. It’s that the authors have zero capacity to write from a reader’s perspective. They suffer from the “curse of knowledge,” dumping a massive cognitive load onto the page and expecting you to just magically follow their train of thought.
I used to brute-force my way through these, getting increasingly frustrated. But recently, I started using a new strategy: I just get the spoilers first.
Whenever I hit a paper that feels like trying to read a barcode, I stop trying to be an academic hero. I use an LLM like a cognitive crowbar. I drop the text in and ask: “Strip the jargon. What did they actually do?”
I’m not doing this to skip reading the paper. I’m doing it to strip away the author’s academic camouflage and establish a “prior” in my brain.
Here is why this is actually powerful, beyond just saving time. Academic reading usually forces you to do two highly intensive cognitive tasks simultaneously:
By the time you reach page three, your working memory is completely fried. But by getting the AI spoilers first, you completely decouple these tasks.
Once I already know the core argument, the proposed solution, and the results, I don’t have to hold a dozen abstract concepts in my head while hunting for the main verb in a 60-word sentence. I already know the plot.
When I go back to the actual text, the dynamic completely changes. I’m suddenly reading it from the author’s perspective. I am no longer a confused reader trying to decrypt a secret message; I am reverse-engineering their thought process. Instead of being lost in their narrative maze, I am looking at the maze from above.
I find myself nodding along, thinking, “Ah, you shoved this dense math section here because you were terrified of Reviewer 2, but it completely ruins your flow. I see what you were trying to do. You just did a terrible job doing it.” It fundamentally turns reading from an exhausting decryption exercise into a highly satisfying, forensic teardown. You stop asking “What does this mean?” and start asking “Why on earth did you choose to explain it like this?”
Now, the AI trick is great for surviving bad writing. But what about the good papers? The important ones that you actually need to understand deeply and remember?
For those, I rely on a set of 10 questions originally shared by Prof. Xiangyang Shen.
The harsh reality of a PhD is that you will completely forget a paper three weeks after you read it. Answering these ten questions forces you to process the information deeply. Think of the answers as creating an embedding of the paper to store in the knowledge graph of your own brain (or, you know, your Notion workspace).
If a paper is important, I make sure I can answer these before moving on:
Anyway, that’s my current survival strategy. If you are also drowning in literature review, I highly recommend adopting the “get the spoilers first” method. It saves a lot of headaches. Back to the lab.