Choosing papers to read and write about

There are many academic CS conferences every year, with more papers than I could ever hope to read and write about - The Morning Paper was published every day, and I don’t know how Adrian did it! The volume of research, coupled with learning topics that I often knew little about beforehand, means that I have to select a smaller subset of papers to dig into.

As someone asked on Twitter, I figured I might as well share what my algorithm is!

First, I make a list of all of the conferences I wanted to read papers from. I did this in 2021 as a reference for myself, and I’m working on a list for 2022. I start this list by seeing which conferences from last year will repeat, checking the main organizer’s site (like USENIX), or looking at the list of the “top” CS conferences.

Next, I skim the list of “Accepted Papers” from a conference when it is published, bookmarking papers based on (in no particular order): my familiarity with the research topic, my interest, group / company that published the research, familiarity with authors, or general buzz. As mentioned above, I often don’t know much about a research area before reading the paper - expanding what I know is a big factor in choosing which papers to read.

After making a list of papers I’m interested in, I download the research into Zotero, where I store and mark up papers. Conferences publish the list of accepted papers in advance, but don’t always include links to the papers themselves - many authors publish preprints on their personal or research group sites if the conference doesn’t make the paper available. When I take notes and outline articles, I use Roam.

Last, I take the preliminary list of interesting papers I would want to read for a conference and filter down to what I can actually manage (based on how much time I have with other committments). Sometimes there are papers I want to read, but they get nudged farther down my list because another conference comes along.