An Ode to Reference Management Systems
A long while ago someone recommended reference management systems (RMS) to me. It ended up saving me lots of time in uni, improving my marks1, and it’s useful even beyond academia.
RMSs are software which collects websites, articles, book chapters, reports, and other documents used as references in one place. Their primary use is to manage bibliographies for scientific publications. Ignoring that for a second: What they are is searchable, taggable, groupable collections of files (mostly PDFs and website snap shots) with metadata (author, date, title, abstract, …). The metadata can be entered by hand, but is mostly extracted automatically from websites (!!). Put all of that together, and you get an amazing tool.
I’ve used an RMS on six uni papers, and also on some other things described below.
My “Stack”
Before diving in, it will be easier to talk about the work flows I use with reference to specific software. I expect other RMSs to provide similar functionality, but they might have different terminology and quirks.
I use Zotero, in combination with the Firefox plugin Zotero Connector and the Zotero plugin Better BibTEX. I use zathura instead of the integrated PDF viewer, trading the ability to edit for ultra-fast scrolling and vim-like key bindings and position marks.
Normal Usage is Awesome
First of all, RMSs really excell at supporting the creation of academic publications. I’ve used two main flows for finding and “processing” relevant literature on a topic.
- Reference chasing. If I already have a paper A that is relevant to what I am writing about, I will often want to look through some references R of that paper to learn more. I will usually have the screen split between a window for Firefox and a window with the paper A. I will go through A, look for specific papers R in the browser, and import them to my RMS with one click using the plugin. I will either look at those references R right after importing them, and make notes in a third window, or have a dedicated Zotero collection for this information gathering effort, and look through them all in bulk (and make notes). I take notes mostly outside of Zotero, as I have found the built-in system not to be very übersichtlich2. I’ll take notes either in a separate document for notes, or directly in the document I am working on. I don’t have a great solution here, and I would expect the note taking to evolve quite a lot if I was writing 10s or 100s of papers.
- Exploration via keywords. Another way to approach a body of literature is to figure out the relevant key words and use e.g. Google scholar to identify a lot of potentially relevant papers, and then work through their abstract. Much has been written on this subject, by people a lot more seasoned in research than I am. I refer for example to Gwern. The point I’m making here is: I take the list of potentially relevant papers, put them all into a Zotero collection (for Google Scholar results, this is literally <10 clicks for the entire results list) and then work through them there. Note taking works as in 1.
Another level of integration comes from using a makefile to build the document I’m writing, in combination with Better BibTEX, which allows me to fetch the most recent bibliography from Zotero by calling curl
from the makefile every time I save my LaTeX source file (using entr
).
All of this makes for blazingly fast work, held up only by paywalls and my inability to focus for too long.
Other Usage is Awesome, Too
Beyond academic research, I have so far found two other ways of using RMSs. I am sure I will think of many more that will both improve existing work flows and make new workflows possible by reducing faff.
For Job Hunting
I’ve recently put my first proper organized job hunt behind me. One challenge I’ve had in the past is to remember the job details for a given position I was interviewing for… At the time of the interview, the job posting had often been taken down meaning I only had my own motivational letter to use for preparation.
This time around, I put every job posting I was applying to into a Zotero collection, and then wrote down the Better BibTEX citation key in a spreadsheet I was using to keep track of the whole process. It worked like a charm and I referenced the stored postings many times throughout.
For Reading
When Reading the Internet I often come across an interesting article that I don’t have the time to look into at that moment. Similarly, when listening to talks or lectures I sometimes note down papers to skim through later. I found that maintaining a Zotero Collection called Reading Queue is a good first step towards actually reading those materials, mostly because it’s very easy. For webpages it’s <5 clicks, for papers it’s <1 minute of searching for the paper and grabbing it. Once it’s in the queue, it is very low-effort to open it up and read it, which is good for habit formation.
Creating a habit is of course one of the hard parts of this, retaining what you’ve read being the other. But whenever I find myself with spare brain power, I try to keep in mind that looking in the reading queue for something that sounds fun is something I can do, and it’s worked reasonably well.
If one of the things in the queue isn’t interesting anymore, it gets a strike. After two strikes it gets deleted. This way reading doesn’t feel like a chore, because I’m genuinely interested in everything in the queue.
When I’ve finished something, I move it to a separate Done collection, for later reference and a nice feeling.
Caveat
RMSs work really well to support certain workflows, both within and outside their intended usage. But it’s the workflow itself that is central, not the tool. Looking at shiny new tools is a great way to avoid doing actual work. Textfiles, spreadsheets and MVPs FTW.
-
Probably. Some of the papers I had to submit for examination had “Context” as one of the marking criteria. Context means a ton of legit-looking references, and I’m not sure I’d have had the patience to write 6 essays with ~30 references each if I didn’t have some system to help me. I might have figured something out without RMSs, or not. ↩
-
Visually clear, such that it is easy to grasp everything that’s going on at a glance. Open to suggestions how to put that sentence purely into English. ↩