Reading papers

Academia is a special place when it comes to ideas. There is real value placed on uncovering new ideas, new theories, new solutions. It is also a vast chasm of specialisation. I have employed strategies to keep on top of a few areas of interest, and spend an hour each week reading more broadly. There is no way that I can read all of the chemistry literature, and even less possibility of reading all science literature. This means that shortcuts have to be employed.

My strategy, for the most part, relies on having done a deep-dive on an area (starting with a review article), and then setting up a Scifinder alert for the same topic. I additionally have alerts for the top researchers in the field of interest. This keeps me on top of major developments. However, this strategy, combined with broader journal based browsing, is still time consuming. Now that I am no longer working from home, I have far less undistracted time available to devote to the maintenance of my knowledge. I have found myself taking further shortcuts - only skimming the abstract to see if there is anything that interests me.

The problem is that the abstract only summarises the sales-pitch results. These are important, and importantly are what made the paper pass the ‘important results’ criteria of peer review. However, sometimes the most impactful finding hasn’t yet been recognised as such. A paper that describes a new route for a compound might choose to focus on the results of a biological activity screening. This is the important bit for the biologists and medicinal chemists. Buried in the paper, there might be a sentence¹ about the unexpected instability of an intermediate, and that might be the seed needed to make a breakthrough in a completely unrelated problem.

I am guilty of this as well. Science is not as incremental as I would like, and results have to be sufficiently impressive to be published. Papers don’t have means to make minor updates. A typographical mistake in a published paper cannot be corrected without the paper being corrected, and there is a real stigma around corrections. The assumption is that the scientist made some egregious mistake, when sometimes it is an accidental oversight.² There is no way for a mistaken number in a table to be updated without a correction issued.

When I find a mistake in a paper, I note it on my PDF for my future self. But there is no way to share that finding with the broader community, unless I contact the publisher with my concerns. I have only done that with ASAP publications, as the mistake can be rectified for the final version - no correction needed. The other way that science gets corrected is to have a discussion of the mistake in subsequent papers.³ These often do not meet the ‘important results’ criteria, and so get mentioned in passing.

These side-notes, initially underappreciated results, and mistakes eventually - if relevant enough - get discussed in a review article. However, to be across them before a major review is published (and you read it…), the original papers need to be read completely. I have had the experience of skimming a paper or two for the salient details, trying an experiment or two, and then finding that the outcome was already known but buried as a footnote or even supplementary discussion.

I’m not sure of the best answer to this conundrum. I think one way to work might be to recognise the problem and address it as best as I can. The problem: papers are extremely information dense. The solution: they require considered, attentive, reading. My idea to achieve this is to narrow my field of expertise, while accepting that the broader areas will suffer as a consequence. I can schedule time-blocks for reading literature, and make sure that I read sufficiently deeply. One method I am finding helpful to ensure the latter part is to create literature summaries of the important papers - a subject for another post.

The academic literature is a bottomless pit. Rather than skim stacks of papers, focus on a few and really read them.