lili's musings

do we need stories for good science?

For the past decade or so, I have been immersed in academic culture. In this culture, in public presentations, it feels like so much centers around the stories that we tell with we research. At public talks, the ones we remember and praise are the ones with a clear narrative.

Until recently, I took this for granted. Then I came across a paper that just had no story at all. It was the complete antithesis of a Nature-style paper, in which the few main article pages are devoted to crisp narrative. No, here was a paper that eschewed any pretense of narrative in order to focus on the methods.1

The authors built a new model that integrated research across several other papers. I expect the model itself to be very useful, but by itself it doesn't make for a good story. Usually, this kind of paper has an additional figure or two showing an application of the model. This kind of figure gives it a proper story arc and allows the paper to be published in a "top" journal, but my impression is that practicing scientists then ignore this additional figure and focus on applying the model to their research. The focus of this kind of paper is definitely the model.

So I've been wondering, does a paper really need a story format in order to convey the science? Does the story format even benefit science to begin with?

Indeed, the mark of a good scientist is in how they are able to unravel the surface level story that a paper has spun and get at the actual novel finding.

For decades now, scientists have been struggling with a replication crisis: key results in psychology and medicine turn out not to be replicable. At the heart of this crisis, there is the pressure that scientists face in having to spin a captivating story out of their findings. It is often not the findings or methodology that are judged directly, but the story around them that is conveyed in the research articles, conference proceedings, and job talks.

The problem is that we are naturally keen to tell each other stories. This whole system has self-assembled from our intrinsic drive for storytelling. We cannot help ourselves really. I must admit that I too am drawn to a paper with a good story, especially for the (majority of) fields outside of my (narrow) expertise.

What can we do? The whole system seems structured for stories. Scientists participate in a strange double sided arms race and are trained both in how to tell stories and how to disassemble them. I can't tell if this is helpful for the general public. On one hand, the stories help in engaging non-scientists on science. On the other hand, they often conceal some important caveats.

So in the end, I'm not sure where this reflection is going. I'll probably keep telling stories in order to satisfy the system. I wonder if there's an alternative means of sharing results without forcing them into a story. The best I can think of is to share detailed methods (including code), make the inputs to your experiment/model widely available (such as chemical reagents or specific viruses), and share the raw data from your study. When building on a past study, those are what I rely on the most.

  1. I'm not gonna name the paper here, as the bias for storytelling in academia is so strong that calling a paper "storyless" feels like an insult.

#science #story