The most annoying things about communicating effectively

by Tiwa Aina

20.109 has been my first foray into conducting practical work in the natural sciences. As a Course 18 without any lab experience, I’ve already had a lot of novel experiences—in terms of both experimentation and scientific communication. The latter presented some unique challenges as I worked towards creating the Data Summary.

One thing I found very difficult was managing verbosity. In preparation for the Data Summary, my drafts for each section were pretty robust. I tend to use a lot of words to describe what I want to say, and my sentences can get needlessly complex. In math, sometimes you have to use a lot of words, but as long as they precisely communicate your ideas, people generally don’t mind—most prefer something precise, enlightening, and wordy to something terse but vague! My writing style has naturally evolved to prioritize precision of language, even at the cost of concision. So of course, once we were tasked with adapting this content into a condensed format, I had to do a good amount of revision. In particular, I had to think about the most efficient way to distill my findings into (say) half a page’s worth of bullets.  Additionally, it was a struggle to make sure my writing was robust while still retaining some semblance of elegance—all without descending into rambling or purple prose. This meant pondering things like “what’s the best way to describe these DNA damage trends we found, without repeating damage  a billion times?”. This got pretty exhausting! It was definitely for the best though—when I give my oral presentation, I’ll have to be very conscious of verbosity and the ways it can affect content clarity. 

Another big thing for me was writing about limitations of certain approaches. For instance, the foci counting method is highly sensitive to noise, cells’ proximity to each other, and even the thresholding values that the researcher elects to use. It’s a lot of work too, so researcher fatigue might even affect the measurements one gets from it. I wanted to mention this in my summary, but I could imagine a reader thinking “if I knew that, why would I have used it as an analysis approach during experimentation?” It was a bit time-consuming to think about the right words to talk about this—I needed to use language that provided an honest review of drawbacks without delegitimizing the entire set of data I collected! I think discussion of the inherent and/or practical issues associated with a certain experimental method is probably best for the scientific community as a whole, but clearly there’s an incentive not to be too transparent, lest one undermine one’s results.

It makes me wonder how transparent researchers are about experimental issues in the “real world.” In the machine learning research community, a type of model that has gained popularity in recent years is the Bayesian neural network. I wanted to do some research with them (in fact, I wrote a blog post about it here), and it turned out that nearly every single popular open-source implementation is riddled with issues and instabilities—and yet, these are considered state-of-the-art! Could it be that many biological researchers are doing foci counting (or some other approach fitting this analogy) but not revealing the issues and errors associated with it? What methods are out there that seem reliable, but only to the researchers who haven’t personally struggled with them?

In any case, it’s been a while since I had to think carefully about the way I communicate and present ideas, and it’s a bit exciting to finally do it in a technical context. I look forward to getting more opportunities to sharpen my scientific communication skills!