05 July 2010

Losing the scientific lede

ResearchBlogging.orgOver at SEED, Dave Munger reflects on how online publishing and dissemination methods can strip the nuance from scientific news:
I thought I was being careful to explain the results of several studies, showing that suicide is a difficult problem with many potential contributing factors and confounding variables, including mental illness, depression, and the seemingly contradictory influences of intelligence. Yet on social-networking sites, many readers latched on to one finding: That countries with higher average IQ tend to have higher suicide rates.
Munger suggests that this problem can be mitigated by careful consideration of both the nut graf sent out via Twitter and RSS and the audience receiving them, and that's clearly right. But I think it's also worth considering whether some subjects are less appropriate for blogs.

Consider your medium! Photo by K!T.
Blog posts are best when they're less than 700 or 800 words long, and their contents are readily summed up in a headline and only slightly expanded upon by the first paragraph. Think newspaper, not magazine articles. Do people read posts longer than that? Sure they do. But the longer a post is, the more possibility there is that some fraction of the readers will quit reading before the end, and maybe even pass on links or comments based on that incomplete understanding. I realize I'm not in the majority of online science writers in taking this position, but I think this better reflects how the average online reader reads.

Posts about individual, straightforward results work well in that context. For example, my colleague Jeanne Robertson recently discovered that desert lizards under divergent selection for camouflage have also become confused about visual mating signals. It's simple—one lizard population moved to white sand dunes and evolved lighter coloration, so now light males think that dark males from the ancestral population look like females—and it supports a lot of catchy headlines that don't sacrifice accuracy. The title of the talk at Evolution 2010 in which Robertson presented the discovery was "Dude looks like a lady." I'd say the Wired Science article I linked to above captures all the interesting details.

Complexity doesn't work so well. Scientific papers based on broad surveys of the literature, or many interrelated experiments, are inevitably going to lose some potentially important nuance when translated into an RSS-suitable post title, and explaining them accurately may take a lot more than 700 words. I've run into exactly this trying to write about complicated papers—either I go on for longer than I think my readers are likely to follow, or I have to omit detail and rely on readers to follow up with the links to the literature.

Mind you, this length-versus-content balance is a universal problem in disseminating scientific results—just look at the short-form journals Science and Nature. Some results are perfectly suited to the three-pages-and-online-supplement format, like an experimental result showing that sexually-reproducing lines of the worm Caenorhabditis elegans maintain more fitness in the face of mutation than asexual lines [PDF]. (I've posted about another result in this experimental system.) It's a simple result easily understood even without getting into the Supplementary Material. Compare that to a recent statistical survey of evolutionary trees that concluded species interactions weren't important in the history of life [$a]. That, too, fits into three pages of Nature, but the result is deceptively simple—even after delving into the Supplementary Material, the statistical reasoning underlying the core result isn't clear, as the comments thread on my post about the piece reveals.

Which isn't to say that online science writers should stick to covering simple experimental results or flashy natural historical notes, any more than scientists should never tackle complicated projects. They do, however, need to consider the limitations of the medium in which they report scientific results. Is a topic too complicated to fit in a single post? Maybe it's suitable for a series of posts. I like how Slate handles this, building collections of interrelated articles that can stand alone, but link into something like a long-form magazine article—see Will Saletan's great series on memory manipulation for a recent example.

And now I've blown through 700 words in the service of an extended, hopefully nuanced, discussion. Take from that what you will.

References

Morran, L., Parmenter, M., & Phillips, P. (2009). Mutation load and rapid adaptation favour outcrossing over self-fertilization. Nature, 462 (7271), 350-2 DOI: 10.1038/nature08496

Venditti, C., Meade, A., & Pagel, M. (2009). Phylogenies reveal new interpretation of speciation and the Red Queen. Nature, 463 (7279), 349-52 DOI: 10.1038/nature08630

2 comments:

  1. I agree that breaking complicated stories into individual posts is often the best approach. I'm not sure this would stop the kind of thing Dave Munger is worried about, though, because people might just link to the one post out of the series which discusses the spicy data.

    I suppose the difference is that if you devoted a whole 700 words to the spicy stuff, you could make most of those words be caveats. That might not stop the problem entirely, but if it doesn't, nothing (short of not talking about spicy stuff at all) would.

    ReplyDelete
  2. Well, and, really, we as writers can't blame ourselves for every instance where readers oversimplify something we've written. Even the most carefully-constructed, thoroughly referenced piece is going to suffer in dissemination through the Facebook-o-sphere. We just have to cover the bases as best we can, and after that, it's up to the readers.

    ReplyDelete