Reading research outputs online is a lousy experience and I’m salty about it
Remarks at the launch event for Screens, Research and Hypertext
I don’t think it’ll surprise anyone who knows me that I find the state of research communications on the web to be frustrating.
This is me right around the time my youngest colleagues were born. It was about six months after the worst of the dot-com crash, right before the web really started to become mainstream.
I’d done pretty much all the research for my degree the old fashioned way — from paper books and photocopies of bound journal articles. It was a kludgy and labor-intensive way to research.
And yet, it was still way less clunky than trying to work with the few things that were digital at the time. (Anyone else remember when the EBSCO database lived on a CD-ROM?)
I assumed back then that as the web matured, it’d get much easier to conduct research directly on a screen. But here we are 30 years into the web, and if I want to seriously engage with a text, it’s still easier to print it out than it is to read it off of a six-figure think tank website using my fancy design agency MacBook.
I wrote Screens, Research and Hypertext as a guide for a better web for longform research outputs. It’s meant to be both an argument for those changes and an example of what those outputs might look like. There are a number of things going on in the book, but they’re all in support of two overarching points.
- The document is a bad model for research outputs.
- The web is optimized for distributing documents.
I want to talk about each of those at a very high level.
Documents as research outputs
My freshman year at Hampden-Sydney college, I took an honors course on the Industrial Revolution. It was team taught by two literature professors and an economist. In a given week, we might read a section of Dickens’ Hard Times, an essay from Thomas Carlyle, and economics texts from John Stuart Mill or David Ricardo.
The class really drove home to me the realization that the boundaries between academic disciplines are pretty arbitrary, because knowledge is all connected. Or, as Ted Nelson pithily puts it:
Everything is deeply intertwingled.
Most of what we call research or scholarship is about uncovering those connections between ideas.
In the public policy sector where I’ve spent most of my post-academic career, research can mean finding the patterns in hundreds of individual interviews. It can mean fitting a theoretical model to observable data. It can mean showing that pulling on lever A doesn’t just move B. It shifts entire systems.
It’s about finding common threads weaving their way through lots of different pieces of information.
In short, we generate new knowledge, new ideas in the spaces that lie between texts.
Ideas live in the spaces between the texts.
Then we dump this amazing new set of ideas — this thing of wonderous and bewildering complexity and interconnectedness — into a bland, formulaic output.
Introduction. Methods. Results. Analysis. Discussion.
This new web of ideas, confined to a format that allows for a single set of spatial relationships. Before and after. Anything else relegated to some clunky footnotes or the occasional text box.
What’s worse is that we write these things as if communicating research results is our only goal, when we know that today’s research outputs are tomorrow’s research inputs.
When we’re conducting research, we aren’t reading things front-to-back. We’re jumping around. Cross-referencing things. Comparing different texts.
Documents suck at all of those things.
That brings us to my second theme.
The web is optimized for distributing documents
That starts by understanding that HTML is a terrible hypertext system.
Yes, HTML literally has the word “hypertext” right there in the name. But despite the name, HTML is not a language for communicating hypertext. It’s a language for marking up documents.
HTML is literally a subset of SGML, or Standardized General Markup Language. That’s a protocol developed explicitly for the purpose of creating and distributing electronic documents.
Because HTML is a subset (rather than an extension) of SGML, we end up with weird things like using *unordered list* to mark up menus. When your language is built for displaying documents, any non-document things end up with some janky, unsemantic markup.
But it’s not just the language that’s a problem.
- The hyperlinks that encode the connections between texts run in only one direction. If you land on the target of a link, there’s no way to know about its source.
- Hyperlinks document references, not relationships. There’s no way to say that this kind of link shows a source and that kind of link documents an aside and this other kind of link shows the previous version of a thing.
- URLs — the addressing system for referencing things on the web — largely store pages rather than arbitrary text strings, making it nearly impossible to link directly into specific passages.
- Modern responsive web design privileges a single column of text, with everything else shoved above or below, setting millions of pixels and gigabytes of graphical processing units to the task of imitating a 1.5p sheet of A4 copy paper.
So what can we do about all of this?
Well, I’m not going to tell you. I mean, if I could put it all into a 10-minute speech, I wouldn’t have bothered writing a book.
But I can suggest three broad directions of travel.
Researchers should adopt better writing tools
Researchers need to embrace writing tools that reflect the way research really happens. That means ditching Word and Evernote and Mendeley and trying out things like Obsidian or Roam Research or Logseq. These are notes built on top of a knowledge graph and they allow you to do things like create bidirectional links at the paragraph level.
Communications teams need to embrace lean-forward reading
As research communicators, we must resist the urge to turn research outputs into lean-back reading. Or more bluntly. Stop trying to make Snowfall happen. It’s not going to happen.
A paper called “COVID sickness absence — transition from enhanced provisions to application of regular sickness absence arrangements” is an important piece of work. But no amount of great design or whizzy web features will turn that into something people want to read over coffee on a lazy Sunday afternoon.
Consultants should press for radical change
As content strategists, designers and developers, we can think beyond iterative improvements to existing things. The internet has the power to transform communications every bit as much as the printing press or broadcast media. But we’ve mostly been content with using it to package up things that were created for other media. Too many of our websites are the equivalent of filming radio scripts.
What’s next?
Now this past week, the entire Soapbox team gathered in London for the first time since the pandemic began. I’ve watched my colleagues apply their passion and creativity and intelligence to a some really hard problems. I’m reminded again that this is an unimaginably talented bunch of people. And just in case that wasn’t enough, for today’s launch we’ve also invited some of our smartest and most forward-thinking clients.
In this room is the potential for some terrifying innovation.
And I don’t say “terrifying” as hyperbole. I’m going to build a pocket computer and then rip off the keyboard is an idea from a lunatic. Unless it works. Then you’re a genius who has transformed an entire industry.
Writing a book with no central theme, no order, no table of contents and no way to meaningfully track analytics on account of how it generates 10³⁷⁵ possible URL combinations? Almost certain to end up on the lunatic side of the spectrum.
The truth is that I don’t know the right way to put research outputs online.
But I look around at the people in this room and I can’t help but think that it’s going to be a great ride figuring it out.
