Making science writing smarter
Sometimes to really go forward, we must first go back. I tried to find famous scientists born in Boston, and one name (and only one) kept appearing everywhere, the name of Benjamin Franklin. Now normally I’d be quite disappointed to only find one person, but this is Benjamin Franklin we’re talking about. He’ll do just fine. So take my hand and let’s travel back in time, to a different age, 300 years ago.
Benjamin Franklin, 1749
Franklin, who had just invented the lightning rod, was very accomplished in many things, being a scientist was one of them, and that’s what we’re focusing on today. Particularly, how did Benjamin Franklin do science? How did he write about his scientific experiments and how did he report them?
Well, he had some pen and paper and recorded whatever he did, and then sometime later combined those notes with the world’s scientific knowledge, which he read almost in its entirety and knew most by heart, to produce paragraphs like these:
In his correspondence on the nature of electricity, cited here are Thomas Burnet’s Theory of the Earth and William Whiston’s New Theory of the Earth, both of which were probably in Franklin’s library and he had read and even reread them. Now, a bit beside the point, these books were not considered in accordance with contemporary scientific insight and fairly controversial even then, but received support from prominent scientists like Newton and Locke, and at that point in time it would surely be a leap of faith to call them either science or not. In any case, Franklin’s modus operandi was to do experiments, combine their results with the world’s knowledge in his head, and finally write conclusions down on paper with a pen.
Let’s step back into the time machine and travel exactly 200 years into the future from Franklin’s time, to meet this brilliant gentleman:
Albert Einstein, 1949
Notice the knowledge producing tool of choice Einstein is holding in his right hand? And the surface that tool is being used on? That, my friends, is a pencil writing on paper. Einstein, like Franklin, knew most of the relevant scientific knowledge by heart, with the number of all papers in physics reaching low thousands in the beginning of the 20th century. And again, Einstein’s way of advancing science was quite similar to Franklin’s: do experiments (Einstein’s experiments happened almost exclusively in his head), record the observations, combine them with an all encompassing database of relevant knowledge that was his memory, and write the results down on paper with a pencil.
Fun fact, one of the first people to sell pencils in the USA was … wait for it … Benjamin Franklin.
So in a span of two centuries, we merely added 3 letters to the state of the art tools for writing science: we went from pen and paper in Franklin’s time, to pencil and paper in Einstein’s time.
What about the present? Take my hand one last time and let’s set our time machine to the present. What tools are our researchers using to write science today?
Modern tools, 2015
We now have these portable devices called computers, capable of holding the entire corpus of scientific knowledge (1.25 TB, from an average of 5000 words per paper at an average of 5 characters per word and an estimated 50 million papers). We now have the internet, enabling access to all of knowledge in an instant (if we disregard the tragic fact that we purposefully, in the spirit of the olden days, invented ways to limit access to this information, such as paywalls).
And how have these indisputably revolutionary technologies changed the way we write science? They have not, at all.
How do we write science today? This should sound familiar by now: we do experiments, record the results, combine those with an internal model of knowledge obtained from papers we’ve read, and write it all down by pressing keys that cause letters to appear on pixelated paper.
Some are of the opinion that this kind of stasis is exactly what science needs, and I tend to agree, historically, but I cannot disagree more, if we take this present day into account, as nothing else in this knowledge production system has been stationary. One of the more sobering statistics is the number of published papers per year, which has risen exponentially from thousands at the turn of the 20th century, to millions in the 21st. There is no human brain that can manage that amount of information, and yet, virtually no infrastructure has been developed to deal with this tsunami of knowledge.
And that brings us to the the Collaborative Knowledge foundation (Coko for friends). In short, we’re an open, community-driven project, building open source software that aims to make the production and dissemination of scholarship easier.
One of the things we’re exploring is knowledge production support, or in other words, how can we lend a hand to researchers when they’re writing scientific papers, how can we make their writing smarter. We’re at the beginning of our journey, but we do have one quick vision of the future for you today:
PubSweet Science Blogger using the Lens editor, searching CrossRef’s API and data mining found papers on the fly.
Pictured right is a prototype running on a open source (repo details coming shortly) publishing framework we’re developing. It focuses on modularity and is extendable and completely customizable, because of its unique architecture. At the core sits a smart backend, supporting server-based features, such as persistence and authentication, but also custom server processes, such as the text mining process. And on the front-end sits a custom component-based web application, where each component can be separately developed and switched out, easily replaced, if the requirements change. We’re crazy excited to show you more about this soon!
To conclude, a change is coming in the way we write science and the tools we use to do it. Machine-based support of scientific writing will soon no longer be a luxury, but a necessity, and our framework will play a part in the global effort to introduce machines into our scientific writing workflows (legal efforts are an important part too).
Oh, and one more thing – Einstein, despite having a brilliant mind, did say that you should:
“Never memorize something that you can look up.”
Perhaps we should listen.
Post by Jure Triglav, based on his talk at the CrossRef 2015 meeting.