I thought I was making a big confession: “I do not read other peoples articles” I whispered. Strange looks around the coffee machine. “Off course you don’t, nobody does!”. “maybe the odd abstract, but I never read articles front to back.”
Nice. I toil and sweat for three months to write te perfect academic article. All academic prose, passive voice, unreadable garble, just so that nobody will read it. Ok, not nobody: one editor and three reviewers will read it. Most likely they will find that I failed to read some other unreadable paper which showed that what I did was already done in the 1970’s. Or they will find my writing style “too colloquial” and demand that I bore it up even more. Just to make sure that indeed, anyone that starts reading it will try to commit suicide around page three.
Why do I even bother? There are three reasons for writing an academic article:
to save the knowledge I gained, to add it to the great library of human knowledge. This is the "archive" function.
to prove to funders, the general public, but mostly to future employers that I am a good researcher. This is the "accountability" function.
to inform other scientists and the general public about my research. Fellow scientists in my field need to know what I am working on so they can make sure we’re not doing the same work1. Or they can spot opportunities to either work with me or base their work on my results. Using article to inform other scientists is the "news" function.
For historical reasons, we choose to serve all these functions using the two centuries old “academic article” format. For those three functions:
It’s as if we are archiving all the human knowledge, but forget that we invented computers and databases and require that every piece of knowledge is written on a cardboard card to be stored in a large cabinet with a big lock2.
It is as if we forbid future funders and employers to look at my LinkedIn profile or Youtube channel, instead forcing them to rely on letters of recommendation written by people that may be in the same field but barely know my work.
I hate it, yet I work within it, accepting it as a fact of life. Until a former student of mine opened my eyes to another possibility. Wouter3, wrote a blog on the current state of our knowledge of evaporation: the great unknown in the watercycle. The target audience of the blog is fellow scientists: at no point does Wouter dumb down his analyses. Yet the style is modern: it is first and foremost a blog. In a few minutes, I was up to date on the current knowledge, and lack thereof, surrounding evaporation. It got me thinking: if every academic article came with a blog as readable as this, I would start reading other peoples work again!
So I am arguing that scientists should be writing blogs to communicatie with each other. Blogs will address the “news” function that, for me, academic articles currently fail at. A blog would consist of up to a thousand words4 on why your work matters and what the results mean, both for fellow scientists as well as for the general public.
“Yeah right, even more work” I hear my fellow scientists already say. Not really: the blog would not be in addition to the article, it would replace parts of it. The blog combines the abstract, introduction and conclusions5 of a classic article. So we should take those out of the article. What remains are the methods and results of the article. I strongly belief that for methods and results we do not need to write long winding academic prose. In the scientific field of genetics people are building computer programs that parse academic prose to analyse what was studied and store it in a database. Why not directly add our results to such databases? Technologies are currently emerging that allow researchers to store, and get credit for, different parts of a classic article. Research data can be stored in open data repositories. Research methods can be stored in dedicated methods journals. Software used to analyse data can be stored on Github and figures on Figshare. And all these services allow researchers to claim credit for them using Digital Object Identifiers (DOIs).
So let’s cut up the classic article and publish the parts separately: data on a data-repository, code on Github, figures on Figshare and a blog that informs fellow scientists in an academic magazine6.
The devil is, of course, in the details. This new way of publishing raises a lot of questions:
Nice idea, but academic magazines do not exist yet: how do we get there from where we are now?
There is currently no academic magazine that accepts blogs. So let’s just start publishing our blogs ourselves and see what happens. Next time you want to publish your work consider publishing it in chunks. Send the data to a data repository. Publish your figures on Figshare, your code on github. Than write a blog about it. Have the blog reviewed: hire a journalist or editor to have a look at your writing. And here comes the tricky part: do not publish it yet. Save it and rewrite a version into a classic article. This should be easy considering you have already written the blog. Translate and expand your blog into the academic style7. Your data section should be a reference to your own data. Your figures are already on Figshare. Any code used is already on Github. Submit this to a journal of your liking. As soon as you have your work published in a classic journal, than publish your blog8. Publish it on your own website, or at a topical site like younghs.com. Make sure to add a reference to your classical paper at the end of your blog. Use Zenodo to get a DOI for your blog so people can cite the blog as well as the underlying article.
For as long as academic journals do not accept the blog as a valid form of article, we may have to take this route. But blog-based academic magazines might just start emerging. I know I will be setting up one very, very soon: stay tuned.
What about peer review?
Part of the unreadability of academic articles stems from the review by peers. And although the process has its flaws, I do not want to abandon it: peers should look into your methods. They should test whether the conclusions follow logically from the results. So when a blog is submitted to a magazine, the editors should still invite fellow scientists to look if the data, methods and conclusions behind the blogs are sound. What they should not do, however, is ask the scientists to comment on the (quality of the) writing of the blog. Scientists are generally not trained to be excellent writers. Therefore, editors should take it upon themselves, or ask a science journalist10, to judge if the paper is readable. If not, the editor or journalist can either help the scientist to improve the writing of the blog, or reject it as garble.
But I need to publish in Nature to secure my career!
And you should. My PhD supervisor told me that at a visit to the Chandon-Moet factory11 he was told that if cork was ever to be completely replaced by plastic or screw caps, that Chandon-Moet would be the last to change, being dragged kicking and screaming into the new age. The same goes for really high ranking journals. They have all the incentive to stick to the old model. Change should come from the “normal” journals. And change can come from the journals that scientists own and control, like the EGU journals, or we could start new journals (and call them magazines!). If you agree with it, we should try to publish in journals that support this vision as much as possible. But I can’t, and won’t, blame you for submitting anything to Nature: I know I did, and will keep doing until Nature realises that this new blog based way of publishing is the way to go. And we might be surprised: Nature and Science were both magazines before they become the journal powerhouses we know today.
If everything gets a DOI, what should people cite: my data? my results? or my blog?
That depends: are they using your data for their own analyses? Then they should cite your data. Are they using your method on their own data? They should cite you method. Are they building their hypothesis on your conclusions: they should cite your blog. By separating the article in chunks we also allow for different specialties within science to shine in their own right. If you are an ace in collecting data in the field, but not the best statistician around, just publish data. If you want to tell people about all the cool data you collected that they can use, write a blog about it and publish that in a magazine. If you are an ace statistician, but are a danger to yourself and your colleagues when taken into the field, you might read about this cool new dataset in a magazine and come up with a way to get interesting new insight out of it. Or you might end up working with the field-scientist because you came up with a hypothesis based on her work that needs more of that excellent field data to be tested.
I like your idea but I do not know how to write a blog!
Code on Github, figures on Figshare, data on a data repo: what about methods?
Ok, you got me there. Although we do have ways to get DOIs for data, code and figures, we do not, to my knowledge, have a solid way to store methods. A few platforms are emerging that want to do this: protocols.io allows one to share (mostly lab based) research protocols. And for building your own sensors or other hardware, like I do, platforms like instructables and Wevolver are interesting. None of these, however, do support getting a DOI. A way to work around this is submitting your method to Github and getting a DOI through Zenodo. This is, however, not the intended use of both of these services(11). There are classic journals that focus on publishing methods in science, however, these journals ask for a classic article. Together with Pete Marchetto I am in the (slow, sorry) process of setting up a new service to allow getting DOIs for novel methods in the geosciences. Stay tuned and for now: employ unintended uses!
But if its not published in a respected journal, how will other scientists find my work?
Do you read all the newsletters from the journals in your field? If you do: wow, you must know how to speed-read and teach me one day because I do not. The way I get my science news is threefold:
through news-sites such as BBC-science-news or the (online) science pages of newspapers. If you want to get your work on these pages, follow the path you would have followed in the past: publish your blog also as a classic article (see above) and write a press release. Send it to journalists that you think might find it relevant for their medium.
through social media. I follow a lot of geoscientists and geoscientific organisations that happily post links to relevant blogs.
through Google Scholar. When starting new research, I spend way too long on Google Scholar, looking for publications related to what I intend to do. If your blog has a DOI, Google will find it. You can even tell Google about your blogs by adding them to your Google Scholar profile page manually.
So in short: please cut up your work in chunks, get DOIs for the different chunks and write a blog that ties it all together. That is something that I would like to read.
1 not too much. There is a good argument to be made for some replication research. Follow Daniel Lakens on twitter to be educated about the importance of enough reproduction research.
2 the lock representing the pay wall that academic publishers erect to shield general access to our shared knowledge. But that is another battle. In the same war, but still: another battle for another day.