Thoughts on the future of academic publishing

Academic publishing is broken. This post is about how things could be better.

The traditional life of an academic paper can look like this:

You write a draft and email it to your coauthors as a text file. Often this is a LaTeX file.
Your coauthors make comments and send back a new file. There’s not an obvious way to work in parallel without risking edit conflicts, nor is there an official and easy way to make inline comments, so the working details are up to you and your coauthors to figure out ad hoc.
Submit the paper to the place you’d most like it to be published, perhaps filtering your choices based on some sense of realism.
Wait for a reply.
If things go well, rewrite based on reviewers’ comments; if they don’t go well, take two steps back and resubmit to your next-best choice of journal or conference.
Wait more.
If more comments are provided by reviewers, repeat the rewriting cycle.
If things go well, you may now need to pay a publication or membership fee. Some publication fees are as high as $5,000, though there is a wide variety of prices.
Congratulations! Your paper will eventually appear somewhere. Could be several months.
After this, it is extremely difficult to change your paper. Any errors in the published version are likely to be seen by readers for the rest of time.

Once that process is complete, this is what reading a paper looks like:

If you’re at an academic institution, you can usually perform a simple search and begin reading a pdf file. You may not think much about it, but it’s likely that your institute pays as much as $2,000,000 every year for this kind of access. To get a sense of the expense, a single journal subscription may be as much as $20,000 annually.
If you don’t have institutional access, then you’re effectively out of luck. Yes, you can sometimes pay around $30 to read a single article. The problem is that it’s difficult to tell if a paper will be useful to you until you’ve had a chance to at least glance through the entire paper. So you often pay for articles that end up not being useful. If you’re writing a paper with 35 references, and you read 70 papers in order to find those 35 references, then you’re paying $2,100 to publishing companies for your work on a single article. The authors of those 70 papers do not receive any of this money.
Many papers are read as pdf files viewed on a computer, the suboptimality of which is examined below.

Next are some ideas on how things could be better.

Treat all papers as living documents

Since most paper experiences are already digital and network-based, the technology is ready to support papers which change slightly over time. Just as a wikipedia article constantly has the opportunity to improve, so may a paper.

Work with version control

Many startups use a version control tool called git along with github, a host for git-based repositories. Although these tools were designed for source code, they can work perfectly well for all the files behind a paper. This would make it easier to collaborate with coauthors, to ensure that no data was lost regarding any version of the papers, and to communicate explicitly and in a standardized way about your work.

Publish in html first

Pdf files are great when printing is needed, but physically-printed papers are a thing of the past. Virtually everyone uses a laptop, desktop, or mobile device to read. Most of these devices have browsers which are built specifically for displaying network-based text-heavy content. I cringe at having to spell out what a browser does; my point is that html is a superior format for papers. There are questions of reliable formatting, but these questions are being addressed over time — for example, the MathJax library enables nice-looking math expressions written in a familiar subset of LaTeX’s syntax. Read this blog to understand the importance of converting PDF files into machine-readable files.

Publish before peer review

I see the main benefit of peer review as a way to catch errors and improve a paper before it’s widely read. These are genuine benefits that shouldn’t be lost — but often a paper can provide great value as soon as the authors are ready to share it. Yes, there may be errors, but there may always be errors. By providing complete transparency into the level of peer review performed, there need be no false sense that a paper has been endorsed by anyone but the authors. This works well for established researchers whose papers come with some credibility, and it doesn’t remove the value of eventually editing based on peer review after a paper has been posted online.

All journals are overlays

The idea behind an overlay journal is that an editorial group may highlight excellent content to be featured with their seal of approval. Instead of being the original distributors of the content, an overlay journal expects the articles to have already been published. This way articles can be immediately posted online by authors while allowing for the opportunity that the article later receives the prestige currently associated with the most well-regarded journals or conferences.

All articles are open access

This statement is worth many elaborations, but there is not good reason for closed access publishers to continue existing. The arXiv costs around $10/article/year to run, while other means of distribution may have operating costs as high as $40,000/article [source]. Let’s go with the $10/article choice. This cost range makes it reasonable for institutions to offer hosting, or for a third-party host to subsist based on grants or author memberships that are closer to $10/article than the $5,000/article mentioned earlier.

Here’s something I believe in strongly, yet will only touch upon momentarily here:

There is incredible value to humanity in the idea that all knowledge should be free to everyone, everywhere.

We have the technology and the economic means in place to make this a reality. It’s primarily a matter of content creators and influencers pulling themselves away from the siren song of closed access publishers and toward the new and admittedly rockier ground of open access.

All articles are open source

This last idea is a reflection on the very purpose of papers themselves. The career path of many researchers depends greatly on metrics surrounding the papers they’ve published. As a result, papers are currently not just about their content or about teaching the reader. They are badges for authors, and stepping stones on a career path. Such a perspective is skewed. I’m not sure how to modify the system which encourages this point of view, but certainly the real long-term value in sharing knowledge is about communicating an idea clearly to readers.

To this end, it may be worth de-emphasizing the role any individual author has with a paper, and instead thinking of a paper as something closer to a wikipedia page or a multi-contributor github repository. If papers are living documents that are already expected to change over time, why not allow anyone to suggest changes? In case this model is new to you, it can be seen as the original author changing their role to an editor who may accept or reject edits suggested by anyone. This model has worked well for many projects, and I’d expect those benefits to transfer directly to the world of academic papers.

This post is built on blue-sky thinking about what’s possible in the future of academic publishing. The sad reality is that changes of this scope tend to happen slowly or not at all — especially when billions of dollars are solidly invested in maintaining the status quo, and when those with the most power to initiate change are also insulated from the negative impacts of the current state. Although my pessimism has grown over time, I’m happy to see many smart and dedicated people with similar visions who are actively pushing the community in promising directions.

Our suggestion is that you take a look at SciSpace if you want to simplify research workflows. Your research writing tasks, including literature searches, can all be completed through a single platform. It helps you assess the quality of a source by providing an overview of its references, citations, and performance metrics.

By combining writing and publishing tools, copyright detection technology, and searchable indexing, this platform will let you display and manage your scholarly output.