If you ask an academic about their experiences publishing in peer-reviewed journals, it’s a safe bet they’ll have at least one story of frustration; inattentive editors, low-quality reviews, poor copy-editing, difficulty removing or changing already-published work, and more. The time between submission and final publication can also be unbearably long, in some cases a year or more, with these waits being studied in fields including fishery studies and biomedicine. This all falls on top of the reality that mega-publishers like Elsevier are raking in billions of dollars in annual profits. Given the apparent low quality of publishing, it is difficult to see these earnings as anything other than rent-seeking. Worse, these issues relate only to the publisher; lab-side problems like poor reproducibility or lack of data- and code-sharing also exist.
Given these woes, it is no surprise that there is a growing push for reform in academic publishing. The idea of reform is maturing, and some attempts at solutions to specific problems already exist: open-access articles ensure research is freely accessible to readers, preprint servers like arXiv get results out sooner, data- and code-sharing practices improve reproducibility, and catch-all journals like PLOS ONE highlight the waning need for topic-specific journals in the internet age.
While each of these approaches have merit, they all ultimately feel like adaptations. Each still implicitly treats the peer-reviewed journal article as the de-facto standard. To me, this is unsatisfying. Scholastic output was not always published in journals – journals as we would recognize them today first appeared in the 1800’s, and peer-review was not the norm until the 1970’s – and there is no fundamental reason that the peer-reviewed journal is and always will be the optimal solution. The system has changed before, for the better, and it can change again, for the better. Do not misunderstand – I am not suggesting that scholastic output should not be reviewed by other scholars; I suggest that the way which we go about it can be improved, and such improvement might require starting from first principles.
It is difficult for people to imagine things wholly outside their experience. Science fiction authors keep making aliens that are almost-humans, people tend to assume that their way of doing things is the best or only way, all art is derivative, etc. For scientific publishing, however, the exercise of imagining a system from the ground up can be useful. I mulled over this idea and tried to imagine what properties an effective scientific publishing platform should possess, given current technology. It was helpful to imagine what I would want to see in a publishing platform if I was rebuilding humanity’s scholastic endeavour from scratch. Here is the list of properties I arrived at:
Scholars must be empowered to share their work in a format they deem suitable
At first glance this might seem obvious, a sort of academic publishing zeroth law. Obviously a research sharing mechanism needs to enable sharing research. The tricky part is in the latter half of the requirement: …in a format they deem suitable. Historically the written word, possibly interspersed with helpful figures and tables, has been the overwhelming standard. This need not be the case, especially in an era where many (most?) researchers will be consuming the research product on a computer. If a researcher deems a movie, image, poster, infographic, lecture, or any other format to be the best way to communicate their results, they should be able to communicate it in such a format. In some cases there is growing acknowledgement and acceptance of alternative formats. My alma mater, for example, notes in its doctoral thesis guidelines that alternative formats might be accepted. Formats like video have also been widely successful for popular science and math communications, with YouTube channels like 3Blue1Brown and Numberphile using effective videos to explain complicated mathematics topics at a high level to wide audiences. Many journals also accept multimedia files as supplementary material for articles, so researchers can technically include additional information, but in these cases the files remain supplementary and the author is clearly expected to treat the written article as the primary communication.
I have also seen some points raised about traditional forms of indigenous knowledge like oral knowledge. I am not well-enough versed in such matters to give meaningful consideration. However, I view the existence of such discussions as further reason to permit broader definitions of what counts as meaningful knowledge-sharing. Not everyone finds the sterile, clinical academic paper to be the best form of communication. Personally, I also suspect that alternative communication forms will be helpful for those living with any of the myriad neurological states that can make a written paper difficult to parse.
Other people should have a chance to review the work and offer feedback
This is another that at first glance is self-evident. It’s just peer-review. While the peer-review system in its current form is flawed, the core idea of fielding feedback on one’s own work, and providing feedback to others’ work, is foundational to science operating as a community effort. There are a few common properties of the current system of peer-review, however, that should be permitted to expire.
First, peer-review should be applicable to the entirety of a research project, not just the final publication, wherever possible. Currently, prior to publication only labmates and existing colleagues might have a chance to see and comment on a single project as it is unfolding. At best, a paper might get released as a preprint, but this is still effectively a finished publication. If researchers start sharing their work before it’s finished, which I discuss more below, then there is an opportunity for peers to provide comment earlier in the process, potentially avoiding the headaches of reanalysis, additional experiments, or flaws that go unrecognized until the project is near-completion. This also encourages review of the work itself rather than just the final communication.
Second, non-peers should be able to provide review. This is why I say Other people instead of Other scholars in the header. Public trust in science and scientists is endangered. Encouraging comment from non-experts, or even just experts from other fields, should serve to improve trust and improve the work. A comment from a layperson showing confusion over a result, figure, or passage might mostly reflect their non-expertise, but it could also capture instances of ineffective communication. Making results accessible to non-experts can also make results accessible to junior researchers who lack jargon fluency (or senior researchers who lack jargon fluency). To anticipate a criticism of fielding public review: not all comments need to result in a change to the study or the way it is communicated. Sometimes a simple reply to a comment can inform a reader. In practice, this could look like a comments section attached to a project ( inappropriate comments can be removed).
Taking the second point further, helpful review can come not just from the public, but from experts in other fields. To provide an extreme example, consider the case where a medical researcher published a method of calculating the area under a curve. In other words, basic numeric integration. Fielding open comments from out-of-field experts, like mathematicians or statisticians, would have prevented this. To avoid the fear of nothing new under the sun preventing anything from being published lest some other expert say it is old news in their field, such a criticism need not entirely prevent the publication – even if something has been done before in another field, it can be useful for to see someone from your own field share examples or a method or idea being applied in familiar use-cases. The numerical integration example is trite, but cross-discipline collaboration is an oft-praised and little-practiced boon for practically any field: statisticians can help scientists devise better tests or avoid common pitfalls; professional programmers can help researchers improve their analysis code and find sneaky bugs; writers can help improve communications; graphic designers can critique figures; the hypothetical list of collaborations is effectively endless.
Access for authors and readers should be free, forever
Scholastic output should be shared widely and equitably, and the fastest way to ensure this happens is to remove financial barriers. Nobody should need to pay to publish their research or to read others research, especially since so much academic work is publicly-funded. Keeping research free would further the above-stated goal of broadening public engagement. With current journals often requiring subscriptions, members of the public would need to pay exorbitant fees to participate, meaning essentially only members of institutions maintaining a subscription to a variety of journals are permitted to participate in the research community.
While journals are increasingly moving towards free open-access models, this system is still broken because authors often need to pay a fee to publish. These types of open-access simply move the financial barrier from the consumer to the producer.
This requirement is one where there might be cases that it should not be maintained. In an idealized, post-scarcity world, scholars could pursue their craft without concern for compensation. In the real world, there might be cases where a scholar would expect to be paid for access to their results. Such cases might include private enterprises conducting research or developing software without public funding, researchers publishing large bodies of work or educational material like textbooks, or other cases where work cannot be publicly funded. Despite these caveats, most research results should be free to share and consume.
Authors should have control over their work, forever
Science evolves and sometimes it is necessary to change existing work. Scholarly pursuits are by their nature evergreen. Methods and ideas evolve and we find better ways of doing things. In practice this usually looks like a researcher starting a new project incorporating whatever new improvement they have in mind. Other times the improvement fixes a prior mistake. In some cases there might be benefit from or need to revisit past work. Researchers should not feel trapped by the need to forever edit their past work, lest they never do anything new, but currently the choice is often out of the researcher’s hands. In the existing system, revisiting past work could look like a correction, errata, corrigendum, retraction, etc. However, when publishing in peer-reviewed journals, authors relinquish an alarming amount of control to the journal and publisher. Stories abound of journals or institutions refusing to implement a change or withdraw a paper.
When one thinks of corrections or retractions their mind might go immediately to misconduct. Certainly such cases exist. I do not dwell on them here. Such cases are dramatic and naturally attract attention and have thus been discussed ad nauseum elsewhere. In a system including the requirements I have listed here the opportunity for community self-correction should arise naturally. I focus instead on other benefits to retaining control like genuine progress and privacy.
Retaining control over one’s work permits both continued ownership, for purposes of credit, copyright, or profit; and permits the opportunity to update things should that become desirable. Perhaps one wants to update a figure to be more colourblind-accessible, to update some analysis code, fix an incorrect or awkwardly-phrased sentence, or add some new data or additional results that do not themselves warrant a whole new publication. To expand on one of these cases, let us consider research involving analysis code. Put frankly, most scientists are poor programmers. In my experience I’ve met few scientists that have formal training in good programming practices beyond one or two undergraduate programming classes. I am personally no exception. There is good reason to believe that bugs are common in the scientific literature, and there are solid steps that scientists can take to improve their programming practices. Participation from professional programmers could help. Still, in the world of software, bugs still happen (constantly) and software is patched (constantly). In scholastic research, then, we should also allow for updates, patches, bug-fixes, pull requests, etc. In the current system, what would this look like? Squash a bug, and if results change, correct a paper? As mentioned, corrections are often resisted or might not even be worth the hassle. If, instead, scientific results were published in a system more akin to open-source software, even minor changes can be easily added, tracked, and explained. This would also apply for non-code parts of research projects, like the mentioned updates to improve a figure’s colours or fix a sentence.
Finally, there are legitimate reasons unrelated to the work itself that might lead a researcher to want to remove their work from the internet. Privacy or safety come to mind, especially in an era where people are increasingly aware of their online footprint. Individuals, including researchers, should have a right to rescind their consent to share. There are also other, trivial reasons for removal: things that are no-longer or were never needed, are now outdated, or that have been moved to a different location might all benefit from simple deletion.
The platform should inherently encourage sharing work from inception
Research is more than just a final paper of results. There is planning, funding acquisition, experimentation and data collection, data management, formal analysis, figure creation, writing, etc. With research sharing via the internet there is no need to restrict ourselves to summary papers – these can and should still be written, since I doubt many scholars have the time or patience to sift through raw files, but sharing work via a paper alone made more sense when the real estate in physical journals was limited.
Sharing raw data, planning notes, discussions, analysis code, and other details will also improve research transparency and reproducibility. This can also create more opportunities for review and participation from those in other fields and non-scholars, as I’ve discussed under some other headers here, since to participate they’d probably need access to the data, software, drafts, etc. anyways.
All of this, however, places an additional burden on researchers. It is extra work, and I do not begrudge researchers for avoiding extra work if the benefit is dubious or not guaranteed (and again, I am not free of sin here). Thus any publishing platform should make sharing easy. A good publishing platform should permit researchers to upload any sort of file without hassle. Updating files should also be easy; the ability to easily update research content is important for reasons I discussed above, but it also eases the mental burden of sharing work early when researchers know they can always come back and make changes. With easy updating, there would be less worry about any file being ready-for-sharing/final/complete before uploading.
Sharing work from inception would improve research culture. Creating a project might be as simple as planting a stake and uploading a planning or funding document. It is a way for researchers to say “I am here and I am studying this”. This alleviates the need to rush out half-baked results to avoid being scooped – everyone will know you were working on something because they can see that you created your project some time ago and provided regular updates. A counterpoint might be that this system would encourage a sort of academic patent trolling: researchers staking out projects with broad ideas without actually acting on them, simply to say they had the idea first. This should not be a problem in practice, however, because such an effort should be immediately obvious. If someone started a project but added little or no data or did not appear to be seriously working on it, as visible through regular activity and updates, then that project could be safely ignored. Legitimate research, even if scooped in the sense that another group finished their project first, would remain obvious and citable through the presence of regular activity and clear, serious analysis; this would require, however, that projects become citable as soon as they are posted. Another cultural improvement would be that researchers would simply be better-informed on what their peers are working on. This could ideally encourage greater collaborations – why work in isolation when you know your peer in another institution or country is working on a similar project? Pool your data and analysis and create something greater than the sum of its parts.
Non-scholars should have an opportunity to participate in research projects
I have already mentioned that non-scholars (i.e. members of the public) should be able to access and comment on research work, including while the research is in progress. This should extend to participation in the research itself.
Some scientists have already experimented with citizen science. This is a type of research project that includes non-expert members of the public. In my field of air quality studies, this could look like citizen scientists helping gather measurements with low-cost sensors. With the participation of the public, measurements can cover a wider area than would be possible with only a few lab researchers (i.e. students). In the case of air pollution specifically, a citizen carrying a sensor on their commute can measure what their actual pollution exposure is, enhancing the validity of the measurement over some idealized or estimated commute acted out by lab members. With a good research-sharing system, these citizen scientists can also participate in the uploading and sharing process and receive appropriate credit for their work.
Public participation should also include the examples I’ve mentioned previously, where professionals in fields not typically viewed as scholastic or academic can offer their expertise: graphic designers and artists can help create visualizations, writers can write, programmers can program and debug, etc.
One might question why members of the public would want to participate without benefit to themselves. First, I would argue that there is a benefit: with appropriate credit (described below), participants receive the benefit of this credit for their contributions and can honestly claim their participation in the research process. A designer, writer, programmer, or citizen participating alongside professional scientists themselves become scientists. Secondly, voluntary participation in efforts to better one’s community, local or at large, already exist. Large open-source software can have thousands of contributors. People volunteer in their communities. I would posit that the public wants to participate in science, but simply lacks the means to do so in the present system. With more of the world having access to the internet, an online-first publishing system has the opportunity to bring many more people into the scientific community, building trust and making science better along the way.
There should be no institutional or reputational barriers to access for authors or readers
This requirement is essentially a subrequirement that appears in multiple of the above requirements: the requirement that non-experts should be able to review work, publishing and consumption should be free, and that non-scholars should be able to participate. Since this implicit requirement appears multiple times, I’ve highlighted it with its own entry.
Science, and scholarly pursuits in general, should not be limited to a select group of insiders. With the current system, this is the case. While anyone is permitted to submit a paper to a journal or publish a preprint, realistically the barriers in place make it so most serious scientific work is only completed in institutional laboratories. Even if a university strives to ensure equity in admissions and hiring, this still discriminates against those who cannot afford to attend in the first place. I also suspect that it would be difficult for a scientist without affiliation to receive public funding.
Even with an ideal publishing system following all these requirements, I expect that most significant work would still be conducted by institutional laboratories. There is no reason, then, that the means of publishing should also place restrictions on access. This would just discourage public participation. You should not need a .edu email or institutional login to participate.
Research items should be easily indexable and searchable
It would be pointless to share work if nobody could ever find it. Any system needs to be searchable. There exist many scholarly search tools and databases like Google Scholar, Semantic Scholar, Scopus, EBSCO, PubMed, and Web of Science. Some of these also offer researcher profiles, linking scholars to their work and fellows. Each has their own specific advantages or fields of focus, and any new system should learn from these examples.
The research community needs to be able to search for works so as to find them in the first place, and work should be indexable and linked to author profiles so as to provide fair credit to the author. Citations and links should also be indexed.
Scholars must get credit for their contributions
This final point might, like some others, seem self-evident. The existing article-based paradigm also provides credit. However, the current approach is flawed. Too much attention is given to who earns (first) authorship, giving the appearance of one or a few scientists being responsible for most of the work. If one provides a small amount of aid they might be mentioned in an acknowledgements section, but since these are not indexed they are effectively useless for providing credit. Some fields are already recognizing the problematic nature of existing authorship tropes and are working to address these issues.
A new system of publishing should allow project originators, owners, and leaders to identify themselves as such, while also giving credit to all contributors for their specific contributions. To again reference open-source software, a contribution system like that present on GitHub would provide a good starting point. In this system, contributions are tied to specific contributors, while project owners retain the right to identify themselves as such. This would require extension, however, to credit participants who contributed in ways that do not appear in the project’s repository, such as attending planning meetings or helping collect measurements while someone else uploads the data.
Participants in research projects should also receive credit for contributions to the publishing process itself – that is, those who volunteer to act as editors, reviewers, or communicators should also be credited within the project. This would ensure that these roles are still beneficial in a system without financial barriers.
While this specific approach provides a starting point, in the end an ideal publishing system would provide credit to contributors for each of their specific contributions, no matter how minor. In practice achieving this to perfection might be impossible, but we should still aim to asymptotically approach this goal.
Final thoughts
This list is likely incomplete. I might revisit it. There are also likely criticisms for some of the points I have included that I have not considered.
As I mentioned near the start of this post, many existing improvements and reforms to publishing are really just patches to the status quo. Truly adopting all the requirements I list here would probably require an entirely new mode of research publishing. Widespread adoption would be difficult since journal articles are entrenched as the standard. What a new system could look like and how it could be adopted are beyond the scope of what I hoped to write in this post, however – I have briefly mentioned using a system akin to open-source software, and I feel this should work as a starting point. I believe that adoption of even a subset of my list here would dramatically improve the state of scholastic work-sharing.