Frustration with the process of publishing academic papers seems to be widespread in the scientific fields.
Major publishers charge exorbitant amounts to read each article, keeping them out of the hands of all but those in universities able to purchase access in bulk. This system is also inextricably linked to the death of Aaron Swartz. Open access publishers charge exorbitant amounts to publish each article, allowing only those working or studying at universities that pay be printed. They have also resulted in a large number of fake academic journals, propping up the careers of fraudsters and simultaneously defrauding honest scientists.
On another level, the method for having a paper accepted into a reputable journal is outdated. Countless scientists lament the comments of reviewer two, often blamed for holding a paper back from publication. Results are delayed years through the revision process, resulting in academic science being far slower than private research and development. Experiments are rarely replicated, and the original papers are rarely updated or retracted when there are conflicting results. References are beneficial in looking at past research, but there is no way to look forward. The entire process puts power in the hands of those at the publisher, rather than in the hands of scientists.
The end result is an expensive document that is difficult to read and has often taken years to reach the scientific community. This system can and should be changed. The following is a thought experiment with regards to how academic publishing could be reimagined.
Rather than submitting a paper to a specific academic journal, a scientist would upload the paper to a research pool privately. Natural language processing would analyze the submitted documents for similar papers, allowing the scientist to do a bit more reading if necessary. It would also notify the scientist of potential unintentional plagiarism, so that modifications could be made prior to submission. Potentially appropriate editors would be notified of the article having been submitted based on the analysis of the paper and the primary topics of research as indicated by the scientist. One paper may have ten appropriate journals, so a scientist could get up to ten offers for review.
The scientist could choose the offer they feel is best. The submission would then be published in a new draft format. Importantly, any paper that cites a draft paper must clearly indicate it as having not been fully reviewed and marked with a tracer. Rather than simply getting comments from two or three reviewers, the scientist could also get entire sentences, paragraphs, and replications that further develop the paper. These could come from anyone who has seen the draft online. The scientist could choose to reject or accept these modifications, adding the person who provided the information as a contributor, a new category that would be different from an author. The scientist would make additional modifications based on the suggestions of the official reviewers and the scientific community, if necessary, before submitting it for final approval.
If the paper is approved, anyone who has cited the draft with the tracer would receive a notification that the version of their paper available has been deprecated. This, like the draft status, would be available for the community to see. Removing this status could be done by updating the paper or confirming that the modifications to the cited paper do not require an update. Either of which would need to be confirmed by the editor of the journal in which it is published. Any other papers that cited this paper would then be notified of the version change, with the changes highlighted, and their own deprecation status. This may seem as though it would be a lot of work, but keep in mind that it is rare a paper would cite another that is still “in press” and be published before the original.
Now that it is published, that is the end of the whole process, right? Wrong. The trouble with academic publishing does not end with simply getting the paper out. It should also include annotated forward citations. The forward citations would be easy to implement due to the nature of the Internet. While we are on the topic of citations, they should be annotated (where necessary) and in the margins of an article rather than at the bottom of the page or, worse, at the end of an article. This is for efficiency and clarity, and it is one area that all major style guides get wrong.
Finally, the end consumer of the research should be considered. Rarely does one need every detail contained in an article, but rarely is the abstract enough for more than a decision to continue reading. An effective summary could be provided by the scientist, a contributor, or another member of the community after or concurrent with publication. This could allow for much quicker dissemination of the 20% of information likely to be important to 80% of the interested scientific community. It is at this point that we can say that we have done our duty in publishing.
Journal editors, reviewers, and authors are generally not paid. The primary expenses come in printing the academic journals and in hosting them online. In 2015, print journals are an anachronism. The cost of developing, hosting, maintaining, and updating a service such as this would not be small. It could be on the order of hosting Wikipedia or Facebook, although the scientific community is admittedly smaller than either of these. Frankly, this cost should be borne by the public. The total cost would be less than a dollar per taxpayer (in the United States). Making all of this scientific knowledge available to the entire world would result in a boon for science and for education.