Reimagining Academic Publishing with Game Theory

When Han Zhang published his first research paper as a PhD student at Duke, he couldn’t help but think that there must be a better way. After looking into the pitfalls of the current model of academic publishing more, he came to the conclusion that all the problems with the system point to a common root cause—poor incentive structures.

The lack of accountability or compensation for peer reviewers means that the research community is relying mostly on self-policing and altruism for that form of quality control. And disincentives in the form of resource costs and lower citation counts removes much of the will to do replication studies across academia.

This poor quality control atmosphere then leads to academia depending on weaker markers of quality, such as metrics like impact factor, or even the brand names of labs or journals, leading to the perverse scenario where academia correlates prestige with quality and pays prestigious private journal companies to disseminate their work whilst giving away exclusive distribution rights to that company.

Over time, journal companies like Springer Nature and Elsevier have built empires with annual revenues of $5 billion from publication and subscription fees, built off of monopolies of pieces of information that were discovered with mostly taxpayer funds.

“I was so taken aback by the whole process that I had to investigate whether or not what I experienced was normal—and it was,” said Zhang, who is in his third year as a PhD student in the laboratory of Cate Brinson, the Sharon C. and Harold L. Yoh, III Distinguished Professor of Mechanical Engineering and Materials Science (MEMS) at Duke. “I started wondering whether there might be mechanisms from the startup world that could be ported to academia. After all, startups do seem to efficiently incentivize collaborative work amongst employees and returns something to stakeholders if successful.”

After graduating from The University of British Columbia with a degree in engineering physics and a minor in mathematics, Zhang spent five years working for startups. He spent a year at Neuralink—making brain implants for Elon Musk—before finding his way to designing and building instruments for a Duke Engineering spinoff called Celldom. Launched by former MEMS Professor Benjamin Yellen, Celldom produces microfluidics chips for single-cell genomics experiments.

With that experience in mind, and after taking some classes at Duke in game theory and graph theory, Zhang isn’t content to just complain about the system; he’s working to create a new one. His advisor, Cate Brinson, is completely on board. “There is an increasing pressure to publish open-access, for good reason,” she said. “But with the commensurate $1,500 – $3,000 price tag per article to publish our own research, I was eager to work with Han toward a new model.”

It started as a simple concept. What if, instead of the current authorship system, credit for work done was represented by shares of a paper and those shares could be given to other academics in exchange for peer reviewer and replication services? After fleshing out the details more, he applied for funding through a Bass Connections project to turn his idea into a reality.

Launched in 2013, Bass Connections is a university-wide academic program that supports collaborative, interdisciplinary research at Duke. Through year-long research teams, intensive summer programs, semester-long courses and student research grants, faculty, graduate/professional students and undergraduates work together to explore big, unanswered questions about major societal challenges drawing on perspectives and methods from multiple disciplines.

The result of this collaboration is Liberata, an open access academic publishing platform with incentive structures for peer review and replications. Brinson says there are enormous advantages to a platform that can achieve open science, quality review and incentivized replication without spending precious grant dollars to unlock open access for any researcher to view the work and use its findings.

“The first key difference with the existing publication system is that the research paper is chopped up into contribution shares,” Zhang said. “That allows for more accurate pictures of what each author’s contributions were.”

In the current publication model, authors are listed in a convoluted order, usually with the first and last positions seen as generally more prestigious. Besides not giving an accurate view of each author’s contributions, many researchers take prime author positions and their associated credit where little is actually due.

To replace this, Liberata has the authors fill out an anonymous survey that attributes a percentage of work to each colleague. It then uses averaging and the discarding of outliers to bring the results together into a final reflection of contributions amongst the team.

After this step, the paper is released into an “academic marketplace,” where “shares” of the paper are offered to other researchers in exchange for their efforts in peer review and replication studies.

If a research team gave six percent of the credit for their paper out to three reviewers, and that caused their paper to receive at least six percent more citations, then it would be worth the trade for the authors. If a potential peer reviewer thought that two percent of the paper’s credit was worth more than the time it would take to do the whole experiment by themself, then it would be worth it for them, too. Similarly, researchers could offer perhaps 10% of a paper’s credit to those willing to replicate the results.

Ultimately, market forces between the availability of qualified scientific labor and the demand for good quality control results in equilibrium share costs for peer reviews and replications for different fields of science.

“Unlike the current system, where peer review is on a purely voluntary basis, this approach gives researchers an incentive to review and replicate more papers and to do a good job, because the better the paper ends up being, the more citations it will accumulate, and the more credit they’ll end up getting,” Zhang explained.

Brinson is especially excited about the replication aspect. “The ability to incentivize replication studies is a huge innovation and will lead to more open data sharing, which is great for science,” she said.

There is more nuance and mechanisms at play within Liberata than this short description, but that is the main idea and driving force behind the effort. The result is that, instead of traditional metrics (think H-index) based off crude numbers like citation count and authorship position, each researcher builds a sort of stock portfolio containing the weighted sum over all the papers they’ve been involved with in any role. These portfolios then can be analyzed meaningfully by almost all the existing financial metrics, leading to a much richer and more accurate picture of productivity and impact for individuals, labs, institutions, time periods and fields of science.

Whether or not this idea and system catches on with the larger academic community is yet to be seen. But Zhang is giving it his best shot.

Thus far, it has been a purely volunteer effort in the little spare time that can be mustered, but the group has already produced a cohesive explanation of their concept in the form of a website, video and technical papers at liberata.info. Moving into this summer and the coming academic year, the team plans to develop the demo platform while creating a marketing campaign to spread the word to academia—and to investors.

Since the announcement of the Bass Connections project, Zhang has formed a team of over 25 people: a mix of students, faculty and seasoned industry software engineers to accelerate the project.

It will be a long uphill battle for the idea to gain traction over the entrenched system, or even the free, open-source platform arXiv, but Zhang has high hopes.

“I’ve gotten feedback on Liberata from dozens of professors, PhD students, librarians and other academics, and it ranges from neutral to overwhelmingly positive,” Zhang said. “The primary concern people seem to have is whether the prevailing academic culture would be open to change, which is a promising sign in favor of the fundamental mechanism design. If the equilibrium position can be set at the right spot, and friction reduced enough, the system will change, even if slowly.”