Should academics boycott Nature Machine Intelligence Journal?

Cover of Nature Machine Intelligence Journal Volume 1

Originally answered on Quora: Should academics boycott Nature Machine Intelligence Journal?

Nature Publishing Group (NPG) first announced Nature Machine Intelligence (NMI) to the public on April 30, 2018. The announcement generated a lot of discussion, as the machine learning community has largely embraced the principles of Open Science and Open Access (OA), which stand in stark contrast to NMI’s pay-wall model. As a strong advocate for OA, I am inclined to support a boycott against NMI. However, I also recognize that a simple boycott may not effectively persuade NPG to alter their policies, and critically, could limit the dissemination of artificial intelligence and machine learning research. Below, I discuss why publish in a journal such as NMI and propose a strategy of applying pressure to NPG by creating more effective competition.

Why would someone read Nature MI?

The NPG family of journals are curated and formatted with the goal of publishing highly impactful research that also appeals to a broader audience: those who would be interested in the topic but may or may not be an expert. This means that the main text is going to be necessarily less technical in comparison to what you’ll find in current top-tier AI/ML journals. That could be a pro or con, depending on who you are or what you prefer. They likely allow more detailed methods as an online supplement so it’s not really a con.

Why would a researcher (academic or tech) publish in Nature MI?

People publish in Nature journals because of the name. Yes, the name is prestigious and that’s a big component, but crucially, the name has reach. Publishing in a Nature journal will make it more likely for your research to be read by a greater number of people and by a larger variety of readers. That’s the Nature brand. If you feel like your AI research has broad appeal and you want to reach those people, you’ll want to publish in a journal that both specialists and non-specialists will read. Sadly, it’s not going to be JMLR, NIPS, ICML, or any of those IEEEs.

NPG knows this so they’re confident that researchers are going to publish with them anyway—OA or no OA. Besides, they allow authors to upload and distribute preprints of their manuscript. From NPG’s perspective, they probably think we’re a bunch of whiners trying to bully them with a threat of boycott! They are not going to budge.

What can be done?

Big tech-related companies are well-situated to make the largest impact. If a broadly influential company like Google were to create an accessible, academic OA journal in AI/ML, with the right system in place, that’s going to have both name and reach. The main text shouldn’t be overly technical so that it can reach a wide audience. In short, a new OA journal in AI/ML would have to match the influence of a Nature publication and market to a broad readership. When OA journals increasingly compete with NMI readership, researchers will have the option to choose their publishing platform without sacrificing reach.

That’s my 2 cents.