�պ�����

About four years ago, editors from Nature Portfolio, the team that publishes the journal Nature, reached out to �պ����� professor of computer science Laurent Hébert-Dufresne. They knew of his leading work using networks, contagion theory, and co-evolution to explore many kinds of problems—from wildfires to Covid, honeybees to hate speech. They wondered if he’d lead the formation of a new journal on network science. He didn’t think this was needed.  “What is needed?” he asks. “A journal that is designed to be interdisciplinary, that creates dialogue across truly different disciplines. That’s what many real-world problems need; they’re complex.”

On April 17, the journal  launched, with articles on “phase transitions of civil unrest,” how “adaptive link dynamics drive online hate networks,” and, yes, an article on honeybees and how they aggregate. Hébert-Dufresne is the editor-in-chief. He co-authored — “npj” stands for “Nature partner journal” and complexity is, as Hébert-Dufresne and his colleagues write, “a perspective that embraces uncertainty and the need for multidisciplinarity in the face of large interconnected systems.”

�պ����� science writer Joshua Brown had a talk with Hébert-Dufresne to learn more about this journal and to find out why “complex” is more than just complicated.

Why did you launch this new journal?

If you talk to people about interdisciplinary journals, they may bring up Nature or PNAS or Science. But those journals, they're not focused on interdisciplinary work. What people mean is that they publish on almost any topic. But there are very few venues specifically designed to be interdisciplinary—to create dialogue between disciplines.

What’s really missing is a journal that publishes new research with that goal. That’s why we formed npj complexity. My job is to curate it enough to be interesting—so that people go see what's new on the page! It’s a gap that should have been filled 30 years ago: having journals dedicated to interdisciplinary dialogue.

Can you give me an example of what that looks like?

Yes, our first batch of papers is now out. In our second batch, we have a paper from �պ�����—led by Juniper Lovato and Jeremiah Onaolapo, who are both in our CS department, and Randall Harp, a philosopher—that uses an observational experiment where subjects look at videos. This is a good example because they’re studying misinformation and deep fakes and they're curious about people's abilities and biases when it comes to detecting deep fakes and false information. So, it has a lot of computer science—and connections with cybersecurity and whatnot, and there's models and typical computer science material in the paper. But a big component of the paper is an experiment. And so they borrow more from psychology experiments than they do from computer science. When first trying to get that paper published, they got feedback like, "well, why wasn't this pre-registered the way psychology does a lot now?"

They were pushing disciplinary boundaries! It didn't affect the science in any way, but some reviewers wondered: why did you not follow the tradition of psychology if you're going to do something inspired from psychological methods?

That's a good example of hurdles that face actual interdisciplinary work.  In the reply to one reviewer, I remember Jeremiah told me, “I didn't know that pre-registration of experiments was a thing!”  He didn't know it existed. Perhaps the study was reviewed by people whose perspective is too ingrained in their own discipline?  Npj complexity was very happy to take it!

Preregistration is a great tool in social science, but we shouldn't harm well-done research because disciplines have different cultures and tools that they rely on. That's the kind of paper that we hope to publish, with a mix of methods that may be hard to publish anywhere else. If researchers get reviews that say, “it's a good study but a weird approach— that may be good sign for us."

How wide a net would you be willing to cast for research?

In terms of scientific scope, there's almost no limit. We've commissioned pieces from philosophers who study theory of mind, and we asked them to review advances in the biology of cognition and computation, and they're happy to do that. We think that's important dialogue.

We’re also gathering collections of studies—open for submissions now! — on information warfare, networks and space, and the dynamics of interacting contagions. We're going to do some more based on ecology and biology in the next few months. In these, we specifically ask for work that tries to create dialogue between disciplines but written in a way to be potentially advantageous to scholars trying to do the same thing across other disciplines. We have a strong editorial line about the readability of the content across disciplines. And other than that, it's really open.

We're not trying to be the catchall for weird papers. We're trying to be the catchall for great papers that make an impact but are having a hard time getting published because they're weird. Our editors are more in communication with authors than at many journals to help writers tune their introduction, make them aware of literature that they didn't know about, making papers more readable. I mean they’re acting the way editors should—not just a machine sending papers to peer reviewers!

OK, so, what is complexity?

Let’s talk about complex systems instead. They’re easier to define than complexity! Complex systems are opposed not to simple systems, but to separable systems. And often different parts of these systems have different natures that we study with different disciplines. They may be sociotechnical systems that involve both algorithms, technology and engineering, and humans and social biases and so forth. And so, by their nature, we need to study them with tools from different disciplines and with dialogues across different expertise.

Now, I don't think that “complexity science” is a science. I think we use the phrase a lot and it depends on how you want to define a science. But complexity science doesn't have a shared set of questions, systems of interest, or tools. It's all over the place.

So, I use “complexity” almost as virtue signaling—to group the subfield of scientists interested in studying these complex systems and open to doing it with tools and perspective from different disciplines. Identifying with “complexity” is a signal that you're open to dialogue and weird ideas more so than a defined science. So that's why it's not npj complexity science, it's npj complexity. It's more of a perspective or a mentality and less of a science. That's how I see it. And I think that's the identity of the journal. No particular systems of interest, but more about a perspective with which we do many kinds of science.