Research News Briefs highlight new studies from the SFI community published in the last quarter. The following briefs appeared in SFI's Fall 2020 Parallax newsletter.
Ecology for pandemic prevention
What would it cost to prevent the pandemic from happening again? In a policy forum article for the journal Science, ecologist and SFI External Professor Andrew Dobson (Princeton University) and co-authors estimate that $260 billion over ten years — about 2% of Covid-19 economic damage — could substantially decrease humanity’s risk of contracting and spreading a pandemic viral pathogen from wildlife. The researchers’ suggestions for mitigating pandemic risk include activities such as stepping up the monitoring and regulation of the wildlife trade, and preventing the deforestation and fragmenting of tropical forests.
“There is substantial evidence that the rate of emergence of novel diseases is increasing and that their economic impacts are also increasing,” the researchers write. “Postponing a global strategy to reduce pandemic risk would lead to continued soaring costs.”
Read the paper at https://science.sciencemag.org/content/369/6502/379
Leadership for a sustainable future
Sustainability science focuses on ways to meet the needs of the present without jeopardizing the future. It’s a deep and complex area of research that encompasses many systems, from the environment to economics, and draws on the strengths of a spectrum of disciplines that meet at the intersection of nature and culture.
In the journal Sustainability Science, leaders of sustainability organizations including Christopher Boone, Dean of Arizona State University’s School of Sustainability, and Jennifer Dunne, SFI’s VP for Science, report on valuable lessons they’ve distilled from their own successful leadership experiences. The work describes pathways to success for tomorrow’s organizations and highlights five areas that can contribute to success, including intellectual resources, institutional policies, financial security, a physical space, and governing boards. Together, the authors propose, these resources can form a foundation for interdisciplinary and transdisciplinary researchers to identify solutions to complex problems.
Read the paper at doi.org/10.1007/s11625-020-00823-9
Little sets of nodes stick together
Within real-world networks, groups of nodes often have similar patterns of connections. People in a social network may have many friends within a group/community but few outside it. While there are multiple ways to divide the network into communities that do a comparable job summarizing its structure, those divisions can look drastically different.
In a recent Physical Review E paper, network scientists Maria Riolo, SFI Omidyar Fellow, and External Professor Mark Newman (University of Michigan) reported that while working on a community detection algorithm that sampled possible divisions of the network, they noticed little sets of nodes often stick together — even if that set of nodes was never identified as a separate community. “In many cases, the various good divisions of the network could be assembled using a small number of ‘building blocks’ of nodes that rarely get split up,” Riolo says.
Read the paper at doi.org/10.1103/PhysRevE.101.052306
Predicting links in online networks
“Networks are a powerful tool for modeling complex biological and social systems. However, most networks are incomplete, and missing connections can negatively affect scientific analyses,” write External Professor Aaron Clauset (CU Boulder) and coauthors in a new study published in Proceedings of the National Academy of Sciences. After applying 203 link prediction algorithms to 550 real-world networks, the researchers show that social networks are easier to predict than biological ones, and that no link prediction algorithm is best or worst overall. The techniques build on SFI Professor David Wolpert’s famous “no free lunch” theorem for machine learning, which proves that if a machine-learning algorithm excels at solving one type of problem, it has to fail at others.
Read the paper at doi.org/10.1073/pnas.1914950117
Are scientific papers too mainstream?
It’s possible to determine whether scientific papers are out-of-the-box, mainstream, or interdisciplinary. From the papers authors cite, you can infer what they read and which “crowd” or “specialty” they belong to. In a recent PLOS ONE paper, SFI External Professor Stefan Thurner (Complexity Science Hub Vienna) and coauthors present a bibliographic coupling network to visualize the position of every paper with respect to these crowds.
If your paper falls within the center of a cluster, it’s mainstream. If it’s at the surface of a cluster, it’s an out-of-the-box paper. If it’s between clusters, it’s interdisciplinary. “Our measures allow us to see general trends during the past 100 years within physics literature,” says Thurner. “Surprisingly, we found a trend across the past decade toward more mainstream papers and less willingness to publish interdisciplinary and out-of-the-box papers — at least within the physics domain.”
Read the paper at doi.org/10.1371/journal.pone.0230325