To connect or not to connect (and if so, to whom)?

We have just had a perfect start of the New Year of 2025 by getting our paper published in npj Complexity, a new Nature Portfolio journal focusing on interdisciplinary topics of complex systems. The paper, published online on January 9, 2025, presents key discoveries of our 7-year-long interdisciplinary research project on how human collective teams perform in a collaborative setting and how their performance would be affected by social network structure and functional diversity of people. This is a topic that is quite relevant to today’s highly networked collaborative work environments, yet it is still largely underexplored.
This was a rather complicated human-subject experimental research project that involved lots of factors and components. We were able to complete most of the experiments within the first three years by 2020, but then the COVID-19 suddenly hit, and our post-experimental work and data analyses were significantly delayed for a couple of years due to the pandemic. We managed to complete the first version of the manuscript and submitted it to a different journal in July 2023. Since then, we had gone through what seasoned researchers would all know too well – the usual series of rejection after rejection, but this time for us for as many as 17 months. This was not necessarily the worst case in my academic career, but still it was long. The reason why this work had such a hard time in finding the right place to get published was probably because it was an extremely interdisciplinary study – involving experimental social sciences, network and complexity sciences, statistics, computer and computational sciences, machine learning and artificial intelligence (because we used AI to quantify semantic contents of the text data), and even ecology and evolutionary theory to some extent. It was very challenging to satisfy all the reviewers coming from all kinds of scientific disciplines at once, but we are very happy now that it finally finds a perfect venue (npj Complexity) in which this research fits very well.
What we did in this study was to set up a custom-made Twitter/X/Bluesky-like online collaboration platform and let people participate anonymously in a text-based collaborative ideation/design task (such as creating a marketing slogan). Their social network structure, which determined whose posts would show up in one’s timeline, was configured by us researchers in two different ways. In one condition, everyone was connected to everyone else, representing an intensely well-connected team. In another condition, the participants were arranged in a ring-shaped structure and each person was connected only to the nearest and second-nearest neighbors on the ring (that is, everyone had just four connections – two to the right and two to the left), representing a sparse, localized social network. These conditions were hidden from the participants. We were interested to see which condition would make the team perform better.
In modern days when technologies and connections are everywhere, we all tend to believe “the more connected, the better.” To our surprise, however, our experimental results showed the opposite – the teams structured in a sparse, ring-shaped social network tended to produce more diverse ideas with higher quality than those in a fully connected network. Meanwhile, people who participated in fully connected teams tended to express more self-confidence and greater overall satisfaction. These results paint a complex picture of human collaboration – that is, how our communication/interaction pathways are structured can affect our overall collective performance, but each one of us individually may not be able to perceive the effects accurately. Rather, having more connections and getting exposure to more information may put us into an illusion that we did great, while we actually didn’t.
Moreover, we also found that who the people were connected to could affect the performance of the teams as well. We collected a self-introduction essay describing their backgrounds from each participant and quantified its content and its distance from others’ essays using semantic analysis AI tools. In the sparse, ring-shaped social network, when people were connected to others with similar backgrounds to theirs, the teams tended to produce more diverse ideas. In contrast, when people were connected to others with dissimilar backgrounds to theirs, the teams tended to produce better quality ideas on average. And most puzzlingly, when people were connected completely randomly regardless of their backgrounds, that’s where the best ideas came out most frequently (even though they might not produce good ideas on average). We must note that these findings were statistically not significant because we were able to run only four experimental sessions for each condition. Nonetheless, these patterns were seen consistently in all the sessions, indicating that who people are connected to can influence their collective performance.
So, how can we make sense of all of these rather strange, sometimes counter-intuitive, experimental findings? We hypothesize that the ideas people generate are actually going through an evolutionary process in social network environments, just like biological organisms do in natural environments. In this view, each human individual is serving as a “habitat” in which many ideas live and get selected for or against by the individual’s personal criteria, and when someone shares an idea with their social neighbors, that means the idea successfully migrates from one habitat to other habitats, where it will be subjected to evaluation by other selection criteria. This is a rather non-traditional perspective in which human beings are pushed aside as the environmental context and the ideas are brought to the center stage as actively evolving agents, but we believe this perspective greatly helps interpret all of our experimental findings. We plan to continue conducting further experiments to test this hypothesis and explore ways of how we can utilize it to improve human collective creativity and productivity. Check out our paper for more details and let us know what you think!
Follow the Topic
-
npj Complexity
This is an open access, international, peer-reviewed journal dedicated to publishing the highest quality research on complex systems and their emergent behaviour at multiple scales.
Related Collections
With collections, you can get published faster and increase your visibility.
Information Warfare
Publishing Model: Open Access
Deadline: Jun 30, 2025
Networks and Space
Publishing Model: Open Access
Deadline: Apr 30, 2025
Please sign in or register for FREE
If you are a registered user on Research Communities by Springer Nature, please sign in