Think about your last month. With how many different people have you communicated? Are there some people with whom you usually write messages and others you prefer to talk to? Do you talk to different people differently?
Who we are determines who we contact and how we exchange information, integrating us into our social network. But what about the reverse? If an observer would just see our way of communicating, what could they learn about us?
Honey bees are similar to humans in that regard: They have many different ways to communicate and exchange information and a bee might communicate differently with her peers depending on the tasks they perform in the hive.
The complex social structure that guides the task allocation in honey bees makes them a great model organism for understanding how social systems work: some have to go out and search for food, while others care for the brood - and all that has to be organized without central control. What role does the social network of a bee play there? This was the question we pursued in our work.
Yet while observing bees over their lifetime is a lot easier than doing the same with humans, it’s still tremendously hard.
Researchers have observed individual bees by glueing little, numbered markers on their thoraxes or by painting them for many decades. But even the most diligent team of researchers cannot watch thousands of bees 24 hours a day for several months. We had to find a different solution: Over the years, we developed a custom marker akin to a QR code that optimally fits the curved body of a bee  together with a machine learning solution that can efficiently localize and decode all these markers [2, 3] and track the individual bees .
After marking thousands and thousands of bees, as well as spending years developing software, we could proudly say: Oh, how we underestimated the effort! Still, in the end, we had an incredibly detailed dataset containing the complete lives of thousands of bees in a mostly natural setting. But then we faced the next problem: How to make sense of this huge amount of data?
To answer the initial question, how much information a bee’s social network gives us about her role in the hive, we still had to find a way to handle the diverse social networks: Each network consists of hundreds of thousands of interactions and the position and behavior of a single bee can change from one day to the next. Eventually, we found a method that takes all that information and compresses it to a single number for each bee and day. This number characterizes the social role of the bee on the given day and lets us predict her current and future task allocation. We can even tell a lot more accurately when that bee is going to die than if we just knew her biological age!
We call this number the “network age” of a bee: while the biological age gives us a rough idea of what she might be doing, her network age allows us to follow her dynamically as she ‘ages’ through the social network and takes on different roles throughout her life.
Check out all the details in our paper !