Navigating computational ecology when you also like science-fiction
The thoughts and work of a scientist do not exist independent of that individual. Long before we become formal researchers, the way we reason and perceive reality is shaped by our experiences, which are in turn emergent from the culture and place into which we happen to have been born. For me, reading science-fiction as a young adult cultivated thought processes I don’t see as distinct from my formal scientific education.
In this talk, I describe two mental models I use to understand how speculative hard science-fiction relates to the practice of science: 1) the science-fiction horizon; and 2) decision tree search on the future. The first of these I argue is not real, and instead emerges from a form of psychological bias, with some ideas arbitrarily deemed beyond a horizon of science. The second I argue is actually how all of us perform science, albeit on a longer time-scale. Using examples from my own work in computational ecology, I describe how I use both the science-fiction horizon and decision tree search to further my own research. For the first, I argue that rapid recent developments in multi-modal large language models, as well as concerns on AI safety and control, have pushed back the science-fiction horizon. This shift has opened up space for computational ecologists to speculate on research—within the formal structures of science—irrespective of lagging technology. For the second, I argue that early career computational ecologists should imagine reducibly complex long-term future goals, and then apply an informal decision tree search to help realise those goals. I conclude by arguing that decision tree search at the new science-fiction horizon might be a path to solving problematic dimensions of biodiversity change, guarding against the risks of our own conservation technologies, simulating ecosystems at large scale, and better understanding the phenomenological laws of ecology.