Governing What Exists In a (Modelled) Forest: Darwinian Demons and the Algorithmic Life of Numbers

• In Event: Governing Natures by Data

Abstract

Attempts to better represent the world's vegetation through computer simulations have led models to include adaptation and evolution dynamics in their innerworkings. These dynamics, though, reintroduces classical ecological issues to these models, especially on how evolution of fitness relates to diversity. In vegetation modelling, this re-emerging figure is the Darwinian Demon, a speculative organism that, by becoming so well adapted to its environment, is able to outcompete any other organism and ends up colonizing the world. If this conceptual organism cannot exist in the natural world, due to biogeochemical limits, the digital environment is a much more favorable place. A central activity of modeling, thus, is to produce spaces opened enough to unexpected adaptation and able to suppress perfect solutions. This is regulated by tradeoff mechanisms, algorithms based on the notion that resources are limited, that establishes negative relations of proportion between disparate, and usually incompatible, measurement datasets of plants characteristic. Based on my ethnographic work about the development of a vegetation model for the Amazon Forest, what I proposed is to explore the making of these tradeoff mechanisms and how they produce presuppositions of ecological relations based on possibilities of connecting different dataset. First, I present a recollection of the Darwinian Demon concept, its relation to problems of counting and its material effects. Then, I explore how modelling practices connect different datasets through algorithm building and, to conclude, I present how these computational-ecological transformations affects the very numbers they rely on to describe what can exist inside the model.

Author

• Felipe Mammoli, UNICAMP