CREF Talk, March 22nd 2023, 2 PM Aula Fermi

Andrea Tacchella

The Language of Innovation

In the definition of Stuart Kauffman, the Adjacent Possible is the set of novelties that lie just one step farther than the knowledge that has already been explored. It is an ever-expanding space, as its size increases with new possibilities every time a new idea is explored.

In this sense, the concept of adjacent possible can’t be separated from the dynamic process of its exploration. This exploration, however, is not a single process happening simultaneously over all branches of knowledge but rather an emerging collective phenomenon. Every agent performs its exploration and has its own adjacent possible, a function of its personal knowledge (i.e., past exploration). The sum of these explorations and their interactions shapes the adjacent possible. This vision implies that no agent has a holistic view of the adjacent possible. Their exploration can be guided by a “science of innovation” that can extract emerging patterns from the collective exploration and anticipate its dynamics.

Here I propose ideas on how to structure a quantitative framework for this science of innovation that relies on ideas borrowed from computational language models and the concept of recombinant innovation to infer the landscape of the Adjacent possible and chart new possible ways through it.