Dilution of expertise in the rise and fall of collective innovation

Abstract

Abstract Diversity drives both biological and artificial evolution. A prevalent assumption in cultural evolution is that the generation of novel features is an inherent property of a subset of the population (e.g., experts). In contrast, diversity—the fraction of objects in the corpus that are unique—exhibits complex collective dynamics such as oscillations that cannot be simply reduced to individual attributes. Here, we explore how a popular cultural domain can rapidly expand to the point where it exceeds the supply of subject-specific experts and the balance favours imitation over invention. At this point, we expect diversity to decrease and information redundancy to increase as ideas are increasingly copied rather than invented. We test our model predictions on three case studies: early personal computers and home consoles, social media posts, and cryptocurrencies. Each example exhibits a relatively abrupt departure from standard diffusion models during the exponential increase in the number of imitators. We attribute this transition to the “dilution of expertise.” Our model recreates observed patterns of diversity, complexity and artifact trait distributions, as well as the collective boom-and-bust dynamics of innovation.