Innovation adoption and diffusion when there are competing innovations

Firms face a choice between two innovations without knowing which of the two is better. Each firm has a prior that one of the innovations is better, which is updated by Bayes rule as trials with the innovations are made. A simple decision rule indicating which innovation should be tried, given any posterior, is derived and used to define an expected diffusion curve for the better innovation. This curve has the commonly observed ogive shape, even in the absence of any external demonstration effect, when the distribution of firm priors belongs to a general class.