Innovation is not generally part of standard economic modelling. There exists no widely accepted method to quantitatively model innovation, beyond the standard simple learning curves at the micro or macro scales for process innovation, and nothing for product innovation. There exists, however, a wide range of established and empirically well-validated qualitative models of innovation systems that describe different stages of knowledge and product generation going from basic science to marketable products, and different functions of agents. Here, an overview is given of qualitative innovation theory, seeking to develop the basis for a quantitative model. A quantitative theoretical complex networks model of process innovation is then proposed that can be applied to patent network data allowing to derive learning curves as well as productivity growth in the aggregate economy. This model could potentially explain differences in rates of learning observed between sectors and countries, according to how the topology of the network of ideas is organised.