Breakthrough Markets is a novel approach that combines human and machine intelligence to coordinate scientists and researchers towards scientific breakthroughs. This mechanism extends the concept of prediction markets, where participants can bet on the relationship between two events as opposed to just single events. The key aspect of this mechanism is that it focuses less on minimizing mistakes in estimating the likelihood a particular research path will lead to a breakthrough, and more on generating a real-time map of the most promising combinations of research concepts for scientists and researchers to investigate.

In this model, humans and bots trade against each other until the market price for an event stabilizes. The data generated by this process is then fed as the input to a neural network that calibrates its aggregate estimate of all the participants over time. The objective is not to estimate the accuracy of a specific hypothesis, but rather to crowdsource the weight of the edges of a Bayesian network mapping the combinatorial relationships between different concepts within the scientific domain.

The Hybrid Prediction Market aims to identify opportunities for interdisciplinary breakthroughs by crowdsourcing a Bayesian network for scientific domains through a prediction market that merges human intelligence with artificial intelligence. Having a hybrid-prediction-market-backed ranked list of the most promising combinations of scientific concepts and results can accelerate the formation of interdisciplinary teams, whose efforts in turn are directed at research whose results are intersubjectively expected to lead to the biggest update in the Bayesian network.