King Digital Entertainment

Client: Vince Darley (Vince.Darley@king.com)

Vince made two suggestions

Automated%level%design%and/or%automated%game%play

Simulation%and%Modelling%of%a%Player%Population%

Alan's suggestion:

Humans are still better at making real world decisions than robots are - self-driving cars are notoriously bad at judging novel situations. The concept for this project is to use an abstract version of a real-world scene to collect and aggregate judgments from millions of users. Using the game mechanics of Candy Crush Saga, take the pixels from some part of a real-world image (say a blurry traffic light or road sign), and give prizes when large numbers of players agree on which action is best - left, right, go or stop. Player motivation can be increased by rendering the pixels as candy, so players don't feel they are being exploited, but individual intelligence will be aggregated by an algorithm that applies the results back to real images - perhaps by training a neural network.