Ned Gulley and Karim R. Lakhani presented The Dynamics of Collaborative Innovation (description, audio/video) last week at the Berkman Luncheon series. I had previously recorded and blogged Karim's similar, less detailed Open Innovation presentation back in May for the Berkman@10 Conference. Karim and Ned have been measuring various aspects of collaborative innovation around a programming contest that seems ideally suited for this purpose. I suspect that few real-world environments have such a good built-in mechanism for objectively measuring the strength of innovative contributions.
I found their insights into the differences between novel, game-changing submissions and incremental improvements particularly interesting. Within this programming contest, each individual user submission is objectively measured for performance against a desired outcome (e.g. algorithmic best fit), and the current best performing code submission is highlighted for all to see. This structure may create a problem in that innovative new approaches often do not immediately yield the best result as compared to an incremental code improvement. The social reward of being highlighted as the current best may encourage incremental improvements over novel approaches, potentially having the overall innovation outcome stuck in a local max.
Assuming that introducing novel ideas increases the chance of an eventual best outcome, then innovation environments like this might benefit from better incentives to reward novelty. Additionally, this contest environment has no inherent mechanism for identifying novel and potentially useful knowledge. Ned and Karim highlighted a specific example where a novel, under-performing programmatic approach introduced early was eventually adopted later in a programming contest and provided the conceptual foundation for the winning and final submission. Had this novel approach been overlooked, it is unlikely that the winning code would have performed as well.
I would argue that incentives designed to encourage the introduction and eventual sharing of novel information would prove useful, especially considering our human tendency towards only exchanging shared knowledge and withholding unique (and potentially important) knowledge in many social circumstances. Cass Sunstein explores this hidden profiles phenomenon at length in his book Infotopia.