Monday, February 11, 2002

Collaboration Games

An intelligent and coordinated response to business change seems to assume a considerable level of trust between the collaborating parties. Many companies are not willing to share commercially valuable information with suppliers, and prefer to keep business partners at arms length. (This is of course another form of decoupling, and is also supported by web collaboration technologies.)

What these technologies also need to support are the various games that people play. One of the popular early forms of the booking problem was the diary synchronization problem. The idea was that everyone would publish their diary, and this would allow meetings to be scheduled automatically. It would also allow meetings to be dynamically rescheduled, and appropriate text messages sent to your mobile phone. Although applications of this sort have sometimes been successfully implemented, they are widely resisted in practice. People adopt all sorts of games to regain control of their own diary, including creating fake meetings to prevent other people booking them for real meetings.

With supply chain, the procurement game often involves placing more orders than you actually intend to take. This gives you a higher level of volume discounts, and some protection against the vagaries of supply. You can then cancel excess orders as late as possible, while trying to minimize any cancellation charges. Travel agents play similar games when they try to make provisional bookings and block bookings against anticipated demand. While such tactics may be regarded as antisocial, they may be perfectly well supported by web collaboration technologies.

Game-playing and other selfish behavior can be regarded in terms of a deficiency of trust between the collaborators. If we regard trust as a binary property (either you've got it or you haven't), then this results in seeing a large class of collaborations as only possible with perfect trust (which is a pretty restrictive condition when you think about it). But it is much more realistic to assume that most business collaborations involve imperfect levels of trust, with some conflicts of interest, and to find a way of building collaborations around this assumption.

Thus instead of designing a simple collaboration that assumes perfect cooperation between the collaborators, we can start to think about complex collaboration that allows for the possibility of selfish behavior and game-playing. This possibly requires more than one type of collaboration model. A single player may want a model that helps to determine the best (selfish) strategy to adopt. (Such a model could be used to build software that dynamically adjusts the strategy in real time, in order to exploit the observed behavior of the other players.) A self-regulating community or external regulator may want a model that helps to detect and prevent or mitigate anti-social behavior.

Amended extract from my article From Web Services to Web Collaborations (CBDI Journal, November 2002)

No comments: