* The Enterprise changed the way IT projects were funded;
* The Enterprise focussed on Infrastructure first; Processes second and Applications last;
* The Enterprise adopted a Product approach to IT;
* The Enterprise recognised the need to change the way they did business, not just IT;

The biggest barriers that all of these Enterprises encountered in realising Shared Services were; finding a way of funding the initial Infrastructure and Application Migration costs; and addressing the disconnect between the IT organisation and the Business.

In cases where Enterprises realised greatest success, there was strong support for the inititives at board level, the business was transformed in-line with the IT transformation, and very strict funding controls were put in place, to prevent the delivery of competing solutions and non-compliant projects across the Enterprise.

Where Enterprises were not able to exercise such tight funding controls, Shared Services were only ever realised at the Infrastructure level, where commoditisation is relatively simple and painless to acheive.

Possibly one of the reasons why Cyclopaths behave as they do is due to the difficulties in enhancing and extending existing services without breaking existing interactions – one of my goals in participating at the W3C is to break away from the copy and corrupt another stovepipe model for services – after all HTML and HTTP’s extensibility is why we have a Web. Getting agreements from the commons to change is very difficult, that’s why even in Open Source, a single entity such as Apache having ownership of a resource can make changes a project where each contributer maintains individual ownership rarely can.

Finally, wrt the Prisoners’ Dilemma, you might enjoy the Perl coding contest – some of the tactics, but possibly not the result fell outside of classic theory:
http://hackvan.com/pub/stig/articles/prisoners-dilemma/