Please see: http://en.wikipedia.org/wiki/COLLADA

COLLADA provides a standard to exchange assets. The assets may vary in actual content, but the core data deals mainly with geometry and physics. Such core data makes up the heart of what virtual worlds are about.

Instead of the need for virtual worlds to reinvent the wheel to support a variety of asset information, like OBJ, 3DS, JT, Sculptied prims, etc, the protocols of the virtual world integrate the exchange methods of COLLADA in order to exchange such information in a standard way. Once the data is at its destination, it may be put trough filters that best match for an individual implementation. Hence, the burden for all destinations to carry the same implementation becomes nil. This also allows a particular implementation filter content once and then just use the optimized version from the filter output. The implementation may use the actual core data of the COLLADA format or the filtered output.

As for protocols that deal more with the exchange of the COLLADA format, they mainly need to understand the metadata of the format in order to transfer complete asset information, but they need not know about the actual core data.

Action: As a new feature request, this jira is intended to build consensus and collaborate resources about COLLADA. Most likely this will appear as a meta issue, but that is due more to the wide scope that assets affect. Immediate implementation is not requested in any particular area, but this jira may serve more to influence the current development to allow such collaborative design.