I agree with you, history is pretty much full of committee designed specs that failed, but full of defacto standards from commercially successful products that

I agree with you, history is pretty much full of committee designed specs that failed, but full of defacto standards from commercially successful products that are later standardized. A lot of the features, like group video calling, first need their use cases polished by user trial, then hopefully, they can be implemented on top of either existing standards, provide inputs to change those standards to satisfy the requirements. OpenID/OAuth is an instructive example. The specs were built very much driven by committee oriented thinking, and the resulting UX implementations out of the gate were much more complicated than the Facebook Connect experience. I think actually focusing on the "design" aspect of how users use it, and then extracting the relevant technical requirements into a spec works better.