I have found that it is a great desire of people within IT to try to find a one size fits all approach to innovation management. Unfortunately, as you have to play to the highest common denominator this can mean that innovation gets lots in a mountain of governance paperwork.
Luckily, with the advent of standards enabled enterprise SOA it is possible to keep control AND encourage fast innovation. This is possible by keeping tight control of the service producers (backends) so you know that the atomic business logic is sound and having governance over the various tools used to combine the services into Composite Applications so you know that the final application will fit into your infrastructure.
Once you have this in place you can start to allow projects to use the tools with the appropriate level of governance for the project, not in terms of the budget but in terms of the new service producers and new tools that it uses. If it creates no backend services and uses known tools it should require less governance than a project that delivers a large number of new services and uses a new composition environment.
Let me use yet another house analogy : Think of this in the same way you govern changes to your house. A quick paint job requires little planning and/or control, a major kitchen refit does not require external control but needs carefully planning and budget control, the extension you put on the front needs formal planning consent and if you knock the house down to build 3 smaller houses you might have to go to a planning committee to check all the infrastructure can cope.
Taking these concepts to enterprise SOA means we can relax the controls of some of the projects that are redecorating the existing landscape with composites that open up the system or automate collaboration. We can do this because the atomic backend services ensure that we still update the critical business objects correctly.
So in an enterprise SOA world one size does not fit all.