Agile Ideation and Lean Innovation?

More on "agile futures" from some of my earlier posts on globalization 3.0 and extreme competition and how the only way to stay competitive will be to innovate faster and more frequently than the competition ...

So does that mean that the "most valuable features" to implement first will be the ones that are considered "innovative"? Before we can execute on doing agile development for innovative features we have to have some kind of initial "innovation clearinghouse" in the organization where we have a buffer of potential innovation idea-candidates. Those "gestational" ideas will need to undergo some kind of evaluation to decide which ones to toss, which ones to develop a business case for, and which ones to do some early prototyping of, which ones to "incubate", etc.

Eventually, I see two queues, one feeding the other. The "Candidate Innovations" queue will need to "feed" the agile/lean software development queue. Things on the candidate innovations queue will have to go thru some equivalent of iterating, test-first, pairing/brainstorming, refactoring, and continuous innovation integration so that the queue can take "raw" and "half-baked" ideas in the front and churn out fully-baked, concrete, actionable ideas to then feed the development queue.

So the one queue will exist to create "actionable knowledge" (ideation) and will then go into the queue that cranks out "executable knowledge" in the form of working software. Given this two-queued system, how does this work where the "software queue" has both a request (product) backlog and a sprint (iteration) backlog. Lots of things on the product-backlog might be viewed as waste. And yet if they have made it thru the "ideation" backlog to produce an actionable concept and business-case, then that will indeed have value (but it will be perishable value).

What would Lean+TOC say about how to unconstrain and eliminate waste and maximize flow of the innovation flow that feeds the agile development flow? (I'm assuming the innovation flow would be a bigger bottleneck than the software development flow)