Expanding a bit on my rationale for this, one less obvious opportunity would be changing the role of ongoing research and experimentation on established curriculum. What I mean by "curriculum" in this case extends from overall objectives through unit plans to individual activities, but is not strictly prescriptive and not necessarily a textbook. It also assumes that the overall approach of the curriculum is sound (if it isn't, there's not much to be done other than start over).
Currently. there is no technical or social mechanism, in the US at least, for a teacher (or school, or district, or state, or random university) to undertake research around a particular facet of a specific curriculum. I'm thinking of thorny problems like division of fractions. What happens if we use a certain kind of manipulative here? What about this piece of software? Sooner? Later? A little faster? Slower? The kind of things studied in lesson study.
There is lots of research around these issues, but the relationship between this research and actual changes in published texts is ambiguous at best. And, there is (as far as I can tell) little incentive for third parties to do research based around specific curricular implementations -- it is not your job to improve someone else's commercial product, and you can't redistribute a modified version, and I'm thinking of more finely grained research than would generate profitable stand-alone commercial products. Is there a market for supplements to commercial K-12 texts? I doubt it.
If we are talking about open curricular resources, however, everyone has more capacity and incentive to undertake, and particularly to publish, this kind of research, whether it is semi-formal action research by experienced classroom teachers, public employees in other administrative layers, university researchers, or other interested parties. There is an infrastructure for continuous, if not necessarily tidy and linear, improvement of curriculum. It would promote the ongoing development of already deployed curriculum to be approached as real science, not just commerce.