Institutions are crucial organizational hubs within the science system. However they are notoriously tricky to change. It is interesting that to change the EU science funding landscape, rather than try to add a new function to existing EU research platforms, instead a new ERC (European Research Council) was created in 2007 which is highly independent and was able to start a new type of organization culture which it has been able to maintain.

Discussing the decline of science in France, then Germany, then UK, then USA from strengthening organization and more conformity, Hollingworth [http://faculty.history.wisc.edu/hollingsworth/documents/Nature.7-24-2008.The_end_of_the_science_superpowers.pdf] suggests “when organizational isomorphism is high, there are strong pressures for organizations to converge in their behavior and culture.” When organizations in different countries peer closely at each other, none takes the lead. If organization is “...more weakly developed, organizations generally have had greater autonomy and flexibility to develop new knowledge and to be highly innovative“.
In my view it is also competition (as well as organization) that removes diversity, because it prevents more risky experiments in how to fund and support science, for risk of failure. They noted that “isomorphic pressures are especially strong when actors in highly saturated environments are competing for the same finite resources“.

In their study of 271 biomedical discoveries, what helped was:
1) fairly high scientific diversity
2) capacity to recruit scientists who internalize scientific diversity
3) leaders who integrate, encourage, critique, fund scientific diversity, with good vision and background
4) flexibility and autonomy associated with loose coupling with the institutional environment
Constraints observed were departmental boundaries, hierarchies, funding focus, bureaucracy, and hyper-diversity.