Sir

As Luke Georghiou urges in his Commentary (Nature 452, 935–936; 2008), 'Europe's research system must change' — but most certainly not in the direction he proposes. He wants to strengthen strategic and applied research within the EU Framework Programme, advocating direct political influence on research into “problems that society recognizes as central” by creating even larger directed research programmes than we have now. But problems can arise from forced collaborations and top-to-bottom direction in large research consortia, as your recent Editorial and News Feature on the subject make clear (Nature 452, 665 and 682–684; 2008).

Almost every significant breakthrough in the history of science has come about by serendipity — not as a result of strategic planning or problem-oriented and directed research. To foster real innovation in Europe's science, he should be advocating bottom-up research by small and innovative groups.

Young investigators, even those with brilliant ideas and a novel approach, are unlikely to be invited to participate in huge directed consortia. These mostly involve laboratories that have already established their name and fame, and are now often comfortably operating on well-worn tracks or working opportunistically on headline-grabbing problems or fashionable topics.

Science and innovation are chaotic, stochastic processes that cannot be governed and controlled by desk-bound planners and politicians, whatever their intentions. Good scientists are by definition anarchists, who don't want to be managed by what Gottfried Schatz of Biocenter Basel calls 'chronoclasts' — people whose bureaucracy steals their research time and blunts their creative potential. Good science has an inherent potential for self-organization.

Let's have non-bureaucratic, generous research support for the broadest possible palette of basic science problems. That is what would change Europe's research system for the better.

See also European research must not go bananas Large projects can create useful partnerships