Sir

Your News Feature 'Group theory' (Nature 455, 720–723; 2008) highlights the dwindling contribution of single authors in the face of the productive increase in collaboration among scientists. Sociologist Brian Uzzi is not alone in questioning whether a lone researcher can still capture the moment when lightning strikes.

Uzzi uses some general metrics to suggest a correlation between the number of citations and the number of authors per paper. But before we rush to embrace these heuristics as a basis for designating best practices for collaboration, our social scientists need a better understanding of how high-impact discoveries in the life sciences come about, for example at different stages of the drug-discovery process.

Moving along this continuum from early discovery, through the translational stage and on to clinical trials, we would expect to see increases in the aggregate levels of funding and in the number of authors and papers published per year. Consequently, epidemiological evaluation will produce metrics and heuristics that more heavily reflect the practices employed in those later stages of the discovery process.

But social collaboration patterns probably vary with each stage. In the discovery phase, for example, projects are often driven by sole investigators, with support from a tightly knit group of colleagues from universities and biopharmaceutical enterprises, frequently generalists who move readily across disciplines and are at ease with counterintuitive interpretations. These small groups act as the initial nucleating agent, but they are then radically transformed into a complex web of interdependent specialists as the drug, vaccine or biological candidate is developed and moves towards the clinic.

With the explosion of available information, publishers and granting organizations are understandably desperate, seeking tools that more effectively predict high-impact science. But in searching for these tools through the lens of social network analysis, we must not lose sight of the key contribution by a project's pioneer. We shall then see more clearly how collaboration patterns alter as an area of scientific discovery matures.

Readers, reviewers and grant administrators should not be biased against early-stage papers with just one or two authors. Lightning can still strike the solitary explorer whose mind is prepared.