Support for a pet notion.

"Successful research groups are those that grow and evolve on their own over time," Bejan said. "For example, an individual comes up with a good idea, gets funding, and new group begins to form around that good idea. This creates a framework where many smaller groups contribute to the whole."

However, extremes at either end of the spectrum are not conducive to productive science, Bejan said.

"If an institution is made up only of solitary researchers, it would have many ideas but little support," he said. "On the other hand, a group that is large for the sake of size would have a lot of support, but would comparatively have fewer ideas per investigator."

Such an extreme example would be that of the old Soviet-style research, where the government decreed the goal and scope of research and populated its monolithic structures with like-minded scientists. The more efficient laboratory model would be one that grows naturally, without dictates from above, Bejan said.

For these reasons, Bejan said there is not an inherent conflict between research empires and the individual, but rather a balance that serves the greater good. Institutional administrators therefore should not worry about such conflicts since the structure follows a natural design as described by the constructal theory, he said.

"I would argue that those administrators who coerce their colleagues into large groups solely to attract more funding, to beef up their curriculum vitae or to generate more papers, are acting against the self-organizing nature of the institution and its research," Bejan said. "Complete coalescence into large groups does not happen and will not happen."

On a larger scale it isn't just collaboration of free agents within institutions that yields results, it is also between them and others in the world wherever they are found and whatever they study. In the past this has been so on a scale limited by transportation and communication technologies as well as national chauvinism. I think that these semi-permeable barriers are melting away due to infrastructure advances, though it's still early days.

Update: Margin Call

My guess is that at the margin, scientific communication would benefit from more blogging and less effort put into journal articles. We have a tenure mechanism that exalts the peer-reviewed journal article. This distorts resource allocation, leading to inefficiency in scientific communication and discovery.