Implicit in this recent article is the often heard argument that for science to be more relevant, or more clearly benefit society, individual research projects need to be scrutinized for their societal impact. In other words, if we want to increase the benefit of science to society, we need to increase the direct benefit of individual research projects to society.

That this is true is far from clear:

Science as a whole is meant to benefit society. However, it doesn’t follow from that that it will best do so when each piece *individually* directly benefits society (nor does it follow that the people producing those pieces must be directly motivated to benefit society).

Why? Because science is a collective endeavour that is made up of many parts, parts which interlock in unexpected ways, and which are frequently re-purposed in entirely unexpected ways.

A simple example is Boolean algebra. Boolean algebra involves representing logic with zeros and ones. It is at the heart of the estimated 2 *billion* computers in the world. Yet when George Boole developed it in the mid 19th century, none of that was remotely on the horizon. Likewise, what George Boole thought and felt about what he was doing at the time is entirely irrelevant to those 2 computers.

There is a wider principle here. Imagine a structure designed to shelter from the rain (say, a bus shelter). The structure doing this successfully is not dependent on all its parts individually sheltering from the rain. In fact, the most effective shelters may well have parts that merely support the functions of other parts (walls, foundations for those walls, all of which prop up the sheltering roof). And there are countless examples like this.

What is in operation here is, arguably, the so-called fallacy of division: the mistaken inference that what is true of a whole also has to be true of its individual parts.

It may, in the end, turn out to be the case that maximising the direct benefit of individual research projects such as PhD’s is the best way to maximise the overall benefit of science to society, but actual evidence for this will be required.