A claim from Lux research (reported in Small Times here) that China is now second only to the USA in its output of academic nanoscience papers is being met with some scepticism over on Nanodot. While there is clearly a real and important story about the huge recent growth in nanoscience capability in China, I’m also a bit sceptical about the central claim of this story, about China’s publication share. Of course, I don’t know about the detailed methodology in the publications study the Lux report cites. But I do know how a study which reached a very similar conclusion, commissioned for the UK’s science funding agency EPSRC, was done. Essentially, a database search was done for papers with “nano” or some compound thereof in the title.
I can do this too. If we look in “Web of Science” at papers published in 2004 and 2005 with “nano” or a compound thereof in title or abstract, we find that from a total of 59,938 papers, 10,546 – 18% – have at least one address from China. This is still behind the USA, with 28%, but is ahead of Japan, at 11% and Germany, at 8%. The UK is futher behind still, at 4%. (actually, the UK shows up only a pitiful total of 27 papers – 2370 are listed under England, with Wales and Scotland adding a further 487. I never realised British science had such separatist tendencies!). Of course, working out the sums this way will give a set of percentages that add to a total of more than 100%, since many papers have coauthors from different countries.
What’s wrong with this is perhaps only clear to scientists who are working in the field. When I think of what I believe to be the most significant papers in nanoscience, most of them simply don’t mention “nano” anywhere in the title. Why should they? Unless they are actually about carbon nanotubes, their title and abstract will generally refer to something much more specific than the rather general and all-encompassing “nano” label. We can get some feel for the fraction of significant and relevant papers that are excluded by this methodology by asking what proportion of papers by leaders in the nanoscience field would actually show up in a search like this. For example, taking a few more or less random US nanoscientists, only 24% of Whitesides’s papers would show up, 50% of James Heath’s, and even the rather radical and hardcore nanoscience of Ned Seeman and Fraser Stoddart still only pass the “nano” test 54% and 31% of the time respectively. Mark Ratner, despite being a prominent “nano” author, similarly would have nearly 70% of his publications slip undetected through the “nano” net.
And here in the UK, are we lagging behind quite so badly? Maybe, but again if we look at the output of some of our most prominent nanoscientists, we find most of their output is missed by this kind of bibliometric analysis. Of Richard Friend’s 35 papers, only 20% show up in this kind of search, while my Sheffield colleague, quantum dot guru Maurice Skolnick, similarly produced 35 papers, of which precisely 1 passed the nano-test.
I’m labouring the point now, and I’m sure the Lux people would say they’ve done their search in a much more sophisticated way. But I’m still convinced that any kind of mechanistic, keyword based search on the scientific literature is likely to lead to a highly distorted result, simply because what counts as “nanoscience” is so ill-defined. What you are seeing is not an accurate measure of nanoscience output, but a reflection of how strong is the fashion for attaching a “nano” label to ones work. This, of course, is somewhat unfair to people who are studying nanotubes, for example, who can hardly avoid putting “nano” in their titles and abstracts, but one’s strongly tempted to view the ratio (nano papers/total papers) as a kind of “nanohype index”. There is clearly genuine growing strength in China’s nanoscience output, and there is probably cause for concern in the UK, but these rather crude measures need to be taken with a substantial pinch of salt.
(And how do I score myself on the nanohype index? 7% on a total of 15 papers, I’m perversely proud to report).