We’re running a series of guest posts from the judges of the 2013 Chemistry World science communication competition. Here, science writer and Chemistry World columnist Philip Ball considers the place of chemistry in open science initiatives.


In the energetic current discussion about openness in science, chemistry has been largely absent. With the one obvious exception of drug trials – how can we encourage pharmaceutical companies to be more upfront with their findings? – chemistry seems to have been lost somewhere in the space between the life sciences, where the focus is on the accessibility and intelligibility of huge data sets, and physics, where open-access and participatory crowd-sourcing are already well advanced in projects such as the arXiv preprint server and Galaxy Zoo. Perhaps another way of saying this is that it is less obvious what is at stake for chemistry. Might it have continued to thrive on the basis of old models of how science is done, if left alone to do so?

My own view is that, among other things, a preprint server for chemical papers is long overdue, and I would rejoice if some enterprising institution were to initiate one. Partly this is selfish – for a science writer like me, the physics arXiv is an absolute boon for searching out interesting stories at the early stage, although of course this relies on the reader possessing some mechanisms for selectivity and discretion that do not depend on traditional peer review. But it is also invigorating to see how the arXiv has fostered a culture of active debate and engagement among physicists, in which responses to claims and controversies can be rapidly disseminated. That is something any science needs.

A preprint server for biology, called bioRxiv.org, has just been launched by Cold Spring Harbor Laboratory Press, and one for chemistry surely can’t be far behind. But that’s not to say that the model established by the arXiv has to be copied by the other sciences – there’s no unique way to go about disseminating and discussing preprints. I’d be interested to know what chemists in particular need and might look for in such a thing (I’d rate graphical abstracts as a must, for instance).

On the issue of data, I have encountered many debates and discussions about specific results and claims that require access to crystal structure data or simulation code. There’s no longer any argument for why such details cannot be made available, both during peer review and on publication. What’s more, computational tools appear to be moving towards greater standardization, so that for example raw data can be checked using off-the-shelf software. And the rise of well informed and well subscribed science blogs offers a growing forum for debating the issues free from the sometimes cumbersome procedures of traditional publishing.

Developments like this do seem to be cohering into a genuinely new way to do science – to forge collaborations, analyse data, share resources, communicate and assess results. No one yet knows what that will mean for time-honoured mechanisms of funding, networking and publishing, although one hopes that it might at least remove some of the entry barriers experienced by smaller laboratories or by researchers in developing countries. I’d love to hear what visions people have!


Philip Ball is a freelance writer. He previously worked for over 20 years as an editor for the international science journal Nature. He writes regularly in the scientific and popular media, and has authored many books on the interactions of the sciences, the arts, and the wider culture. Philip also writes for Chemistry Worldand has a regular column – ‘The Crucible‘.

VN:F [1.9.10_1130]
Rating: 0.0/10 (0 votes cast)
VN:F [1.9.10_1130]
Rating: 0 (from 0 votes)
Digg This
Reddit This
Stumble Now!
Share on Facebook
Bookmark this on Delicious
Share on LinkedIn
Bookmark this on Technorati
Post on Twitter
Google Buzz (aka. Google Reader)