A breakfast policylab with Michael Nielsen and Geoffrey Boulton at the Royal Society, my old patch. The topic was open science, about which Nielsen will publish his much-awaited book in October. Geoffrey Boulton chairs the Royal Society’s project on science and openness.
Nielsen tells a great story. (Have a look at this TED talk). He began with one about Galileo, who semi-published his discovery of Saturn’s rings in the form of an anagram that he would later decode. As with Robert Hooke fifty years later, this allowed him to take credit for discoveries but not fully reveal them until he was ready.
Nielsen’s retelling of the scientific revolution puts the arrival of scientific communication right alongside the scientific method (yes, I know there’s no such thing…) as vital innovations. And with the arrival of scientific communication, journals and peer review, the tussle between scientists’ individual interests and the common good begun. Early modern scientists were eager to advance knowledge, but they were just as eager to protect their reputations. It was often their patrons who imposed openness upon them.
As science moves online and data turns digital, the communication of science and its conduct are now overlapping. Open science, learning from the successes of open source computing, is acquiring more and more advocates. The opportunities seem clear. Nielsen described examples such as the Polymath project, Github and Stackoverflow. We might also point to Open wetware and Climate code as well as micro examples like Sharon Terry, the mother of a child who worked with university researchers on her child’s rare genetic illness and ended up patenting its gene. Open science is efficient science, able to produce knowledge more quickly and more innovatively. But open science is surely also democratically preferable – more transparent, with greater public involvement and easier rooting out of error and fraud. We might call this the ‘climategate’ factor.
And yet progress has not been as fast as hoped. Scientists might talk a good game about openness, but can they be trusted to deliver it? As Geoffrey Boulton described, scientists continue to talk about data as personal property rather than a shared resource. Some disciplines see the advantages of promiscuity, while others are more guarded. Incentives clearly matter, and for scientists the rewards system is well-established. If, for example, we could make data citable, so that scientists could refer directly to it as they do to papers, we would expect to see sharing improve.
But my question, drawin on Nielsen’s point about science’s early paymasters, was whether research funders also needed to apply more stick. If open science matters so much, shouldn’t we mandate it? The response, which seems absolutely correct, is that a stick would make scientists share, but would it not make them share well. If data, code, models etc. are to be useful, they need to be shared in such a way that they are usable. A grudging release of information is unlikely to be helpful.
There are examples in which a mandate has helped a research community get over a hump, after which they see the benefits of sharing and innovation begins. The Human Genome Project, with its Bermuda Principles requiring open access, is one example. The Royal Society group have received their evidence and are meeting as I write. They will have to decide what mix of carrot and stick will work for open science.
P.S. Michael Nielsen will also be speaking at the Science Online conference tomorrow, where I will also be saying some stuff about science policy.
P.P.S. I should declare that, in my previous life at the Royal Society, I initiated the policy study that Geoffrey Boulton chairs. I also asked Michael Nielsen to come and help that institution throw open its grand old windows.