Chris Lintott assesses an idea that unites individual scientific expertise with collective intelligence.
Reinventing Discovery: The New Era of Networked Science
- Michael Nielsen
Imagine sitting down at your computer each morning and choosing what to work on from a list of problems submitted by scientists from around the world. The requests cut across academic boundaries, yet have been selected with you in mind as the person best qualified to solve them. What would such a system achieve? What would your career be like as a consequence?
This scenario is the vision of writer and quantum-computation expert Michael Nielsen. In 2007, he announced on his blog that he felt he could have more impact by developing new scientific tools than by pursuing his physics research. Reinventing Discovery, his thought-provoking call to arms, suggests he was right.
Nielsen is convinced that, as a result of our growing ability to share information and ideas, we are living through a revolution in how knowledge is constructed. He argues that, just as the widespread adoption of the scientific method in the seventeenth century made discovery a realistic aim for anyone who could devise the right experiment, so the rise of data-driven discovery will transform the ability of the average scientist to contribute. Although there is room for the lone genius, the rest of us need all the help we can get.
Researchers should take advantage of the emergence of this collective intelligence to tackle more difficult problems, Nielsen suggests. In one chapter (on which I gave feedback), he describes a variety of web-based 'citizen science' and other case studies to marshal his points. Some have research as their goal, such as my Galaxy Zoo project (http://www.galaxyzoo.org), which enlists the public's help in classifying galaxies, and the collaborative mathematics carried out by participants in the Polymath projects (http://www.polymathprojects.org). But a strength of Nielsen's book is that many more of the examples come from beyond academia.
For instance, a virtual game of chess in 1999 pitted the then world champion Garry Kasparov against 'the world', as represented by the votes of some 50,000 online participants. It was more keenly fought than expected. The game hinged on a critical move by 'the world' that was proposed by US women's champion Irina Krush. Moving the black queen to fork two pawns, Krush exploited her knowledge of a previous situation she'd studied to open up a novel position. She wasn't in Kasparov's league, but she had the ability and knowledge to exploit the situation in which her team found itself.
This idea of micro-expertise underlies Nielsen's vision of networked discovery. If enough people are exposed to a problem, he explains, the odds of one of them having the special insight or analytical ability needed to solve it increases, and so does the speed of discovery.
The key is that, as with Krush, the right person is faced with the right problem at the right time. What links the successful case studies in the book is that each has a system for directing the attention of the multitude of participants. Posts on the Polymath blog, for example, are structured so that each contains only one insight, making it easier to direct your contributions. It is this sort of systematic focusing of attention that moves us beyond the traditional 'wisdom of crowds' to a more powerful and subtle collective intelligence, capable of solving even complex problems.
It is tempting to rail against the lack of romance in Nielsen's collective vision, but a division of labour has been under way in science for decades. Many scientists are happy to leave the creation of large data sets to those who specialize in it: modern astronomers, for example, are more likely to write a database query than to visit a telescope. Last year saw the launch of ScienceExchange.com, which allows professional labs to sell their spare experimental capacity in a market tailored to that commodity. An exchange for inspiration, or at least for mental capacity, may not be far behind.
In a plea for an “open science imperative”, Nielsen argues that scientists need to do more than swap data and code. They need to share speculative thoughts — including half-finished ideas and open questions. Some are already doing this: take Rosie Redfield's open-resource blog chronicling her microbiology experiments and bench-based musings from her lab at the University of British Columbia, Canada (see http://go.nature.com/ae82wp). But it is still a lonely endeavour, and until a substantial number of leading researchers take the plunge, it is likely to remain so.
Nielsen ends simply, with the suggestion that we all try openness. The lack of a step-by-step programme for the revolution comes as a disappointment, but it is to his credit that he convinces us that radical change is a real possibility. Reinventing Discovery will frame serious discussion and inspire wild, disruptive ideas for the next decade.