Reinventing Discovery by Michael Nielsen

This coming June will mark ten years since I started this blog (using Blogger on our own domain– here’s the very first post) and writing about physics on the Internet. This makes me one of the oldest science bloggers in the modern sense– Derek Lowe is the only one I know for sure has been doing this longer than I have, and while Bob Park’s “What’s New” and John Baez’s “This Week’s Finds” have been around longer, they started out as mailing lists, not true weblogs.

As such a long-term denizen of the Internet, I’m pretty much contractually obliged to have an opinion about Michael Nielsen’s new book, Reinventing Discovery: The New Era of Networked Science, now out form Princeton University Press. I know Michael from back in my misspent youth on Usenet, and I’ve heard him talk about this a few times (in person and via the Web), and I’m mentioned briefly in the book, so I was happy to get a review copy.

The central thesis here is an idea that’s become almost cliche: “The Internet changes everything!” The book argues that with the recent revolution in information technologies, it’s become possible to do science in entirely new ways. The ability to connect to vast numbers of other researchers, and even ordinary citizens, who might have useful expertise to bring to bear on a scientific problem opens the possibility of solving problems that would be completely intractable in the traditional mode of scientists working alone or in small groups. Web-based scientific enterprises like the Polymath Project and Galaxy zoo have generated impressive scientific results, proving new theorems and discovering new classes of astronomical objects by using the Internet to network together large numbers of people to tackle the problem in a distributed way. These projects are potentially the leading edge of a revolution in the way science is done.

In other hands, this can easily tip over the line into the sort of Internet triumphalism that I find really grating. What I appreciate most about this book, then, is what it isn’t. It’s not a grand, sweeping revolutionary manifesto taking this idea and running with it past all sensible limits, a la most of what’s been written about the “Singularity”. While it does highlight the revolutionary potential of the networking of science, it also clearly spells out the limits of the method. The Internet is changing everything, true, but some things will change more than others, and the big payoffs for networked science will come in a handful of fairly specific fields where the research problems have the right characteristics. As someone whose scientific background is in one of the areas least likely to be transformed by these tools (experimental AMO physics), I like seeing these limits acknowledged.

The book is also worth praising for what it is, though, which is a wide-ranging and engagingly written description of some of the coolest things that have been done using Internet technologies. These include things that are obviously scientific, like the projects linked above, but also activities that seem more frivolous, like the distributed chess game Kasparov vs. the World in 1999. These aren’t necessarily pieces that I would have put together, but once identified, they clearly fit, and make a convincing argument for the potential of networked science.

The other aspect of this that will draw a lot of discussion is the chapter on Open Science, which argues for more sharing of scientific information, from research papers (a la the arXiv) to raw data and analysis code. Again, I’m less convinced of the immediate benefits of this, largely because my own field of research is one of the least likely to show benefits, but it is notable for having one of the best arguments for open-ness that I’ve heard. Anticipating objections that giving work away “for free” is against the interests of scientists, Nielsen points out that we already give our work away for free– nobody gets paid for submitting scientific papers to journals, after all. The reward for publishing a paper is ultimately a social construct: we have agreed to use journal publication as a measure of scientific publication, and thus the reward for “giving away” data to a journal is the accumulation of prestige within the discipline. While this does indirectly lead to financial rewards through job promotion and so on, the immediate benefit of publishing a paper is status, not money.

As Nielsen notes in the final section, once tools became available to track citations to arXiv preprints, large swathes of theoretical physics shifted to using those as the marker of professional success, with journal publication becoming a secondary activity in some cases. There’s no reason that, given the right tools, we couldn’t also agree to treat the contribution of data to a public archive, or code to a community repository as professional activity in the same way that we count preprints as professional activity. All that’s really needed is a willingness on the part of the scientific community to expand our definition of professional activity. As I’ve written stuff along these lines before, I liked seeing that mentioned explicitly.

This is a fairly short book, so there are, of course, places where I would’ve liked to see more detail. One of the necessary conditions for successful networking of science is a kind of community management– you need ways to make sure that the useful contributions are brought forward and highlighted, which is a tricky problem. While the book highlights some successful examples– the Polymath Project and Kasparov vs. the World being the best examples– a quick scan of comment pages on the Internet will show that this is not a trivial matter. I would’ve liked to see more discussion of the failure modes of this sort of thing, and maybe some connections to the ideas of people like Teresa Nielsen Hayden who have put a lot of work into understanding what it takes to make a great commenting community on the Web. It’s not immediately clear to me that these sort of networked projects scale well– if their success depends on heroic community management efforts by people with both relevant expertise and the right sort of personality, then the effects might end up being less sweeping than anticipated. I think it’s too early to have good data for how this affects scientific projects, but above a certain threshold of popularity, blog and media comment sites tend to degenerate rapidly into an absolute sewer, and that gives me a little pause.

All in all, though, I highly recommend this book. It’s engagingly and persuasively written, while still being measured in its approach to the subject. If you have any interest in the way science is done in the modern age, and how it will be done in the future, you should pick up a copy.