Reinventing Discovery_ The New Era of Networked Science - Michael Nielsen [21]
To put it another way, the big advantages of online collaboration over offline conversation are in scale and cognitive diversity. Imagine that the people at ASSET India had gotten together a group to brainstorm ideas for wireless routers. Unless they were extremely lucky, the group would not have contained anyone with the same kind of expertise as Zacary Brown. By increasing the scale of collaboration, online tools expand the range of available expertise, reducing the chance that the group will be blocked by a problem that no one in the group can solve. Ideally, designed serendipity and conversational critical mass will occur, enabling the group to explore in depth a far wider range of ideas than is possible in a small group, with its limited expertise.
How do online tools enable conversation to be scaled up? The obvious answer is that online tools make it easier for experts around the world to get together as part of a group. That is important, but it’s only a small part of what’s going on. In fact, by using a carefully designed architecture of attention, online tools enable collaborations to involve far more people than is practical in offline conversation. Let me describe how this worked in the Polymath Project. Superficially, the format of the Polymath Project, based on comments on blogs, seems similar to discussions of mathematics in face-to-face conversation. But it goes further in three important ways. First, when working online people pre-filter their comments more than in ordinary mathematical conversation. In offline conversations even the best mathematicians have long pauses, need to backtrack, and occasionally get confused. In the Polymath Project most comments distilled one point in a relatively sharp way. Second, as a reader it’s easy to skip rapidly over blog comments. When you’re face to face, if you don’t understand what someone’s saying, you may be stuck listening to them speak incomprehensibly for ten minutes. But on a blog you can glance at a comment for a few seconds, take note of the general idea, and move on. Third, when you skip a comment you always know that you can return to it later. It’s archived, and easily findable using search engines. The overall effect of these three differences is to scale up the number of people who can participate in the conversation. By increasing the scale of conversation the blog medium gives us access to the best ideas from a more cognitively diverse set of participants, and so designed serendipity and conversational critical mass are more likely to occur.
There is, however, an inherent trade-off in scaling up collaboration. On the one hand, a collaboration should involve the largest and most cognitively diverse group of participants possible. On the other hand, once the collaboration gets large enough participants cannot possibly pay attention to everything that’s going on. Instead, they perforce must begin paying attention to only some of the contributions. Ideally, the architecture of attention will direct participants to places where their particular talents are best suited to take the next step—where they have maximal comparative advantage. So each participant sees only part of the larger collaboration. As a simple example, InnoCentive classifies Challenges into subject areas, to help participants find the Challenges of most interest to them. In the next chapter, we’ll see some more sophisticated ways of helping people decide where to direct their attention. In this way, it’s possible to scale beyond the point where each participant must pay attention to the entire collaboration. Put another way, the art of scaling is to lter contributions so each participant sees only the contributions they personally will find most valuable and stimulating; the important thing isn’t what we see, it’s what we get to ignore. The better the filters, the better our attention is matched to opportunities to contribute. In a nutshell, an ideal architecture of attention enables the largest,