One of the more interesting discussions at the Perimeter Institute-sponsored conference Science in the 21st Century involved the role that Web 2.0 technologies could play in the dissemination of tacit knowledge in science.

Sociologist of science Harry Collins made the argument that there are three kinds of tacit knowledge that scientists share:

1) "Contingent tacit knowledge" is knowledge we don't talk about. There can be several reasons for this. We may not realize it's important, because it's enshrined in custom or everyday routines, and thus is never absent. We may not share it because we don't know for sure what others know or don't know. It may remain ambiguous because different groups have different meanings for common terms. Contingent tacit knowledge could be formalized and shared, but it's not.
2) "Somatic tacit knowledge" is physical knowledge. Riding a bicycle is the canonical example: describing all the physics of bicycle-riding doesn't make it possible for a novice to stay stable. (Surgery is another great example of a body of practice that has a strong intellectual component, but is also very physical.) You have to learn by doing.
3) "Collective tacit knowledge" is knowledge that you can only get by immersing yourself in a culture or society. As Collins put it, if riding a bicycle is somatic tacit knowledge, knowing how to ride a bicycle in traffic is collective tacit knowledge. The rules for successful riding are different in San Francisco, Copenhagen, and Mumbai: even if you're successful riding in one place, you can fail in the others.

All three of these forms of tacit knowledge, Collins argues, are important for the success of science. This suggests that there are some hard limits to the gains or transformational impacts of formal publishing efforts like arXiv. As Collins put it, "The Internet is really useful because we've already got all the socialization necessary to use it." Circulating papers more efficiently offers advantages to communities that are already connected, but is not going to be sufficient to build science in the developing world, or support the creation of new scientific communities.

What difference could Web 2.0 technologies-- in particular, social software, but also short-form video, web cams, and other relatively light media-- make for the transmission of tacit knowledge? We've tended to think of the Web's impact on science in terms of displacing publishing, or making the sharing of data easier; but social software tends not to be about moving around large bodies of formal knowledge, so much as facilitating the sharing of small amounts of knowledge, at highly specific times, and in particular contexts. In other words, social software edges toward tacit knowledge.

While more conventional kinds of publishing are useful for communicating formal knowledge, published results are the tip of the iceberg; most of science consists of informal and tacit knowledge, and it's that knowledge that defines and binds together communities of practice, specialties, and other functional groups within science. In the long run, then, it may be that social software is as significant as arXiv, or open publishing, or new modes of peer review, precisely because of its disorganized, informal, contextual nature.