Distributed literature search

In his classic book "Laboratory Life: The Construction of Scientific Facts", Bruno Latour observed a huge amount of energy spent by researchers in the discussion of scientific manuscripts, their own and the ones and from others. This discussion created a rich knowledge environment focused on their area of expertise, allowing participating researchers to absorb this distributed knowledge shared within their community. This discussion of the literature had implications not only for how their new manuscripts were crafted, but also on how their experiments were designed and interpreted.

With our transition to online environments, it is still unclear how such an environment for literature review and discussion could be created without face to face interaction. So far collaboratories have largely failed in that they could not re-create the same level of interaction. This leaves the question of what we might be missing? Not trying to be comprehensive, but here are a few possibilities:

• Culture still seems to be the major barrier. Our current researchers did not grow up in the Web age, and so do not feel as comfortable with virtual interaction as our teenagers. This will probably change over time as we dinosaurs die out.
• We need better tools to store and compare different publications. Thus far most Web applications have been focusing on the management of references for manuscripts, such as Zotero. Although effective for their intended purpose, they do not allow for knowledge construction, or the ability to create knowledge by combining portions from different articles, contrasting and criticizing them.
• The environment should be dynamic, with researchers having a well-defined personality and visible track records so that they have an incentive to "score points" while contributing to the discussion.
  
• Resulting data should be movable, so that researchers can not only move data across different platforms but also apply data mining tools for analysis and graphical display. Of course, for this to happen efficient ontologies have to be in place, created specifically with these potential applications in mind.

Too much to wish for? Well, modular experiments could be started now to see what works and what doesn't ...