On May 25-28 I had the privilege of attending the knowledge infrastructure workshop at SI in Ann Arbor. The workshop was a great opportunity to meet with colleagues from many different fields interested in better understanding the manner in which knowledge is generated and how that is changing in our digital culture. Hopefully the workshop will generate a community interested in studying this area and understanding the fundamental research questions necessary for the future. I’ll leave it for the official notetakers to report on the workshop, but I found the workshop both fascinating and intriguing, as well as frustrating. The frustration stems from the complexity of the issue and the problematic nature of getting major funding agencies such as the NSF to accept research the steps that a standard disciplinary boundaries. Thus far, the track record of the NSF funding research in this area has been less than stellar, with programs such as DataNet placing too much focus on immediate technical results without adequate attention on longer-term sociotechnical issues.
We were asked to submit short position papers for the workshop. Here’s mine; a few scratchy thoughts on the issues of boundary breaking and knowledge infrastructures.
Recently I’ve been reading about, becoming intrigued with, and/or beginning research in three areas I find important as we work to broaden our thinking about knowledge infrastructures (in my interpretation of that term). While each of these stands alone, there is a common thread among them; a blurring of both the boundaries of infrastructure and of the roles of its participants or agents.
Human Computation Humans were the first computers (the women, by and large, who performed computations for math intensive efforts such as WWII artillery trajectory tables or the Manhattan Project). We are currently seeing increased interest in re-integrating human effort back into the computational process, in a complimentary relationship with machines. Some prominent examples are Amazon’s Mechanical Turk, recaptcha’, and the many citizen science projects (more about this below) such as eBird and Galaxy Zoo. This has been called “Human Computation” and the book of the same by Edith Law and Luis von Ahn provides a wonderful overview of this area. It describes through many examples the effectiveness of combining machine ‘intelligence”, adept at effortlessly and efficiently perform large-scale well-defined tasks, with human intelligence, uniquely equipped with the ability to perform nuanced, context-dependent tasks that require ‘universal knowledge’. Tasks such as in-situ bird identification, which requires pattern recognition and context awareness abilities far beyond that of computers, fall into the class of tasks that presently, and for the foreseeable future, can only be performed by humans. Apropos to the workshop topic, this area adds an interesting dimension to the now familiar sociotechnical dialog on infrastructure. In addition to our accounting for human (social) influence on technical trajectories, we need to incorporate the technical/computational integration of humans and machines and the effect of that sociotechnical integration as we design infrastructures and predit their effect.
Citizen Scholarship The ubiquity of the networked knowledge infrastructure has had a dramatic effect on the producer/consumer and professional/amateur divide in the information realm. This is exemplified by well-known phenomena such as blogs (anyone can be an author) and YouTube (anyone can be a tv news reporter or producer of drama, comedy, etc). The manner in which this has disrupted established institutions such as newspapers, the entertainment business, traditional media etc is well-known. In the same manner, science and scholarship are no longer, and probably have never really been, a domain reserved for those who certified to do it (e.g. by degree, or license). In scholarly fields ranging from civil war history to climate change to birding to astronomy, the contributions of so-called amateurs are becoming not only accepted but necessary for full observational coverage. This presents challenges and tensions to knowledge infrastructures that, prior to the digital age, incorporated assumptions of role division, quality and integrity, and authority based on professional certification. Loosening these assumptions and creating new ones that retain important features of earlier infrastructures, such as the ability to distinguish quality from junk, is a significant knowledge infrastructure challenge.
Extended Mind and Extended Cognition I’ve recently been reading Andy Clark’s book ‘Supersizing the Mind’ that builds on his paper “The Extended Mind” with David Chalmers. Clark’s basic thesis is that human cognition extends beyond the physical brain to incorporate the extra-physical objects we use to make cognitive action possible. The canonical example he uses is an early dementia man, Otto, who has seamlessly integrated a notebook into his daily cognitive life and employs it in the same manner as his pre-disease memory. I am skeptical about many aspects of this theory and the looseness with which Clark employs it, but components of Clark’s explanation are worthy of consideration in our thinking about knowledge infrastructure. Especially compelling are the notions of “fluency” and “unmediated action” that are essential for the integration of external objects into the cognitive process. Clark explains these concepts by calling on Dourish’s notion of ‘inhabited interaction’; the natural, effortless, and immensely efficient integration of extended (extra physical) cognitive devices into our established cognitive processes, in contrast to the “disconnected control” of those external objects that, by their design, resist integration into this extended mind. While we don’t need to fully accept Clark’s thinking and turn infrastructure into part of extended cognition, I believe it is critical that infrastructure design not only account for task integration, but also for cognitive integration.
A. Clark and D. Chalmers, “The extended mind,” Analysis
E. Law and L. von Ahn, Human Computation. Morgan & Claypool, 2011
S. Kelling et al., “eBird: A Human/Computer Learning Network for Biodiversity Conservation and Research,” in Twenty-Fourth Annual Conference on Innovative Applications of Artificial Intelligence (IAAI-12), 2012