It's no accident, I would suggest, that the latest wave of technological innovation, the one prompting Yancey's "new model", is also preoccupied with the social. In the past few years, the phrase "social software" has emerged, transformed by its adherents from a small subcategory of software (designed to assist users with the planning of social events) into what may turn out to be a new paradigm or era in the history of software development.
As Christopher Allen's history of social software reveals, sociality is hardly a new element in our use of technology. What has changed, however, is the centrality that social considerations now possess when it comes to the design of technologies. From a certain perspective, this change is a negligible one. Take email, for example. Without putting too fine a point on it, email accomplishes a simple function--it provides a nearly frictionless way to transport data from one person or place to another. This simplicity is one of the great strengths of email, and also one of its weaknesses (one that spammers have exploited relentlessly in recent years). With less information than you would need to put on an envelope, you can send an email to anyone with an address.
And as a result of this simplicity, email is put to a variety of social uses, from sharing information in the workplace to keeping in touch with friends, and much more besides. And yet, none of these uses are intrinsic, nor even really necessary, to the primary function of email, the transfer of data from one machine to another barring a direct connection between them. This function implies equally simple criteria with which we gauge the effectiveness of email--if it transfers this data accurately and quickly, it has served its purpose.
The development of social software, however, emerges into a much more complicated conceptual space. In "Expertise and Agency: Transformations of Ethos in Human-Computer Interaction," Carolyn Miller draws on the work of historian Paul Edwards as a means of locating ethos in our interactions with technology. Edwards delineates two "related discourses" that emerge in HCI during the cold war: closed-world discourse and cyborg discourse. She explains:
In this essay, I will explore these two modes as the rhetoric of machine control and the rhetoric of computational subjectivity, using the specific examples of expert systems and intelligent agents, two technologies in which the role of ethos is foregrounded. Both expert systems as intelligent agents blur the boundaries between human and machine, creating "hybrids" (in Latour's term) or "cyborgs" (in Haraway's). Such human-computer hybrids transfer to the computer some aspects of human character and require some adaptation by a human interactant, creating a "system," or dwelling place, where both must abide (199).
Expert systems are relatively self-explanatory; they are computer systems designed to operate at the level of a human expert. According to Miller, they
- depend on a database of specific knowledge,
- are designed to "learn,"
- provide accounts of their reasoning, and
- provide reasonable responses even when knowledge is uncertain or incomplete (199-200).
Amazon provides a perfect contemporary example of an expert system, in that it contains not only a vast database of bibliographic information, but an extrapolatory search function and because it tracks searches and purchases, the site learns in a way that provides expert information on books. This is not all that the site accomplishes, but at its core, Amazon provides users with a kind and breadth of expertise about books that even a human expert would be hard-pressed to duplicate. And this expertise is where the ethos of this system is located; in a more strictly organizational context, an expert system allows relatively minor informational tasks to be automated, freeing employees to devote more time to tasks requiring human interaction, judgment, etc.
Unlike expert systems, which centralize the knowledge of an organization into databases, an intelligent agent (usually a piece of software)
- interacts with its environment,
- must be able to perceive and initiate action in the environment, and
- is autonomous, "to some degree." (208)
Rather than mimicking (or even exceeding) the capacities of an expert, agents simulate a more casual form of information-gathering. For example, I may ask a friend for a book recommendation, a query based not (necessarily) on that friend's expertise, but on my experience that she and I have similar tastes. Depending on how closely our tastes coincide, I may consider particular movie reviewers (and indeed, they hope that I consider them) as agents.
Technologically, there are a number of intelligent agents devoted to searching. For example, Google allows you to initiate ongoing searches (Google alerts) that applies the search terms you set to pages as they are posted or published on the web, and sends you results via email. Although these queries don't offer much more functionality than the search engine itself, it is certainly possible to imagine that, were they to become more sophisticated, they might track which results tend to attract a user's attention and/or track other keywords that show up with some frequency. For instance, if I set a query to email me articles about "social networks" and Duncan Watts's work shows up in a number of my results, this more advanced agent might include Watts as an additional category of interest for me.
Rather than expertise, then, intelligent agents carry ethos insofar as they are worthy of our trust, an ethos that is much more aligned with pathos, according to Miller, and that can be distinguished from the logos offered by expert systems. They work partly because of the relationships that users develop with them.