Ontological Computing

Communications of the ACM, February 2002. Viewpoint column.
A project of the Information Design Atelier

Felipe Castel
Information Design Atelier
Fort Lauderdale, Florida


The general issue of ontologies, viz. ‘How best to structure our concepts for effective computation’, leads more fundamentally to the philosophical issue of ‘Just what is computing?’. In seeking an answer, we will be touring some well-known areas within computer science.

A current issue within information design today is whether we should focus on designing information or on designing interactions [see Nathan Shedroff’s book Interaction Design, 2001]. In other words, should we design experiences or artifacts? Information Design as a field of activity intersects graphics, cognition, communication and computing, and it is a field which is about to bloom as we concentrate ever more on the notion of information.

But let’s get a better grip on the issue by considering it in the more familiar field of HCI. This component filed of computer science has struggled to establish a user-centered design framework over the years, one which considers the needs and the tasks of the users of a system, but yet it cannot [and must not] forsake the challenge of designing artifacts, products that users interact with.

This dynamic that underlies HCI generalizes to computing generally. Is it best to view what we are doing as product development or as process facilitation? The issue is a deep philosophical one, with on one side categorical ontologies of being, and on the other process philosophies a la Whitehead – although the issue goes all the way back to Aristotle and Heraclites, and before that to the wisdom of the venerable Tao Te Ching. In effect, everything can be viewed as thing or as process, depending on the perspective we decide to adopt, according to how we wish to view the phenomenon. The two views are complementary – although each leads to a certain take on the issue.

We can see this in considering information and cognition. The field of AI was built on the notion of artificial cognition – in effect the processing of conceptual elements [structural information tasks]. This very ambition is pursued today in the more trendy field of agent technologies, themselves a nice combination of artifact and process. Indeed, processes that operate on other information artifacts – as does once again all computing. This ties nicely into the traditional consideration of specifications, both structural and functional.

It all brings us back to ontologies and to the core nature of information. Ontologies are but the ways in which we carve up reality in order to understand and process it. Information, that still vague and generally misunderstood concept, is the product of that carving up. It is the representation we make of the world [in all its representational complexity]. It is the model we ascribe to in computing, the structure we create in order to make sense of the world and communicate among ourselves. Information, as must be realized, is functional, artifactual, and designed.

Why is this important? Why is it more than just academic [in the pejorative sense], more than merely an interesting philosophical issue? The crux of the matter in a nutshell is that we are moving away, rapidly and inexorably, from human computing. Computing is spilling out from the confines of user-centeredness, from its human focus.

What’s this? A scenario of robots gone wild, of computing run amuck? No, that is not the intent – although that possibility is seen as becoming ever more real, as seen through the eyes of futurists like Bill Joy [dark clouds on the horizon] or Ray Kursweil [clouds still white, but ever more potent].

The land of agents illustrates the trend well. Agents act on behalf of humans but interact not only with humans, but also with other agents as well. A program that initiates a sell of a given stock upon a certain condition acts on behalf of humans but also within a perhaps unforeseen set of circumstances that go beyond human foresight, in effect a context that is larger than the human one.

American HCI has yet to recognize this transformation within computing [see the ACM book Human-Computer Interaction in the New Millennium], as European HCI is beginning to do [xxx], and as we must all do sooner or later as computing continues to expand beyond its traditional user-centered confines.

A new perspective on computing is needed, one that re-centers on processes rather than tasks, and one that examines more closely the nature of information. Continuing references to information theory from mid-century engineering [a la Shannon & Weaver] show starkly just how dinosaurian we are in this respect.

It is important to really come to grips with information architecture – not in terms of system efficiencies or web site navigation, as is often the case, but in terms of designing information for use in computing, communication, and just plain enjoyment within entertainment and the arts.

Information is malleable. Ontologies are man-made frameworks. Computing does not just process information, it commits to a certain representation of information. There is active design going on within this process.

And information is structure, it is organization. Books and documents collect and archive that ‘knowledge’ for later use by others. But yet, the structural view remains diffuse – there is a continuum from data and information to knowledge and even wisdom, but no general scientific agreement on what is what.

These issues become very real as soon as we enter very applied domains like e-learning, for instance. We do design online learning materials for processing at a distance by interested and goal-oriented students. But can we define the learnability of these materials? Or is that too much of a stretch – putting too much emphasis on the artifactual dimension rather than the process one in which learning is traditionally viewed. Sadly enough, learning theory itself is currently felt sorely lacking within the field of instructional design.

And so, the issues continue to draw our attention, even though our ‘just get it done’ attitude generally prevails and we do indeed produce respectable materials and programs, albeit without often understanding why or how they might have been better.

The questioning currently going on in information design will continue and in the end help us better come to grips with the nature of information – and by extension with the nature of computing. The rapid development of agents and the coming of the singularity [the point at which progress reaches beyond our ability to keep up with it – see Kursweil] bring us face to face with the environment of computing and its new context beyond the human one. This will revolutionize HCI and much of computer science with it. The new face of computing remains to be drawn. Ontologies are key to this task, but they need to reach deep into philosophy, not just parade around the square.

        Philip Duchastel via e-mail