The Cybernetics of HCI: A Pragmatic Approach
Paul Pangaro, Sun Microsystems
Seminar on People, Computers, and Design
Stanford University February 15, 2001
A discussion of cybernetics for HCI is pragmatic by nature because cybernetics is a "science of interactions" for which human-computer interaction is a subset. The ability of cybernetics to model the progression and limitations of human communications holds insight for difficult problems such as HCI design.
As we all know, semantic information processing by machine hasn't yet been achieved and today's technology is incapable of conversing with humans. Human-computer interaction is a means to the end of human-to-human exchange (messaging), human-to-team (collaborative work), or human-to-self (learning). One could paraphrase the goal of HCI therefore as supporting effective human-through-computer interaction. Interactions of hands and eyes to input and screen are necessary but unfortunate intermediations. What we want are our wants to be manifest. With its practical tools for modeling purpose, feedback, and autonomy, cybernetics has something to offer.
Early cybernetics was responsible for the now-common but once startling and revolutionary word "feedback", along with the first electronic "turtles", and the first "learning machines." From the 1940s it held an even-handed consideration for either organic or machine embodiments of intelligence. Neural nets arose here. In the 1950s, with the ascendancy of the programmable computer and its theoretical equivalencies to the nervous system, the fathers of cybernetics begat the fathers of artificial intelligence. Then cybernetics turned inward to consider the purpose, feedback functions, and limitations of cybernetics itself -- that is, to observe human observing. With the 1960s it grew into an applied epistemology, a discipline whose modeling tools would crack the conundrum of subjectivity in a scientific frame.
With a sketch of theory and a smattering of screenshots, this talk will position cybernetics as a means to sharpen understanding of human-through-computer interactions in terms of:
- conversation, defined as the rise and evolution of distinctions in a semantic space; and its dual,
- intelligence, defined as a property of a system that can assert, achieve, and modify its goals.
Today's technology isn't doing much talking, but there is no a priori reason why conversations with an artifact cannot be achieved. The key is a prescriptive model for innovation, which is possible in today's cybernetics at least at an architectural level. At the implementation level, it is a matter of when, not if.
Paul Pangaro is a software designer and performer. At MIT he programmed computer-graphics on vector displays and one of the earliest color raster-scan machines, the latter at Negroponte's research lab in the mid-70s. The disconnect between symbolic programming and human expression forced an inquiry that led to a Ph.D. in cybernetics from Brunel University (UK). He founded a bi-continent software development company to build hypertext browsers with fine-grained adaptation to user goals and learning styles, under contract to US and UK agencies in the mid-80s. Pangaro was engaged by clients such as Du Pont and Lotus Development Corp to consider wider questions about the role of information technology in what is now called "the new economy." In dot-com startups he worked as CTO and prototyped desktop knowledge navigators. He is currently Senior Director at Sun Microsystems where he manages the group that delivers content and services to developers.
View this talk on line at CS547 on Stanford OnLine
Titles and abstracts for all years are available by year and by speaker.
For more information about HCI at Stanford see