Designing human-machine interaction involves a close consideration of many of the less glamorous aspects of both people and computers. Among the most unpleasant of these is the problem of the perception of time. On the computer side, the processing speed of the machine and the level of its responsiveness may limit the complexity of interaction. Meanwhile, a human may have unrealistic expectations and begin to get annoyed if the computer does not provide a meaningful response within an arbitrary period of time.
Everyone has seen a person hitting an inanimate machine, whether in the form of an ATM, a computer monitor, or a vending machine. Likewise everyone has encountered an error message from a computer. Rather than signaling the devolution of the human species, the aggression points towards a fundamental inconsistency between the two. The human generally interacts in a sophisticated manner with other bipedal earthlings and sometimes with animals. The computer generally interacts efficiently with other machines.
When someone takes another person’s money and refuses to give it back, the wronged person is likely to get angry and hit the thief. Very rarely does the thief give an error message in response to a threat. Conversely, when two computers are transferring data via their serial ports, it’s a cold day in hell that one of them willfully transmits a nasty voltage spike down the line in response to a lost packet at the other end.
Designers have attempted to overcome the human-machine disparity in a number of ways. Myron Krueger in “Responsive Environments” typifies these efforts when he says that the machine should try to obtain as much information as possible about the person. By being informed, the machine would presumably be able to provide meaningful output. The question is, “how much information is enough”? In many ways, this dilemma in the design of interactive environments draws parallels to work on artificially intelligent robots.
Until very recently, most AI developments were concerned with acquiring an accurate internal representation of the real world. The work was based on an idealistic belief that an exact internal world model would facilitate more sophisticated and life-like behavior. Yet, as computers grow increasingly more powerful and memory gets cheaper, we still haven’t seen any convincingly life-like robots. Getting a computer to recognize a potato is rare enough.
Almost all supposedly cognizant man-made creations rely upon carefully controlled conditions and assumptions. If these conditions aren’t met by the world, the computer malfunctions. Like the mind of an arrogant human, the computer’s use of a precise internal representation of the world limits the capabilities of that machine to only those situations that can be matched against something in memory. But unlike an arrogant human, the computer isn’t capable of bullshitting its way out of a bad situation.
As technology-based interactive designers, we share many techniques, tricks, and hacks with the field of robotics. We can also maybe learn from the work of Rodney Brooks, the director of the MIT AI Lab. Faced with the lack of meaningful behavior in robots, Brooks began creating robots that had very little or no internal representation of the world. These simple robots were designed to react to the conditions of the here-and-now. Without a reliance on memory, adjusting to unexpected conditions proved no more difficult than the normal operating procedure. This made the robots more adaptable and versatile than traditional robots, and this technique has since become mainstream.
Absorbing these trends in robotics could be useful for us as students of design. In the experience of being human, behavior takes precedence over any technologically or philosophically inspired ideal of what “should be”. As neuro-cognitive scientists map out crude descriptions of our mental processes, we tend to forget that their results have absolutely nothing to do with the unchanging subjective experience of existence. This generation’s analogy of the brain as a parallel-processing computer is just the latest incarnation of an age-old tendency to compare the mind with the latest development in popular technology, be it a sewing machine, gearbox, or whatever. In other words, as Brooks showed, there is no need to rely on dated models of mental computation or behavior if those models are more adept at exhibiting their faults than at providing any positive result.
Similarly, in trying to design meaningful works here and now, why rely on technology for technology’s sake? As Bill Buxton proposes in “Less is More”, meaningful interaction and complexity are inversely proportional. To combine Brooks and Buxton’s ideas with a contemporary aesthetic, it’s easy to view most contemporary interactive art as clumsy and uninformed. In a society where technology ends up on discount shelves in a matter of months, it seems reasonably urgent to emphasize ideas and to use technology only in an incidental manner. The most fundamental rule of “supply and demand” economics holds that rare commodities are valuable, and commonly available commodities are cheap. As the availability of technology (and the people who know how to use it) increases and its price goes down, ideas are a better investment.