I met Cliff Nass in late 1994. He and his colleague Byron Reeves were consulting on a project at Microsoft called “Utopia”, often cited as an example of one the company’s greatest failures—Microsoft Bob. However, that attribution is unfortunate because Cliff (and the history behind Microsoft Bob) contributed in a very positive way to Hoaloha’s design philosophy and to the entire technology industry.
At that time, Cliff and Byron were both professors of sociology at Stanford University, initially researching how we humans interact and the almost innate, unconscious patterns of behavior we exhibit. For example, their research that human social behavior such as our natural tendency to give more positive feedback when face-to-face than when relating that information to another party, that participating within a group makes tend to feel more positive about our team members, that we hold stereotypes based on the gender of voices, and that we tend prefer personalities more like our own can also be applied in the design of interaction with technology.
Karen Fries, the smart, talented program manager for Utopia/Bob, brought these guys in as advisory consultants. Karen had been working on a fresh approach to interaction design to deal with the increasing complexities that were a part of conventional interfaces. The idea was consider how interfaces could make things easier by assisting the user through social interaction, somewhat like a seasoned waiter might help guide his customers through the choices on a menu, rather than allowing them to struggle through a plethora of options. Her vision was that “Bob” would present an approach that would cut a pathway through the dense forest of choices by understanding their intent and guiding, personalizing, and automating the process for them, making user interaction simpler, more efficient, and effective. In many ways, her motivation was similar to how Amazon and Netflix provide recommendations to help you in making a purchase selection.
In their design for creating an interface that would be simple and fun, the Bob team created a colorful interface populated with animated on-screen cartoon-like assistants. Unfortunately it came across as too simplistic and too child-like, making many users uncomfortable, as if they were being asked to learn to ride a bicycle by starting with a tricycle. Further, Bob did not fully isolate the user from Windows and non-Bob applications, so the transition between the two worlds may have been too jarring (somewhat like the complaint frequently cited between the desktop and tiled interfaces of Windows 8). Whatever the reasons, it did not matter, because the product was shunned and quickly withdrawn from the market by Microsoft.
Despite the failure of Bob, the Microsoft Office team believed that there was a kernel of a great idea in what Karen had set out to do and could help address the increasing complexity of their applications. So they modified the Bob code, integrating it in the form of a new Help interface called the Office Assistant. Initially the Office Assistant did not include the annoying paper clip character that helped lead to its demise from Office’s interface, but a variety of other characters.
Around that same time, I was equally inspired by Karen, Cliff, and Byron’s approach, but I hoped to side step the mistakes taken by the Bob team in the creation of a technology called Microsoft Agent. Agent also presented animated characters on-screen, but displayed them more integrated within the existing Windows interface. The animated agents could point to specific interface elements within an application, enabling the ability for the agent to walk a user through the steps to perform an operation. Agent also provided a simple application programming interface that enabled application and web developers outside of Microsoft the ability to include these agents in their own interfaces. I also added speech input and output as options, so interfaces could be designed to enable users to “talk” to their on-screen agents.
The Office team subsequently replaced their Bob implementation for the Office Assistant with Microsoft Agent. Unfortunately they also failed understand how to apply the technology effectively, creating not a helper, but an annoyance that slowed users down rather than improved their use of applications. In addition, the limited quality and accuracy of speech technology meant that speech could not be relied on as a part of a successful interface. This and the stigma of Bob’s past failure and mediocre response to the equally poor implementation in the Office Assistant resulted reluctance by the Windows team to include Agent as a core UI element; and Agent too faded from the scene, though did end up being used in some very innovative interface projects and has given me a better design perspective on what I am doing now.
All this suggests that while social interfaces can be attractive, it takes much more work than designing a conventional user interface in that you had to consider more than basic human cognitive abilities in terms of how they can interact with devices and technologies. You have to understand how humans interact with each other, and that the nuances of the choice of a word, how it is said, or a facial expression when spoken give meaning to its understanding. It is one thing to come up with design guidelines for how to present information effectively on the screen, but quite another to script social interaction to feel natural.
Cliff and his colleagues have demonstrated over and over again that we are social animals and that we often bring a social interpretation to the world we live in, even with what seems like non-social entities, including the myriad of technologies we use today. Helen Greiner, former co-founder of iRobot, often related about how Roomba and Packbot users gave names or spoke of their robots as if they were alive. This is further reflected in Dr. Sherry Turkle’s book, Alone Together, where her research also confirms that the human tendency to apply social behaviors in their interactions with robots. Cliff’s research demonstrates that it doesn’t take much to stimulate this form of response, but it does take a lot of work to deliver it effectively.
As a result Cliff’s research continued to be considered of important value and he ended up consulting with a number of companies, becoming one of the world’s foremost experts on the design of the social interfaces. In addition, his research into social-oriented user interfaces helped popularize important research being done with robotics at places like MIT, Georgia Tech, CMU, and USC.
The importance in the design of a companion robot was obvious. So soon after founding Hoaloha I invited Cliff to be a member of Hoaloha’s academic advisory board as I believed that his research and insights would be valuable. Cliff initially declined because he felt he couldn’t offer the time it might require. I have no doubt that would have been a reasonable argument as Cliff worked on a number of papers and books, consulted with a several companies, and pushed forward in his research. In addition, I knew he was a key member of the faculty at Stanford that helped welcome incoming freshman. However, because of the importance of his work and our shared interest in human-technology social interfaces, I continued to pursue him. After a year of discussion, Cliff accepted my invitation (though I have been delinquent in getting that posted to our website). I looked forward having him more involved in what we are developing, as the focus of his research was so well-aligned to what we are doing. Still his research will continue to inform and inspire our design.
So long good friend. You will be missed. There was so much more we could have done together to improve the world, but I greatly appreciate what you brought and know that it still lays an important foundation for what is yet to come.