Patterns are one form in which an individual or community can create and share knowledge. Patterns are typically more concrete than principles but more abstract than stories. I am not sure that Patterns are everyone's "cup of tea."
A speculation is that the level of concreteness and generality of knowledge depends on how quickly a field changes and how knowable it is. In the field of Newtonian physics, for instance, we can reasonably move all the way to very general equations. This makes sense because the relationships are relatively knowable, simple, and they presumably do not change much.
On the other hand, in a field where things change very rapidly and where it is very difficult to know what is really happening, stories may provide the right level of generality for knowledge sharing.
Patterns occupy an intermediate place.
Collaborative Innovation Tools
John C. Thomas
1. Importance of Collaboration: Practical and Scientific
We live in an increasingly interconnected world. In reflection of this trend, the field of human-computer interaction has shifted focus from individuals to teams and large organizations [35]. From a scientific perspective, we learn most about the object of study during transitions. Thus, a learning test is generally more diagnostic of brain function than a test of stored knowledge; a glucose tolerance test tells us more than a resting blood sugar level; a stress test reveals more about the heart than does resting heart rate. Similarly, this century's rapid transitions should allow us to learn a great deal about collective human behavior. At the same time, we face enormous planetary problems including global fouling of the ecosphere, inequity in economic opportunity, increased chances for catastrophic disease, and international terrorism. These problems arose with current approaches and limitations to collaboration and will only be solved via breakthroughs in collaboration.
From a more mundane viewpoint, similar challenges exist today for large, international organizations. For instance, the world is changing more quickly but creative design ability has not increased. As a result, there is a widening gap between the degree of flexibility and creativity needed to adapt and the capacity of individuals and organizations to do so [12]. Design problems are often extremely high leverage for organizations. For instance, errors in design, whether in software, drugs, business processes, or automobiles are extremely costly. Conversely, effective and innovative designs can be extremely lucrative; are a hallmarks of long-lived companies [7, 10]. Even a modest increase in the ability of organizations to create more effective designs could greatly increase profits in existing markets and create whole new markets. Increasing design effectiveness will require collaboration breakthroughs.
Human beings evolved natural language as a method of collaboration among small groups of people who generally shared context, goals, experience and culture. Under those circumstances, sequential human speech served fairly well, e.g., the telling of stories for sharing experiences [34]. However, unaided speech is not well-suited to large-scale collaborations; particularly not when the people involved have vastly different assumptions, cultural backgrounds, goals, contexts, experiences and native languages. We have not yet invented an entirely effective replacement of natural language for large, diverse groups though storytelling can be useful in bridging gaps among groups when incorporated into the appropriate process [3, 4, 37]. Can we further extend such techniques to facilitate communication among larger, more diverse groups? Or, should we limit such interactions to "dry" interactions [2]?
One of the special challenges offered by collaboration today is that often it involves remote participants; sometimes, worldwide[25]. In many conversations and papers, an implicit assumption is that remote collaboration is limited by bandwidth alone and that the current superiority of face to face over remote collaboration will disappear once bandwidth becomes large enough. Such an analysis overlooks two additional and potentially quite important aspects of face to face collaboration.
First, face to face collaboration allows people to see and experience the physical and social context of their collaborators. Perhaps they see the building where others work; try the same food; find out whether they work in a quiet or noisy environment; what the moods are of those that pass by in the hallways. Second, sharing an actual physical space allows the possibility of much deeper interaction and that possibility may well affect trust even if the possibility never materializes. Consider two rather extreme examples. First, two people sharing a physical space may be subject to a natural disaster such as an earthquake and one may save the life of the other. Although obviously a very low probability event, the mere possibility may well put people's perceptual and emotional apparatus into a heightened state of arousal. Second, if two people share a common physical space, one could physically injure the other. Since A's trust of B is enhanced by situations wherein A could hurt B but in fact, does not, the typical face to face interaction may enhance trust in just this way.
It is not only the medium and context of communication that impact collaboration, but also the content. In particular, we argue that expressive communication may offer an opportunity for collaborators to gain more comprehensive models of each other than instrumental communication alone. Instrumental communication is communication that is required to accomplish the current task. Expressive communication is communication that tells about the communicator as well as the subject; it is communicated more because the communicator wants to than because they need to.
Zheng, Bos, Olson, and Olson [38] showed that collaboration and trust can be, in effect, "jump-started" with social chitchat. Stories can also help people develop more trust than the exchange of information per se. A story is not simply an objective recounting of events; it always implies a number of revealing choices. The storyteller chooses which events to talk about; where to start; tone; viewpoint; which details to describe and so on. Through such choices, the storyteller inevitably reveals themselves as well as the subject. So long as collaboration proceeds along predictable lines, models built from expressive communication may be unnecessary. But, if standard procedures break down, then collaborators who have developed more complex models of each other will be able to react more effectively and efficiently as a team. Of course, there is also a danger here. As perhaps hinted at by Azechi [2], stories might also reveal characteristics of the storyteller that other collaborators might find quite negative while purely instrumental communications are unlikely to do so.
A challenge for knowledge socialization is to determine the conditions under which it is better to keep communications "dry" or "instrumental" and when it is desirable to include more expressive or "wet" modes of communication. If the latter is necessary, we also need to develop methods of progressive disclosure that minimize friction and maximize empathy.
2. New technological possibilities
Recent advances in computing power, interface technologies, bandwidth, storage, and social engineering provide a broad field of possibilities from which novel solutions to large-scale collaboration may be designed, tested, and improved. In the "real world" effective on-line collaboration systems both at a distance [16] and face-to-face [17], are already being facilitated by technology. We believe further advances can be made by incorporating creativity aids, suggestions for processes [33], and by providing tools for alternative representations [31].
Failure to innovate is not random, but can be ascribed to one of several main difficulties: 1. Individuals or groups do not engage in effective and efficient processes of innovative design. 2. The necessary skills, talents, and knowledge sources are not brought to bear on the problem. 3. Appropriate representations of the situation are not used. Laboratory [6, 15, 29] as well as field research [24, 36] has established that the major process difficulties are mainly due to a limited number of preventable errors.
An appropriate overall structure may facilitate groups through steps of innovation and help guide these separate steps; distinct guidelines are appropriate within each of these steps [28, 33]. A common problem is that people typically fail to spend sufficient time in the early stages of design; viz., problem finding and problem formulation [27]. A common failure during a specific stage of innovative design is that people often bring critical judgment into play too early in the idea generation phase of problem solving. As another example, unlike Newell and Simon's [22] normative model of ideal problem solving, in fact, people's behavior is path-dependent and they are often unwilling to take what appears to be a step that undoes a previous action even if that step is actually necessary for a solution [29].
Regarding the second issue (bringing to bear necessary skills, talents and knowledge sources), while software tools cannot fully substitute for human experts, evidence suggests that individuals have a large amount of relevant implicit knowledge which they often will not bring to bear on a problem and that giving appropriate strategies [29], or knowledge sources [30] can help.
Regarding the third issue of appropriate representation, controlled laboratory experiments have shown that subjects did significantly better, for example, in a temporal design task when they used a spatial representation; yet, very few subjects spontaneously adopted such a representation [6]. The impact of good representations, however, is not confined to laboratory demonstrations. Speech research advancements accelerated greatly when waveforms were largely replaced with speech spectrograms and Feynman diagrams allowed breakthroughs in atomic physics. By providing people with a variety of potential representations and some processes to encourage the exploration of various alternatives, we could probably improve performance significantly.
Advances in speech recognition, combined with natural language processing and data mining raise the possibility of large-scale real time collaborations. Speech recognition can turn raw speech into text. Statistical techniques can automate the formation of "affinity groups" that share various interests, values, or goals [23]. Speech recognition, in this context, need not produce perfect transcripts of what is said but only transcribe enough content to enable natural language processing software to cluster segments of text. Additional benefits stem from a speech to text to clustering system. In the past, conversations were transient. There was no "objective" evidence of their content or structure. It often happens, e.g., in a group meeting that the first person to raise a new idea is not recognized as having done so. Instead, the second or third person to mention the idea if often credited with it, quite possibly because the first mention is unassimilable by the current mental model of the listeners but causes a change in mental models so that a subsequent mention is comprehensible. The more general point is that computerized records of group meetings and larger scale collaborations allow the possibility of feeding back to the participants various visualizations of behavior, making the computer an active participant in group communication [32]. In conjunction with effectiveness metrics, such feedback mechanisms may allow groups to improve effectiveness. At IBM, we recently engaged in a corporate-wide experiment called "WorldJam" wherein all IBMers worldwide were invited to a three-day electronic meeting to discuss ten issues of interest to IBMers including employee retention, work-life balance, and working remotely. Over 52, 600 employees participated and posted over 6000 suggestions and comments. Each topic had a moderator and facilitators. Each moderator, in turn, had been asked to assemble a topic-knowledgeable "Board of Advisors" to provide references, websites, and other relevant materials ahead of time as well as participation during the on-line conference. In addition, the set of moderators and facilitators communicated with each other through a system called "Babble" which was designed, developed, and deployed at IBM Research. The Babble system blends synchronous and asynchronous text communication. Individuals in the system are represented as colored dots. The position of a dot within a simple visualization called a "social proxy" allows each participant to quickly see who else is present and which topics are being discussed. When a user of the system types an entry or scrolls through recorded discussion, their dot moves to the center of the social proxy for that topic. Several "Babbles" are now active within IBM including one for "Community Builders"; that is, people in various organizations throughout IBM interested in the process, tools, and methods for community building; "KM Blue" which includes a similar cross-organizational group interested in knowledge management and "Designers" which brings together people whose primary professional identification is as a designer. In the case or WorldJam, Babble enabled the moderators and facilitators to trade best practices and engage in joint problem solving in a timely manner. Additional information about the features, functions, design rationale for and empirical studies of Babble is available in [13, 14]. In earlier work, we showed that the introduction of aids to problem solving designed to break psychological set increased performance and creativity [30] and that instructions to take on multiple viewpoints increased the number of problems found in the heuristic evaluation of a software design [11]. The use of multiple viewpoints has been quite consciously used by the Iroquois (and other cultures) for thousands of years [36]. Other writers on creativity have suggested similar methods [9, 28].
3. Work of the knowledge socialization group
The work of our own group obviously relates to a tiny area of the vast space outlined above. Our work comprises several interlaced threads. In one thread, we are conceptualizing, designing, and building tools to support the creation, capture, organization, understanding, and utilization of stories as a method for groups to build and share knowledge. In the "Value Miner", e.g., natural language processing methods are used to find values as expressed in text. This could be applied to conversations, documents, and web-sites as well as stories. The Value Miner finds value-related words and phrases and tries to categorize these. A related, "Point Of View" tool shows the value similarities and differences of participants. We are also working on story visualizations aimed at helping individuals and groups create, understand, and find stories relevant to a situation at hand. For example, in one line of development, we are showing timelines of plot points and character development. In another line of representation research, we show a top level view of the kinds of attributes that are used to describe characters. By clicking on a top level view, the user may zoom onto the value associated with that attribute and ultimately to the underlying text. In addition to visualizations, there are guidelines and measures based on known heuristics of story writing that can be incorporated into groupware [18, 21].
In order to provide a common underpinning for the various story related tools that we have developed, we have proposed a first pass at a "StoryML"; that is, a markup language specifically geared toward stories. In this representation, there are three different but related "views" of story: Story Form (what is in the story); Story Function (what are the purposes of the story); and Story Trace (what is the history of the story). In turn, the Story Form can be broken down into dimensions of Environment, Character, Plot, and Narrative. The idea of the StoryML is that it is expandable according to purpose. For some purposes, the user (e.g., a student studying mystery plots) may be satisfied with minimal detail concerning Function and Trace but need to expand certain aspects of the Story Form in great detail. In another context, a different user (e.g., a historian comparing certain themes across time and cultures) might have a very high level view of Story Form and Story Function but want to provide a detailed description of Story Trace. At this point, the meta-data in StoryML must be supplied by a knowledgeable human being.
Once a base of potentially useful stories becomes large in any one collection or domain, it can become a challenge to find the "right" story or stories. If one is looking for stories with particular objects, people, or places in them, "keyword in context" searches are generally sufficient. But, if one is looking for stories about activities, a more subtle approach is required. In response to this challenge, we have developed a script-based story browser. The "script" is a default set of parameters about an activity; it may specify roles, goals, objects, and a sequence of events. In the story browser, a user may choose an activity and find stories related to that activity or related activities through a combination of searching and browsing. Although this activity-based search works at a higher level of semantics than typical searches, in many cases, a person is searching for a story that illustrates a particular kind of very abstract point and even the particular activity is not that important. For instance, the story of Odysseus hiding his warriors in The Trojan Horse may be applicable in a wide variety of domains such as disease control or computer security. In such cases, to find stories that are potentially applicable, we really need a system based on abstract planning and problem solving strategies. In our lab, Andrew Gordon [20] has developed such an ontology for abstract planning and problem solving by interviewing experts and reading strategy books in a wide variety of domains and then formulating these strategies in abstract terms. In the next step, these terms can be used to categorize stories according to the strategies that are utilized. This will enable individual problem solvers, educators, and teams to find stories that are potentially applicable to improving specific situations or solving particular problems.
We are also engaged in attempting to extend the architect Christopher Alexander's [1] concept of a Pattern Language to stories. A Pattern Language consists of a lattice of interrelated patterns. Each pattern has a Title, a description of a context in which a problem is likely to occur, a description of opposing forces, and the basic outline of a solution. A pattern also often contains a diagram illustrating the basic solution, and may contain references or other evidence about its efficacy. Each pattern also includes links to higher level and lower level patterns. The notions of patterns and A Pattern Language have been applied to a variety of fields besides architecture including object-oriented programming [19], project structure [8] and human-computer interaction [5]. Typically, a Pattern Language is developed by a community of practice as a way to create, organize and reuse knowledge.
Our attempts to provide additional knowledge sources are focused mainly on teaching stories [34], particularly during specific stages of problem solving. For example, the story "Who Speaks for Wolf" by Paula Underwood [36] is a story especially well-suited to either problem formulation or to a last minute check that all stakeholders' concerns are covered before significant resources are committed to a particular plan. In other cases, the individual, team, or organization will need to use a story browser whose expanding capabilities are outlined above.
In this paper, we have attempted to do three things. 1. Convince the reader that improving and understanding the ability of individuals, teams, and organizations to innovate more effectively is key to our collective survival. 2. Outline how recent advances in science and technology offer a promise to enhance collaborative innovation. 3. Describe in outline the small contributions along these lines of the IBM Research Knowledge Socialization Group.
References
1. Alexander, C., Ishikawa, S., Silverstein, M. Jacobson, M., Fiksdahl-King, I. and Angel, S. A pattern language. New York: Oxford University Press, 1977.
2. Azechi, S. Social psychological approach to knowledge-creating communities. In T. Nishida (Ed.), Dynamic knowledge interaction. Boca Raton: CRC Press, 2000.
3. Beyer, H. and Holtzblatt, K. Contextual design: defining customer-centered systems. San Francisco: Morgan Kaufman, 1998.
4. Bodker, SA. Scenarios in user-centered design: setting the stage for reflection and action. Presented at the 32nd annual Hawaii International Conference on System Science, January, 1999, Maui, Hawaii
5. Borchers, J. A patterns approach to interaction design. New York: Wiley, 2001.
6. Carroll, J., Thomas, J.C. and Malhotra, A. Presentation and representation in design problem solving. British Journal of Psychology, 71 (1),. 143-155, 1980.
7. Collins, J. and Porras, J. Built to last. New York: Harper, 1994.
8. Coplien, James. http://www1.bell-labs.com/user/cope/Patterns/Process/index.html, 2001.
9. De Bono, E. Six thinking hats. Boston: Little, Brown, 1985.
10. DeGeus, A. The living company: habits for survival in a turbulent business environment. Boston: Harvard Business School Press, 1997.
11. Desurvire, H. and Thomas, J. Enhancing the performance of interface evaluators using non-empirical usability methods. Proceedings of the 37th Annual Human Factors Society Meeting, 1132-1136, Santa Monica, CA: Human Factors Society, 1993.
12. Drucker, P. Managing in a time of great change. Truman Talley Books: New York, 1995.
13. Erickson, T., Smith, D. Kellogg, W., Laff, M., Richards, J. and Bradner, E. Socially translucent systems: Social proxies, persistent conversation and the design of "Babble." In Human Factors and Computing Systems: The proceedings of CHI'99. New York: ACM Press, 1995.
14. Erickson, T. & Kellogg, W. "Social Translucence: An Approach to Designing Systems that Mesh with Social Processes." Transactions on Computer-Human Interaction, 7(1), 59-83, 2000.
15. Farnham, S. et. als. Structured online interactions: Improving the decision-making of small discussion groups. Proceedings of CSCW 2000. 299-308. New York: ACM, 2000.
16. Finholt, T.A., and Olson, G.M., "From Laboratories to Collaboratories: A New Organizational Form for Scientific Collaboration," University of Michigan, Ann Arbor, January 1997.
17. Fischer, G. Domain-Oriented Design Environments: Supporting Individual and Social Creativity", in J. Gero and M.L. Maher (eds): "Computational Models of Creative Design IV", Key Centre of Design Computing and Cognition, Syndney, Australia, 83- 111, 1999.
18. Frey, J. How to write a damned good novel II. New York: St. Martin's Press, 1994.
19. Gamma, E., Helm, R., Johnson, R., and Vlissides, J. Design Patterns: Elements of Reusable Object Oriented Software. Reading, MA: Addison-Wesley, 1995.
20. Gordon, A. The representational requirements of strategic planning. Fifth symposium on logical formalizations of commonsense reasoning. New York University, May 20-22, 2001.
21. McKee, R. Story: Substance, structure, style and the principles of screenwriting. New York: Harper, 1997.
22. Newell, A. and Simon, H. Human problem solving. Upper Saddle River, NJ: Prentice-Hall, 1972.
23. Nishida, T. Dynamic knowledge interaction. Boca Raton: CRC Press, 2000.
24. Olson, M. and Bly, S. The Portland experience: A report on a distributed research group. International Journal of Man-Machine Studies, 34, 211-228, 1991.
25. Olson, G. and Olson, J. Distance matters. Human-Computer Interaction. 15, 2-3, 107-137, 2000.
26. Poltrock, S. and Englebeck, G. Requirements for a virtual collocation environment, Information and Software Technology, 41(6), 331-339, 1999.
27. Sobel, D. Longitude: The true story of a lone genius who solved the greatest scientific problem of his time. New York: Penguin, 1995.
28. Stein, M. Stimulating creativity. New York: Academic Press, 1974.
29. Thomas, J. An analysis of behavior in the hobbits-orcs problem. Cognitive Psychology, 6, 257-269, 1974.
30. Thomas, J. , Lyon, D., and Miller, L. Aids for problem solving. IBM T. J. Watson Research Report, RC-6468. New York: IBM, 1977. 31. Thomas, J. and Carroll, J. The psychological study of design. Design Studies, 1(1), 5-11, 1979.
32. Thomas, J. C. The computer as an active communications medium. Proceedings of the 18th Annual Meeting of the Association for Computational Linguistics, 83-86, New York: ACL, 1980.
33. Thomas, J. Problem solving by human-machine interaction. In Gilhooly K.J., (Ed). Human and machine problem solving. London: Plenum Publishing, 1989.
34. Thomas, J. Narrative technology and the new millennium. Knowledge Management Journal, 2(9), 14-17, 1999.
35. Thomas, J. An HCI agenda for the next millennium: Emergent global intelligence. In R. Earnshaw, R. Guedy, A. van Dam and J. Vince (Eds.), Frontiers of Human-Centered Computing, Online Communities and Virtual Environments, 2001.
36. Underwood, P. Three Native American learning stories. Georgetown, Texas: Tribe of Two Press, 1994.
37. Van Der Heijden, K. Scenarios: The art of strategic conversation. New York: Wiley, 1996.
38. Zheng, J., Bos, N. Olson, J. and Olson, G. Trust Without Touch: Jump-Start Trust With Social Chat. Proceedings of CHI 01 (Conference Companion), New York: ACM, 2001.