These pages contain information from Los Alamos reports: LA-UR 97-1200, 98-489, 98-2227, 98-1150, 98-2549


Symbiotic Intelligence and the Internet: A Deeper Overview

(to the main page)
 
Contains information from LA-UR-2549, S. Rasmussen & N.L. Johnson
Presented at the 6th Santa Fe Chaos in Manufactuing Conference April 1, 1998

 

Self-Organization in Societies

Why are ants, wolves and humans organized into societies? The simplest answer is that these societies enhance the survival of individuals and, therefore, the populations as a whole. At the very foundation of the human societies is their ability to solve vital, complex problems for the individuals that make up the societies. We as individuals contribute to, as well as adopt to, these hyperstructures because they make life easier in one way or another, yet we, individually or collectively, often do not understand the process involved. Despite the lack of understanding, our society has changed dramatically, from small hunting tribes to a highly technological, global society.

One of the apparent paradoxes associated with human society is its diversity in both goals and means. Subgroups often tend to disagree significantly. Different individuals, different political parties, different companies, employers and employee, etc., create a complex network of contradicting means and goals. We shall argue that this divergence is not a liability or even an inconvenience, but is of vital importance and inherent survival value.

In addition, many problems of the survival of human society, or even confronting large companies, are of a nature where the complexity and the magnitude of the issues involved are of too large of an extent be defined and understood by a single coherent organization - and much less by the individuals. An example is the dynamics of the world economy. Nobody fully understands what is going on, even though it impacts us all, and the need for understanding is great. A loosely coupled network of cooperating and competing organizations has for thousands of years sorted out these problems and eventually generated solutions - our survival is proof of the success of the process. Individuals of course play a vital role in this process, but the emergent dynamics of the society is, more times than not, out of the control of any individual. Our society is a self-organizing system which involves many individuals and organizations at the "controls."

Traditional Problem Solving

Most of our day-to-day technical developments and problem solving are achieved in a linear, premeditative approach:

  1. Understand the system in question
  2. Represent the system behavior in an approximate model or simulation
  3. Prediction of the system behavior based on the model or simulation
  4. Apply the model or simulation to optimize and control some aspects of the system to meet defined goals
  5. Repeat the process for corrections or failures

As long as we, as a society of technocrats, can "understand" the system, the above approach works well and we can optimize it for our needs. This approach is desirable because it has the potential to achieve optimal and rapid solutions.

An Alternative Paradigm for Problem Solving

Distributed systems (biological evolution, immune systems, human societies, etc.) take a very different approach to problem solving. When the system is too complex for analysis (as in global economics), an "understanding" of the whole is not possible. Without this understanding or when a centralized control is absent, the linear approach to optimization of a system is interrupted at the onset. Instead, a solution by distributed systems relies on a broad diversity of potential solutions at any time, combined with an emergent system dynamics. Then, given a change of the overall dynamics of the system - one of the potentials is selected. If a solution is not found within the current diversity, the system fails and "dies." If a solution is found, then the dynamics of the system will "optimize" the solution, but not by sacrificing the diversity of the system, only by changing the relative dominance of subsystems. Thus, a solution is found without premeditation or centralized control.

Social structures that take advantage of our inherent, self-organizing social dynamics will be best enabled to cope with our increasingly complex world {Abraham, 1994}. Indeed, we argue that this has happened in modern, overly centralized governments, such as the USSR, and is the reason that democracy and capitalism provide the most robust solutions in modern times {Slater, 1964}. There are also trends towards decentralized corporate management {Anderson, 1988} {Youngblood,1996}.

Herein lies the thrust of our idea. What we are suggesting will never replace the linear problem solving for less complex problems, in fact, the idea requires traditional problem solving at a local level. But for systems that defy timely understanding, due to the size or variety of information, only a distributed approach to problem solving can discover the solution space.

We believe that in the future, emerging solutions on distributed networks will hold a respected place in problem solving and in science, once we accept the limits in our understanding and once the capability of emerging solutions is demonstrated. This way of solving problems has been around since the origin of life and now our technological capabilities allow us to further understand and improve this capability in a clear and precise manner by the use of the Net.

Past Views of Social Evolution

The view of human society as a collective organism is not new. George Dyson, in his recent book, "Darwin Among the Machines" {Dyson, 1997} surveys the works of great thinkers over the last 500 years who have touched on visions which involve evolution in the collective intelligence or awareness of humans and machines as symbol processors. Despite the long history of interest in these ideas, it has only been in the last decades that there is now promise of a quantitative theory of social dynamics. This new foundation was driven by the dramatic success of the application of complex systems methods to biological problems as expressed, for example, in the Artificial Life movement {Artificial Life}. In the last two decades there has been a virtual explosion of interpretations or dynamical theories of social and economic systems (e.g., citations in {Abraham, 1994}).

Unique Capabilities of the Net (Internet plus humans)

The Net has three significant, arguably unique, capabilities beyond prior human-technological systems: (1) the ability to integrate quickly heterogeneous systems (the breadth of systems), (2) its ability to capture detailed signatures of the access and use of information (the depth of systems) and (3) its ability, with relatively minimal loss, to relate and transmit information (the accuracy of communication) The Net has the ability within one hyper-system to integrate:
  1. Information storage: both in the form of simple data and complex text and images. This was earlier done in off-line libraries and a variety of data banks.
  2. Communication: Communication was earlier done either by the relatively slow movement of people or documents or, in recent times, by telephone or other electronic technologies. However, complex documents, simple data and images can now be transported instantaneously and close to cost-free from anywhere to everywhere. Geographical barriers are virtually gone.
  3. Traditional computing: the automated (simple) information processing of huge amounts of data.
  4. Human processing: The human ability to analyze, understand and process limited, but highly complex information.
    Until very recently (1), (2) and (3) were physically separated processes, all combined by human intervention (4). Now (1), (2) and (3) are integrated in a more standardized medium so (4) is no longer the slowest link and now can be active at a more efficient and useful level. Thus, the time scale for knowledge organization and creation is drastically shorter. This new integration has been overwhelming to humans, but tools are readily being developed in this infant hyper-structure to overcome this initial shortcoming [see firefly.net, amazon.com, alexa.com].
  5. The Net can capture the complexity (the depth) of how information is associated by retaining all references between data on the network. A simple example of how much of this relational information is currently lost is the use of scientific publications. While papers contain citations that connect the current paper with the information in other papers, the more immediate and detailed information about the numbers and types of readers of the papers could be only obtained in the past at great expense. Now with the advent of online publications, such information is explicitly available at effectively no cost. In general, the Net can capture all traces of the use of information. As more and more of the human activities occur on the Net, these traces will capture the full complexity of our interactions and activities that were formerly lost. While currently not used to any extent (except in the examples noted above), these traces represent implicit knowledge of how we interact and how new knowledge is created. In the past it has been difficult to do experiments involving the access and formation of knowledge, but by conducting experiments on the Net, as demonstrated in a later section, the situation has been reversed and now is data rich, instead of data poor.
  6. Traditional human-to-human communication results in a rapid loss of information a bit removed from its creator (the children's game of passing a phrase around a circle is a telling example of the high noise-to-signal ratio of verbal communication). By contrast, information exchanged or related on the Net, whether in web pages or emails, suffers minimal loss of information during transmission or linking, in the same way that the content of a book is not altered when exchanged or a citation to a journal is not degraded over time. The advantage on the Net of reduced loss or noise is at a sacrifice of bandwidth or dynamic range with current technologies: the elimination of vocal, facial and gestural expressions which do not accompany one-to-one communication. In this discussion, we do not include the misinterpretation that still occurs in understanding the exchanged information; this source of miscommunication occurs independent of the mechanism of exchange.

The degradation of verbal communication is one possible explanation for the maximum critical number of participants for effective problem solving in all human systems, from social functions to organizations. Why does a committee cease to function above a certain size, even though greater resources are available?

With the stronger presence of these three capabilities in human use and interactions on the Net, we propose that an informational "phase change" will occur: for the first time we can capture our creation, manipulation and rejection of knowledge, encompassing a much broader range of the complexity of our existence - information which was earlier lost in the traditional person-to-person interactional noise or never available. With this phase change of information processing capability comes the possibility of an enhanced dynamic that, while an essential part of human organizations, has been limited in extent and responsiveness.

Self-Organizing Knowledge

Large problems are sometimes "solved" without our explicit intervention and may only be seen as problems in retrospect. A "solution" emerges and takes care of the problem without any single person identifying the problem or orchestrating the solutions.

A simple demonstration of a self-organizing solution is the formation of a path in a forest. The undefined goal is a path that is usable and of minimum length between two destinations and is adaptable to variations in local conditions. The self-organizing path evolves over time and includes contributions from a variety of people and animals with a broad diversity of goals (mode of travel, destinations, security, etc.) and changing forest conditions (weather, tree falls, dangerous locations, etc.). A comparable "designed" path would require a thoughtful process of premeditated selection of the goals to be optimized, an analysis of the forest layout and dynamics, a plan to meet the goals, and the implementation of the solution.

Self-organizing solutions have existed long before humans became capable of alternative approaches to problem solving. And as we continued to develop our individual, organizational and governmental abilities as problem solvers, the self-organizing societal dynamics continued to function in the background of our social dynamics. For example, the shift from an agricultural-based society to an industrial-based society occurred without any group of individuals being responsible. The society self-organized into another form. Only in retrospect, it is clear that a number of key factors had to be in place to make such a shift possible. Only in retrospect, do we understand and model the process with the traditional process of knowledge creation. A simple example illustrates the impact on emergent problem solving of the unique capabilities of the Net described in the last section.

Because informational systems have not been studied as self-organizing systems, some explanation is necessary of the nomenclature borrowed from other fields. We distinguish "self-organized knowledge" as being different in origin, not necessarily in kind, from "traditional knowledge". We define traditional knowledge as being derived from the premeditative actions of people or predetermined processes of automated information processing to associate facts and draw conclusions. It represents a condensation of information and is localized in time and space - in books, speeches, memories. Self-organized knowledge, in contrast, originates from the self-organizing dynamics of human hyperstructures, without the action of a central coordinating processor. Once self-organized knowledge is formed, it can either be observed and transformed into traditional knowledge, e.g., through writing.

Self-Organization on Networks: An example

A simple experiment was done by Bollen and Heylighen {Bollen, 1996a} in Belgium that illustrates the concept of self-organizing knowledge through human interaction. The idea was to use the Net to make hypertext webs more intelligent {Drexler, 1991} {Bollen, 1996b} by reinforcing paths of people using the various sites, similar to the perceived reinforcement of pathways in the brain by frequency of use {Gaines, 1994} {Heyligen, 1996}.

The researchers created 150 web sites, each identified by a word from the 150 most common words in English. Under the word, a list of 10 randomly chosen words from the list of 150 words were displayed. Upon entering one of the 150 sites, the user was asked to pick the word from the list that most closely is associated with the header word. Upon choosing a word, the order of the list is recalculated based on the frequency of selection and then the user is taken to a new site that corresponds to the selected word, and the process is repeated. The researchers found that the lists stabilized to a fixed order after about 4000 selections in a site. The resulting ordered lists determined a common semantics despite the heterogeneity of users.

This simple task of ordering is easy for an individual but of little utility due to large individual variation in semantic differences. Alternatively, the task is difficult to complete for a committee of experts, while the result would be more useful. The network solution achieved a result representing useful collective knowledge, but with minimal instruction and effort for the collective group of individuals. This example captures the essence of developing an self-organizing knowledge system that combines the advantages of both human and computer networks to quickly solve a semantically complex problem. From this example, one can imagine a host of previously challenging, if not intractable, problems that could be addressed once the methodology is developed.

Conclusions

The above text presents preliminary arguments on the possible future of "problem solving" or collective decision making in our society and organizations. We have argued that a dynamic process underlies all life: the ability of self-organizing systems to "solve" essential problems, will take on new functionality as our society increasingly utilizes the Net for human interaction.

In the development of the ideas presented in this paper, we have realized that the experience gained in studying complex behavior of biological or physical systems is not generally applicable to knowledge systems. Self-organizing knowledge formation represents a system that appears to be unique: it is not dissipative, nor constrained by limited resources, in the sense as are previously studied examples of self-organization in constrained systems. No equivalent laws of conservation or dissipation exist for knowledge, as do for material resources: the act of giving knowledge does not erase knowledge in the giver and new knowledge is not created at the expense of prior knowledge.

These unique characteristics of self-organizing knowledge systems are further enabled by the unique capabilities of the Net: in addition to providing improved temporal and spatial system dynamics, the Net has relatively loss-less transmission and linking of information, unlike our predominant verbal communication. The symbiotic intelligence of the combined human-hardware-software system, which constitute the Net is believed to be able to operate at a higher level of functionality, both in numbers of individuals and the complexity of capability.

The implications of the arguments presented in this paper are far reaching to future dynamics of organizations, society and economics. With the increased use of the Net in our economy and society, many of the problems that have challenged traditional forms of management may find solutions. The main consequence will be for management and governments to facilitate the use of collective problem solving when problems arise in systems that are not understood. Organizations may find that to function at a higher level of performance, policies will be pursued that encourage diversity, increased access to information and decentralized control.

Bibliography

Abraham, R. 1994. {\it Chaos, Gia, Eros}. San Francisco: Harper.

Anderson, P.H., Arrow, K.A. and Pines, D. (Eds.) 1988. The Economy as an Evolving Complex System. Santa Fe Institute Studies in the Sciences of Complexity V, Addison-Wesley.

Artificial Life Proceedings, Volumes I-III, Santa Fe Institute Studies in the Sciences of Complexity, Addison-Wesley. Volumes IV-VI MIT Press.

Bollen J, and F. Heylighen 1996a. "Algorithms for the self-organization of distributed, multi-user networks. Possible application to the future World Wide Web." In: Cybernetics and Systems '96. R. Trappl (Ed.). Austrian Society For Cybernetics Press, pp. 911-916. (Also see, "The Adaptive Hypertext Experiment", http://pespmc1.vub.ac.be/ADHYPEXP.html)

Bollen J, and F. Heylighen 1996b. "Learning, Brain-like Webs". WWW document, http://pespmc1.vub.ac.be. (Also see, F Heylighen 1997, "Correspondence between Organism and Society", http://pespmc1.vub.ac.be/COMPTABL.html)

Drexler, K.E. 1991. Hypertext Publishing and the Evolution of Knowledge. Social Intelligence; 2-22.

Dyson, G. 1997. "Darwin among the Machines: The Evolution of Global Intelligence." Addison-Wesley Publishing, 1997.

Gaines, B.R. 1994. "The collective stance in modeling expertise in individuals and organizations." International Journal of Expert Systems. Vol. 7, no. 1, pp. 22-51.

Heylighen, F. and J. Bollen 1996. "The World-Wide Web as a super-brain: from metaphor to model." In: Cybernetics and Systems '96. R. Trappl (Ed.). Austrian Society For Cybernetics Press, pp. 917-922.

Johnson, N., Rasmussen, S., Joslyn, C., Rocha, L., Smith, S. and Kantor, M., "Symbiotic Intelligence: Self-organizing knowledge on distributed networks driven by human interactions." In press, Artificial Life 6, (Eds.) C. Adami, R.K. Belew, H. Kitano and C.E. Taylor, MIT Press, 1998.

Koza, J. 1994. Genetic Programming II. Cambridge, MA: MIT Press.

Mitchell, M. 1996. An Introduction to Genetic Algorithms. Cambridge, MA: MIT Press.

Mosekilde, E. and Rasmussen, S. 1986. Technical economic succession and the economic long wave. E.J.O.R. 25, pp. 27-38.

Schement, J.R. and Lievrouw, L.A. (Eds.) 1987. Competing visions, complex realities: Social aspects of the information society. Norwood, NJ: Ablex.

Slater, P. and Bennis, W. 1964. Democracy is Inevitable. Harvard Business Review, March-April (reissued September-October 1990).

Youngblood, M. 1996. Life at the Edge of Chaos. New York, NY: Wiley and Sons. 

Learn More about the Project

Detailed Summary | Related Documents | Mailing List
FAQ |
Glossary | Revision History | Webmaster

Learn More about the Project

Detailed Summary | Related Documents | Mailing List
FAQ |
Glossary | Revision History | Webmaster

Additional Links for Symbiotic Intelligence Project
Good summary: http://www.computerworld.com/managementtopics/management/story/0,10801,64912,00.html
http://www.fathom.com/media/trailDemo.html
http://itri.loyola.edu/ConvergingTechnologies/
http://www.edu-cyberpg.com/Technology/trends.html
Relevant workshop: http://www.santafe.edu/~dynlearn/colcog/
For the symintel site:http://www.man.ac.uk/sociologyonline/course/soccyber/self.htm
Long list of referrals: http://www.mikro.org/Events/OS/wos2/Levy-pp/liensIC.html

Google



Search WWW   Search this site