Self-organizing Systems in the Information Age


Professor William Murphy
Franklin Pierce Law Center
Concord, NH 03301


Ever since Darwin the notion of self-organizing systems in biology has gained nearly universal acceptance(1). In biological terms a self-organizing system, like a flock of Canadian geese flying in perfect unison or school of fish darting and weaving among the coral as one, unites the disparate, independent actions of individuals into a complex and coherent whole without centralized command and control. The individuals, following simple “rules”, develop structures and behaviors that were not those necessarily intended by the individuals.

In human society we often insist that “someone” or “something” must be controlling and directing the organization and operation of orderly, complex, and interacting systems(2). But recently an awareness of how decentralized systems can also form orderly, complex aggregations and interactions without centralized command and control has emerged; first in biology(3) , now in economics(4) and even in the production of digital domains(5).

The reason that free market capitalism has thrived is that it self-organizes to produce, at least in comparison to alternatives, beneficial results(6). In the industrial age economy, the decentralized ideal of the capitalist model works, for the most part, because when individuals, pursuing selfish interests (like the goose seeking an easier flying burden), interact freely in the marketplace the resulting complex and ordered system produces benefits that exceed its costs. On ethical rather than economic terms, this self-organizing free market satisfies three moral criteria:

  1. justice (distributional justice under the capitalist criterion whereby each receives according to contribution and according to voluntary bargained-for exchange),
  2. rights (right of free consent, voluntary exchange, and freedom of choice), and
  3. utility (resources are allocated, used and distributed efficiently).

There is little doubt that a truly free market economy is a self-organizing system. By following the simple rules of individual self-interest the collective interactions in the free market produce a dynamic and complex economy that answers the four basic economic decisions(7) in a manner superior to command and control systems. Many have argued(8) that the internet should be left alone, and that a rough and tumble free market will produce that best outcomes in terms of efficiency and consumer benefits. The operation of the free market in the industrial age has had this effect so why not apply the hard-won lessons of the industrial age to the information age?

Potential for Self-Organizing Monopolies
The question that the paper will discuss is whether the economic environment that is emerging in the information age is sufficiently different from that which prevails in most of the industrial age sectors, that the same self-interested behavior by individuals and organizations, will lead to undesired consequences or structures where the long-term benefits might be outweighed by the costs and burdens(9). Specifically, is the environment of the internet such that the activities of individuals and organizations will self-organize into monopolistic structures? The paper will examines six separate but often interrelated factors that are found in many business sectors that are emerging in the information age economy. When one or more of the first five factors examined are present in sufficient strength it is suggested that there is the danger of self-organizing monopolistic structures, while the sixth factor examined may have a mitigating effect(10).

The Six Factors to be Examined

1. Scale economy effects
In the high-tech economy of the information age the ratios between fixed costs and variable costs have changed for many products. Now the percentage of fixed costs to variable costs can greatly exceed those common during the industrial age. Because of the high fixed costs associated with the development, production and distribution of many information age items relative to their variable costs, economies of scale are even more important today than in the industrial age and can lead to industry structures where a single firm can dominate a market by exploiting a scale economy derived cost advantage(11).

2. Learning/experience effects
Learning curve and experience curve effects express the mathematical relationship between the accumulated output of a product at the unit costs of the product. The effects are thought to be underlying natural characteristics of organized activity and have long been observed in industrial age industries(12). The relationship is generally expressed in terms of a percentage decline in cost for each doubling of output. For most standardized manufacturing industries cost declines of 10 to 30% are common(13). It has been suggested that learning curve effects are even more pronounced in knowledge and technology intensive industries(14). In addition, many information age products have extremely low (or potentially no) marginal costs so accumulation of output can be accomplished at little or no expense(15). To the extent that learning curve or experience curve effects play a significant role in information age industries there is a marked tendency to favor the first mover, the company that is able to accumulate output quickly(16).

3. Superstar effects
Superstar effects(17) are related to the scale economy effects but differ in important aspects. In times past there were limits on the capacity of individuals or firms to satisfy the demand for their products or services. As a consequence, after the capacity of the most desired supplier had been reached customers would, of necessity, turn to their second best choice. This excess of demand over supply would also have the effect of permitting the suppliers of the preferred product or service to charge a higher price. This again would have the effect of customers shifting to their lower-tiered choices. Unlike scale economies and learning curve effects, superstar effects are based on specific consumer preference (other than favorable prices which can result from the scale and learning effects) for the perceived “best” product. As a consequence, the fact that a winner-take-all marketplace may arise due to superstar effects may not necessarily be bad for the consumer since in such a winner-take-all marketplace each consumer can have the most desired product or service and need not settle for second or third choices(18).

4. Network Effects and Standardization
Network effects or network externalities(19) exist when the amount that one is willing to pay for access to the network is dependent upon how many other parties are also connected to the network(20). Network products or services become more valuable as their use becomes more widespread(21). In a related manner standards are often required to allow the efficient functioning of a network. As a consequence, achieving a critical mass of acceptance is significant in a number of information age markets. The first firm to achieve a critical mass of acceptance sets the de facto standard, and success then breeds more success. When a critical mass is reached on a standard, the beneficiary of the standard receives increasing benefits as more and more individuals switch to the dominant standard, which as a result increases the standard-setter’s dominance(22). The importance of first mover advantages may go beyond critical mass of acceptance. First movers will also be farther down the learning curve, and may enjoy greater scale economies(23).

5. Information Search and Retrieval Burdens
The fifth factor contributing to self-organizing monopolistic structures in the information age focuses on the burdens of information search and retrieval. This stems from two causes. First is the complexity of the products and services. The informational asymmetry between buyers and sellers is magnified in technology driven markets. Individual consumers are often ill-equipped to comprehend the salient technological features of competing products and system. As result, consumers place increased reliance on reputation and prior experience to make judgments. A type of herd behavior can evolve, further entrenching market leaders. It is safer to stay with the grazing herd, even if the herd is grazing in a sub-optimal area. The second cause of the information search and retrieval burden is a direct by-product of the information age. The increasing amount of information available can overwhelm even the most conscientious consumer(24).

6. Non-rivalry and Reproduction Costs
Like Pandora’s box the story ends with one new factor of the information age economy that may provide a ray of hope against self-organizing monopolistic structures, although this characteristic poses ethical and economic dilemma of its own. One of the fundamental characteristics of tangible property in the industrial age economy is that possession is rivalrous. When the possibility of non-rivalrous possession is coupled with the costs of reproduction a serious problem arises. In the industrial age economy social welfare was maximized because consumers were willing to pay a price for tangible goods that was greater than the marginal cost of producing another copy. It was unlikely that you were going to cheaply “clone a copy of your Ford” for a friend. The marginal cost of reproduction (often near zero) for many digital goods is below the marginal cost of producing and distributing the product. Will anyone have an incentive to produce digital goods where the costs of production and distribution exceed the marginal cost of reproduction? If the answer is no, all of society will be worse off. The producers of digital goods have begun to attack the situation by trying to increase the marginal costs of reproduction through the threat of legal prosecution for infringement or through technological changes (such as digi-marking)(25).

Ethical and Ecoonomical Considerations
The self-organized monopolies that may arise in the information age economy appear to pose an ethical and regulatory dilemma that was not present in the industrial age economy(26). As noted in the beginning the industrial age economy satisfies three moral criteria(27). The potential winner-take-all nature of the many information markets can cause a disassociation between these three moral criteria(28). Distributional justice may be threatened in that rewards may exceed one’s contribution. In addition, individual rights of choice may be limited by the absence of competitive alternatives. On the other hand, one could argue that utility is actually enhanced as costs are driven to their lowest levels. Of course, low costs do not guarantee low prices, so the benefits of the effects listed above may be captured and internalized by the producers rather than shared with the consumers.

The possibility for self-organizing monopolies to evolve in these winner-take-all markets also makes them extremely attractive because of their potential remuneration for the winners. As a result participants can become engaged in an expensive “arms race” to achieve dominance. This type of competition can result in escalating expenditures that may not produce any additional security of success since security of success is relative to the threat(29).

  1. In the early 1970’s Professor Keller at MIT demonstrated that slime mold cells, a simple life form without specialized cells, formed clusters or aggregations simply in response to a chemical secreted and sought by each slime mold cell. This “aggregation without a leader” is but one example of self-organizing behavior whereby complex structures emerge as the result of simple interaction “rules” followed by free acting individual organisms. [These experiments were conducted by Evelyn Fox Keller, a professor of mathematics and humanities whose work has been in mathematical biology and in the history, philosophy, and psychology of science]
  2. Although the self-organizing nature of biological systems is widely accepted (as noted earlier) there are some who hold to a command and control explanation of biology. This disagreement is at the core of the creationism debate. Our tendency to centralized control explanations may also help explain the persistence and popularity of conspiracy theories. See also, Michale Resnick, Changing the Centralized Mind, Technology Review, July 1994.
  3. Benno Hess and Alexander Mikhailov, Self-organization in living cells, Science (4/8/94) at 223. Howard T. Odum, Self-organization, transformity, and information, Science (11/25/88) at 1132.
  4. Larry D. Browning Janice M. Beyer Judy C. Shetler, Building cooperation in a competitive industry: SEMATECH and the semiconductor industry, 38 Academy of Management Journal 113 (1995); Peter Nijkamp and Aura Reggiani, Non-linear Evolution of Dynamic Spatial Systems: The Relevance of Chaos and Ecologically-Based Models, 25 Regional Science and Urban Economics (April 1995); Raghu Garud and Suresh Kotha, Using the Brain as a Metaphor to Model Flexible Production Systems, 19 Academy of Management Review 671 (1994); Friedrich Hinterberger, Self-Organizing Systems, appearing in The Elgar Companion to Austrian Economics, Peter J. Boettke, ed. (Ashgate, Brookfield, Vt, 1994); and, Michael J. Radzicki, Institutional Dynamics, Deterministic Chaos, and Self-Organizing Systems, 24 Journal of Economic Issues (March 1990)
  5. Michael Rothschild, Call It Digital Darwinism, Upside, December 1991. 20,000 bytes under the sea, Economist (6/13/98) at 81. See also, M. C. Yovits and S. Cameron, Eds., Self Organizing Systems (Pergamon, New York, 1960); H. von Forster and G. W. Zopf, Eds., Principles of Self-Organization (Pergamon, New York, 1961).
  6. For a discussion of this point see A. Michael Froomkin and J. Bradford De Long The Next Economy? at
  7. The four fundamental economic decisions that must be made by a society are: 1. What products to produce. 2. How much of each to produce. 3. What resources shall be used in producing these products. 4. Who will get what. Or to express the four in a single sentence -Who will produce what and how for whom?
  8. John Perry Barlow, Selling Wine Without Bottles: The Economy of Mind on the Global Net,, and John Perry Barlow, Stopping the Information Railroad, Keynote Address, Winter 1994 USENIX Conference, San Francisco, California, January 17, 1994 at The freedom from control and regulation is often part of the desire for preservation of the freedom of speech on the internet. See for example, the web site of Internet Freedom at
  9. The fact that something is different about information age competition, and the internet in particular, has not escaped notice. According to one commentator, “Something very unusual is going on here. É There is something about the information industry in general and the Internet in particular that makes the application of normal antitrust rules problematic, whether you approach them from the point of view of classical antitrust scholarship or of the Chicago School.” Mark A. Lemley, Antitrust and the Internet Standardization Problem, 28 Conn. L. Rev. 1041 (1996)
  10. Some authors have characterized the “new” economics of the information age as the proliferation of winner-take-all markets. Robert H. Frank and Philip J. Cook, The Winner-Take-All Society, (Free Press, 1995)
  11. For a discussion of this point and its application see Joseph Kattan, Market Power in the Presence of an Installed Base, 62 Antitrust L.J. 1 (Summer 1993)
  12. Wilfred B. Hirschmann, Profit from the Learning Curve, Harvard Business Review, January-February 1964. The first documented observation of the learning curve effect was made in 1925 by the commander of what is now the Wright-Patterson Air Force base in Ohio.
  13. Pankaj Ghemawat, Building Strategy on the Learning Curve, Harvard Business Review, March-April 1985. The unit cost declines of 10% and 30% for each doubling of output are generally described as 90% and 70% learning or experience curves, respectively.
  14. A. Michael Spence, Competition, Entry, and Antitrust Policy, Strategy, Predation, and Antitrust Analysis 45, 65-66 (Steven C. Salop ed., 1981). One way of summing up the learning curve effect is “practice makes perfect”. In information age industries where flexibility and willingness to change are present the benefits of learning curve effects should be significant. Additionally, some have argued that the more complex the undertaking the greater the rate of learning. [Wilfred B. Hirschmann, Profit from the Learning Curve, Harvard Business Review, January-February 1964] As a consequence, in high-risk, high-reward winner-take-all markets any incremental benefits, such as those from a learning curve effect, may be sufficient to determine the winner. Another author claims that “self-organizing systems are learning systems”. David H. Freedman, Is Management Still a Science? Harvard Business Review, November-December 1992
  15. In most tangible products diminishing returns eventually limit the positive cost effects of the learning curve.
  16. This may also help explain why market share is profitable and why pursuit of market share is a rational strategy for many organizations. In the 1970’s and 1980’s the Boston Consulting Group used the learning curve in support of strategies seeking significant market share positions. John S. Hammond III and Gerald B. Allan, Note on the Boston Consulting Group Concept of Competitive Analysis and Corporate Strategy, Harvard Business School Note #175175, 1983. Learning curve effects have also been used to explain some of the results from the PIMS (Profit Impact of Marketing Strategy) database which show a correlation between market share and profitability. Robert D. Buzzell, Bradley T. Gale, Ralph G.M. Sultan, Market Share – A Key to Profitability, Harvard Business Review, January-February 1975.
  17. For a general discussion of the concept see When Winners Take All, The Economist, November 25, 1995 at 82. Other discussions on the topic can be found on the web at and
  18. If there are significant entry barriers this elimination of lower-tiered competitiors may increase the riskiness inherent in the market. If the “superstar” winner subsequently becomes less desired the concentration of decision making in fewer hands may result in sub-optimal results. (All your eggs in one basket problem.) Markets with numerous competitors are more robust than single firm marketplaces, even if the single firm is currently the most desired or lowest cost producer, because of the risk posed by a single firm’s mistake is more easily corrected by consumers shifting to competitive products in a contested market.
  19. Technically, network effect should be used to describe situations where there are increasing returns as the number of network participants increases and network externalities should be limited to situations where the network effect creates less than optimal conditions where a decision maker does not bear the full costs of his or her decision. S. J. Liebowitz and Stephen E. Margolis, Network Externality: An Uncommon Tragedy, 8 Journal of Economic Perspectives 133 (1994).
  20. Nicholas Economides, The Economics of Networks, 14 International Journal of Industrial Organization (March 1996).
  21. Economists refer to this as increasing returns. In the physical world one could think of this as a virtuous circle, where feedback from the growth becomes self-reinforcing.
  22. Often cited examples of this phenomenon are the well-known Betamax/VHS, Windows/Macintosh, 8 track/cassette audio tapes situations.
  23. The first mover advantages may also be related to the marketplace advantages enjoyed by early or quick success in the movie and television industries.
  24. A similar phenomenon is beginning to appear with searches on the internet. So many “hits” are returned that an exhaustive search may no longer be effective. Automated agents and filtering mechanisms may help but there is a concern that those who control the filters (Microsoft Sidewalk is an example) could
  25. Digi-marking is the electronic tagging of copyrighted work with identifying information that can assist in tracing the origin of the work. Copy protection schemes, at one time widely employed in the software market, may also reappear – and already have done so for high-end software products.
  26. If self-organizing systems orchestrate the formation of undesirable structures and do so without leaders, is it fair or correct to look for culpable leaders who are the architects, the master planners, of these self-organizing structures? Is the queen termite “responsible” for the construction of the termite colony mound on the plains of Africa? No, and it is an example of the centralized mindset to even call her the “queen”, which implies leadership.
  27. justice (distributional justice under the capitalist criterion whereby each receives according to contribution and according to voluntary bargained-for exchange, 2. rights (right of free consent, voluntary exchange, freedom of choice), and 3. utility (resources are allocated, used and distributed efficiently).
  28. Some have even argued that computers are the largest driver of wage inequality in the emerging winner-take-all information age economy. By these accounts the concentration of greater economic rewards in proportionately fewer hands will lead to a substantial income and wealth gap, and since one “votes with dollars” in allocating resources in the free market economy those with more dollars get more “votes” in this resource allocation process. Whether or not this is “fair” is left for others to decide. Neil Munro, For Richer and Poorer, The National Journal, July 18, 1998. For a discussion of the moral legitimacy concerning these competitively determined profits and incomes see Derek Bok, The Cost of Talent: How Executives and Professionals Are Paid and How It Affects America (Free Press, 1993).
  29. In fact the opposite may occur. If each side keeps increasing its “arms build-up” it is possible that each is less secure (or at least no more secure) than it was before the race began.