Understanding Complexity and Knowledge Requirements at Mobil

Ted Lumley, Mobil Oil

The Knowledge Advantage Chicago November 17th, 1995

Introduction: Complexity, Knowledge and Energy

[VG1] I want to share some ideas with you based on a study of consistencies across some of our successful engagements with complexity, and their "fit" with adaptive models coming from the field of artificial life. While no final answers are in hand, some interesting clues and insights are emerging.

In preparing this material for presentation, one of the things I struggled with was how to communicate the consistent positive energy associated with those cases where we are really starting to get a handle on dealing with complexity. Analytical frameworks provided by mathematics and physics are useful for sharing the structure of such experiences but they don't speak to the emotional content. So what looks like a sound rational solution to a complex problem may be unworkable when it's implemented in a world of real people. That is, analytical models are not able to account for the complex relationships between people. Conversely, solution approaches which tap the positive energies and emotions of people can deliver spectacular results, even if they might be the last thing to come out of a disciplined design effort.

On the flight back to Dallas last Friday, they happened to be showing the film "Apollo 13", and it struck me that the extraordinary effort which brought the Apollo crew home is a very rich and appropriate metaphor for what I'll be talking about today. The basic elements of the Apollo experience are generic to our most complex oil industry challenges. That is, in complex environments, seemingly minor events can spawn "deterministic chaos" in which a small event is amplified over time, through a multitude of complex relationships, to blow away predictability and threaten the viability of the encompassing system.

In the case of Apollo 13, "chaos" was spawned by the failure of a small coil in an oxygen tank manufactured two years prior to the mission. Through a cascade of events, this small failure grew to huge proportions bringing the sophisticated Apollo support systems to their knees, threatening the lives of the crew and challenging the knowledge and resolve of the entire NASA complex. An extraordinary effort on the part of NASA and the Apollo crew brought them safely back home.

[VG4] Some of the generic properties of this remarkably successful engagement with complexity were as follows:

1. The "management by parts" tradition was abandoned, solutions were not generated via designated individuals but were drawn from a resonant shared mindspace

2. "Procedures" and "means" became the slave of purpose, rather than vice versa

3. Roles and responsibilities freely "morphed" according to environmental need

4. Solutions were developed by simulation rather than being engineered from basic principles

5. Competition ceded to cooperation

6. Implementations harmonized with human values

It seems that when such a combination of system attributes come together, huge coherent energies are unleashed which can give birth to spectacular results.

For my money, a videotape of Apollo 13 has to represent the best educational value available on the topic of complexity and knowledge management.

While the business cases I am basing this talk on [1] won't make it to the history books or network news as did Apollo 13, they were characterized by a very parallel set of attributes. Positive energies in these complex, unpredictable problem spaces, seemed to derive more from tension in the gap between necessity and possibility than from expectations of success. Body language of the participants interviewed during these initiatives captured this energy far better than words are able to.

The data I'd like to share with you today, relates to the growth of complexity on an industry basis. Both the nature of the problems and of the standout successes in this environment can be linked to a particular feature of this complexity, "morphing", which I'll define in a moment.

Complexity in the Upstream Petroleum Industry:.

So, with these introductory thoughts out on the table, I'd like to move on to discuss three topic areas which impact information management needs; (1) the nature of rising complexity in the upstream petroleum industry (UPI); (2) alternative evolutionary models (behavior-oriented versus parts-oriented) for dealing with this complexity, and; (3) the knowledge requirement implications of these alternative models.

Complexity in business has a lot to do with evolution and metamorphosis, so I'll preface my further remarks by putting up some images which are insightful to me in thinking about the practical impacts of this "morphing" type of change.

[VG4] This viewgraph raises some problems in my mind in trying to use the traditional "managing by parts" approach in an environment which is undergoing complex change. If I am asked to work one of the parts in this "organization", it is clear that I will have to work very closely with my colleagues to help adapt the overall organization so that it deals appropriately with the environment and fulfills the shared organizational goals in the process. If we all do a good job on this, we may end up with an organization that looks something like the one on the right.

But there is a problem here in that the notion of a "part" is a very transient thing since most parts on the right look very different from the parts on the left. In essence, as I work on a particular part, I do not know what I do not know about the shape of things to come, nor do my colleagues, but its clear that we're all going to have to shuffle territory, roles and responsibilities as we go along.

So these images is to provide a metaphor for the term "morph" which I'll be using in subsequent discussion. This effect is particularly meaningful to me as I started off as a geophysicist and ended up as an I/T person without really being sure of how that happened.

[VG5] These images help me remember that in a complex environment, no unique forward predictions are possible and the current state can evolve into some very different future states, depending on initial circumstances and environmental factors. Because of this ambiguity, it will be doubly important for those of us working the parts to continually share our visions of the future, passing the reins back and forth as needed.

Those two viewgraphs illustrate the dual problems associated with "management by parts" in a complex, "morphing" environment; i.e. the transient identity of a part and the multiple alternative futures which can grow out of similar sets of parts. The rising complexity we are struggling with in business seems to me to be strongly related to these indeterminacies. The data I'm about to share with you on complexity in the upstream petroleum business makes this point convincingly in my opinion.

Let's now look at complexity in the context of technology, culture and economic change.

[VG6] This simplified viewgraph indicates the technical complexity we have to deal with to find and describe subsurface hydrocarbon reservoirs; to assess and predict the producing dynamics of the reservoir; to design, build, commission, operate and dispose of surface producing facilities, and; to engage the consumer market and transport product. Each one of the technical "spheres" shown on the viewgraph is undergoing its own technological revolution.

Managing assets across this highly dynamic, far-from-equilibrium structure requires creative mental interaction amongst all the participants.

However, in the linear tradition, we have all tended to make of our technical function or department, an independent fiefdom employing its own language, culture and performance metrics. While we've used information technology effectively to allow basic business and specialty information free passage through this complex network, our knowledge assets have been progressively decoupled, and we have failed to develop a collective wisdom on an overall organizational basis.

[VG7] I am showing this next viewgraph to compensate for the understatement of complexity in the prior viewgraph. We have more than thirty areas of technical specialty in the upstream petroleum business and each of these is spawning an associated archipelago of sub-specialties. Facilities engineering, for example, currently explodes into more than thirty islands of sub- specialties. This situation changes continuously as important new specialties come into play.

[VG8] But complexity isn't fully defined in the contemporary plane. If we look along the time axis, we can see that technical specialization has been undergoing exponential growth. It is not going to stop. In addition, as we refocus on core competencies, outsource non-core technologies, and form more joint interest alliances and consortia, we are experiencing exponential growth in the number of stakeholder organizations involved in a single project. In the North Sea, a Norwegian consortium (NorSok) established that a recent major offshore producing project involved the participation of 350 different organizations and over 10,000 people.

An evolutionary view of the industry shows that this rising complexity is not confined to the progressive specialization and subdividing of existing specialties. It is characterized by the continuing adjustment of roles and behaviors in the overall dynamic of the business ecology. That is, the rising complexity we are facing is of a "morphing" nature in which each activity element is being modified via influences coming from its interconnections with a web of other activities. The result is that there is continual upwelling and subduction in the form and function of the business.

Like other complex markets, the UPI appears almost as a "lifeform" [2] in its own right. On a global industry basis, the dissipative energy it consumes to "stay alive" is on the order of eight hundred million dollars worth per day (54 MMBOE) or three hundred billion dollars per year (20 BBOE). We all continue to feed its huge appetite, even though we are sometimes disgusted with its personal hygiene.

One of the nominal goals of the Norwegian consortium just mentioned is to eliminate by 1998, via a standards strategy, 20-25% wastage in total costs associated with information management. Looking at this savings in terms of 42 gallon barrels of oil which cost roughly twelve dollars each to produce, this comes to about three dollars per barrel. If these savings were equally applicable to worldwide petroleum production, this would equate to $60 billion per year, or twelve times the $5 billion per year currently being spent on upstream information technology. This magnitude of economic incentive can induce a great deal of change in the industry.

However, there is an awareness of an even larger, less quantifiable prize which relates to another "morphing" pressure. By bringing their knowledge resources out of their silos and engaging them in the joint search and discovery of better ways of doing business, they hope to go even farther in extending the life of their existing assets, finding ways to make currently sub-economic prospects viable, and delivering major economic uplift to the regional business ecology.

[VG9] In the cultural complexity domain, stakeholder interests and relations seem to be taking on new importance as people become more informed on the interplay between their own needs, interests and actions, and those of others. In fact, as the various stakeholders in the industry increase their degree of collaboration, the traditional but arbitrary "we-they" distinctions are morphing into a unified "us". In response, our information infrastructures are having to accommodate knowledge sharing across the full stakeholder ecology.

[VG10] While the prior viewgraph implied a traditional view of stakeholder needs and roles, this viewgraph, prepared by British Petroleum for a joint venture underway in the Azerbaijan - Kazakhstan region of the CIS, suggests something more. It suggests that new stakeholders are continually coming into play with new needs which impact the overall business dynamic. As a result, roles, behaviors and investments are "morphing" in a rather complex way. Our information and knowledge sharing requirements are consequently being profoundly effected.

[VG11] On the economic front, our industry's tradition has been to intensively assess and risk manage the key, performance influencing variables associated with a given asset, starting from the time of acquisition of the asset. The variables include; the recoverable reserves, the cost of facilities, operating costs, supportive infrastructure costs (pipelines, tankers etc.), oil prices, royalties and taxes. Our economic assessments have leaned heavily on discounted cash flow analysis and associated NPV, ROR etc. indicators.

[VG12] A retrospective analysis of many assets has shown, however, that such economic assessments have more often than not been superseded by unpredicted, emergent events (positive or negative) which eclipse the economic predictions derived from traditional DCF assessments.

For example, on a North Sea property acquired in 1971, the tax regimes have undergone unanticipated major change (which first more than halved and later doubled after-tax profit margins) as governments or government policies oscillated. Technological breakthroughs enabled the identification and recovery of far larger quantities of hydrocarbon than were initially assessed. Subsequent satellite discoveries in the region, ordinarily below the threshold of economic viability, were able to share emergent producing infrastructure and thus become economically viable. Industry cooperation led to the development of regional gas collection infrastructure which made gas delivery to markets in continental Europe economically feasible. An unanticipated pipeline from Russia to Europe also perturbed the regional economic scene, and so on.

If we were to visually simulate the global evolutionary history of petroleum assets, it would not present itself as decoupled patterns of growth and decline, but would appear more like a bacterial culture, coherently swelling here and recoiling there in response to technical, cultural and economic events. That is, there is a whole lot of morphing going on here and there's going to be a whole lot more. Our information infrastructure needed to support it is having to become "business ecology aware", as are our business models.

New economic assessment techniques (e.g. Option Pricing Theory), are emerging to capture the rising value of investment flexibility in the face of these evolutionary dynamics. Ad hoc, frictionless, information-rich assessments are needed to support this flexibility. Since these new needs represent much more than enhancements to existing approaches, a new design paradigm is also called for.

In sum, the apparent growth of complexity we have been struggling with in the upstream petroleum industry is not just due to asset lifecycle and technical, cultural and economic factors, it is due to the complex interactions and changing patterns of roles and behaviors. This in turn is complicated by the use of a "management by parts" schema which tends to preserve the identity of the "parts" as it seeks to enhance them. The data that I have been exposed to suggests that an evolutionary model which can better deal with this type of complex "metamorphosis" is needed.

Alternative Models of Evolution and their Knowledge Implications:

After struggling to understand complexity for the past several years, I was surprised to discover, a few weeks ago, that colonies of simple "minded" digital organisms seemed to be making more progress than myself. As you may know, digital organisms are artificial life forms composed of computer codes that reproduce, mutate and evolve their skills through a Darwinian selection process. Hearing about their progress had me searching for an explanation as to how these little bit-heads could read and respond to complex situations so rapidly. The secret to their success seems to be connected to their...what else...ignorance! (see footnote). In any case, these artificial life findings appear very relevant to the models which we deliberately or implicitly use to evolve our organizational responses to the changing business environment.

In artificial life as in biology, there appears to be two contending models for evolution which differ with respect to "fitness" selection; a "purpose-oriented, behavior scoring" model (POBS) [Note: terminology and abbreviations are arbitrary and for convenience only], and a "means- oriented, component scoring" model (MOCS). The business case studies I've been exposed to, and the Apollo 13 experience mentioned above, which relate to complex environments, support the superior performance of the POBS model. Recent publications by Larry Fogel, of Natural Selection Inc [3] provide some convincing supportive arguments for this choice, which have been seconded by their digital organisms.

The MOCS model follows the idea in biology that natural selection is on the basis of individual genes (i.e. components), while the POBS model follows the alternative proposition in biology [4] that natural selection is on the basis of aggregate behavior of the organism or group of organisms. This does not mean that the genes don't count, since they are a primary determinant of behavior.

[VG13] The MOCS model seems to have been the implicit model of choice in the traditional business organization. It seeks to evolve the system by selectively rewarding the components of the organization which are nominally involved in successful business engagements. Thus the manager, team or employee who is involved in a successful venture will be rewarded in such a way as to increase their influence over organizational resources. In a sense, this is akin to being allowed to reproduce. This model assumes that over a large number of engagements with the business environment, all other things being equal, this rewards and selection process will optimize the performance of the overall organism (organization).

What the model seems not to consider is that "all other things may not be equal" and unmeasured behaviors emanating from the many interactions of components may outcrop at another time or place, far removed from the domain of the engagement in question. When they do, they will have been sufficiently well laundered by chance and metamorphosis that there will be no way to explicitly trace their origins. To state this another way, this model does not account for any "deterministic chaos" that the players may be injecting into the system, which could drastically alter the future state of the organization.

In the billiards examples shown on the viewgraph, the MOCS model would reward each player according to the number of opposition balls sunk. This could lead to the dilemma (similar to "the prisoners dilemma" [5]) wherein sinking the maximum number of balls would be in conflict with leaving the table in the best shape for one's partner. A player's intent would be hard to "read" since the goodness of positioning of the balls is not quantifiable but only qualitatively estimable (due to deterministic chaos) and accountability is "laundered" away by the evolving play.

The POBS model, on the other hand, would reward each player on the basis of how his/her play (behavior) contributed to the shared purpose of sinking all balls prior to (more efficiently than) the opposition. Thus the POBS model would not specifically reward players for sinking opponents' balls, but would reward for the overall "goodness" of the table configurations bequeathed to one's partner/s. Such a model encourages team cooperation rather than "betrayal", placing fulfillment of shared purpose ahead of individual goals fulfillment.

In undertakings which are sufficiently short in time duration or involve sufficiently few variables and interdependencies, that is, in stable regions of space and time where the practical effects of deterministic chaos are small, the MOCS model may be appropriate. As such regions are shrinking in response to the rising tide of complexity, part of our problem seems to be the persistent use of the MOCS model where it is no longer appropriate.

In the case of Apollo 13, while the MOCS model may have been acceptable for the less complex environments within the project segments, it did not appear to be appropriate for mission conditions where the number of intermeshing parts and procedures (i.e. the number of variables and interdependencies) grew enormously. Since the full complexity only emerged as the whole ensemble of parts and procedures came together during the mission, the needed level of fault tolerance was never designed in, as would likely have occurred via a POBS approach.

In systems which are engineered to be fail-safe (i.e. where each player or component is independently optimized as in the MOCS model), the exposure to failure by "betrayal" rises with the number of players. However, systems which are fault tolerant (i.e. where players or components cover for each other as in the POBS model), have a resilience against failure which rises with the number of players. In the case of Apollo 13, the original engineering appears to have followed the MOCS model, while the crew and support staff supplied compensatory resilience as they shifted into POBS mode.

It is the POBS model which appears to be associated with emerging information age team successes. This model is organizational purpose focused [3] and selects (i.e. rewards) on the basis of behaviors which associate with business success. Unlike the MOCS model, it leaves much flexibility in the choice of means and gives far more attention to the definition of purpose (i.e. it goes well beyond a vision and mission statement). It considers a full spectrum of possible outcomes, "from utopia to catastrophe" and their relative desirability (see Fogel, "Valuated State Spaces" [6]). The rewards systems take into account the preferential concerns and behaviors which lead to successful engagements with a complex environment.

Since the rewards are linked to behavior rather than organizational components, there is an inclusionary aspect to the POBS model, with respect to who can contribute and what is considered a contribution. That is, the rewards are not constrained to go to administrative, organizational or disciplinary components bureaucratically linked to the outcome, but may go to anyone who puts their shoulder behind the initiative through whatever behavior is deemed helpful. By not constraining the "means", this POBS approach engages the energies of anyone with a better idea, yet discriminates against rewarding "components" solely by virtue of their position in the organization, or those components whose behaviors support immediate goals at the expense of long term organizational health.

This POBS type of model was the implicit model of choice in several exceptional team successes recently documented within Mobil.

[**Note: Similarities between the Apollo 13 approach and the POBS model can be seen by reconciliation with the six attributes listed earlier; i.e. the "management by parts" tradition was abandoned, solutions were not generated via designated individuals but were drawn from a resonant shared mindspace, "procedures" and "means" became the slave of purpose, rather than vice versa, roles and responsibilities freely "morphed" according to environmental need, solutions were developed by simulation rather than being engineered from basic principles, competition ceded to cooperation, and implementations were harmonized with human needs.]

In sum, the MOCS model which seems to have been the implicit choice of industrial age organizations, tends to promote independent optimization of organizational units and individuals, rather than overall systemic performance. In complex environments where "morphing pressures" abound, managing by the sum of optimized parts seems to accrue large overheads and lead to confusion, suboptimization or worse. An optimizing schema which frees internals as well as overall organizational behavior to "morph", such as the POBS model, seems better suited to today's complex environments.

The proponents of the POBS model [3,4] claim that "nature scores behaviors, not components or organs". Scoring components, as in the MOCS model, tends to promote competition rather than cooperation and simple enhancement, rather than "morphing" of the "parts".

Knowledge Implications:

[VG14] In environments or in eras where we have had relative stability (near equilibrium), and morphing effects have been small, knowledge requirements have tended to emphasize increasing specialization. Local context possessed by individual knowledge workers has not been seen as important since roles and deliverables are agreed in advance. The goal, instead, has been to improve the cost, cycle time and precision of the parts. Unfortunately, the knowledge management tools in this "clockworks" environment often seem to be of the "strip-mining" variety [8], which tap only a small fraction of an individual's knowledge potentials.

As we go deeper into the information age, the impressions of a world "far from equilibrium" continue to emerge and "morphing" effects seem to grow stronger and more pervasive. There appears to be at least three salient knowledge implications of this type of environment, as drawn from the anatomies of successful and unsuccessful business initiatives, and from autopsies on successful and unsuccessful digital organisms. While the whys and wherefores of these findings may be more fully explainable in computer, maths and physics terms, descriptive generalizations are as follows; They are; (1) the need to solve complex problems "in toto", in the shared mindspace of the team; (2) the need to solve complex problems iteratively with the participation of those with in-context knowledge of alternatives, and; (3) the need to view the problem from the inside, positioning oneself within the connective web of stakeholders.

1. There are too many variables and too many interdependencies in many of our information age business problems to solve them by breaking them into "parts". Instead, they need to be solved "in toto".

In both computer and human terms, this means that you have to have a lot of memory available. Neither individual computers nor individual people have sufficient memory for many of these problems [7], so that gaining new knowledge when dealing with problems of this class requires memory sharing and/or adaptive solution approaches.

This is a familiar requirement in team sports such as hockey, soccer or basketball where the overall team response does not fit into one player's head since he cannot by himself see and retain all of the subtleties of movement and intent of the full complement of players.

This requirement is implicitly being addressed via the successful team practices referenced in this discussion, which employ a combination of visualization and historical dialogue. Similar approaches are being used in petroleum industry collaborative efforts (e.g. in the business workshops and forums of the Petrotechnical Open Software Corporation.).

2. The type of problems we face are subject to "deterministic chaos" or "sensitive dependence on initial conditions". This means that there are no unique, predictable solutions and a vast number of solution options are involved. The choice of the optimal organizational response will depend on organizational purpose and preference relative to the environmental dynamics. It means that the solution will have to be developed iteratively and adaptively, tapping, in each iteration, the knowledge of those individuals who understand the alternatives in their environmental context. That is, "strip-mining" of the knowledge base will not suffice. The sports metaphor above applies equally to this requirement. Team responses are required, management proxies are insufficient.

3. Since unique, predictable solutions are precluded, predictability in this class of problem is qualitative and arises from a knowledge of patterns and inter-relationships. In the embryo morphing example, the number of acceptable futures could be greatly reduced if there was knowledge of internal relationships (e.g. the vertebrate structure, circulatory system structure etc.). In the business context, the gathering of internal relationship data requires a departure from the top-down decompositional view of the problem and the adoption of an "inverted perspective".

In the terminology of the successful teams studied, this involves "shifting from a 'boss-focus to a 'customer-focus'" and viewing the system (i.e. gathering knowledge) as a member of the extended web of interconnected customers and suppliers. One can visualize this in the form of a large roller skating surface in which suppliers and consumers clutch each other in a web like configuration (everyone is connected since everyone is both consumer and supplier to different parties). This view avoids the constraints of organizational boundaries and proxy communications and opens up to non-discursive indications of imminent "morphs" which can then be exploited or avoided.

These three knowledge implications; memory sharing, participatory decisions, and inverted perspective, are suggestive of a very different type of information infrastructure and design philosophy than has been developed to support the MOCS model. Ad hoc information and knowledge sharing capability on an any-to-any basis, including external stakeholders, becomes more important. Information rich communications, incorporating both discursive and non-discursive information [9] also rise in importance.

Internal business success cases studied in an information management context within Mobil showed that some form of shared visualization of the time-evolution of problems accompanied by dialogue (particularly same-time, same-place) was an important contributor to knowledge sharing. Sophisticated technology was generally not, in this context, a make-or-break requirement.

The need for closely knit, shared memory teams is as much indicated by the nature of the problem as by their demonstrated successes. However, the associated abandoning of predictability and embracing of trust, as is demanded by memory sharing, is a non-trivial requirement. The team viewed in this context has more of the aura of an integrated life-form than a mechanical meld of components sustained by managerial incantations.

Nonlinear Knowledge Transfer Issues:

Communications in the linear, machine age problem solving mode have been well supported by linear word structures, high order bit patterns and "bottom line" statements. Complex problem solving environments involving many dynamic interrelationships appear to be far more demanding on communications. For example, simple word structures (noun, verb, object) are too narrow band for efficiently transferring the subtleties of complexity, without resorting to the use of metaphors.

If an image is worth a thousand words, the dynamics implicit in a metaphor may equate to a thousand frames of imagery. This represents information compression of the order of one million to one. Thus, a five minute discourse containing ten metaphors could take months to deliver in simple word structures. Sharing knowledge of the Apollo 13 experience, for example, which involves complex feelings as well as complex structure, would be impractical to deliver in non-metaphorical text. While books (employing metaphors) may compete time-wise with video knowledge transfer, the ambiguity of interpretation is far higher (i.e. actors and directors interpret books, reducing ambiguity while incorporating their own personal "spins").

Since images and metaphors stimulate recall of personal experience, the use of these forms requires some sort of normalization to ensure objectivity (or consistent subjectivity). This is being accommodated in successful team environments through the combination of visualization and dialogue.

The suggestion is, in making note of these issues, that the need to solve problems of increasing complexity is placing significant new demands on human communications.


The above perspectives on complexity and knowledge requirements were drawn first from actual business situations and subsequently extended and restated in the more insightful, unifying terms of the mathematical models of complexity and the simulation models from artificial life.

The suite of consistencies between the business experiences and the model and simulation data appears to provide a basis for some corollary insights on open systems, open data standards, distributed computing, investment flexibility valuation, and the genesis and maintenance of teams (i.e. knowledge infrastructure).

These consistencies also give a sense of high potential value in discriminating between our structured, components-oriented view of the world needed for mental housekeeping, and the purpose-oriented, behavior-scoring models needed for actual responses to complex environmental and human dynamics.

An outstanding attribute of the POBS model usage is that it begs the re-engagement of our full natural knowledge potentials. An aspect of the organizational use of POBS which the artificial life researchers are unable to discern (as yet?!?) is the positive emotions experienced by the POBS practitioners in fully engaging their rational and intuitive skills in the shared pursuit of solutions. This is not only a question of actualizing more of one's potential, it is one of working in an environment where natural selection pressures seem to favor inclusion, diversity and cooperation over exclusion, "clubism" and internal competition.

The extension of knowledge requirements in the POBS environment into the domain of non- discursive information (roles, gestures, silent signals etc.) and issues of trust [9] are important needs which have not been delved into in this discussion. However, the need for new perspectives coming from the design methodologies and conceptual frames of human systems is clearly indicated.


[VG16] Like entropy and time, information goes only one way; it increases. How we perceive the continuous growth of information (e.g. "complexity") seems to determine whether or not we accept it as being natural or whether we find it troublesome. When we associate it with the continuous unfolding of nature (e.g. parent, children, grandchildren), which we feel a part of, it can be very pleasing. When we see it as an out-of-control growth of alien "parts", it becomes problematic. As collaborating stakeholders in the petroleum industry are discovering, traditional "we-them" distinctions are arbitrary. We clearly have a choice.

We seem to have a parallel choice on the organizational plane; to manage everything by parts, or as purposeful, self-determining collectives whose health and sustainability is inseparable from their environments. We appear to be at such a crossroads of choice in the upstream petroleum business. The emergence of collaborative groups such as the Petrotechnical Open Software Corporation implies that we may be opting for the latter path.

As our exposure to unpredictable events emerging from technology, stakeholders and the marketplace rises, our work and management practices must support an increasing organizational resilience. Evolutionary models provide insight as to how we can improve resilience, via orientations and rewards systems which are well matched to the natural selection process in the business ecology.

Of the alternative models discussed in this talk, the "purpose-oriented, behavior-scoring" (POBS) model seems to best match the anatomy of information age success. Within this approach, we elevate purpose above means and open the door to inclusion, innovation and "morphing" in our response to a barrage of emergent events.

As studied success cases have shown, and as is particularly evident in the Apollo 13 "archetype", one of the greatest sources of resilience is the purpose-directed, behavior-aligned shared mindspace of knowledge workers. Traditional knowledge management infrastructure, which has grown up around the "means-oriented, component-scored" (MOCS) model, will need to undergo a major "morph" to move into harmony with the POBS model requirements.

Our ability to learn how (or to "learn how to learn how" [9]) to create an ethic of purpose and people-based synergy and subordinate our traditional control and "management by parts" ethic, appears to be the greatest challenge. The common catalyst in making this cultural shift, as seen in the success database utilized for this discussion, is the sense of emergency felt by the team. This emotional trigger opens up team sensibilities to new options which emphasize natural, systemic insight over engineered solutions.

Since evolution suggests that an encounter with emergency is just a matter of time, limiting our focus to potting the next ball may be less than an optimal strategy. And as long as we persist in scoring "parts" instead of aggregate behavior, who knows how many Apollo 13 teams are being deprived of their right to emergence?

Return to Home Page .....................


[1] Mike Beers, Ernst & Young Center For Business Innovation, "Knowledge Transfer at Mobil, Using Success Stories for Organizational Learning", May, 1995

[2] Michael Rothschild, "Bionomics, Economy as Ecosystem", 1990

[3] Lawrence J. Fogel, Natural Selection Inc., "Evolutionary Programming in Perspective: The Top-Down View:", Plenary paper presented at the IEEE World Conference on Evolutionary Computation, June, 1994

[4] Robert Wesson,"Beyond Natural Selection", 1991 MIT Press

[5] William Poundstone, "Prisoners Dilemma", 1992

[6] Lawrence J. Fogel, Natural Selection Inc., "The Valuated State Space Approach and Evolutionary Computation for Problem Solving", Preprint of keynote address to 1995 IEEE Conference on Evolutionary Computation, Nov. 29-Dec. 2 1995, Perth

[7] Jim Dowe, Excalibur Technologies Corp., "Digital Organisms, Another Step in the Information Ecosystem", Bionomics Conference presentation, San Francisco, October 20th, 1995

[8] Ted Lumley, Mobil Oil, "Computing Tools for an Information Age Organization", Bionomics Conference Presentation, San Francisco, October 20th, 1995 (emlumleyl@onramp.net)

[9] Hadley Smith, The NEWORK Group, "The NEWORK Primer: An Evolutionary Learning System for Designing Someplace Else" (Work in progress. Hadley Smith hsnework@aol.com)


Footnote: The "Ignorance Advantage" of Digital Organisms:

Digital organisms approach their work with the same sort of unquestioning trust that characterizes relationships within the most adaptive of teams. It can be likened to a child's innocence or the innocence that an adult might only experience in a dream state. In this state, information is taken in without question, bypassing, in the case of mature adults, banks of bs filters built from years of defense against a barrage of bad information and bad advice. So it would appear that "innocence", in one sense, implies full bandwidth receipt of information.

Innocence without consciousness, simplifies into ignorance, which is perhaps the more appropriate attribute for this full bandwidth quality in digital organisms.

What do these banks of reject filters do to the pristine, flat response of innocence? My own experience with geophysical filters has taught me that if you want to remove 60 hertz noise, in practical terms you are going to have to cut out a swath from 55 to 65 hertz. This gives an exposure to the loss of valuable information in the "shadow zone" which surrounds the unwanted input.

In the context of information and knowledge, the implication is, for example, that legitimate archeological data which may be interwoven with stories about "Bigfoot" may be discarded, or more generally, legitimate scientific data in the shadow zone of myth or mystic themes may not make it through our involuntary bs filters. Unless, of course, it is "innocents" who are engaged in these learning exercises. Deterministic chaos which is giving rise to much of today's complexity is neither myth nor mysticism, but a phenomenon consistent with the laws of classical physics. However, the "mysterious" emergence and subduction of "morphs" associated with it, are in the shadow zone of our bs filters.

Compounding the problem with the fidelity of our input filters is the fact that complexity often comes in the form of subtle, low level signals. The difference between a virtuoso performance of Chopin and a mechanical rendition is not in the high order bits. Such signals may be difficult to pick up if the input spectrum with bs filters in, looks like a gap-toothed haircomb. That is, reject filters seriously degrade the fidelity of subtle signal components.

In times of greater stability and less complexity, we needed only focus on the strongest signal components; the high order bits, or "the bottom line". So the paradox that our human intellect seems frighteningly slow to adapt as we race forward on a collision course with complexity, might just relate to our industrial age input filters. Digital organisms, which share our appreciation of crisp logic but not our lost innocence, are in the process of building and demonstrating a better complexity mousetrap, and are going to have to be heard.

In fact, since they have no guilt or sensitivities, their behaviors are far more unconstrained than our own, and since they are not conscious, we do not have to take their word for anything, we can simply cut them open and examine what makes them successful or unsuccessful. And if we "listen" closely to their current message, what they seem to be saying is that complexity is bringing us an opportunity to kick the four hundred year old Cartesian, "management by parts" habit and come home to a more natural, less mechanistic engagement with the world.