Emergent Order and New Science 'Macro' Leadership Dynamics: Strategy, Microcoevolution, Distributed Intelligence and Complexity in Firms
Complexity Study Group and Seminar
led by
Professor Bill McKelvey
John E. Anderson Graduate School of Management at UCLA
A report by Malcolm Peltu
London School of Economics and Political Science,
Houghton Street, London WC2A 2AE
email:
E.Mitleton-Kelly@lse.ac.uk|; Web: http://www.lse.ac.uk/complex|
Overview
Bill McKelvey is an organisation scientist whose current main research interest centres on applying complexity science and computational agent-based complex adaptive learning models to the development and implementation of successful business strategies. On 14 March 2000, he led two meetings at the LSE:
1. The Complexity Study Group (in the morning), which highlighted detailed research that explores whether concepts from complexity science can provide a coherent theoretical framework for all physical, biological and social systems.
2. The Complexity Seminar (in the afternoon), which focused on showing how the ideas discussed in the Study Group could be applied to help CEOs keep their jobs by maximising shareholder value.
A total of 28 people participated in the two events, although not everyone was at both (see Appendix 1). There was considerable overlap in the topics covered in the two meetings. The first part of this report summarises the main themes presented by McKelvey. This is followed by summaries of the main issues highlighted during the lively interactions between McKelvey and participants. More in-depth background to McKelvey's presentations is available on the LSE Complexity Website (McKelvey 1999, 2000). Appendix 2 provides a glossary to key terms highlighted in italics in the report.
This report was written by London-based Editorial Consultant Malcolm Peltu.
Objectives of the Study Group and Seminar
Bill McKelvey explored a number of key questions relating to the application of complexity science to business strategy:
* Is there a coherent theory of emergent order in social systems which covers all levels from the quantum to organisational, as well as encompassing all complexity science disciplines (such as quantum theory, chaos, biology, neural science and adaptive learning models)?
* Can complexity science produce any tools, techniques or rules that will give practical help to CEOs in developing successful business strategies?
* What kind of leadership does a firm need to execute novel strategies which get firms to change their internal way of functioning to produce continuing success, through generating emergent structures capable of complex adaptive behaviour?
Key concepts presented by Bill McKelvey
Towards a coherent theory of emergent order
McKelvey emphasised his belief that complexity science is fundamentally aimed at explaining order creation and the links between entities (Prigogine 1962; Cohen and Stewart 1994; Mainzer 1996; Omnès 1999). However, he said his research had found that complexity science has not yet formulated a coherent theory which explains where order emerges from in physical or social worlds — or why it emerges in one way rather than another. Evolution, for example, is not an adequate answer in biology because it presupposes an existing order, in the form of selection. Nevertheless, he believes he has identified the beginning of a theory which could help to develop successful corporate business strategies (McKelvey 1999; 2000).
His approach is based on using adaptive tension (McKelvey 1999, 24; 2000, 12-15) as a force managers can apply to generate emergent order likely to produce profit above the industry average (rents). The concept of adaptive tension is derived from Lorenz’s (1963) deterministic model of turbulence in weather systems, which is also applicable to the Bénard cell, a device which creates movement in a liquid sandwiched between a hot and cold plate. An energy differential is created by the hot surface of the earth and cold upper atmosphere, or between the hot and cold plates of a Bénard cell, causing energy transfer via bulk current in the air or liquid in-between. This involves the formation of strange attractors, where the system oscillates between paths without settling at an equilibrium — in contrast to the point attractors that exist in systems which have some kind of equilibrium position. In organisational terms, McKelvey likened strange attractors to the notion of 'herding cats' (Bennis 1996) as a way of moving people in a generally desired direction, without determining precisely where they have to go and how they must get there. He frequently referred to GE as an example of a company which has successfully implemented simple rules that have generated effective levels of adaptive tension aimed at meeting key corporate goals (see section Learning from GE's successful strategy).
The emergence of higher-level structures is said to take place at the edge of chaos, in the region of emergence between two critical values of the imposed energy differential. Below the 1st critical value, heat transfer occurs via conduction — molecules transfer energy by vibrating more vigorously against each other while remaining essentially in the same place. Between the 1st and 2nd critical values, entities move elsewhere in bulk, causing turbulence in the atmosphere or cell. Above the 2nd critical value, a transition to chaotically moving molecules is observed, leading to the emergence of bifurcated structures, such as tornadoes.
McKelvey related these ideas to organisations: the hot surface could represent a firm’s current position and the cold surface where the firm should be positioned for improved success. This kind of adaptive tension neither defines what people do nor constrain the new ideas which emerge, and the technologies and systems involved in them. A CEO's strategy should aim to set an appropriate level of adaptive tension to get the organisation between the two critical values, where constructive emergent behaviour is more likely to move the firm from where it is now to where it should be. If the adaptive tension of the energy differential is not high enough (below the 1st critical value), nothing much happens and the status quo in the firm remains. If the tension is too high (above the 2nd critical value), the firm behaves chaotically.
A 3rd law of thermodynamics is a key missing piece in the development of a coherent complexity science theory of emergent order, according to McKelvey. The first two laws assume order already exists. The 1st law holds that energy is conserved; the 2nd that order and the energy which causes it dissipate over time into disordered randomness. McKelvey’s (2000, 15-19) early formulation of the 3rd law proposes that it should be generalisable across the physical, biological and social worlds. It should explain how order comes about if it doesn't come from nature, God or some other form of pre-existing order. He suggests the new law should include the need for an environmentally imposed energy-differential effect to set into motion the emergence of order in the form of coarse-grained higher-level structures.
Managing adaptive tension to meet corporate goals
McKelvey (1999, 22-37; 2000, 21-23) described a number of approaches to applying knowledge about adaptive tension to get well-directed innovative behaviour in a firm. He said simple rules can set up the tension to create the required energy differential, but stressed that adaptive tension should be regarded as a steering mechanism. It can't guarantee anything, but can increase the rate of blind variation in the right direction — rather than a million different directions, most of which are useless.
Precise indicators of tension that will drive innovation cannot be predicted. There is also no mathematical formula, or other measure, to indicate when the tension moves beyond the 1st or 2nd critical value. This means managers must be aware of the symptoms of emergence (McKelvey 1999, 30-33), such as 'bureaucratic' and 'chaotic' behaviour (Brown and Eisenhardt 1998). Firms that don't have any emergent behaviour are exhibiting symptoms of being a bureaucracy and are below the 1st critical value. Much frenetic behaviour, tension and anxiety are symptoms of chaos, indicating the tension is too high. Managers should then back off and get the tension better focused. Firms should aim to avoid either extreme.
The critical value for triggering chaotic behaviour can vary greatly between firms, McKelvey added. It can also change as firms become more skilled in dealing with adaptive tension. Firms with an ongoing history of change can comfortably handle a relatively high chaos level, whereas an organisation that hasn't changed for a long time could produce chaotic behaviour at a small level of adaptive tension. If there are symptoms of oscillation between bureaucratic and chaotic behaviour, either the region of emergence isn't big enough, or the energy differential is so high it generates bifurcated behaviour where people are always veering between trying new ideas or falling back into resistance to change.
Entanglement and adaptive tension
McKelvey stressed that managers need to do more than just set adaptive tension in order to create the conditions for effective emergence. They must also deal with entanglement ties (McKelvey 2000, 19-24). In complexity science, entanglement is what exists before order emerges. The role of quantum entanglement as the precursor to emergent order is much discussed in physics (McKelvey 2000, 2-8). For instance, Gell-Mann (1994) defines an entanglement field as a 'fine-grained structure of paired histories among quantum states'. The notion of the proverbial primordial pool which existed before the origin of life is also much discussed in biology (Kauffman 1993).
McKelvey has found that an understanding of entanglement from quantum theory can throw useful light on the nature of ties among people (Granovetter 1973, 1982; Burt 1992) and their impact on emergent order in organisations. In terms of human behaviour, he explained that a high correlation between the paired histories of people would mean they think in similar ways; a low correlation would mean they go in different directions. He stressed that in quantum physics the entanglement field is relatively 'pure' and 'uncorrupted', which means emergence is more likely to be constructive. In organisations, however, order usually emerges in the context of existing sets of strong ties, such as cliques. This results in biased emergence, which indicates why fostering uncorrupted entanglements in firms is important.
McKelvey observed that social entanglement ties are inherently unstable and deteriorate toward weak or strong ties over time. Strong ties occur typically when people meet at least once or twice a week; weak ties when they meet a few times year. Bridges across social groups are important because ties between existing cliques can help to bridge differences between functionally specialised 'silos' in firms. This concept of a social entanglement is analogous to Granovetter’s (1973) 'strength of weak ties' finding that innovation and novelty tends to come from weak ties, as strong ties generally favour the status quo and are therefore not as adaptively efficient as emergence from weak ties.
Advice provided by McKelvey (2000, 21-22) on how to foster entanglement ties likely to produce effective emergence includes: create denser networks of ties in the fine-grained structure; bring in employees with diverse backgrounds and interests; and create diverse task groups and social mixings.
Improving distributed intelligence by accelerating human and social capital formation
In discussing the historical development of ideas about effective management strategy, McKelvey (1999, 1-6) highlighted the importance to strategic business leadership of balancing human and social capital. Classical economists saw income purely as a function of capital and labour, which explained why firms can get advantage from investing capital in cheap labour areas. Becker (1975) emphasised the significance of human capital because competitive advantage often comes from ideas in people's heads, not just buildings and muscle. This has become more evident in today's service-intensive and knowledge-based business world. Burt (1992) suggested that competitive advantage is also a function of social capital — the networks of one kind and another established between people.
McKelvey argued that speeding up the formation of both human and social capital in a changing world is a vital way of generating rents. He said coevolutionary theory suggests that a firm may best enhance its adaptation to a rapidly changing competitive context by speeding up the microcoevolutionary processes within the company. This can also be a way of generating an effective system of distributed intelligence to produce teams capable of getting the firm to develop new ideas, new product concepts, new vendors, new relationships between marketing, production and engineering, and so on. Despite this need, he felt most executives spend too much time on changing the shape of the organisation on paper, instead of trying to make changes inside the firm.
McKelvey described the speed at which a firm works as its organisational metabolism. He noted that smaller firms with a high metabolism are generally better at rent generation from within its own competitive activities; efficient large companies are better at rent appropriation by acquiring rent-generation companies. For example, smaller biotechnology firms who are good at producing new technology have been bought by traditional large pharmaceuticals with the expertise and large-scale administration, manufacturing and sales systems needed to bring products successfully to market. McKelvey said large firms can appropriate rents in this way, but won't get sustainable advantage unless they change their internal operations to produce an organisation capable of generating rents. He likened attempts to speed up a firm's metabolism through acquisition to trying to speed up the metabolism of a dinosaur by grafting onto it a chicken's leg. He warned that the dinosaur's metabolism usually wins because novelty can't be bought.
Distributed intelligence, he noted, arises from interactions between nodes in the corporate brain, the network of communications among people in a firm. This view that 'intelligence is in the net' emphasises the value of social capital and the need to accelerate networks to create an organisational 'brain of brains', with people rather than simple on/off neurons at its nodes. It is the approach favoured by GE. On the other hand, a firm which regards human capital as the most important factor will be concerned more with the capabilities of individuals than networks, so will tend to look to buy genius 'stars'.
In some activities, such as biotechnology, the fastest growth is achieved by firms with the best research stars. GE, on the other hand, has built success by systematically avoiding prima donnas and focusing on speeding up microcoevolution to transmit the new idea across the network as quickly as possible. Finding the right balance between human and social capital (an optimising function which McKelvey called 'D') is a critical management challenge for most enterprises. If a CEO has a certain amount of money to invest, a decision must be made about how much should be spent on nodes versus interconnections; stars versus networks; isolated geniuses versus 'networked idiots'. McKelvey said neither extreme is good. He realises searching for the right D is difficult, but believes it is an important strategic issue.
Learning from GE's successful strategy
McKelvey used GE as an exemplar of a company whose CEO has developed and implemented an extremely successful strategy based on an apparently simple set of rules, which could be regarded as defining the borders of a Bénard cell without prescribing the detailed behaviour expected within the cell. He said GE’s rules act as strange attractors that move people in the desired direction, but attempting to speed up the emergence process without having the right adaptive tension is like herding cats without knowing which direction they should go. Adaptive tension, he added, can help managers to 'seed the clouds' a little to avoid relying on pure blind chance that a few people in marketing will meet up with a few people in production to create a winning new idea. McKelvey observed that GE's rules create a level of adaptive tension which moves people in the direction most likely to optimise shareholder worth. This can also avoid the agency problem, where people go off in different directions doing their own things (McKelvey 1999, 35-37).
McKelvey believes GE’s business success makes it an important case to study. It has grown faster than any other company since 1990; has a P/E ratio of 44 compared to a mean of 19 for diversified firms generally; makes an average of at least three company acquisitions a month; and is No. 1 in many business areas (The Economist 1999; Stewart 1999). The CEO responsible for these achievements, Jack Welch, has added more shareholder wealth than any other Chief Executive.
GE's strategy has two main targets: the development of its corporate brain and the creation of a 'boundaryless' organisation where information is shared. This is achieved through the articulation and implementation of a few clearly-stated simple rules. These are backed by strong incentives that make employees rich if they apply the rules effectively, but could get them fired if they break the rules. The rules aim to promote some desirable general behaviour patterns, rather than seeking to constrain detailed behaviour. There is nothing secret about the rules.
McKelvey described two significant GE rules which help to embed its human and social capital in a hospitable organisational culture:
1. Don’t hoard valuable ideas. One of Welch's main devices for breaking the organisational boundaries which inhibit information sharing is the 'anti-hoarding' rule: anyone in GE who discovers a valuable idea or practice must spread it through the rest of the company as quickly as possible — or face being fired. The good idea cannot be saved by the discoverer solely for his or her own career advancement, or for the benefit only of a local group. The underlying belief is that every idea can be abstracted to a generalisable level that can be applied successfully anywhere in the company. The 'not-invented-here' syndrome is avoided by stipulating that an idea can be rejected only if it fails after being tried in practice for at least six months. People also move frequently between different company activities, for instance from Turbines to Plastics divisions, to further facilitate sharing and knowledge exchange. One of the effects of the anti-hoarding rule is to make expertise readily available throughout the company, thereby enhancing the corporate brain and making its distributed intelligence work well. This is significant because GE’s success has been described as "the story of how GE leverages its intellectual capital" (Stewart 1999, 124) and how it uses its "collective brainpower" (McKelvey 1999, 12). The rule also speeds up microcoevolution within the firm (McKelvey 1999, 11), and it has been noted that "there is nothing special about [GE’s] changes except the speed with which GE does them" (Stewart 1999, 127).
2. Your business unit must be No. 1 or 2 in its industry — or else it will be divested. Each division, department or other group has its own set of criteria and ranking table against which competitive performance is assessed. McKelvey sees this as an effective way of focusing adaptive tension on the difference between where the company is now and where it needs to be to stay ahead of the competition. This focus helps managers to bring the rule to life by exploring what it means for everyone in that group to be number 1 or 2 in its sector.
McKelvey also emphasised that GE systematically creates many learning opportunities and puts people in positions where they might fail, for example by moving them around the company as part of its strategy of encouraging innovation and breaking down barriers to information sharing. Experimentation is explicitly encouraged through the establishment of popcorn stands, where new ideas can be tried out without affecting the rest of the business. If a popcorn stand doesn’t go well, it is killed off quickly and effectively. If a stand succeeds, the anti-hoarding rule ensures the good idea is spread very quickly. The aim is to have lots of popcorn stands where new ideas can be tried out in safe places.
McKelvey knows of no other firm which has got as close as GE to behaving like a complex adaptive learning model, in which each agent tries to learn incrementally, no agent goes to sleep and all agents have incentives to continue learning incrementally. He believes GE’s most valuable lesson for other CEO is that new ideas, their acceptance and practice can be introduced quickly into the corporate network by using simple rules that help design the firm’s ‘Bénard cell’ box for setting adaptive tension — rather than having top management spending much effort in trying to design the whole firm in detail. However, he noted that nobody had yet been able to copy GE, although it had widely promoted its strategy and basic rules. The GE case, and the difficulties of copying its approach, generated much audience debate (see section Main themes in audience discussions).
Summary: A New Science approach to leadership
McKelvey believes complexity science offers valuable fresh insights into the process which sets up the conditions for emergent order in firms, offering the potential for developing tools, techniques, processes, structures and rules to support a New Science leadership approach. This could help managers to create the conditions for the likely emergence of effective rent-seeking behaviour, without creating the stultifying barriers that often result from a command-and-control management style.
He explained how organisation science literature has moved away from an early 1990s emphasis on 'distributed leadership', involving employee empowerment and self-directed teams, to a preference for 'upper echelon' management headed by a charismatic visionary leader and supported by multiple management layers (McKelvey 1999, 17-22). There is also a belief that improved performance in today's current volatile environment needs the kind of heroic, charismatic visionary leader who Bennis (1996) describes as creating a "compelling overarching vision" about doing the right things and specifying the behaviour which fits that vision.
McKelvey warned that such leadership is likely to freeze attempts to promote distributed intelligence and is antithetical to human and social capital formation. He identified some key principles on which more effective New Science leadership advances could be built (McKelvey 1999, 37-42). These include:
* Economic rents and competitive advantage depend on human and social capital.
* High-velocity and hypercompetitive contexts require rapid microcoevolution of human and social capital.
* Rapid microcoevolution of distributed intelligence forms the basis of novelty that delivers competitive advantage.
* Current leadership theories are more likely to suppress than enhance distributed intelligence.
* The critical values of adaptive tension in firms define the complexity region that stimulates the emergent social capital networks necessary for improving distributed intelligence and rent-seeking behaviour.
Main themes in audience discussions
1. The GE success story, and why it has had no imitators
Is Welch's leadership style at GE a good exemplar?
Many participants were concerned that Jack Welch's management style at GE is of a 'visionary leadership' style that McKelvey generally criticised. The apparently ruthless treatment of people who step out of line at GE was mentioned as a sign of a rigid approach.
McKelvey accepted that some aspects of Welch's strong charismatic leadership could be seen in this way. However, he believes Welch is significantly different to a stereotypical autocratic leader because he establishes a proper process and incentives, rather than trying to command-and-control by specifying detailed behavioural rules.
Some participants felt McKelvey's emphasis on Welch might overstate the role of leadership. It was pointed out that the selection process through natural evolution means we may be winners at one moment, but doesn't necessarily mean we are the leaders always. One leader might emerge in a situation, but someone else could have done an equally good job. The duck at the front might be said to be the leader because it is the duck in front, not because it is the leader.
There was also a general consensus that a command-and-control management approach kills emergence. However, if formal structures collapse there can be no leader or sense of sense-making. Leaders are, therefore, faced with a dilemma: they must create structure to foster sensemaking, but if they create structure through command-and-control they will shut down emergence and suppress distributed sensemaking. This can be overcome by setting up only a small amount of structure, which McKelvey said had been done at GE.
Are some kinds of behaviour inimitable?
GE has been open in explaining its approach, but McKelvey is intrigued that GE’s approach had not yet been applied successfully within any other firms. One participant pointed out that cultural anthropologist have known for a long time that behaviour as fine grained as how a tribe lives, or how a football team plays, is inimitable. The reason why other teams can't play like Manchester United is not just because they don't also have great players. Manchester United is a community of practice built over many years of learning-by-doing the fine detail of how players act on the field to make the rules work for them. Simple rules may be important, but they matter most in context. Transferring them between contexts is a very different question to, say, the way consultants routinise certain types of knowledge and standard processes.
Another participant highlighted the as yet unresolved difficulty of capturing the difference between the rules contained in a CEO's strategy and those that arise through natural selection. They are similar and interlinked, but slightly different. The significance of context was also mentioned, for example in what was said to be the 'dire' situation at GE when Welch took over. This gave people in the firm only a 'change or die' choice, which left them no alternative to going along with Welch when he asked them to start a new community of practice by following new sorts of rules. This also highlighted the significance of initial conditions in any context. The emphasis given by participants to treating systems as highly contingent on context was seen by some as indicative of a general European scepticism about visionary leadership.
Problems about communicating rules were also explored in relation to fast-food franchises, in a discussion triggered by the question: 'If McDonalds can communicate its rules effectively across its franchises, why haven't its competitors also learnt the rules?' One response was that McDonalds' rules were difficult to copy because they are embedded holistically throughout the organisation, not just at operational level. Its franchises get substantial capital resources, together with much marketing and logistics support across a routinised value chain. This gives McDonalds' corporate culture a differential which can be matched by few fast-food operations. McKelvey noted that rules at McDonalds are aimed mainly at maintaining the status quo, but at GE they seek to create an adaptive learning system.
2. New Science leadership: opportunities and constraints
Is anyone really trying to tune organisations with complexity science?
McKelvey highlighted a survey of books about applying complexity science to management (Maguire and McKelvey 1999). This found that most books started with a hook about firms facing uncertain, nonlinear, rapidly changing environments. As complexity science and chaos theory deal with such environments, the books argue that managers should look to these disciplines for solutions to their strategic challenges. The books then typically discuss emergence, equating it with empowerment. From then on, McKelvey commented, the books generally read like organisation theory of the past thirty years — and could be more useful if they took out all references to complexity science, which they used primarily as a metaphor. However, one participant felt strongly that few specialists who apply complexity science to organisations were actually trying to tune whole organisations. Instead, the participant's experience was that they were more concerned with applying complexity science to issues such as tuning a conversation, dialogue, workshop or planning meeting, with an emphasis on speed-of-change rather than nonlinearity.
Limitations on what managers can achieve through adaptive tension
A participant asked how far the adaptive tension level is determined by the turbulence of the general market, to which firms can only react. McKelvey said the market defines adaptive tension, but the position of a firm in the market is determined by rules within the company. Nevertheless, managers still have much scope to maximise the space in firms where emergent behaviour is likely to occur. McKelvey noted that all organisations aren't equally capable of producing emergence after setting up the required adaptive tension. However, there are many consultants and internal agents who know how to coach people to generate networks, team skills and other structures and activities which foster effective emergence.
Although the GE rules mentioned by McKelvey might help to suggest how complex adaptive behaviour can be promoted, one participant said there is still relatively little detailed knowledge about which rules give rise to such behaviour in physical systems — and that we are even further away from knowing the answers in social systems. The participant added that it takes a long time to locate the region of emergence in physical systems, which is sometimes very small and difficult to expand in a realistic way. This view raises significant questions about whether, and by how much, people can influence emergent complex adaptive behaviour.
However, a participant who had expressed doubts about the ability of managers to change adaptive tension also emphasised the importance of pursuing the kind of approach advocated by McKelvey. This participant emphasised that managing firms at the edge of chaos in the complex adaptive region is vital when a phase transition is required, but current ways of achieving this are too limited and unimaginative. Complex adaptive systems offer the prospect of a faster, smoother engine of transition that opens previously unthought of possibilities for moving away from linear mechanistic behaviour to a more interesting point in the chaotic region.
Stars versus networks: getting the right balance in human and social capital
There was much debate about finding the right balance between human capital 'stars' and social capital networks. McKelvey said stars are seen as being generally better at creating new products or services, but effective networks also create innovation and offer vital support to stars. He noted that a CEO can relatively easily identify the need for a star and know how to get one. However, there is no similar toolkit to facilitate the development of an efficient network. In a discussion on team working, it was suggested that teams can be seen as a way of making ordinary people become spectacular, which is equivalent to a team becoming a 'genius'. There were warnings against allowing team or individual 'geniuses' to become isolated from the rest of the firm.
Some participants felt that McKelvey's concept of 'D' as a function to optimise human and social capital goes against complexity science's insight into the way optimising algorithms are generally bounded by relatively linear, nondynamic systems which greatly under represent the potential to deliver change and variety. Focusing the search for a proper complexity-based management approach on trying to find an optimising algorithm seemed to be a contradiction in terms to some participants.
The emphasis on D was also questioned in terms of the way the right balance between human and social capital is contingent on historical background and the functions performed. For example, a small biotech company can generate rents primarily by stars producing new patents and other innovations. In customer service, on the other hand, star-driven entrepreneurship may be the opposite of what is needed because customers should be responded to consistently at all times. As firms grow and move into different markets or business phases, they may also behave in different ways.
McKelvey accepted these were valid points which will make him re-assess his use of D. He agreed that there is a danger that D could be seen purely as a point attractor based on a prescriptive set of rules. However, he stressed his main aim is to encourage firms to use point attractors only as a mechanism to set adaptive tension that acts like a strange attractor in moving people towards a desired direction without prescribing their detailed behaviour.
The importance of sensemaking
A number of participants highlighted the value of having mechanisms which make people in a firm believe they can make sense of their environment (Weick 1993; 1995). This was illustrated by an example quoted by Weick. During the Austro-Hungarian war, soldiers were lost on a snowy mountain pass until they found a map in the rucksack of one of them. When the soldiers were being court marshalled, it was claimed their story didn't hold up because the map was of a different mountain.
The key point for management highlighted by this story is that the soldiers needed to believe there was map around which they could organise with a sense of order. The role of leadership can be seen as giving people this belief that there is a way of finding a way out of any current problem by organising around what they know. A good leader should be able to do that without a lot of structure. Symbolic communication in visionary leadership was seen as a potent force in helping people to make sense for themselves because symbols are capable of interpretation in many ways, without imposing any structures.
McKelvey noted that managers can create the urgency needed to generate the right level of adaptive tension by being highly visible: walking around and meeting colleagues to discuss what the strategy is trying to achieve, ensuring people know where they need to be in the future, and finding out what they are doing to get there.
Making the most of local knowledge
One participant also stressed the significant contribution which can be made by individuals and groups with detailed local operational knowledge and expertise. This enables experienced people to know which 'levers to press' to make something change. People with this expertise have deep insights into the way things work and how to move the behaviour of a team to a new point, so a leader who sets a new goal for a business unit should acknowledge that the people who live and work there know best how to achieve them.
Minimising risks when moving companies into the region of emergence
McKelvey said moving a company from where it is to where you want it to be is risky: you can never know beforehand at which critical values a firm will tip from a region of emergence into chaos and become a completely disrupted organisation. It was also pointed out that a phase transition in the form of cancer can mean death, not just new forms of life. This suggests there should be 'pockets' in an organisation where experiments can be made with different adaptive tensions, for example by implanting a set of rules in a new subsidiary that can change them to see what happens. Such experiments, as with GE’s ‘popcorn stands, shouldn't disrupt the parent organisation, but if successful they could move the whole business to new levels of performance.
Many participants urged caution in deciding where attempts are made to induce complex adaptive behaviour in a firm. For example, it was suggested that such behaviour would not be welcomed in naturally bureaucratic activities, such as payroll. More task-based or person-based cultures where innovation and policy-making happen might be more appropriate places for breaking all the moulds by making complex adaptive behaviour the norm. However, it was pointed out that great savings and competitive gains can still be made in apparently dull areas, but these are likely to come by using other techniques. McKelvey stressed that he aimed to steer CEOs towards the most likely rent-generating places. These are usually not in paper-based systems, but in new product areas, new markets, new technologies, and new approaches to markets.
Measuring emergent order and an organisation’s metabolism rate
Participants were interested in knowing how emergent order could be measured. There was general agreement that profits alone cannot be a suitable yardstick, although the results of emergence have to make a bottom line impact eventually. McKelvey said managers should look closely for signs of any kind of new emergent structure or outcome. It may not be easy to make these visible, but any relevant evidence is an important indicator of the response to the adaptive tension within an organisation. Typical questions which can help to tease out examples of complex adaptive behaviour and help to assess and organisation’s metabolism include:
-
How many new teams or groups have emerged in the last month?
-
How many networks have emerged within groups, between groups, between alien and hostile people, between marketing, production and engineering?
-
How many novel collective outcomes have there been in addition to new networks?
-
How many new products are we running each year?
Can strong ties be good for smaller firms?
A query was raised about McKelvey's claim that weak ties create greater novelty. A participant said there was much evidence that the fastest growing, most entrepreneurial and innovative smaller firms are usually those with a clear strategy of fostering strong long-term ties with other stakeholders, such a customers and suppliers. It was important, therefore, not to denigrate strong ties as such. McKelvey pointed to research (Aldrich 1999) which showed that the entrepreneurs who develop networks more quickly are more likely to survive. His own view is that those who worked better in a weak tie framework are most likely to be successful.
One participant said a distinction should be made between innovation and invention. Smaller firms which innovate fastest tend to have very strong ties with the supply chain, often with previous firms where the entrepreneur worked; inventiveness is more related to weak tie networks. To be successful in the long term, a small firm should have enough weak ties to a whole range of other emerging possibilities to allow itself to adapt to changing circumstances over the long term. Strong-tie networks can cement a firm into one supply chain, one set of products and one set of relationships that eventually kills the company as it can do only one thing well.
3. Applying complexity science to social systems
The relevance of physical complexity science findings to social systems
Some questions were raised about McKelvey's application of certain complexity science concepts because their use in the context of social systems did not seem to reflect their original meanings in physical systems. For example, one participant noted that in biology and chemistry the edge of chaos is seen as a region where phase changes just happen. No explanation is given for this transition. There is an acceptance that this is the way things are: the way a tree branches, leaves are made, clouds are formed. Biologists and chemists don't talk of being able to set boundaries on, or change the dimensions of, the region of emergence. The participant also felt oscillating events could be seen as the standard stretch-and-pull mechanics of physical systems and not something to be avoided, which was what McKelvey had suggested when discussing oscillations between bureaucratic and chaotic behaviour in firms. McKelvey agreed, but pointed out that Kauffman's (1993) framework discusses critical values which create 'bureaucratic' types of behaviour.
A participant also noted that strange attractors are defined in the complexity literature as limit cycles describing patterns of behaviour, rather than causing them. McKelvey, on the other hand, seemed to treat strange attractors as causal in terms of creating adaptive tension to achieve a desired goal. McKelvey agreed that his approach to adaptive tension involved setting point-attractor goals. He noted that CEOs like setting up point attractors, so it is an advantage to get them to focus on ones aimed at managing adaptive tension to allow people to decide for themselves how to fulfil the firm's strategic goals. However, he acknowledged that point attractors must be watched carefully to ensure they do not suck a firm into a maladaptive niche that isn't going anywhere.
McKelvey denied his departures from strict physical complexity theory meant he was using complexity science as a metaphor. He said phase transitions in firms happen as real events with symptoms. However, there is no number which will apply in any given firm in the way that Kauffman's cellular automata could identify a region of emergence of around .02. In creating a more flexible definition appropriate to social systems, he said he was giving up only minor details such as this number — which in any case is far too small to give a firm a chance of creating emergent order — but he was keeping phase transitions, energy differentials, critical values, attractors, adaptive tension and other fundamental complexity science concepts.
McKelvey also cited an example from computational modelling to highlight the specific value of using complexity science ideas. He explained that most sciences assume uniform homogeneity for fundamental units, such as atoms or quantum states in physics, molecules in chemistry and rational actors in economics. Yet complexity science accepts that its bottom layer is heterogeneous, stochastic and idiosyncratic. This is vital for organisation science because its fundamental units are people, and many activities start with random idiosyncratic transactions. From this perspective, a conversation could be seen as a fine-grained structure in which participants quickly get immersed in each other's backgrounds, personality, politics and priorities. The emergence of order from these conditions would require the application of an appropriate external adaptive-tension force.
There was some discussion about the degree to which complex adaptive behaviour can be self-generating. McKelvey said there could be a spontaneous emergence of informal organisations and teams, although managers are often not aware of the emergence when it happens. The main thrust of his presentation, however, argued that complexity science can help managers to get a grip on using adaptive tension as an externally generated force to orient emergence towards the directions the organisation needs to go. This approach is supported by Ashby's (1962) explanation that there is no order or organisation until a constraint is imposed from the outside.
Negative connotations of 'corruption' and 'cliques in entanglement discussions
Much concern was expressed at what was felt to be the unnecessarily derogatory implications of McKelvey's use of the terms 'corruption' and 'cliques' in his explanation of the difference between entanglement fields in organisations and the 'purity' of quantum entanglements. Many participants said cliques hold the potential for rapid and effective change, if the strength of the clique can be used to effect change. One argued that the universe is as 'lumpy' in social sciences as it is in physics and biology, so there is no need to think that 'corrupted' entanglement is a distinguishing feature of the social sciences. McKelvey agreed that it is possible to draw on strong face-to-face ties to get rapid changes that harness the energy of cliques, so a more neutral term might be preferable. However, he said that people in firms cannot evolve in ways independent of their past histories because of entanglement corruptions in the form of cliques, departments and other groups.
The value of computational complex adaptive models
McKelvey believes it is important for social scientists to appreciate the benefits of computational models because they include assumptions which fit well with behavioural science. For instance, computational models can deal with idiosyncratic heterogeneous agents on a stochastic basis. Modern computers can also run increasingly complicated models, with a million or more agents in various sophisticated forms of connections. Agents can be defined at multiple levels, such as neurons, molecules, conversational elements, teams or any other entity. They can be made responsible for incremental learning and adaptive improvements, although that kind of behaviour is not easy to get with real people. However, he acknowledged that at present his approach could be seen as a 'black box' solution because the equivalent of a flight simulator has not yet been developed to help explore what goes on inside the region of emergence for strategic decision making.
One participant commented that computers can do complex adaptive behaviour modelling better than any other approach, but warned that this is still a relatively small part of computational modelling. It is much easier to apply a computational model to a utility function, such as maximising shareholder value in terms of costs, rather than trying to optimise the configuration of an innovative social network to maximise sharing. McKelvey recognised much work still needs to be done, but is confident that computational modelling is rising towards being a rule-based science, despite the huge number of independent variables that need to be modelled. He also stressed that the modelling of social systems should take account of the need to have a dampening function to make a system work, which he said has been forgotten in many social science models.
4. Taking account of a broad range of stakeholders and wider social environment
McKelvey's presentation emphasised the role of shareholders-seeking-rents as a prime selection pressure on CEOs. However, participants pointed out that a more complex set of stakeholders are also involved, such as customers and employees. McKelvey agreed that rent-seeking behaviour is probably too simplistic a focus, although it is crucially important in business competitiveness. He said the aim should be to create a system which learns to satisfy a whole series of stakeholder needs. McKelvey’s own prime interest is the firm and the role of CEOs within that.
Towards the end of the seminar, there was a lengthy discussion about the need for investigations into applying complexity science to the wider environment of non-economic, non-tradable entities that form the context in which firms and CEOs are subsystems.
One participant strongly argued that the vast effort and resources going into the qualitative understanding of the firm and marketplace dynamics is very limited because it rarely deals with the important consequences for the 'mega' system it is part of — where social dimensions can be as constraining as those relating to the physical environment. It is from these broader systems where constraints might be most likely to come from to limit the rapid technological innovation and economic growth that might seem to be an infinitely expanding process from the firm-only perspective. If this narrow focus is maintained, firms could find themselves embedded in dysfunctional societies which will no longer be able to provide a stable system in which products and services can be sold. The participant would therefore like to see more effort going into understanding what economists call 'externalities', noting that the traditional view that these externalities can be ignored is one of the worst flaws of 20th Century economics.
Another participant said the 'mega' social factors could become accelerators, rather than constraints, if firms treat everyone in the organisation as human and social creatures and find imaginative ways of co-producing with service recipients. Social capital is not purely internal to an organisations, so seeking to influence the wider 'community capital' on which it is built is a natural and sensible rent-seeking activity. A number of participants thought there are signs that companies are recognising this by addressing issues not directly concerned with business, such as ethics and reputation, because they are having an impact on competitive performance.
Bibliography
Aldrich H. (1999), Organizations Evolving, Thousand Oaks, California: Sage
Ashby W. R. (1962), ‘Principles of the Self-Organizing System’, in von Foerster H. and Zopf G. W. (eds.), Principles of Self-Organization, New York: Pergamon, 255–278
Becker G. S. (1975), Human Capital, Chicago: University of Chicago Press
Bennis, W. G. (1996), ‘Becoming a Leader of Leaders’, in Gibson J. J., The Senses Considered as Perceptual Systems, New York: Houghton Mifflin
Brown S. L. and Eisenhardt K. M. (1998), Competing on the Edge: Strategy as Structured Chaos, Boston: Harvard Business School Press
Burt R. S. (1992), Structural Holes: The Social Structure of Competition, Cambridge, Mass: Harvard University Press
Cohen J. and Stewart I. (1994), The Collapse of Chaos: Discovering Simplicity in a Complex World, New York: Viking
The Economist (1999), 'General Electric: The House That Jack Built', 352, September 18–24, 23–26.
Gell-Mann M. (1994), The Quark and the Jaguar: Adventures in the Simple and the Complex, San Francisco: W. H. Freeman
Granovetter M. (1973), ‘The Strength of Weak Ties,’ American Journal of Sociology, 78, 1360–80
Granovetter M. (1982), ‘The Strength of Weak Ties: A Network Theory Revisited’, in Marsden P. V. and Lin N. (1982) (eds), Social Structures and Network Analysis, Beverly Hills, California: Sage,105–130
Kauffman S. A. (1993), The Origins of Order: Self-Organization and Selection in Evolution, New York: Oxford University Press.
Lorenz E. N. (1963), ‘Deterministic Nonperiodic Flow’, Journal of the Atmospheric Sciences, 20, 130–141
Maguire S. and McKelvey B. (1999) (eds.), Special Issue on Complexity and Management: Where Are We?, Emergence 1(2)
Mainzer K. (1996), Thinking in Complexity: The Complex Dynamics of Matter, Mind and Mankind, London: Springer-Verlag
McKelvey B. (1999), 'Dynamics of New Science "Macro" Leadership: Strategy, Microcoevolution, Distributed Intelligence, Complexity', Draft #28, November 9, London: LSE Complexity Website (http://www.lse.ac.uk/complex|)
McKelvey B. (2000), 'The "3 rd Law" in Physical, Biological and Social Systems:
Complexity Science vs the Entanglement Trap—in Firms', Draft #8, February 7, London: LSE Complexity Website (http://www.lse.ac.uk/complex|)
Omnès R. (1999), Understanding Quantum Mechanics, Princeton, New Jersey: Princeton University Press
Prigogine I. (1962), Non-Equilibrium Statistical Mechanics, New York: Wiley Interscience
Stewart T. A. (1999), 'See Jack Run Europe', Fortune, 140, 124–136.
Weick K. E. (1993), 'The Collapse of Sensemaking in Organizations: The Mann Gulch Disaster', Administrative Science Quarterly, 38, 628–652.
Weick K. E. (1995), Sensemaking in Organizations. Thousand Oaks, California: Sage.
Appendix 1: Complexity Study Group and Seminar Participants, 14 March 2000
Professor Bill McKelvey (speaker), Anderson Graduate School of Management, UCLA
Steve Nicholson (chair), Director, Delta-Y Shared Learning Group
Arthur Battram, Research Associate, Aston Business School
Tony Bovaird, Director PSMRC, Aston Business School
Julian Burton, Complexity Group, LSE
Didier Clement, Management Consultant, DCmc
Eileen Conn, Independent Academic, Living Systems Research
Anne Deering, Vice President, A.T. Kearney Limited
Mike Duncan, Applied Technology, Glaxo Wellcome
Ted Fuller, Director, Foresight Research Centre, University of Durham
Kate Hopkinson, Director, Inner Skills Consultancy Ltd
Cristian Hormazabal, MSc Political Theory, LSE
Professor Gerry Johnson, Cranfield University
Angelique Keene, Student, University of Essex
Bob Malcolm (Seminar only), Managing Director, ideo ltd, Co-ordinator, EPSRC Systems
Integration Initiative
Ian McCarthy (Study Group only), Principal Fellow, University of Warwick
Simon Morley, Complexity Group, LSE
Maria Papaefthimiou, Complexity Group, LSE
Malcolm Peltu, Independent Editorial Consultant, London
Julian Pratt, Research Associate, LSE
Jyoti Bachani Rahi (Study Group only), PhD Student, London Business School
Duska Rosenberg (Study Group only), Senior Lecturer, Royal Holloway, University of
London, Egham
Professor Jonathan Rosenhead, LSE
Bob Snowdon (Seminar only), Teamware Group
Stephen Sheard (Seminar only), Lecturer in Corporate Strategy, Middlesex University
Sandy Taylor (Seminar only), City Economic Advisor, Birmingham City Council
Ian Turner, Director of Graduate Business Studies, Henley Management College
David Wasdell, Director/Trustee, Meridian Programme
Appendix 2: Glossary
Adaptive tension Imported energy that causes emergent order, for example the Lorenz energy differentials in the atmosphere which generate weather systems.
Agency problem When people in firms go in different directions doing their own things.
Bénard cell Device with a liquid sandwiched between metal plates which creates movement in the liquid via the energy differential between hot and cold plates.
Bifurcated structures What emerges when dissipative structures in a far-from-equilibrium condition at the edge of chaos split into two modes of behaviour because fluctuations affecting the system have become too powerful.
Coarse grained structures Emergent higher-level structures indicating the emergence of order.
Coevolution The joint evolution of two or more entities that have close ecological relationships but do not exchange genes.
Complex adaptive behaviour Exhibited by complex systems which adapt through a process of self-organisation and selection.
Complex adaptive learning. Model in which each agent learns incrementally and all agents have incentives to continue learning.
Corporate brain The network of communications among people in a firm.
Critical values Levels in the energy differential between warmer and cooler surfaces which affect the velocity of the flow of a liquid or gas in-between.
D What McKelvey calls the optimum balance between human (H) and social (S) capital in the formula D = H + S.
Dissipative structure Open system exchanging energy, matter or information with its environment in a way that maintains its integrity while it is far from equilibrium.
Distributed intelligence Result of interactions between nodes in the corporate brain.
Edge of chaos. Where order emerges.
Emergence The appearance of a higher-level structure, typically with novel and unanticipated behaviours, which cannot be directly deduced from the lower-level properties of the system from which it emerged.
Energy differential In a physical system, the difference between the energy levels at hot and cold surfaces. Applied to organisations in setting adaptive tension levels.
Entanglement What exists before order emerges in the physical world.
Human capital The skills, knowledge and experience of individuals in a firm.
Microcoevolution Coevolution within an organisation.
New Science leadership Business management strategy that makes practical use of complexity science, chaos theory and complex adaptive modelling.
Organisational metabolism The speed at which a firm works
Point attractor A characteristic feature of a system which has some kind of equilibrium position.
Social capital The networks established between people in an organisation
Rent Levels of profit above the industry average; rent appropriation Gained by acquiring a company with a rent generation capability; rent generation Profits derived from a firm's own competitive activities.
Region of emergence Space where emergence occurs, between the critical values defining adaptive tension.
Strange attractor Exists in a system that oscillates between paths, without settling at an equilibrium point.