Business at a Crossroads The Crisis of Corporate Leadership_7 pdf

120 Business aT a Crossroads
effective way to run large enterprises. The question we turn to in the
next chapter is whether other forms of enterprise could emerge to chal-
lenge CEO-led companies.
References
1 “Bankers Can Learn from Sport about Fair Play on Pay,” Financial Times, February 28,
2009.
2 Times Books, 2002.
3 “Thain Behavior. Bankers Need to Start Seeing Themselves as Others See Them.” Finan-
cial Times, January 24, 2009.
4 “Why do Dominant Personalities Attain Influence in Face-to-face Groups? The Competence-
Signaling Effects of Trait Dominance.” Journal of Personality and Social Psychology, 2009,
Volume 96, Issue 2.
5 Time, February 11, 2009.
6 “Seven Ways to Attract Analysts and Investors,” Corporate Board Member, Spring, 2001.
7 Simon & Schuster, 1989.
8 “Empowerment: The Emperor’s New Clothes,” Harvard Business Review, May–June
1998.
9 Allen & Unwin, 2006.
9780230_230941_08_cha06.indd 120 09/09/2009 10:01
PART II
Reforming big business
9780230_230941_09_cha07.indd 121 09/09/2009 10:02
This page intentionally left blank
123
7 Leaderless competitors
The hierarchical corporation run by, or rather under the control of, an
omnipotent CEO, is not the only form of enterprise making a decent
living these days. There are well-established traditions of dual leader-
ship in publishing (editor/publisher) and the theatre (director/
producer). Joint ventures, strategic alliances and other inter-firm collab-

orations comprise another class of enterprise in which control is in more
than one pair of hands (see Chapter 9) and the partnership model still
thrives in professional services, such as law and accountancy.
By and large, professional partnerships seem to have done a better
job at adapting to what people want from their work, and in their work-
places, than large joint stock companies. KPMG in the U.K., the U.K.
member of one of the “big four” global accounting networks and an
organization I happen to know quite well, was ranked “the best big
company to work for,” in the Sunday Times 2009 “Best companies to
work for” survey. It was the third time KPMG had come first in the
five-year-old ranking. It has never come lower than third. Two other
“big four” firms, Deloitte and PricewaterhouseCoopers, also made it
into the top 20 big company ranking.
1
To begin our examination of what can be done to adapt big business
to its more challenging environment, this chapter will discuss other enter-
prise forms that are emerging from the undergrowth, which may chal-
lenge and, in some cases, are already challenging, the CEO-led MuBE.
The large CEO-led company is vulnerable to challenge, because it’s
not as solidly based on foundations of contemporary economic logic as
it appears. We have already discussed, in earlier chapters, its accidental
origins and contingency – its emergence, in the mid-19th century, from
a number of fortuitous circumstances that prevail no longer.
Moreover, a case can be made for arguing that the MuBE wasn’t just
the creature of the economic circumstances prevailing at the time of its
birth, but was also a social construction that emerged from the personal
prejudices, interests, relationships and ambitions of its creators, as much
as from those natural economic laws to which Adam Smith attributed
the rise of capitalism.
9780230_230941_09_cha07.indd 123 09/09/2009 10:02

124 BUSINESS AT A CROSSROADS
To paraphrase Khurana when he describes the external CEO market
as a “social construction,”
2
to call the CEO-led MuBE a socially
constructed institution is not to say that it is an institution in the normal
sense. It is an institution in the sociologist’s sense: a pattern of practices,
relationships and obligations so taken for granted that they assume the
status of rules governing thought and action.
This isn’t to say there are no natural economic laws. Any business
institution, whether or not it’s socially constructed, must comply with
fundamental economic logic if it is to survive and prosper. But there is
scope, within that constraint, for a wide variety of enterprise forms, all
of which will be socially constructed to some extent. As Khurana
pointed out, one of the most important lessons of the sociology of
knowledge is that “very little in society had to be the way it is.” An
institution that was socially constructed in one way could have been
constructed in many other ways. “Part of the process of social construc-
tion is to camouflage this fact, since society becomes more stable, when
people accept institutions as simply given and share a common explana-
tion for events.”
Roads not taken
An irony in the evolution of forms of enterprise is that while the early
American railroads were producing Chandler’s MuBE (multi-unit busi-
ness enterprise) they were experimenting with a different form of enter-
prise which, had it been allowed to develop, may well have stopped the
MuBE in its tracks, so to speak.
3
The weakness of the early railroad systems was that roads entering
a terminal city from different directions had no direct rail links and,

because they used different gauges and equipment, the cars of one
road could not be transferred to the track of another. In 1865 the
Boston Board of Trade put the cost of unloading and reloading freight
between Boston and Chicago at over $500,000 a year. These high
trans-shipment costs were in no one’s interests, so the roads got
together and agreed to standardize gauges, equipment and proce-
dures. As a result of this inter-firm cooperation, by 1880 a rail ship-
ment could travel from one part of the country to another without a
single trans-shipment. But standardization increased the intensity of
competition between roads. Because railroads are capital-intensive,
they have steeply declining marginal costs and their economics are
extremely volume-sensitive. A small change in traffic could turn a
9780230_230941_09_cha07.indd 124 09/09/2009 10:02
7 LEADERLESS COMPETITORS 125
profitable line into a loss-maker and vice versa. Railroad managers
were, for this reason, under considerable pressure to poach business
from rival lines by aggressive advertising and rate cutting. Such
competition is dangerous in a highly capital-intensive industry, because
once the line has been built, it is worth taking new business at prices
that cover variable costs. Since a road’s variable costs were only a small
proportion of its total costs competitive rate-cutting, if left unchecked,
would continue to way below breakeven. The solution was for rival
lines to cooperate on setting rates in the same way as adjacent lines
had cooperated on connections.
Federations were established with their own legislative, executive
and judicial bodies to set standard rates. When they too failed to prevent
periodic outbreaks of rate cutting, the major trunk lines signed various
agreements, culminating in the formation in 1878 of the Joint Execu-
tive Committee, chaired by Albert Fink, to approve rates calculated by
sub-committees and associations throughout the country. Fink believed

this cartel arrangement was the only way to prevent the “centralization
and absorption of the roads under the absolute control of one or a few
persons. It makes the separate, individual existence of these roads
possible … puts a check on the consolidation of [the industry, and]
secures all the advantages of consolidation without its disadvantages.”
But he realized this solution relied on “the intelligence and good faith
of the parties composing it” and when he and his allies were unable to
persuade Congress to give his committees’ rulings legal sanction, the
whole system of associations and sub-committees fell apart, and it was
just a matter of time before the “centralization and absorption” Fink
feared became a reality.
Chandler admitted “such co-operation might have worked” –
managers might have been more rigorous in maintaining rates, and
might have worked more closely with Fink in seeking out and fining
violators of agreements. It’s certain that if the cartel agreements had
been legally enforceable the costs of breaking them would have been far
higher. “Given the basic nature of railroad competition,” Chandler
concluded, a “legalization of the cartel arrangements was probably the
only effective method to control competition and so remove the incen-
tive for system building.” Sadly for this early flowering of a form of
enterprise which might have challenged the MuBE for the position of
“boss” institution in business, Congress was in no mood to sanction
what seemed, to most Americans, to be price-rigging. The 1887 Inter-
state Commerce Act outlawed the Joint Executive Committee’s pooling
arrangements and the scene was set for the epic battle between the
9780230_230941_09_cha07.indd 125 09/09/2009 10:02
126 BUSINESS AT A CROSSROADS
speculators, led by the “Mephistopheles of Wall Street,” Jay Gould, and
the “system builders,” led by the Vanderbilts.
Insofar as Fink’s failures opened the way for system builders, the

MuBE itself, like the CEO market (see Chapter 5), can be seen as a
social construction built not by inexorable economic logic, but by a
scheduler’s blunder on the Western Railroad in 1841, the failure of the
American railroads to adopt the telegraph earlier, the lack of price disci-
pline in Fink’s cartel, Congress’s refusal to give legal sanction to the
cartel’s price agreements, and the prejudice against price-fixing, which
led to the Interstate Commerce Act.
The pooling arrangements Americans had objected to as price-fixing
(and Congress duly outlawed) effectively developed anyway, within the
“systems” built by Cornelius Vanderbilt and others. They were required
by the industry’s capital-intensity. When effecting them by a cartel
within a disintegrated industry was proscribed, system building became
the only option.
The U.S. electricity utility industry is also a socially constructed
institution, according to economic sociologists Mark Granovetter,
Patrick McGuire and Michael Schwartz. In their paper Thomas
Edison and the Social Construction of the Early Electricity Industry in
America,
4
they argued that, when the industry was born in the
1880s, three development roads were open: power generation at the
household, or neighborhood level; public ownership of generation
and distribution grids; and private companies, serving large areas
from central power stations. That the U.S. electricity industry took
the third road was not due, according to Granovetter, McGuire and
Schwartz, to any compelling economic case for such an arrange-
ment, but to interactions within a social network consisting of trade
associations, interlocking directorships, and generating equipment
manufacturers.
Another road not taken was the plan to split “Big Blue” (nickname

of IBM, until it was robbed of the PC market it had dominated by a
“multi-agent enterprise”; see under Linux below) into a brood of “Baby
Blues.” Louis Gerstner abandoned the Balkanization plan when
appointed CEO of IBM in 1993, and gained much kudos for turning
the ailing giant into a different, and subsequently successful, IT services
company. But who’s to say the Baby Blues would not have, collectively,
been equally if not more successful?
9780230_230941_09_cha07.indd 126 09/09/2009 10:02
7 LEADERLESS COMPETITORS 127
A fractured landscape
If you asked an alien economist from a distant star system how the
modern, CEO-led company could be adapted, to suit today’s business
environment, the dude might be either cryptic, and say: “Chop off their
heads,” or unhelpful, and say: “If it is adaptation you want don’t start
from here.”
Where would you start? If you had a clean sheet of paper how would
you go about constructing a business institution perfectly adapted to
the modern environment?
The first step, if you’re persuaded that the institution would, as well
as complying with natural economic laws, need to be “socially
constructed” to some extent, would be to list the qualities that most
people want in their work and workplaces. The list suggested in Chapter
1 included “free,” “fair,” “reasonable” and “decent,” and distributions
of primary goods including income, wealth, power and the bases of
self-respect, consistent with Rawls’s “difference principle.”
The second step would be to consider the environment in which
such people with such desires will construct business institutions. One
quality that will be immediately apparent is that the environment is
infinitely more varied and complicated than the environments in
which the U.S. railroad and power industries chose their paths. Many

more roads are open, to many more people, with a wider variety of
skills and aptitudes. And modern communications, particularly the
internet, create a space for the social construction of a business insti-
tution that spans the globe. For today’s business institution builders it
is this enormous increase in complexity and the vastly greater number
of challenges and opportunities it brings with it, that distinguishes
most clearly their environment from that of the mid-19th and early
20th centuries.
To illustrate the importance of this difference and to introduce a set
of ideas, the sciences of complexity, that I believe are vital to any under-
standing of how new businesses will emerge and develop in the future,
consider Stuart Kauffman’s analysis of “patch” size on a “cost surface”
(see box below). Kauffman, whose conjecture about the origin of life I
quoted in Chapter 3 when discussing the origin of Chandler’s MuBE,
is a world-leading complexity scientist who pioneered the use of
complex systems methods to solve business problems.
5
9780230_230941_09_cha07.indd 127 09/09/2009 10:02
128 BUSINESS AT A CROSSROADS
Patches on landscapes
Kauffman says the size of the “patches” (in this context, units of organization or
departments) in a “system” (a business or company) travelling on a “cost surface” or
“landscape” (seeking to minimize costs, or maximize shareholder value), is a crucial
determinant of the entire system’s ability to find the global minimum (the lowest
possible cost, or the greatest possible value).
He says that, on complex cost surfaces, a system can be trapped in a poor local
minimum and that one way to avoid the trap, and allow the system to search a wider
space, is to introduce the equivalent of heat. A
physical system tends to change in
ways that lower its internal energy and take the system “downhill.” But occasionally

a system will change in a way that increases internal energy, and so allow the system
to escape a poor local minimum.
It
follows from this that if an organization wishes to avoid being trapped in poor
local minima, it should try to contrive occasional injections of heat that will make
the system ignore the “minimize cost” imperative and move “the wrong way”
(increase cost).
O
ne way to achieve this, is to change the size of patches that act independently
and selfishly.
S
elfish action is desirable, because it can move the whole system the
wrong way, and allow it to escape poor local minima.
A
s Kauffman puts it “well
chosen partitions can produce markedly enhanced optimization.”
Take
the case of an integrated system or organization, acting as a single depart-
ment.
I
t will accept all opportunities to change that take the system downhill toward
lower costs, but refuse all other moves.
Such systems always descend to a nearby
minimum, and become trapped there.
When the organization is divided into independent departments, the criteria for
screening opportunities change, because opportunities that can take the department
downhill will be accepted, even when they take neighboring departments uphill. An
integrated system in which all departments are “singing from the same hymn sheet”
can’t escape a poor local minimum, but a system divided up into patches will only
remain trapped in a poor local minimum, if it is a local minimum for each patch, which

is unlikely if the minimum is poor.
Th
e maths are difficult, but Kauffman shows that when connectivity (commu-
nication) between patches is low, energy falls as patch size increases. When
connectivity is high, however, systems are better at finding the global minimum
with smaller patches. Th
e degree of connectivity between the patches determines
the size of patch that will maximize the system’s ability to find the global minimum.
As connectivity increases the system approaches a tipping-point or cusp. It
w
orks best at first with large patches, but then suddenly flips and works best with
small patches.
9780230_230941_09_cha07.indd 128 09/09/2009 10:02
7 LEADERLESS COMPETITORS 129
One way to picture the implications of Kauffman’s patch size model
for business, is to imagine that companies travel to their futures across
“fitness landscapes.” Their mission is to climb the highest peak in their
landscapes, because only there will their “fitness,” whether measured in
terms of cost, profits, cash flow, shareholder value or some other vari-
able, be maximized.
In the past, the fitness landscape was even and although the climb
was strenuous, particularly close to the top, the goal was clear. A single
peak with smoothly sloping flanks like Japan’s Mount Fuji beckoned in
the distance. Visibility was excellent, the path was clear and the physical
demands were predictable. Fuelled by a good breakfast, and armed with
intent, and the climbing method known as kaizen (continuous improve-
ment) the firm could confidently set out to conquer its Fuji.
Nowadays, fitness mountaineering is not so simple.
For one thing the landscape’s topography has changed. Its previous
evenness has been puckered by valleys and foothills, and fractured by

gorges and chasms. Fuji’s silhouette is less distinct, although whether
this is because it is no longer there, having been broken and eroded by
the forces that fractured the foothills, or whether it’s simply obscured
by intermediate peaks and the swirling clouds of unknowing that have
enveloped the landscape, the climber cannot be sure.
For another thing the fitness landscape remains active. The forces
that have fractured its former symmetry show no signs of abating. New
peaks erupt constantly. What were heights yesterday are depths today,
and may be heights again tomorrow. The landscape is in a state of
constant chaotic deformation. There’s no terra firma. All is turbulence
and upheaval.
As if that were not enough, climbers are becoming aware of another
new phenomenon; each time they take a step in what seems to be the
right direction the ground quivers. They have become agents of the
landscape’s deformation, and there is no proportion to cause and effect.
A small step can have huge consequences elsewhere, through a complex
series of amplifying resonances that occur far below the threshold of the
climbers’ awareness.
Once quite separate entities, climber and landscape are now locked
in a process of co-evolution. Neither can move without affecting the
other in potentially profound, but intrinsically unpredictable ways.
What happened? What transformed this smooth, stable, single-
peaked landscape into a quaking, multi-peaked morass, where foresight
and resolve seem less important than luck and balance?
9780230_230941_09_cha07.indd 129 09/09/2009 10:02
130 BUSINESS AT A CROSSROADS
Part of the answer is that the fitness landscape of each firm is a crea-
ture of the fitness landscapes of other firms. In the past, there were only
a few climbers and they were too far apart to affect each other. Nowa-
days, the landscapes teem with climbers. Some fall into chasms or are

crushed by rock slides, but each is replaced by several new climbers.
And climbers are communicating more. It’s not just that in crowded
landscapes others are within earshot; it’s also that all climbers are now
equipped with powerful communications equipment. The air is alive
with climbing chatter.
Kauffman’s model showed that, as the number of climbers multi-
plies and the connections between them rise, landscapes become steadily
more rugged and changeable, and reaching the highest peak becomes
progressively harder. Unable to see the entire landscape, climbers have
to aim for nearby peaks. If they’re kaizen climbers, they may reach
them, but might then see higher peaks they could have aimed for had
they been less intent on continuous improvement, or find that the land-
scape has deformed and what looked like the summit at the start has
become a ledge on the flank of another peak.
Having been stranded on low local peaks once or twice climbers may
begin to question the value of route planning in poor visibility, through
a constantly deforming landscape, and even doubt the value of ascent
itself, when all that it achieves is a far from splendid isolation, on a far
from optimal peak.
In this new, confusing world, structure becomes paramount. The
key question becomes: “What enterprise shape is most likely to prevent
the company from becoming stranded in a sub-optimal position?”
Phase transition
Kauffman’s analysis suggests integrated MuBEs, run by all-powerful
CEOs, work better when communication between units is limited,
but that business enterprises consisting of many independent selfishly
optimizing units will work better when communication between units
is high.
Let’s return for a moment to the two business institution builders
living in the late 19th and early 21st centuries. The former faces an

environment in which connectivity is relatively low and favors integ
-
ration and an omnipotent CEO. The latter faces an environment in
which connectivity is much higher and favors disintegration and no
CEO. (Perhaps the alien wasn’t being cryptic after all.)
9780230_230941_09_cha07.indd 130 09/09/2009 10:02
7 LEADERLESS COMPETITORS 131
There’s an intriguing correspondence between this arcane discovery of
complexity science, and the debates about the architecture of a business
enterprise from Adam Smith to Alfred Chandler. Smith believed in the
wisdom of the market’s “invisible hand,” and favored unfettered entrepre-
neurial capitalism. Chandler believed that although it led to market imper-
fections the “visible hand” of managerial coordination was superior to
market coordination, and favored large, integrated companies.
Kauffman says they’re both right under certain conditions and that
what is best in particular circumstances, depends on the degree of
connectivity, or information richness. As connectivity rises, the strengths
of integrated enterprise are apparent. When it increases further, the
system “flips” and the best results are found, “as if by an invisible hand,”
when its units are acting independently and selfishly.
It is a small step from here to conclude that the huge increase in
information richness that has been occurring in the business world in
recent years will reach, and may have already passed, the point where
disintegrated enterprises and entrepreneurial capitalism are superior to
integrated enterprise and managerial capitalism.
If connectivity is seen as a measure of market efficiency, one can say
that, as market efficiency rose in the late 19th and early 20th centuries,
the integrated organization had the edge, but that when market effi-
ciency improved even further in the late 20th and early 21st centuries,
disintegrated enterprise became superior.

Or one could say Smith was wrong to be contemptuous of the
company when he wrote, but would have been right, if he had been
writing a couple of centuries later.
The change in the optimal unit size occurs suddenly when the whole
system flips from order to chaos. When a system is integrated, it is
ordered. When divided into small, independent units, it becomes
chaotic. The fact that problem-solving by small independent units,
none of which is “minding the whole store,” outperforms central
problem-solving for the benefit of the whole system, suggests flat struc-
tures (as opposed to hierarchies) produce better results and “democ-
racy” (see Chapter 10) is a “co-evolutionary problem-solving device,”
as Kauffman called it, which finds reasonable compromises among
conflicting interests.
A business enterprise in which each participant acts independently
and selfishly seems like a nightmare for a CEO, but for people in search
of “free,” “fair,” “reasonable” and “decent” workplaces and distribu-
tions of income, wealth, power, and self-respect consistent with Rawls’s
“difference principle,” it has a lot going for it.
9780230_230941_09_cha07.indd 131 09/09/2009 10:02
132 BUSINESS AT A CROSSROADS
The poverty of strategy
A brief reminder of how the sciences of complexity change our view of
business will be helpful at this stage.
The CEO-system is based on a linear view of the world that assumes
there is a unique effect or outcome for each cause or action. In a non-
linear world a cause or action can have many different effects or
outcomes. In math a linear equation has a unique solution, but a non-
linear equation has many solutions and is thus insoluble.
A linear system can be understood and controlled by understanding
and controlling all of its sub-systems, but a non-linear system is “syner-

gistic”; it has qualities and behaves in ways that cannot be understood
and controlled by understanding and controlling all its sub-systems.
Most managers accept that their companies display a certain amount
of non-linearity, but seem to believe this can be accommodated by
adding “noise” variables to their linear models. They assume large
shocks have large effects on the system, but the infinity of small shocks
they can’t see, let alone measure, can be ignored. In other words, they
don’t believe positive feedback can turn small shocks into earthquakes.
Complexity scientists disagree. They say that non-linear feedback
systems, or “complex adaptive systems” as they are known, are very
sensitive to initial conditions, and small shocks can and often do lead
to fundamental changes in the behavior of the whole system. The
usual example is the so-called “butterfly effect.” When taking wing in
Tokyo, a butterfly can cause a gale in New York and no one will be
able to trace the intervening set of causes and effects. A better example
here would be that a decision by a young homeowner in Cleveland,
Ohio, to spend his limited funds on a ticket for the ball game instead
of his monthly mortgage payment can, by a series of events no one
could have predicted, bring the world’s financial system to the brink
of collapse.
This means that if companies are “complex adaptive systems,” which
seems to me undeniable, they cannot be directed, and the enormous
sums they spend on CEOs and strategy consultants are being wasted.
Complex adaptive systems develop through a process of “spontaneous
self-organization.” Members of such a system, such as the CEO, can
contribute to its development, but no one can control it. Because it
takes time for tiny changes to upset a system’s equilibrium, it is reason-
able to plan and manage for the short term and perhaps do some
scenario planning, but even short-term plans may go awry, and there’s
no guarantee that any scenario examined will be realized.

9780230_230941_09_cha07.indd 132 09/09/2009 10:02
7 LEADERLESS COMPETITORS 133
According to the “linear” view of business, exemplified by Michael
Porter’s “four forces”,
6
successful companies were those that achieved
equilibrium adaptation to stable environments. Complexity science tells
us that “complex adaptive systems” are in a state of constant disequilib-
rium and are not, therefore, constrained by the environment. They
“co-evolve” with it. Members of such systems can do what they want,
but can never predict the long-term outcomes of their actions. The
choice is between stable stagnation achieved by bounding the system
with rules and plans, total (and so terminal) instability, and unpredict-
able creativity achieved by letting the system organize itself.
This challenges a long tradition in management thinking that firms
can, and should seek to, achieve adaptation to a stable, or slowly
changing environment.
This was why there was so much talk about “change management,”
and why the new CEO was often described as a “change agent.” Some
said change was needed, because the previous leadership had allowed the
company to drift out of fit with the environment. Others said that the
company had stood still while the environment changed; that a natural,
institutional inertia had caused the firm to cling to the status quo until
the mismatch with the environment became critical (or, as Laurence
Peter put it, “until long after the quo had lost its status”). Either way, a
CEO-led change program was needed to restore equilibrium.
Peter Senge, for example, although he recognized the complexity of
business and was an eloquent advocate of the “systems” view of the
company, urged leaders to articulate visions, and purposes.
7

Michael
Porter and Tom Peters
8
acknowledged the complexity of business, but
they too believed there’s a discoverable calculus of success. C. K.
Prahalad and Gary Hamel said companies succeeded by matching their
“core competencies” to a “strategic intent.”
9

The successes of the companies all four of these authors described to
support their own theories are just as likely to have been the fortuitous
and unpredictable consequences of the self-organization of complex
adaptive systems.
For complexity scientists, words such as “fit” and “intent” are
meaningless here, because systems only become changeable when
they achieve the instability that develops when the system is far from
equilibrium. Companies can be adapted to the past and try to adapt to
the present, but they can’t adapt or fit themselves to a future that is
unknowable. “The survivors and thrivers,” say David Parker and
Ralph Stacey, among the pioneers in the application of complexity
science to business, “are [companies] that are sustained far from equi-
9780230_230941_09_cha07.indd 133 09/09/2009 10:02
134 BUSINESS AT A CROSSROADS
librium in bounded instability. In the paradoxical state known as chaos
they are inherently changeable [and therefore] capable of continuing
innovation and variety.”
10
The problem with strategies is that they are relatively inflexible
plans for wholly unknowable futures. They’re attempts to stabilize
organizations (and thus make them less agile, flexible, alert, and crea-

tive), by aligning employees with a view of the future that is certain to
prove inaccurate.
Instability is a necessary condition for adaptability, but it must be
constrained, because total instability is fatal. The challenge, for
managers, is to keep their company poised on the edge of total insta-
bility. “The only sensible response to relentless change,” say Parker and
Stacey, “is relentless adaptation.” Managers should not plan long term.
They should try, instead, to make the organization changeable.
The CEO system is a bad system if relentless adaptation is the key to
success, because power exercised dictatorially creates too much stability
and stability is the enemy of agility. But where, if not from the CEO,
can these bounds required to prevent an organization from falling apart
altogether, come from? An enterprise consisting of a number of small,
independent firms and individuals (computer scientists would call this a
“multi-agent system”) will be nimble, flexible, alert and, above all, crea-
tive. But what will prevent it from falling apart altogether?
The answer is self-organization. In a chaotic system, subtle, often
imperceptible, but nonetheless powerful self-organizing forces are
usually at work. Left to itself a chaotic system may collapse, but it may
also organize itself enough to maintain its balance on the edge of the
abyss. It will have no strategy or intent, but, as we have seen, that’s no
bad thing. Without a CEO and a strategy, such an enterprise would rely
instead for its survival and prosperity on the aimless interplay of ideas
and actions on the strategy-free edge of bounded instability.
It will be under control, but no one will be in control.
Self-organizing multi-agent systems don’t try to optimize anything at
system level, but because each agent optimizes selfishly and so takes the
system “the wrong way,” from time to time, the system as a whole (as
Kauffman’s patch analysis showed), gravitates toward the “attractors” of
global (as opposed to local) minimum costs and maximum value.

The lack of an optimizing intent, therefore, doesn’t mean, as some
have argued, that an optimizing CEO system will always outperform a
leaderless system, with no system-level ambition. Enlightened by the
strange truths of complexity science, it’s not hard to imagine a company
focused by its CEO on maximizing shareholder value, with the sights of
9780230_230941_09_cha07.indd 134 09/09/2009 10:02
7 LEADERLESS COMPETITORS 135
all its employees firmly set on some grand strategic goal, being casually
overhauled by an unorganized, aimless group of small firms and indi-
viduals teetering on the brink of collapse. Moreover, complexity scien-
tists will say that, even if the CEO-led optimizer wins, it will be by
chance, not because of any merits of the CEO system itself.
The problem with this conjecture is that it is hard to prove. If a
new kind of unstable, self-organizing enterprise threatens CEO-led
companies (and may, even now, be stealing market share from them),
how can we know? Such enterprises are, by definition, always in a state of
flux; constantly coalescing, changing shape, coming apart and re-forming.
They may not be recognizable as enterprises in the normal sense, and
their own agents may be only dimly aware they’re interacting in self-
organized ways. Moreover, anything more than a dim awareness of asso-
ciation, would in theory, be undesirable, for with awareness of being
part of a group, comes the dangerous idea of a “strategy” and thus the
beginning of the end for the cardinal chaotic virtues.
New roads
It is still early days for the emergence of formidable, leaderless competi-
tors of CEO-led companies, but two multi-agent systems have already
become substantial enterprises through the internet. Both can be seen
as creatures of phase transitions similar to the phase transition to the
auto-catalytic set, which Kauffman suggested may have led to life.
LINUX

In 1983, “open-source” software pioneer Richard Stallman launched
the GNU Project to construct a “complete Unix-compatible software
system” out of free software. (GNU is a recursive acronym, a word-
play much loved by computer experts and very popular at Stallman’s
alma mater, MIT. GNU stands for GNU’s Not Unix). Stallman
founded the Free Software Foundation in 1984, and in 1989 he
published the GNU General Public License (GNU GPL), which set
out the conditions on which GNU software was available. By then,
open-source zealots, eager to liberate the world of computing from
what they saw as the tyranny of Microsoft, were into their stride. By
1990, most of the components of an operating system, including
libraries, compilers, text editors, a Unix shell and a windowing system,
9780230_230941_09_cha07.indd 135 09/09/2009 10:02
136 BUSINESS AT A CROSSROADS
were done. Only low-level elements, such as device drivers, and the
crucial system “kernel,” were still lacking.
Enter Finnish programmer Linus Torvalds. He said later that if the
GNU kernel had been available, he would not have decided to write his
own. In 1991, he turned MINIX, a truncated version of the Unix oper-
ating system used in teaching, into Linux.
The rest is still unfolding history.
The Linux kernel, supporting applications and utilities are freely
available and all the source code, applications, and utilities are published
under GPLs, which prevent any restriction on their “open-source”
availability. Torvalds still oversees the system’s overall development, but
anyone can develop applications and interfaces.
The system is very robust, because problems are instantly aired on
the internet, and quickly fixed. Linux is the flagship of the GNU Project
and the Free Software Foundation. In recognition of their joint achieve-
ment Stallman and Torvalds shared the 1998 Electronic Frontier Foun-

dation’s Pioneer award.
Linux is no plaything for techies. It is seen by Microsoft, which
open-source zealots regard as the enemy, as a formidable rival to
Windows. In November 1998 a leaked internal Microsoft document
(the first “Halloween” document) warned that open-source processes
could command creative resources far in excess of those that were avail-
able to commercial software companies.
11
Mutual sniping between Microsoft and The Free Software Foun-
dation (FSF) became a popular industry soap opera, with the FSF
claiming Microsoft’s production of proprietary software is bad for
software users, because it denies them “their rightful freedom”.
12
The
rivalry became more heated in early 2004, when Microsoft published
a study, Get the Facts, on its own website comparing the costs of
Windows and Linux. Based on case studies, analysts’ research, and
some Microsoft-sponsored inquiries, it claimed that using Linux on
servers compared unfavorably, in terms of reliability, security and total
cost of ownership, with using Windows.
Linux distributors and aficionados, including Novell, IBM, and Red
Hat (among the most successful Linux systems developers) responded
with their own studies, surveys, and customer testimonials refuting
Microsoft’s claims.
Market share comparison is difficult, because open-source software
installations, such as Linux, are not recorded in the same way as instal-
lations of proprietary software, such as Windows. According to IDC,
the most authoritative provider of IT market intelligence, Linux-based
9780230_230941_09_cha07.indd 136 09/09/2009 10:02
7 LEADERLESS COMPETITORS 137

systems in the global server market achieved a 10 percent revenue
growth in the second quarter of 2008, and their share of global server
revenues had grown from 9.4 percent, to 13.4 percent in the previous
12 months.
13

Comparisons in the desktop market where, until recently, Linux
had made only a little headway, are even more difficult, because
Linux systems are rarely paid for. It would, therefore, be possible for
Linux to have no official market share, while all PCs were running
Linux. Estimates of the Linux share in the desktop market in early
2009 ranged from just under 1 percent, to slightly over 2 percent.
All estimates had Linux growing much faster than the desktop
segment as a whole.
Two recent developments help to explain this sudden acceleration.
The first is the appearance of more “user-friendly” Linux desktop
operating systems, such as Ubuntu, which claimed about 30 percent of
Linux desktop installations in 2007. Ubuntu (a Zulu concept, of which
one translation is: “I can only be me through your eyes”) is free, open-
source software available under the terms of a GNU GPL. It is spon-
sored by U.K based Canonical Ltd., owned by South African
entrepreneur Mark Shuttleworth. Canonical’s business model is to earn
revenues by selling technical support. By keeping Ubuntu open source,
the company encourages other programmers to develop Ubuntu
components, in the same way as did Torvalds with Linux itself, without
having to develop the whole system. The Ubuntu system is also used on
some servers, including those of the Wikimedia Foundation.
The other important development adding considerable impetus to the
Linux penetration of the PC market is the emergence of so-called
“netbooks” or “ultra-low-cost personal computers” (ULCPCs), as the

fastest growing segment of the PC market. Typically costing less than
$500, these lean machines have only limited on-board software and rely
instead on applications software running on the internet. Running web-
resident software on netbooks is known, poetically, as “cloud computing.”
Linux has two major advantages for netbooks. It is faster than the
code-heavy Windows Vista, which reduces netbook hardware costs, so
making them cheaper to manufacture. And because Linux is free, the
software costs are also much lower. According to an analyst quoted in a
Bloomberg.com article in November 2008, equipping a computer with
Linux costs about $5, against $40–50 for Windows XP, and $100 for
Vista. The Bloomberg article estimated that Acer and Asustek Computer,
which together claimed 90 percent of the netbook market at the time,
were using Linux systems on some 30 percent of their netbooks.
14
9780230_230941_09_cha07.indd 137 09/09/2009 10:02
138 BUSINESS AT A CROSSROADS
It remains to be seen how this netbook software battle between
Microsoft and the army of unorganized, self-organizing, open-source
zealots will turn out. But Microsoft already knows it has a fight on its
hands. Despite regressing to Vista’s predecessor, XP, for netbooks, it
lost the first round to Linux. It is pinning its hopes of winning the
second round on Vista’s allegedly less clunky successor, Windows 7.
But victory is by no means assured.
There are other examples of open-source challengers to proprietary
software suppliers. Perl, originally developed by Larry Wall, is a popular
dynamic language for Web applications used on particularly busy sites,
such as bbc.co.uk, Amazon.com, and Ticketmaster. It is also used
widely in finance and bio-informatics, because it allows rapid applica-
tion development and deployment, and can handle large data sets. It is
free software licensed under the Artistic License and the GNU GPL.

Larry Wall still oversees Perl’s development.
Tcl (Tool Command Language, alias “tickle”), a popular program-
ming language created by John Ousterhout, is also free and
open-source.
The Apache Software Foundation (ASF), a non-profit company set
up to support Apache software projects, is “a community of developers
and users,” pursuing “a collaborative, consensus-based development
process” (
www.apache.org). Apache open-source software is free and
available under a GPL-compatible license. The Eclipse Foundation, an
open-source “community” extruded from IBM’s Eclipse Project, is a
not-for-profit with a “governance model” that ensures “no single entity
is able to control the strategy, policies, or operations of the Eclipse
community” (www.eclipse.org).
No one controls Linux, but it is under control. And, with the help
of an army of contributors who flock to it and continually explore its
development potential, it’s giving CEO-led Microsoft cause for
serious concern.
Linux and several other user/developer communities within the free
software movement are examples of quasi-organizations interacting and
coalescing into what I call “multi-agent business enterprises” (MaBEs).
Battle has been joined, in the software industry, between traditional
MuBEs and these fundamentally different MaBEs.
Some will say that Linux could not have emerged as a competitor to
Microsoft without the internet and that one cannot generalize from
developments in the internet domain to business as a whole. It may be
true that self-organizing multi-agent systems can coalesce into enter-
prises capable of taking on CEO-led MuBEs, such as Microsoft, in
9780230_230941_09_cha07.indd 138 09/09/2009 10:02
7 LEADERLESS COMPETITORS 139

industries living partly in the net, but that doesn’t mean that CEO-led
MuBEs living elsewhere are at risk from such enterprises.
There are two objections to this objection.
The first is that the internet is an increasingly important domain
in the modern business environment and more and more companies
and their customers are living more and more of their lives in it. To
argue that leaderless MaBEs pose no threat to CEO-led MuBEs
living outside the internet is like suggesting in the 1960s that Japa-
nese car makers posed no threat to Detroit, because they didn’t
make V8 engines.
The other objection to the suggestion that CEO-led MuBEs that
have nothing to do with the internet are safe from attack by MaBEs is
that it is not true. In the late 1980s Microsoft was a member of a MaBE
of small and medium-sized firms (it also included Intel, Dell and
Compaq), each of which was pursuing its own goal, that harried the
mighty IBM to the brink of collapse in the early 1990s.
WIKIPEDIA
When I was a boy embarking on a lifelong love affair with science
fiction, the neatest thing I could imagine was a wrist gizmo that gave
the wearer access to all the knowledge in the world. I don’t think color
televisions were available at the time – we certainly didn’t have one at
home. A sober assessment of the prospects that such a device would
become a reality in my lifetime would not have been sanguine.
But it, or something very like it, is a reality now, because, with a
web-enabled mobile phone, I have access to the Wikipedia. And of all
the many blessings the internet has brought us, the Wikipedia, is, in my
view, the most wonderful. It’s the library of Alexandria realized. It’s the
Aleph in Jorge Luis Borges’s short story.
15

Not the least of Wikipedia’s wonders is that it came to be what it is,
the largest body of knowledge ever assembled, by accident.
That the internet is an ideal environment for assembling knowledge
was recognized very early on. Interpedia was launched in 1993, but
didn’t stay the course. Richard Stallman, doyen of the open-source
movement and founder of the Free Software Foundation (see above),
proposed a “Free Universal Encyclopedia and Learning Resource” in
1999. His idea was realized in the GNUPedia, which went online on
January 17, 2001. It competed with another, earlier project with a
rather similar sounding name.
9780230_230941_09_cha07.indd 139 09/09/2009 10:02
140 BUSINESS AT A CROSSROADS
Nupedia was started by Jimmy Wales in March 2000. It was designed
as a free encyclopedia employing highly qualified contributors and a
rigorous peer review process. Despite the appointment of a full-time
editor-in-chief, Larry Sanger, a graduate philosophy student, progress was
very slow. Barely a dozen articles were signed off in the first 12 months.
Wales and Sanger realized they had to produce content more quickly
and considered various ideas. Quite who was the first to suggest using
wiki software, which enables anyone who visits a web page to contribute
to or modify content using a simple markup language, is not clear.
Wales and Sanger liked the idea, however. Sanger wrote to Nupedia’s
editors and reviewers with the proposal and Wales set up a Nupedia wiki
system, which went online on January 10, 2001.
A number of editors and reviewers were uncomfortable with the idea
of associating Nupedia with a wiki-style website, so Wales decided to
give the project, the only purpose of which was to generate new mate-
rial for Nupedia, its own name, Wikipedia. To distance the new content
feeder more clearly from the Nupedia project Wikipedia was launched
on its own domain, wikipedia.com, on January 15, 2001.

Word of the project spread quickly through the internet thanks, in
part, to hundreds of references a day from Google. An explosion of some
kind occurred. It was as if Nupedia’s content generator had, quite by
accident, touched some switch, the existence of which no one had even
suspected, and released an enormous pent-up supply of information.
Wikipedia attracted 1,000 articles within a month of launch, and had
passed the 10,000 mark by early September. By then the mainstream
media were taking notice. The New York Times ran a piece on the phenom-
enon on September 20 and “network effects” (the more coverage you
get, the more coverage you get) added further to the momentum.
In its first year Wikipedia fed over 20,000 entries to its Nupedia
client and the growth rate was accelerating. The article count had
doubled again by the end of August 2002. The tail was wagging the
dog to bits. Nupedia was laid to rest in September 2003.
By the end of 2008, the English-language Wikipedia (there are many
others) had accumulated over 2.7 million articles, equivalent to a tradi-
tional encyclopedia of close to one thousand volumes. The wiki site,
is well worth
a visit for any readers interested in the growth of and prospects for Wiki-
pedia. It shows that content growth has followed a “log” curve. More
content has led to more traffic, which has led to more content, but
growth reduces potential new content, because that’s finite, and ulti-
mately limited by the combined knowledge of all possible contributors.
9780230_230941_09_cha07.indd 140 09/09/2009 10:02
7 LEADERLESS COMPETITORS 141
Statisticians suggest the maximum number of articles is about four
million. There will always be a trickle of new articles as future events
unfold, but that will eventually become negligible relative to the existing
stock. The peak growth of the English Wikipedia is said to have been
reached in August 2006, at a rate of 60,000 new articles a month. By

the end of 2008 the article influx had fallen to about 40,000 a month.
Quality is likely to improve steadily, as constant editing (a spontaneous
version of the Japanese concept of kaizen – “continuous improvement”)
causes all articles to converge on versions as close to the truth as
contributors can collectively take them.
The Wikipedia, as a knowledge gathering (as opposed to a knowl-
edge maintenance) project, is approaching some kind of completion. In
barely a decade the dreams of the ancient librarians of Alexandria and
Pergamon have been casually surpassed, and a digital Aleph has snapped
into existence.
What are we to make of this breathtaking achievement?
Make of it what you will. Many people have and many more will. For
our purposes, I believe there are four things to be said about the
Wikipedia.
The first is that it has been an achievement by the people and for the
people. There was no intention, no strategy, no directing CEO. A
combination of the idea of a free encyclopedia, the internet, and wiki
software triggered a global orgy of knowledge sharing. No one could
possibly have imagined how eager ordinary people are to tell other
people what they know, and play a part in creating something good and
valuable. We’re all encyclopedists now.
The second thing to say is that the process by which the Wikipedia was
created was spontaneous self-organization within an inherently unstable
system bounded by a few simple rules and by the desire of contributors to
do what encyclopedists do; to tell it like it is to the best of their knowledge.
The system is deeply democratic. Wiki software is, in a sense, a crystalliza-
tion of the democratic idea. (Some say Wiki stands for “What I Know Is,”
but that’s a so-called “backronym.” The word is Hawaiian for “fast.”)
The third thing to say about the Wikipedia is that it would not be
what it has become without some rules. The right to edit had to be

limited, to prevent vandals from fouling the Wikipedia commons. I
have a friend who is a Wikipedia “administrator.” “It just means I have
a couple of extra buttons I can press” he says. His job is to keep watch
for vandals (persistent and vexatious editors motivated by something
other than a desire to improve), order them to desist and withdraw
their editing rights if they don’t.
9780230_230941_09_cha07.indd 141 09/09/2009 10:02