Circular organizing and triple loop learning

Abstract

The organizational learning literature distinguishes different levels of learning (zero learning and single, double and triple loop learning) in order to understand the complexity and dynamics of changes in policies, objectives, mental maps, and structures and strategies for learning. This article explores the case of an emerging new organizational design, the circular organization, in order to understand the role of triple loop learning. The circular model was developed on the basis of ideas about the relationship between organizational structure and behavior taken from theories of dynamic systems. Circular design precepts appear to provide a structural facilitation of single and double loop learning. In this respect, the circular design tends to act as a facilitating infrastructure for triple loop learning, that is, exploring the structural opportunities and key competences people need to participate in making well-informed choices about policies, objectives and other issues.

Full text

Circular
organizing
439
Journal of Organizational Change
Management, Vol. 12 No. 5, 1999,
pp. 439-453. #MCB University
Press, 0953-4814
Circular organizing and triple
loop learning
A. Georges L. Romme and Arjen van Witteloostuijn
Maastricht University, Faculty of Economics and Business
Administration, Maastricht, The Netherlands
Keywords Learning, Organizational design, Electronics, The Netherlands
Abstract The organizational learning literature distinguishes different levels of learning (zero
learning and single, double and triple loop learning) in order to understand the complexity and
dynamics of changes in policies, objectives, mental maps, and structures and strategies for
learning. This article explores the case of an emerging new organizational design, the circular
organization, in order to understand the role of triple loop learning. The circular model was
developed on the basis of ideas about the relationship between organizational structure and
behavior taken from theories of dynamic systems. Circular design precepts appear to provide a
structural facilitation of single and double loop learning. In this respect, the circular design tends
to act as a facilitating infrastructure for triple loop learning, that is, exploring the structural
opportunities and key competences people need to participate in making well-informed choices
about policies, objectives and other issues.
Introduction
Research into the notion and practice of organizational learning has identified
distinct systemic levels of learning: zero, single loop, double loop and triple loop
learning (e.g., Argyris and SchoÈn, 1974; Flood and Romm, 1996; Snell and Man-
Kuen Chak, 1998). Zero learning occurs in an organizational setting when fresh
imperatives or problems arise, yet members fail to take corrective action. Single
loop learning refers to making simple adaptions and taking corrective actions,
whereas double loop learning involves reframing, that is, learning to see things
in totally new ways. Finally, triple loop learning entails members developing
new processes or methodologies for arriving at such re-framings.
Generally speaking: the higher the learning level is, the more complex it is.
Zero learning and single loop learning are widespread in most organizations,
but double loop and particularly triple loop learning are rare. In this article, we
explore the emergence of the so-called circular organization as an example of
triple loop learning in the area of governance, participation and decision
making. First, the different systemic levels of learning are explored in more
detail. Subsequently, the development of the circular organization in Endenburg
Elektrotechniek in The Netherlands is described. Finally, the nature of triple
loop learning in this particular case is discussed in order to identify certain key
conditions under which triple loop learning is likely to occur.
Zero, single, double and triple loop learning
From the perspective of organizational learning, the development of a
sustainable learning ability of (key parts of) the organization is a prerequisite to
The current issue and full text archive of this journal is available at
http://www.emerald-library.com
The kind cooperation of Endenburg Elektrotechniek is acknowledged.

JOCM
12,5
440
survive and succeed in increasingly dynamic and complex environments.
Bateson (1973) and others have distinguished different levels of learning. The
learning is level zero when fresh imperatives arise, yet members fail to take
corrective action (Bateson, 1973; Snell and Man-Kuen Chak, 1998).
Single loop learning occurs when error detection ``permits the organization to
carry on its present policies or to achieve its present objectives'' (Argyris and
SchoÈn, 1978, p. 2). This kind of organizational learning manifests itself as a
consolidation process, that is, changes in the organization's knowledge and
competency base without altering present policies, objectives or mental maps
(Snell and Man-Kuen Chak, 1998).
Double loop learning is achieved if ``error is detected and corrected in ways
that involve the modification of an organization's underlying norms, policies
and objectives'' (Argyris and SchoÈn, 1978, p. 3). In other words, double loop
learning manifests itself as a transformation process, that is, changes in the
organization's knowledge and competency base by collectively reframing
problems and developing new policies, objectives and mental maps (Snell and
Man-Kuen Chak, 1998). For double loop learning to develop, the key actors in
the organization have to be able to create ongoing dialogues, a conversational
process in which defensive reasoning and behavior do not impede free and
open inquiry (Argyris et al., 1985). Double loop learning appears to facilitate the
adaptive potential of an organization, but most organizations seem to have
great difficulties in actually learning in a double loop manner (Argyris, 1996).
Therefore, a third level of learning concerning structures and strategies for
learning is relevant, so-called deutero-learning (Bateson, 1973) or triple loop
learning (Flood and Romm, 1996; Snell and Man-Kuen Chak, 1998). As such,
triple loop learning is about increasing the fullness and deepness of learning
about the diversity of issues and dilemmas faced, by linking together all local
units of learning in one overall learning infrastructure as well as developing the
competences and skills to use this infrastructure (Flood and Romm, 1996).
Triple loop learning manifests itself in the form of ``collective mindfulness'':
members discover how they and their predecessors have facilitated or inhibited
learning, and produce new structures and strategies for learning.
Relative to single loop learning, double loop and particularly triple loop
learning are more concerned with structural patterns (mental maps, facilitating
structures, etc.). Implicit in the distinction between different systemic levels of
learning is therefore the relationship between structure and behavior. In the
following sections, we will explore the so-called circular organization as an
example and outcome of triple loop learning.
The case of circular organizing
In this section, we describe a case involving a deliberate attempt to develop an
organizational system which stimulates and facilitates learning throughout the
entire system. The circular organization case is interesting, because it emerged
on the basis of a systemic viewpoint regarding the interaction between
structure and behavior. This development of the circular organizational system

Circular
organizing
441
was driven by preliminary theories about how organizational structure affects
organizational behavior. The case presented here involves the development of
the circular organization model, also known as the sociocratic model, in The
Netherlands. The development of the theoretical framework behind this model
started in the late 1960s, with the first actual experiments in the practice of a
medium-sized electrotechnical company taking place in the early 1970s. The
development of this model was largely driven by a young engineer, Gerard
Endenburg, at the time acting as the general manager of this company.
Endenburg was an early student of the new dynamic theories emerging in the
1960s in the field of system theory, which are currently better known as system
dynamics (e.g. Forrester, 1961; 1971).
We will first describe the theoretical framework triggering the first
experiments and the experimental development process that ultimately
produced the circular model. Note that the circular model has recently moved
beyond the experimental stage, and has in the past ten years also been
introduced in about thirty other organizations in The Netherlands, Brasil,
Canada and the USA (Romme and Reijmer, 1997). We will focus here on the
developmental process of the circular design in Endenburg Elektrotechniek
from the early 1970s to the mid-1980s, rather than the introduction and
application of the fully developed design in other organizations. We will
describe the case of circular organizing in order to illustrate what triple loop
learning means for organizational design and practice.
Methodologically speaking, the following reconstruction of Endenburg's
experiments with circular organizing in the period 1970-1985 is based on
document study, including written notes, minutes of meetings, so-called social
accountancy reports and other evaluation papers. Most of these documents
provided the data for or were the results of discussions in meetings which
served processes of public testing and inquiry into the new organizational
design introduced and developed in this company. An additional source of data
was provided by Endenburg's (1992) doctoral dissertation, which describes in
retrospect the theoretical background and practical development of the circular
model. In collecting additional data for this dissertation, Endenburg organized
several workshops where key participants, including several outside members
of the top circle (Board of Directors) and observers (e.g. a professor in
organization theory), discussed and reflected on a number of the critical
incidents and situations in the development of circular organizing in
Endenburg Elektrotechniek (Endenburg, 1992).
Preliminary ideas about the relationship between structure and behavior
Several principles in the emerging field of system dynamics provided the initial
framework that Gerard Endenburg developed in the late 1960s. The first
principle involves the pervasive influence of the deeper structure of a system on
the behavior of its constituent elements. The kind of influence of these deeper
structures is a generic one, involving the basic interrelationships that control
behavior. From an engineering viewpoint, such generic structures are the result

JOCM
12,5
442
of an (either explicitly or implicitly) designed and implemented system (e.g. the
design of a machine) which then in turn largely determines the behavior of the
actual system, including direct cause-effect relationships between individual
elements of the system and the latter's immediate environment. Endenburg
realized, however, that the structure of a human organization will be the result
of actions by its (previous) membership as well as the influence of the broader
social system the organization is embedded in. The original ambition of system
dynamics theory, in this respect, was to develop generally applicable models in
order to transfer behavioral insights from one situation to another (Forrester,
1961).
The second principle pertains to the difference between static and dynamic
(generic structures of) systems. Whereas in relatively simple and static systems
information, energy or power flows in one direction only, in relatively complex
and dynamic systems feedback and control loops often produce unexpected or
counter-intuitive results (see Forrester, 1961; Richardson, 1991). In this respect, the
causal links underlying feedback and control relations in such systems are
problematic because cause and effect tend to be distant in time and space. Looking
at and working in his own company, Endenburg found that from a system theory
point of view his organization was severely hampered by its fundamentally static
(hierarchical) management structure, in which power and authority flow in one
direction only. He also found out, by means of the installation of a works council
(which was required by Dutch law), that participation programs which intend to
create more commitment and involvement of workers, fail to work if this
underlying static structure is not reorganized.
The third principle taken from system dynamics involves the dilemma
between global patterns (e.g., corporate patterns and synergies) and local
behavior (e.g., creative actions and findings at the unit or team level). In
complex and dynamic human systems, the best solution to a certain problem
will very likely come out of localness, that is, from the collective wisdom of
those closest to the problem at hand (see Senge, 1990) regardless of their formal
position or group membership. Creativity at the local (team) level benefits from
an atmosphere of equivalence, because then (teams of) individuals will be
motivated to think about and try out unconventional and risky ideas. In this
respect, Endenburg was also inspired by how consensual decision making in
Quaker communities and organizations apparently supported an atmosphere
of free and open inquiry (see Louis, 1994), particularly through his previous
enrollment in a primary and secondary school in The Netherlands which was
organized on the basis of Quaker consensus ideas. In more general terms,
Endenburg believed that only through a fully participative society the
destructive effects of hierarchical authoritarian systems (e.g. the mass murder
of Jews in World War II) could be prevented. Endenburg's personal
commitment to this view was particularly strong given the fact that both his
parents, of whom Endenburg senior had survived a German concentration

Circular
organizing
443
camp, came out of World War II with the vision to set out to change
authoritarian systems from what they thought was the most logical position to
start from: that of the business owner.
First conception of circular design precepts
In 1970, Endenburg had developed the rough outline of a circular organization
design which he hoped, from the three theoretical principles, would solve the
main deficiencies of the hierarchical structure in his own company and other
organizations. This initial circular design involved the following three precepts
(Endenburg, 1992; Romme, 1997):
(1) Decision making is governed by consent, defined as ``no argued
objection''. The reasons and arguments rather than the positions taken
regarding an issue are of prime importance, and a decision is made when
each participant gives consent. In the first descriptions of this precept,
the importance of distinguishing between policy and work is
emphasized because only policy decisions have to be made by consent,
whereas decisions about actual work processes can be delegated to
``functional leaders'' (e.g. supervisors and managers). In general, the
consent rule allows the application of other decision methods, as long as
this is agreed upon by consent. Thus, decisions about, for example, daily
work organization are typically made by hierarchically positioned line
managers or direct supervisors, although other methods such as
democratic majority can also be used (if decided by consent to do so).
(2) Double linking. A larger organization is normally subdivided into a
hierarchy of circles, superimposed on ± and not replacing ± the
traditional hierarchy, in which each lower circle is represented in the
next higher circle (if any) by its functional leader and at least one
representative. The functional leader is elected in the next higher circle,
and the representative(s) is (are) elected in the lower circle.
(3) Election of persons is done by consent after open and free discussion. A
circle assigns its members by consent to the positions and tasks
required by the common work objective after open discussion. This
includes, for example, the selection of representatives in higher circles
and functional leaders of lower circles.
The implementation of these three precepts implied that the hierarchical
structure of the organization would be used exclusively for organizing and
coordinating the work processes, whereas a circular structure would be
superimposed on this hierarchy in order to make policy decisions and learn
about the effectiveness of decisions taken in the past. The term ``policy'' can be
understood as describing the objectives or constraints set for the work
processes. This means that what are policy issues at the level of a lower circle, is
work (implementation of policy) at the level of the next higher circle. Thus, each

JOCM
12,5
444
circle would make policy within its domain of authority and would operate
within the policy constraints set by the next higher circle (Endenburg,1992).
Endenburg's general hypothesis was that the introduction of a circular
design (superimposed on the existing hierarchical structure) would increase the
problem-solving capabilities of the organization as a whole. In this respect, he
expected that the introduction of a circular structure would create the
conditions for free and open inquiry in each (top, general and unit) circle, and
thus would increase the ability to solve problems, both within existing work
constellations (single loop learning) and within ill-structured problem contexts
(double loop learning). Moreover, Endenburg hoped that by this approach win-
or-lose situations could be avoided so that the disruptive processes ``losers''
typically set in motion (e.g. misconduct) could be stopped. The pivotal element
here was the opportunity to switch between the circular and the hierarchical
mode of organizing, with the circular structure as the governing mode of
decision making.
The first experiments in Endenburg Elektrotechniek Inc.
In 1968, Gerard Endenburg took over his father's position as managing director
of Endenburg Elektrotechniek Incorporated, the company his father had
started in Rotterdam (the Netherlands) in 1950. Since its foundation,
Endenburg Elektrotechniek had been developing into a company with about
120 employees, who were working primarily in projects in the area of the
design, production, installation and renovation of electrotechnical installations,
control systems, switching boards, and other electronic instruments. Its most
important clients were companies in the manufacturing, ship building,
offshore, house building and utility building industries. As an explicit
condition for accepting the position of general manager, Gerard Endenburg
requested he would be allowed to experiment, both technically and
organizationally. In the early 1970s, Gerard Endenburg decided to put an end to
the company's growth in order to give intensified attention to organizational
renewal. He then started to experiment with the ideas he had about
remuneration, decision making and organizational structure.
In 1970, the first actual experiments with circles were set up. A circle
hierarchy was formed, representatives were chosen, and the first circle
meetings were held. The circle structure involved a top circle (Board of
Directors), general management circle, and six unit circles. The events
occurring in the very first unit circle meeting set the tone for what
subsequently happened in other circles. This unit's membership included two
employees who were reputed for their obstructionist behavior. At the start of
the meeting, most other circle members did therefore expect the consent
principle to break down, assuming the consent principle would remove all
available (e.g. authoritarian or democratic) instruments to make ``difficult''
employees go along with any collective decision. After the meeting had started,
the two employees in question behaved ``as expected'', trying to undermine and
obstruct the discussion on issues such as the circle's objectives. However, when

Circular
organizing
445
the meeting progressed this typical behavioral pattern appeared to dissolve
and the nature of the discussion changed toward a collective attempt to find
solutions for the issues and problems that were being discussed. The kind of
obstructionist behavior from two of the circle's members disappeared without
any deliberate interventions from the chairperson or any other participant. The
success of this first meeting stimulated this circle to arrange for a new meeting
in order to solve several complex problems. Moreover, this initial success also
stimulated other circles to actually start working with the new design precepts.
In the same period, Gerard Endenburg started to discuss his ideas about new
forms of payment and compensation that should support the newly introduced
organizational design. These new forms of payment and compensation were
derived from the notion of risk sharing and, in addition, from the idea of
providing an adequate mix of fixed and variable rewards to both employees
and shareholders. The base payment of fixed reward, which was defined by
Endenburg as the ``subsistence guarantee'', was at that time largely determined
by collective agreements between labor unions, employers and the Dutch
government (on wages and social security benefits). This kind of payment
would under normal circumstances allow each employee to live according to an
acceptable standard of living, also after (s)he left the company and became
unemployed. The variable rewards involved short-term and long-term bonuses.
The short-term bonus was meant to provide immediate feedback on the
contribution of each individual participant in a given project. Thus, each
participant in a project received a bonus that was directly related to the net
result of the project. Similarly, all participants contributed to bearing the loss
on the project if the net result was negative. The long-term bonus was in effect
a profit-sharing plan. Thus, managers, employees and shareholders received a
share in the net profit of the company in the form of money, shares or share
certificates. On the basis of the result shown in the annual accounts the top
circle decided what amount was to be distributed.
The combination of fixed and variable rewards was viewed to be an
important condition for circular organizing, because it would give each member
of the company a minimum amount of freedom to choose for working in
Endenburg Elektrotechniek or leaving the company, whereas at the same time
above-average performance was rewarded accordingly. However, several
employees initially wondered whether the short-term bonus ``did not boil down
to a traditional kind of pay system, based on piece work''. Once these objections
were openly discussed in circle meetings, it became apparent that a more
fundamental feeling was crucial here, namely the fear of being manipulated by
the pay system. After some discussion in particularly the general circle,
Endenburg's proposals for the new payment system were accepted and
implemented, and by the mid-1970s the short-term bonus system appeared to
have a positive effect on the motivation and commitment of employees.
The first experiments and experiences with the circular structure and the
new payment system in the early 1970s induced several changes in the initial
design (around 1972 and 1973). For one, it quickly became clear that the three

JOCM
12,5
446
initial design precepts had left one important principle implicit: the fact that
every member of the organization should belong to at least one circle. Therefore,
the following precept was added as number 2 (with the initial precepts 2 and 3
becoming precepts 3 and 4): Every member of the organization belongs to at
least one circle, i.e., a functional work unit (e.g. department). The first text
describing this new precept included the definition of a circle as a group of
people (including its functional leader) with a common work objective whose
basic mode of policy decision making is consent. Additionally, the condition was
introduced that consent decision making normally occurs in specially scheduled
circle meetings, where topics are discussed which are relevant to the work
objectives of the circle (and being within the limits of their authority).
A second learning effect pertains to the condition of free access to all
information. That is, the circular structure could only operate if each member of
the company had free access to any piece of information considered relevant to
decision making in the circles. This implied that information on, for example,
investment proposals or results on projects should be easily accessible for those
who wished to obtain this information. By way of a so-called logbook system,
all circles therefore started to record and store information on their work
processes and policy decisions. In a way, the logbook system thus was a vehicle
for systematic data collection. The experiences of the first few years apparently
revealed that the openness implied by the free access to information at the time
did not have any disadvantages in, for example, communicating with
customers and sustaining inter-circle cooperation.
After a while, the way most members of the organization responded to the
introduction of the circular design and the new payment system was quite
positive. Indeed, the motivation to participate clearly increased and members
(of circles) interacted with each other in a more intensive manner than
previously was the case. For many participants the process of solving problems
and making decisions by way of open discussions constituted an emotional and
liberating experience. In this respect, the consent principle appeared to
facilitate and contribute to the breakdown of traditional (e.g., boss versus
worker and male versus female) role patterns, because (female) workers started
to participate more actively in discussions.
Responding to a crisis
The first critical ``test'' of the circular design as a whole emerged through a
crisis situation the company ran into in 1976. This episode can be regarded as
the first real (although unintended) test of the circular model in operation,
particularly with regard to its hypothesized effect on the ability to deal with ill-
structured problems like a crisis situation (we will discuss the implications
concerning this point in more detail in the discussion section). The unusual
chain of events which subsequently occurred, can be reconstructed as follows.
A local shipyard, accounting for more than one-third of the business of
Endenburg Elektrotechniek, suddenly shut down. General manager Gerard
Endenburg saw no other solution but to lay off 60 workers, most of whom

Circular
organizing
447
worked in the Technical Installations Unit. Thus, he decided to start a layoff
procedure for these employees (this decision fell within the mandate he had as
general manager, the leader of the general circle). In the same period, the top
circle was called together to discuss the crisis situation. The top circle at that
time included two representatives from the general circle, the general manager
(Endenburg) and four external non-executive directors. During the meeting two
external members expressed serious concerns with the operation of the circular
model in a crisis. They argued that an adequate response would require things
to be reorganized without sociocracy, perhaps by means of a turnaround
manager. The representatives from the general circle refused to go along with
this proposal because they were convinced that the sociocratic circular method
had to prove itself precisely in a crisis situation. As a consequence, the two
external members decided to resign from their membership of the top circle.
The day after the decision to lay off 60 employees was taken, Jan de Groot,
one of the fitters in the Manufacturing Unit, asked the secretary of his unit circle
to call a special meeting as quickly as possible in order to discuss an idea he had
for a better way to handle the sudden crisis. The circle secretary arranged a
meeting for the next day, and at this meeting De Groot put forward a proposal
about delaying the layoff for a few weeks and shifting everyone who would be
laid off into a concentrated marketing effort. The circle decided to support the
proposal, and appointed De Groot as a temporary representative to the general
circle. De Groot subsequently requested a special meeting of the general circle to
discuss his proposal. In this meeting, which took four hours, the general circle
decided to support the initiative of Jan de Groot. All 60 workers involved would,
after a crash course in acquisition and marketing, work in the next few weeks on
the acquisition of new projects. A final decision could not be made because of
policy limitations on the authority of the general circle to spend the company's
reserves for this purpose. This kind of decision fell within the policy domain of
the top circle, and Jan de Groot was therefore elected as a temporary
representative to the top circle. The top circle, without the two outside directors
who had left a few weeks earlier, decided to support De Groot's proposal to
spend part of the company's reserves for his ``turnaround'' plan, which allowed
the general manager to launch the plan into action. Within several weeks,
enough new projects were acquired to justify further postponement of the layoff.
Only a few workers were actually laid off in this period. The Technical
Installations Unit was sized down, but the accelerated growth of several other
units led to a much more diversified customer base.
Starting again: the role of education and training
In the late 1970s, several studies, including one by a professor from a local
business school, pointed out that the top circle and general management circle
met frequently (the top circle about six times and the general circle about 20 to
25 times per year), in contrast to most unit circles. In fact, several units
appeared to have lost all interest in circle meetings. The general management
circle therefore started to discuss this apparent dysfunctioning of circular

JOCM
12,5
448
organizing. In discussions with members of several unit circles, it became
apparent that most of the unit members were largely unaware of the
opportunities provided by the circular mode of organizing. In several
subsequent meetings in which these opportunities were discussed, it quickly
became understood that there were other problems apart from a lack of
awareness or information. It seemed that many workers had been confronted
with circular organizing too abruptly, and needed a more gradual introduction
into the principles underlying this new method of organizing.
After extensive discussions in the top and general circles, the decision was
made to implement the circular method again, as if nothing was known as yet.
This implementation process was supervised by senior consultants from the
Sociocratic Center, a foundation that was established a few years earlier to
supervise and support the introduction of the circular model in other
organizations. This implementation process also induced the development of a
more systematic approach to training circles in the principles of the circular
method. A key element in the basic training module that all (new members of)
circles since then undergo, is the circular process of leading, doing and
measuring, through which particularly the precept of double linking becomes
more clear (Romme, 1995).
Additionally, the renewed introduction of the circular model also involved
that each circle was made responsible for the education and training of its
membership, including the authority to spend its own budget for these
purposes. These education and training activities cover three general areas:
special functions or tasks (professional skills and knowledge); organizational
processes and structuring (e.g. skills in applying the four ground rules); and
decision making (e.g. skills in chairing circle meetings). With increasing
awareness and skills in operating the circular mode, an intriguing pattern of
group dynamics in circle meetings emerged. This pattern became particularly
evident in circle meetings that started out with sharply opposing interests or
proposals. Several of this kind of decision situations could be reconstructed in
great detail from company documents (including the minutes of circle
meetings) and a number of additional interviews. An exemplary case of group
dynamics occurred during a general circle meeting in 1985, where three major
investment proposals were presented to the general circle whilst the investment
funds for the remainder of the running budgetary year could carry only one
proposal. All three proposals appeared to be rather urgent, two coming from
unit circles and one being put forward by the general manager. One of the
participants recounts: ``When the meeting started, there was no prospect
whatsoever of a decision acceptable to all. The proposals were summarized by
the representatives and leaders of the unit circles, and by the general manager.
Then, a heated debate developed in which proposals were defended and
criticized, and to some extent also discredited. At some point, we became aware
of the common ground under our feet ± that is, the objectives of the general
circle ±, and we started using them to list the proposals in order of priority.
With a number of creative suggestions and ideas, we adapted two proposals in

Circular
organizing
449
terms of the conditions under which they could be postponed. After about three
hours, we reached the decision to support the investment proposal of one of the
units, and also agreed on some of the issues related to the other two proposals.''
This kind of group dynamics, which was evident in several other instances
where circles had to decide on issues involving opposing interests and
viewpoints, too, can be described as follows: first, the participants take position
by giving arguments for their viewpoints, followed by a discussion which
underscores the conflict of interests underlying the positions; then, a stalemate
arises, typically involving an increased level of anxiety; and, finally, the
pressure for convergence, also in view of the circle's common objectives,
increases and the creativity of the group is used to reach a decision that is
acceptable to all. Because this process may not always be successful, or
because certain decision issues require too much time given their urgency, the
general and top circle decided to adopt a rule that would prevent decision
processes from getting stuck (e.g. in a stalemate). This rule involved the
following constraint on the autonomy of a lower (e.g. unit or general) circle: if a
circle does not decide by consent on a certain issue in two meetings, with an
interval of at least 48 hours, the authority to make this decision automatically
moves to the next (general respectively top) circle. In practice, this rule tends to
increase the self-managing capacity of a given circle, in view of the fact that
participants dislike the idea of intervention by the next higher circle in their
own decision domain, and thus will make an extra effort to reach a decision
within reasonable time limits.
Other developments
We have described a number of critical incidents and events in the
developmental process of the circular design in Endenburg Elektrotechniek in
the period 1970-1985. In the same period, the company and particularly Gerard
Endenburg started to work on a new legal form that would support circularity
as a new design principle, because all existing legal forms (e.g. the corporation
and foundation) were based on a linear flow of power from owners to managers
to workers. A detailed discussion of this process can be found elsewhere
(Romme, 1999). In addition to the development of a legal form, the company
started to institutionalize a two-yearly so-called social accountancy audit, in
which the realization of a number of minimum norms for circularity was
controlled. These norms were explicitly incorporated in the company's
constitution, and included, for example, existence of the double linkages (leader
and representative) between two vertically related circles and the minimum
frequency with which unit circles (six times a year), the general circle (15 times
a year) and the top circle (six times a year) would have to meet. The Sociocratic
Center, as an independent organization, developed the procedures and skills
required for such a social audit (offered to other organizations as well).
In the mid-1980s, several other organizations started to experiment and
adopt the circular design, first in The Netherlands and in the late 1980s also in
Brasil, Canada and the USA (Romme and Reijmer, 1997). In some of these first

JOCM
12,5
450
applications of the circular design outside Endenburg Elektrotechniek, Gerard
Endenburg acted as the main consultant to these organizations, but later this
work was taken over by several other consultants. The exchange of
experiences and ideas between an increasing number of organizations (which
started) working with circular design precepts was largely facilitated by the
Sociocratic Center in The Netherlands, and later also through similar centers in
other countries. Moreover, this process triggered an increasingly professional
approach to training and education in the field of circular organizing.
Discussion
The circular design was explicitly developed on the basis of theoretical ideas
about the interplay between structure and behavior in and around
organizations. However, it is at this stage somewhat difficult to determine what
the merit is of the circular model, particularly in comparison with other (newly
developed) organizational models.
Moreover, after more than ten years, it is difficult to determine how
rigorously the newly developed theory and newly created knowledge on
circular designs were tested at the time of their introduction. However, several
critical incidents and experiments in the development of this design in
Endenburg Elektrotechniek were documented rather well by the membership
of the circles at the time and, in addition, were publicly tested (in retrospect) in
discussions with the main participants and several external observers. Thus,
public testing and intersubjective agreement as two critical norms of valid
knowledge were served rather well. Other norms, such as the requirement of
falsifiability of hypotheses, were less well realized.
In this respect, the historical development of the circular design raises an
interesting question as to whether the researcher's and practitioner's role can
be combined in a single person. The initial experiments with the circular design
in Endenburg Elektrotechniek were initiated and supervised by Gerard
Endenburg, who was the main researcher (in developing and testing theories
about the effects of circular designs) as well as the main client-practitioner (in
his position of general manager). In retrospect, this combination of roles may
have been decisive for developing a very unconventional approach to
management and design. At the same time, however, this combination
appeared not to be beneficial for the further development and implementation
of the new model once it started creating momentum, which is clear from the
need to re-introduce the model anew in the late 1970s. In general, we believe
that the role of the researcher cannot be fully integrated into the community of
practitioners being served, but needs to be embedded in an independent
community of researchers as well, which will provide some kind of guarantee
that ``science'' does not suffer too much from ``action''.
Circular organizing and organizational learning
The case of circular redesign in Endenburg Elektrotechniek is interesting
because the circular infrastructure, which co-exists with the administrative

Circular
organizing
451
hierarchy, appears to structurally facilitate single and double loop learning.
Illustrations of how the circular structure tends to increase learning capacity
include the response to the crisis situation in the 1970s, the re-introduction of
the circular system by way of education and training, and more recent
situations characterized by opposing interests.
The following precepts of circular organizing introduce a mechanism which
appears to permit any organizational member, irrespective of her formal
hierarchical position, to correct (the implications of) defensive behavior and
stimulates learning at the individual, group and organizational level:
.organizing into circles, while retaining the hierarchical structure for
managing the workplace;
.decision making by consent (or ``no argued objection''), which facilitates
public debate and free inquiry on the basis of arguments rather than
vested positions;
.double linking between circles, which promotes both upward and
downward communication; and
.election by consent after open discussions.
The key reason for this is that the circular design appears to be inextricably
bound up with a structure that facilitates free and open inquiry. This kind of
design is different from the classical hierarchical organization, by
superimposing a circular structure on the administrative hierarchy. In this
respect, the circular design involves two organizational structures, the
administrative hierarchy serving the needs of coordination and management of
current work processes and the circular structure producing an atmosphere of
open and free inquiry in search of new policies, mental maps, and so forth.
Endenburg developed his organizational model largely on the basis of key
ideas taken from system dynamics. A core notion in system dynamics, as it is
currently understood, is the idea of generic structures (Lane and Smart, 1996).
Generic structures are viewed to be important not only because of their
pervasive influence on human behavior, but also because the leverage effect of
well-designed interventions at the level of generic structures can be much
greater than those at the level of behavior only (Senge, 1990). Most studies in
the area of organizational learning emphasize the need to shape effective
organizational behavior without much regard for (re)designing the structural
context (e.g. power relations and organizational designs) in which this behavior
takes place (e.g. Argyris and SchoÈn, 1978; Argyris, 1996; Romme and Dillen,
1997). As such, the field of organizational learning may benefit from research
acknowledging the complex interplay between structure and behavior, rather
than focusing almost exclusively on ``behavioral treatments''.
The patterns of behavior that emerge from the case study of circular
organizing in Endenburg Elektrotechniek, suggest that the circular design
tends to support and promote processes of open and free inquiry, which reflects
the key condition for rigorous public testing in circle meetings throughout the

JOCM
12,5
452
company. In addition, by incorporating several minimal norms for circular
organizing in the company's legal form and by adopting the social accountancy
audit in order to control whether these norms are met, the company has created
a kind of legal safety net that may prevent the circular structure and the
ongoing process of public inquiry to break down. Again, the focus on
behavioral treatments apparently reflects only part of the story because they
may be insufficient for producing sustainable change toward organizational
learning. With behavioral interventions, change depends on the willingness of
the current group in charge to create momentum in the direction of (double
loop) learning behavior, such as in the case described by Argyris (1993).
However, organizational participants come and go, at all levels of the
organization, and without a structure which (to some extent) institutionalizes
and facilitates learning this momentum easily breaks down.
In this respect, the case of circular organizing incorporates what may be
called triple loop learning. Recall that single loop learning involves error
detection and correction which enable the organization to carry on with its
present policies and objectives, so focusing on the question: Are we doing
things right? Double loop learning involves error detection and correction in
ways that involve changing the underlying norms, policies and objectives,
implying that the question here is: Are we doing the right things?
Acknowledging that rightness is often buttressed by mightiness and
mightiness is often buttressed by rightness, which results in very little learning
at all (Flood and Romm, 1996), triple loop learning addresses the question
whether people really have the opportunity and competence to participate in
making well-informed choices in the process of discussing and managing
issues that concern them. As such, triple loop learning is about linking together
all local units of learning in one overall learning infrastructure as well as
developing the competences and skills to use this infrastructure (Flood and
Romm, 1996). The case of circular organizing in Endenburg Elektrotechniek
shows how such an infrastructure can be built by bringing together in one
overall circular (and underlying hierarchical) structure the inquiry into:
.Are we doing things right (single loop learning)?
.Are we doing the right things (double loop learning)?
.Can we participate in making well-informed choices regarding strategy,
objectives, etc. (e.g. triple loop learning)?
Conclusion
The organizational learning literature describes different levels of learning in
order to understand the complexity and dynamics of changes in policies,
objectives, mental maps, and structures and strategies for learning. This article
explores the case of an emerging new organizational design, the circular
organization, in order to understand the role of triple loop learning. The circular
model was developed on the basis of ideas about the relationship between
organizational structure and behavior taken from theories of dynamic systems.

Circular
organizing
453
Circular design precepts appear to provide a structural facilitation of single and
double loop learning. In this respect, the circular design tends to act as a
facilitating infrastructure for so-called triple loop learning, which explicitly
explores the structural opportunities and key competences people need to
participate in making well-informed choices about policies, objectives and other
issues.
References
Argyris, C. (1993), Knowledge for Action, Jossey-Bass, San Francisco, CA.
Argyris, C. (1996), ``Unrecognized defenses of scholars: impact on theory and research'',
Organization Science, Vol. 7, pp. 79-87.
Argyris, C. and SchoÈn, D.A. (1974), Theory in Practice: Increasing Professional Effectiveness,
Jossey-Bass, San Francisco, CA.
Argyris, C. and SchoÈn, D.A. (1978), Organizational Learning, Addison-Wesley, Reading, MA.
Argyris, C., Putnam, R. and McLain Smith, D. (1985), Action Science: Concepts, Methods, and
Skills for Research and Intervention, Jossey-Bass, San Francisco, CA.
Bateson, G. (1973), Steps to an Ecology of Mind, Palladin, London.
Endenburg, G. (1992), Sociocratie als Sociaal Ontwerp, doctoral dissertation, Eburon Publishers,
Delft.
Flood, R.L. and Romm, N.R.A. (1996), Diversity Management: Triple Loop Learning, Wiley,
Chicester.
Forrester, J.W. (1961), Industrial Dynamics, MIT Press, Cambridge, MA.
Forrester, J.W. (1971), World Dynamics, MIT Press, Cambridge, MA.
Lane, D.C. and Smart, C. (1996), ``Reinterpreting `generic structure': evolution, application and
limitations of a concept'', System Dynamics Review, Vol. 12, pp. 87-120.
Louis, M.R. (1994), ``In the manner of friends: learning from Quaker practice for organizational
renewal'', Journal of Organizational Change Management, Vol. 7, pp. 42-60.
Richardson, G.P. (1991), Feedback Thought in Social Science and Systems Theory, University of
Pennsylvania Press, Philadelphia, PA.
Romme, A.G.L. (1995), ``Non-participation and system dynamics'', System Dynamics Review,
Vol. 11, pp. 311-19.
Romme, A.G.L. (1997), ``Work, authority and participation: the scenario of circular organizing'',
Journal of Organizational Change Management, Vol. 10, pp. 156-66.
Romme, A.G.L. (1999), ``Domination, self-determination and circular organizing'', Organization
Studies, Vol. 20 No. 5, forthcoming.
Romme, A.G.L. and Dillen, R. (1997), ``Mapping the landscape of organizational learning'',
European Management Journal, Vol. 15, pp. 68-78.
Romme, A.G.L. and Reijmer, A. (1997), ``Kringorganiseren en het dilemma tussen centrale sturing
en zelforganisatie'', M&O, Tijdschrift voor Management en Organisatie, Vol. 51 No. 6,
pp. 43-59.
Senge, P.M. (1990), The Fifth Discipline, Doubleday Currency, New York, NY.
Snell, R. and Man-Kuen Chak, A. (1998), ``The learning organization: learning and empowerment
for whom?'', Management Learning, Vol. 29, pp. 337-64.