Butterfly paper

1
A slightly edited version of this was published in VCAT Mediation Newsletter, 9 (Nov) 2003: 19-23.

The butterfly’s wing, bread pudding and justice: the
relevance of complexity theory to ADR processes
By STEPHEN MUGFORD

Introduction1

Complexity theory has been a major scientific development in recent years, extending our understanding of a wide
range of natural and social phenomena. Nonetheless, as the theory and its close cousin 'chaos theory' have entered
into popular parlance, some parts of it are commonly misunderstood.

In this paper, I look briefly at a couple of central concepts of complexity theory and seek to clarify their broad
implications. I then use these concepts as criteria against which to consider ADR processes. Along the way, I draw
upon experiences in ADR work, specifically a major design and evaluation project with the Australian Federal Police
in the mid-1990s and a recent visit to a Peace Project in South Africa.

Some concepts from complexity theory
A full explanation of complexity theory and its mathematical underpinning is not only beyond the scope of this
article but also well beyond the capacity of the author. Instead, this section seeks to outline a couple of key
concepts and their implications.

I begin with the concept of emergence and the related ideas of ‘top-down’ versus ‘bottom-up’ organisation.

Faced with a complex adaptive system—such as an anthill—a common tendency is first to marvel at the intricacy of
it and then to wonder what design principles create this complexity from the top down. Indeed, when scientists
began to explore how such systems might be simulated on computers, the tendency was to develop complex, top-
down rule systems to generate the simulation.

However, we now know from both simulation and observational study that complex adaptive systems, including
life-like simulations, emerge bottom-up. For example, ants have little or no consciousness. Instead they obey a few
simple repetitive rules, yet from the interaction of many ants following these rules emerges the exquisite
complexity of the ant colony. In the words of one pair of writers, “something comes from nothing” (see Nine Laws
of God).

In contrast, top-down systems do not produce complex adaptive systems.

It follows that a simple design criterion for such a system is to find the minimum set of principles from which
dynamic and functional systems might emerge.

A second concept we should examine is the 'butterfly effect', or in common parlance, 'the butterfly’s wing'. This
concept was developed in the 1960s by scientists examining weather forecasting. They found that the weather
system was so complex and so sensitive to initial conditions that if they re-ran the simulation a number of times
with tiny variations in starting conditions, vastly different predictions could occur. It was as if adding the beating of

2
a butterfly’s wing into the equation was enough to tip the forecast four days ahead from ‘fine and mild’ towards a
hurricane.

This is a powerful, but frequently misunderstood concept. It is at times taken to mean one (or both) of two things:

1. that tiny inputs must have massive consequences;

2. that systems are unpredictable and uncontrollable and that we should be cautious about experiments as
unpredictable outcomes may occur (this is a central thesis of the novel/film Jurassic Park).

However, this is not an accurate interpretation of the butterfly effect, mainly because these points apply under
some conditions but not others. It is true that there are systems which are immensely sensitive to initial conditions
or to input perturbations as they unfold. As a simple example, try dropping the ‘same’ tennis ball from the ‘same’
point onto the ‘same’ flat floor time after time and note how no two sets of bounces are identical and how, despite
your best intent to control variation in starting conditions, the ball ends up at different spots when it comes to rest.

In very simple terms, we can call such systems fragile.

There are, however, quite different systems. The doyen of complexity theory, John L. Casti, calls these ‘locally
stable’ and gives a wonderful example: bread pudding (Casti, 1994: 88-89). He notes that if you have a good bread
pudding recipe, small variations in input (the age or type of bread, the exact amount of sugar, etc) or of
perturbations along the way (the exact temperature of the oven, the exact number of seconds the pudding cooks)
will have little effect. Unless you make a major variation in input or process, you will get recognisable and palatable
bread pudding every time.

For simplicity’s sake, we might think of locally stable systems as robust.

From these three simple conceptual points we can construct an argument about the criteria for systems that we
might want to design, namely:

Seek a combination of core principles that will allow complex, adaptive responses to emerge bottom-up in a robust
way and avoid anything that shows itself to be fragile.

ADR systems: butterfly’s wing or bread pudding?
I would like briefly to recount two examples of ADR systems I have been involved with. The first exhibited persistent
fragility—the butterflies were everywhere—while the second is locally stable: a good bread pudding.

A fragile system: attempting to deliver justice to drink-drivers

The fragile system was an attempt to use diversionary conferencing principles to deliver justice to drink-drivers in
the Australian Capital Territory (ACT) in the mid-1990s. This was a part of the larger RISE (re-integrative shaming
experiment) initiated by John Braithwaite at the Australian National University and, under the direction of
Lawrence Sherman and Heather Strang, implemented by the ACT region of the Australian Federal Police.

A colleague of John’s at the ANU, I was (and remain) a strong supporter of restorative justice in general and
conferencing in particular, which I know to be a powerful and effective method. So, when the AFP approached me
to evaluate the drink-driving component on an action-research basis, I was keen to be involved.

3
From the beginning, however, it was clear that this was a difficult task. Although the AFP members who I worked
with were committed to this enterprise and worked with dedication and imagination, it was a non-stop struggle to
make conferences work.

At the core of this, we thought, was the fact that the offenders—drivers apprehended by random breath tests with
mid-high range alcohol readings—often felt unlucky to be caught and thus were resentful, rather than guilty, a
feeling that we could not contradict by pointing to the victim, because there was no victim in the simple and literal
sense.

Numerous interventions were developed—such as showing a purpose-made video incorporating interviews with
police, accident and emergency personnel, relatives of people killed in crashes and an undertaker who had
attended fatal accidents, interwoven with graphic and gruesome material collected by the AFP Video Unit who
attend all fatal crashes. We also had community representatives from the suburbs through which the offender had
been driving and who talked about how they felt to know that they and their families might be at risk.

Many of these interventions were honed to a fine edge and in some cases outstanding conferences emerged. In
these, people appeared to experience genuine remorse for their actions and there were also powerful effects on
the ‘supporter group’ members who, though they had not offended, were once more informed of community
sentiment. When such a conference occurred—and they were not rare—the whole effort seemed worthwhile.

However, these sessions were no more common than those where it was a struggle to establish a worthwhile
discourse and not much more common than the minority that were very unsatisfactory, with offenders and their
supporters seemingly unaffected by the experience.

Moreover, even with the best conferences, the final phase of ‘restoration’ was often strained, and in less
worthwhile conferences bordered on embarrassment.

In a conventional conference where there are ‘real’ victims present, a closing theme of ‘what should happen now?’
runs quite smoothly: most people present, especially the victim and his/her supporters, have a clear sense of what
the harm was and what combination of apology and/or restitution and/or reparative actions would be suitable. So
it might be that a victim who has suffered property damage at the hands of a young offender may be quite satisfied
with an apology, the making good of the damage and perhaps some evidence that the offender will seek to mend
their ways.

In the drink-driving context, however, this phase often lacked direction. Into the vacuum of lack of ideas (many
people were baffled by the request to know ‘what should happen next?’) facilitators eventually came to offer
suggestions, noting that in similar cases, previous offenders had done X, Y or Z.

Since each facilitator’s experience was slightly different and since each had his or her views of a good outcome, it
came to pass that one could guess the facilitator’s identity from an outcome: if there was a donation to the Brain
Injury Foundation it was likely to have been run by A, if it was an agreement to be a blood donor on several
occasions, it was likely to be B, and so on.

At the time, we struggled persistently, even heroically, to make this system work, finding ‘patches’, improving
training, and so on. We interpreted our failure to find a recipe almost as indicative of personal failings: of resolve,
intellect, imagination or whatever. We added more and more ideas and expertise but all to no avail. The process
remained unreliable and unpredictable.

More to the point, we were attempting to solve the problem ‘top down’ (in the sense of trying to create reliable
and detailed procedures).

4

In hindsight, however, it is clear that this was an inherently unstable context and that the idea of conferencing this
category of offender inevitably led to fragile procedures. In the absence of a real victim, the process could not be
relied upon to produce the right types of ‘affect’—that is, initial remorse, guilt, etc., followed by the positive affects
surrounding re-integration.

A robust system: the Community Peace Program in Capetown

Negative outcomes of this kind are, however, not inevitable. By way of a sharp contrast, I would like to recount a
recent experience visiting an ADR process that seemed as robust—or more accurately, locally2, stable—as the
previous one was fragile.

In the latter part of 2002 I was invited by Clifford Shearing (currently of the ANU’s Regulations Institutions Network,
or Reg-Net) to evaluate the work of the Community Peace Program (CPP) in Capetown, Republic of South Africa.

The CPP runs two principal activities: peace making and peace building, and it is the former I will concentrate upon
here.

Peace making—the resolution of local disputes in townships by township dwellers—is carried out by Peace
Committees. These are volunteers who are organised by local workers employed by the CPP. Using funds provided
partly by local government and partly by overseas philanthropic organisations that support community
development in the Republic, for each conference that takes place a donation is made to a local community
development fund (controlled by the Committee) and a matching donation is divided between the members as a
reward for participating. (Interestingly, while the ‘economically rational’ actor would restrict the number attending
in order to increase the per capita return, the actual pattern appears to be to seek wide involvement.)

When a dispute is identified and when parties to that dispute are willing to attend, a peace making ceremony or
‘Gathering’ is undertaken. The direction for Gatherings is always to gather together those people thought to have
the knowledge and capacity to be able to contribute to a plan of action that reduces the likelihood of the conflict
continuing.

A Gathering follows a simple formula in which each party is allowed in turn to state their case fully to the
committee, who may ask clarificatory questions, which leads into a period of discussion and (in the large majority of
cases) an agreed resolution. Often the Committee may be involved in actions to carry out the agreement, e.g.,
organising a meeting where a person pays a debt or returns an item. Matters resolved are often ‘small’ on a global
scale, yet they have the capacity to produce major conflict if left unresolved.

A basic, standard written recipe is provided with some key steps that must be followed and volunteers are trained
to follow those steps3.

This links with a central design philosophy used in developing the CPP processes, namely a concern with
standardisation on a simple and robust recipe (which was often referred to colloquially in conversations with key
staff as the “McDonald’s model” or “making the hamburger”).

Ritzer (1993) following in Weber’s footsteps, identifies the four core processes that McDonald’s employ—and
which were utilised to some extent in the Peace Making ceremonies—as:

• Efficiency
• Calculability
• Predictability

5
• Control

Notwithstanding standardisation of the basic process, however, room exists for local knowledge and variation. For
example, in one township I visited the local people generously interpreted the rules about who could and should be
present at a ceremony so as to include some local opinion leaders as, in their view, this added both weight and
legitimacy to the endeavour. It worked.

In this regard, the program is based upon the notion that local knowledge and expertise is a critical resource and
that this must be elicited respectfully and combined productively with any outside ‘expert’ knowledge. As Toulmin
(1990) puts it, the:

… “modern” focus on the written, the universal, the general and the timeless—which monopolized the
modern work of most philosophers after 1630—is being broadened to include once again the oral, the
particular, the local and the timely.” (p. 186)

Indeed, one of the beauties of the ceremonies is that they are, in a simple sense, expert free—the local people
utilise the resources of the model to create a robust and effective ADR process that fits their needs.

Conclusion
Earlier in the article I suggested a simple design principle which should be applied to ADR processes, namely:

Seek a combination of core principles that will allow complex, adaptive responses to emerge bottom-up in a robust
way and avoid anything that shows itself to be fragile.

As I have shown, it is possible to design an ADR system that meets this principle (such as the Peace Making
ceremonies) but cautionary tales exist of others that do not (such as the drink-driving conferences).

In a sense then, the principle is also a criterion for a good ADR system although it is a criterion of a necessary rather
than a sufficient condition.

Other criteria are also necessary, not least that an ADR system produces healthy outcomes that conform to our
sense of justice. It is, unfortunately, all too easy to imagine a system which is ‘robust’ in the sense intended above
but not just: vendetta pay-back might be an obvious example!

As a very simple rule of thumb, however, the major lesson seems to be that if you are working with any ADR system
that seems to need lots of tweaking and correcting, with ceaseless vigilance to ensure reasonable outcomes, then
there is a good chance that you are in a zone of local instability trying fruitlessly to cope with fragility, almost
deafened by the sound of the butterfly’s wings. If so, the message is simple: stop, this will never work, look for a
good bread pudding recipe instead.

6
Footnotes
1. I am grateful to Clifford Shearing and John Braithwaite who read an earlier draft of this paper and made helpful
comments.

2. The word ‘locally’ is apposite here. In transferring the system to other cultures—such as Argentina—a fair amount of
adjustment has been necessary. This suggests that there is a deal of local knowledge embedded in the system and, as Dave
Snowden (2002) has said in regard to knowledge management, “We can always know more than we can tell, and we will
always tell more than we can write down …[and]… We only know what we know when we need to know it” (Snowden, 2002).

3. For readers interested in knowing more detail about the CPP and its procedures, I suggest reviewing
www.ideasworks.org or contacting Clifford Shearing.

References
Casti, John L. (1994) Complexification: Explaining a Paradoxical World Through the Science of Surprise. Abacus.

Nine Laws of God, see http://www.fs.fed.us/eco/eco-watch/ew950726.htm

Ritzer, George (1993) The McDonaldisation of Society, Pine Forge.

Snowden, Dave (2002) Complex acts of knowing – paradox and descriptive self-awareness. Journal of Knowledge Management
(special issue, July).

Toulmin, Stephen (1990) Cosmopolis: The Hidden Agenda of Modernity. Free Press.

Dr Stephen Mugford is a former academic sociologist who joined the ANU in the mid-1970s where he taught until
1997. He now manages Qualitative & Quantitative Social Research (QSSR), a small specialist consultancy
company that carries out evaluation research and also provides capacity building and change management
services. He is also a Visiting Fellow in military sociology at the Centre for Defence Leadership and Ethics within
the Australian Defence College.

Dr Mugford has a long-standing interest in ADR and restorative justice and is currently evaluating the
introduction of ADR into the Australian Defence Organisation. Dr Mugford may be contacted at
Stephen.mugford@qqsr.com or Stephen.mugford@defence.gov.au

Butterfly paper

Related slideshows

More Related Content

Butterfly paper