Content uploaded by Gerald Midgley
Author content
All content in this area was uploaded by Gerald Midgley on Feb 21, 2021
Content may be subject to copyright.
1
Bringing foresight into systems thinking:
a three horizons approach
Anthony Hodgson and Gerald Midgley
Abstract
A primary goal of systemic intervention is the improvement of the ‘system in question’.
The definition of the system in question is often a function of multiple stakeholders and is
not a fixed object. Boundary critique can be helpful in clarifying the ambiguity,
assumptions and power dynamics around agreeing what the system is that is to be
improved and for whose interests. However, there is another dimension of ambiguity,
which is time. Improvement implies some change from a present condition A to a better
future condition B, which eventually becomes a new present condition B. Where the
environment is about to go through a significant change of pattern (a paradigm shift), the
criteria of improvement will be different, depending which paradigm is being considered.
For example, energy success in an unrestricted environment can be completely different
from energy success in a restricted environment dominated by climate change. In this
paper we will introduce a systemic foresight method called the three horizons. It will
point out three distinct ways of looking at the future, each of which will strongly colour
boundary critique and therefore affect what is considered to be a successful or ‘improved’
system. It is possible to look at improvement to sustain the current system; improvement
consisting of a disruptive innovation, which may reconfigure the system; and
transformative improvement, which may actually result in the collapse or abandonment
of the old system in favour of something new. Choices between futures can be hotly
contested and the source of stakeholder conflict, so the systemic management of
transitions is essential.
Key Words
Systemic intervention, Foresight, Time, Boundary critique, Critical systems thinking, Three
horizons
Centre for Systems Studies, Business School, University of Hull, UK
International Futures Forum
Decision Integrity Limited
Centre for Systems Studies, Business School, University of Hull, UK
School of Innovation, Design and Engineering, Mälardalen University, Sweden
Victoria Management School, Victoria University of Wellington, New Zealand
School of Agriculture and Food Sciences, University of Queensland, Australia
School of Political and Social Sciences, University of Canterbury, New Zealand
2
Taking up the theme of the conference, learning across boundaries, in this paper we wish to
explore, not boundaries in space or connoting inclusion, but boundaries in time; hence the
title, bringing foresight into systems thinking. Although a review of the literature shows some
references in the systems literature to futures thinking, and although a similar review of the
literature shows some references in the futures studies material to systems thinking, the
connection between the two appears marginal. We are dissatisfied with this gap and are
researching and learning across these fields. We would like to share some of our very
provisional thoughts at this stage of the research.
A primary goal of systemic intervention is the improvement of the system in question.
The definition of ‘the system in question’ is often itself a function of multiple
stakeholders and is not a fixed object. Boundary critique (Midgley 2000) can be helpful
in clarifying the ambiguity and the power dynamics around agreeing what the system is
that is to be improved, and for whose benefit. Let us describe some essential features of
systemic intervention:
For a given situation in question,
Systemic means attending to wholes, connectedness and non-linear behaviour, with a
special emphasis on boundaries concerning who and what is included, excluded or
marginalized
Intervention means purposeful action by an agent to create change, and
Systemic Intervention therefore means purposeful action, incorporating reflection on
boundaries, aiming to bring about some improvement.
Clearly, what constitutes improvement is a judgement call (Churchman, 1970), and like any
judgement call depends on the stakeholders making it. Midgley and Pinzón (2011) point out
that one of the most critical judgements prior to settling the specific meaning of improvement
in a given case, for example of conflict resolution, is the determination of the boundary of the
system.
Figure 1 – Basic form of boundary critique
Figure 1 represents a simple, basic form of boundary critique for a single stakeholder, or
where there is agreement between multiple stakeholders. The outer broken line determines
the boundary between what is perceived as relevant to the intervention and what is considered
3
irrelevant. The inner broken line represents the negotiated agreement as to the system in
question, which is subject to the intervention. The peak represents the ‘centre of gravity’ of
the sense of identity and underlying values that make the system in question meaningful. In
the case of stakeholders in contention, it would be necessary to use several of these diagrams
with degrees of overlap. This is a more sophisticated analysis not dealt with in this paper (but
see Midgley, 1992, 2000; and Midgley and Pinzón, 2011). An important element of boundary
critique is not taking boundary judgements for granted, but comparing and contrasting
different possibilities for setting boundaries, in order to explore the likely consequences for
stakeholders and the issues that concern them (Ulrich, 1983).
The boundaries represented in the above exposition are essentially about content, and tend to
be spatial and semantic in character; who and what is included or excluded. Mapping
boundaries helps to surface assumptions and clarify agreement about what is in and what is
out of the system in question. In this paper we wish to suggest that it is also worth
considering the dimension of time within boundary critique.
In foresight practice, we talk about ‘temporal windows’. Typically, forecasts adopt either
short windows (econometric models project between 6 to 24 months into the future, and
strategic planning usually proceeds in 5 year steps) or very long windows (80-100 years, such
as when forecasting climate change). Foresight authors often use the term ‘looming issues’.
Kelly et al (2002) summarize the challenge of the middle region of around 10+ years, where
more established futures methods show diminished effectiveness. Thus
“..businesses must learn to anticipate and adapt more quickly to an increasingly
complex environment in which many political, economic, social, cultural, and
technological forces are shifting, interacting, and even colliding.” p3
However, to relate time to the systems idea it is necessary to go beyond the simple linear,
sequential view of time (short, medium and long term). This is especially true where the
situation of interest is going through some kind of transformative change. We need to
consider the qualities of the temporal window of interest.
One way of doing this is to distinguish between the kinds of systemic improvement that have
a sustaining role and those which have a transforming role, as shown in Table 1.
Table 1 – Sustaining or Transforming Improvement
Sustaining
Transforming
Supporting and reinforcing the
current dominant pattern
Relatively restricted specialist
viewpoint on ‘systems in question’
Innovation that is captured to prolong
the status quo
Shifting to quite a new pattern with
transformed fit to a radically different
context
Wide angle view of the ‘system in
question’, taking a holistic approach
Disruptive innovation which renders
the status quo obsolete
4
Shifting to a new pattern involves some form of strategic thinking and foresight (Miller
2007). I will describe the various disciplines of foresight and futures studies using a
framework introduced by Sharpe (2008). Foresight can be classified into four types according
to the extent to which the decision maker has agency to do things and the degree of
uncertainty they are facing. This distinction classifies four basic modes of futures methods, as
shown in Figure 2.
If the decision maker has relatively low agency, for example when planning in a going
concern, and the operating environment is relatively stable and certain, then the classical
methods of forecasting followed by resource planning in relation to those forecasts can be
effective. These methods generally assume a predictable world where, for example, the
measurement of past trends can be extrapolated into the future without any problem. The
limitation of these methods is that they assume the continuity of a fundamental pattern with
perhaps minor incremental changes. Innovation will tend to be dominantly reinforcing the
status quo.
If the decision maker has high agency in a relatively stable and certain operating
environment, as for example when implementing the rollout of a proven manufacturing
enterprise, then the method of roadmaps into the future applies. Galvin (1998), the former
Chairman of Motorola, summarises their nature and role thus:
“A ‘roadmap’ is an extended look at the future of a chosen field of inquiry composed
from the collective knowledge and imagination of the brightest drivers of change in that
field . Roadmaps communicate visions, attract resources from business and
government, stimulate investigations, and monitor progress. They become the inventory
of possibilities for a particular field. In engineering, the roadmapping process has so
positively influenced public and industry officials that their questioning of support for
fundamental technology support is muted.” p803
Figure 2 – Four Broad Categories of Foresight Method
5
If the decision maker has relatively low agency but faces a very high level of uncertainty,
then the method of multiple future scenarios applies. The origins of scenario planning in
Shell illustrate this (Wilkinson and Kupers, 2013). Although Shell is a massive international
company, its size and impact relative to the total energy market and the geopolitical context
of energy indicates that it has relatively low agency given all the other forces at work. Also,
the uncertainties over a 20 to 30 year exploitation time span surrounding the geopolitical and
geological conditions necessary for economic oil extraction and refinement are considerable.
The basic scenario method researches these uncertainties and distils them down into several
alternative possible futures. These are not forecasts, and nor are they roadmaps of action
pathways to be taken, but rather they are narratives of possible future conditions based on the
imaginative interpretation of a wide range of sources of information. Their application in
strategic thinking is sometimes referred to as ’wind tunnelling’ (van der Heijden 2005), in
which the strategic direction of the company is tested for robustness in more than one
potential future.
However, the range of concepts and practices associated with the term ‘scenarios’ has greatly
enriched since scenario planning was first launched in Shell around 1970 (Wilkinson and
Kupers, 2013), and certain aspects would now appear in all four boxes. Bishop et al (2007)
propose a classification based on eight kinds of technique. Van Notten et al (2003) position
methodologies according to the ways in which they treat project goals, process design and
scenario content. Yet another approach is that of mode-level analysis proposed by Voros
(2006), which is based on a set of thinking modes combined with a series of interpretive
levels to analyse prospective methods in terms of which mode(s) and what level(s) they
operate with or at. Cross-referencing the dimensions of agency and uncertainty as proposed
by Sharpe (2008) and portrayed in Figure 2, brings yet another perspective.
It is interesting that, in the foresight disciplines, there is still relatively little methodology for
the fourth box where both agency and uncertainty are high, and yet this is the area which is
increasingly the operating environment for government, commerce and society more
generally. One expression that describes this fourth area is ‘reflexive futures’. This might also
be characterised as strategic exploration. The high agency component of the decision making
is reflected in a practice of setting a strong vision of a future state of affairs in which the actor
is occupying a desired position (much as Ackoff, 1981, recommends in his Interactive
Planning systems approach). The uncertainty component of the decision making is treated in
qualitatively different time zones, each with its own dynamic. By ‘qualitatively different’ we
mean features like the differences between predictive, transformative and emergent. In this
way the decision maker is operating in an expeditionary mode where progress is made
according to current knowledge but strong feedback is incorporated into adapting any
decision to match the emergent circumstances rather than simply ‘pushing it through’.
‘Reflexive futures’ implies adaptive strategy and adaptive leadership (Pascale, 1999).
Despite the fundamental differences between the high agency / high uncertainty context and
the others in figure 2, there is still a strong tendency to apply methodologies aligned to those
other three contexts, thus leading to inconsistency between theory (figure 2) and foresight
6
practice, and therefore inadequacies and even sometimes disasters in decision making. Miller
(2006) sums up the essence of this situation:
“Three of the many reasons for this inconsistency between theory and practice
were/are: a fear of the future that drives a deep desire to know (divine) what will
happen (clients want predictions); recent (post WWII) relative systemic stability and
the related success of planning in this context; and lack of experience with, and hence
under-development of, the conceptual tools and behavioural conventions that make it
practical to embrace non-predictive approaches to decision making with success. Old
paradigms do not cede easily and the attachment to predictive approaches rooted in
trend analysis, forecasting models, multi-factor calculations, etc. are tightly integrated
with the way risk is managed and decisions taken in industrial society.” p341
Going back to systemic intervention, we argue that the kind of improvement that
organizations are interested in will be a function of how they see the future of the system in
question. If the improvement is to move from a sustaining pattern to a transforming pattern,
then a significant shift is involved, sometimes referred to as a ‘paradigm shift’ or a change of
disciplinary matrix (Kuhn, 1977). This places the requirement for foresight firmly in the
fourth zone. Hodgson (2007) describes the difficulties associated with facilitating strategic
decision making in the transformation zone:
“Part of the assumptions set is the view of time and change held by the strategy owners.
Consider the example of a strategy based on an analysis of trends, and aiming to reach a
certain goal over a time span in which the trend is still valid. Two ways of looking at
this can be called the linear and the sophisticated. The linear is most common and
relates to the cognitive difficulty people have in visualising trend bends driven by
nonlinear dynamics, for example exponential curves. The sophisticated version takes
into account nonlinear trends but still places them within the shape of environment or
context. If a new business ecology emerges or there is a paradigm shift then this
discontinuous view of time leads to conclusions that are far off the mark.” p 290
In other words, even sophisticated methods can be locked into the old paradigm. To deal with
these distinctions in anticipating the future, Hodgson and colleagues have developed a new
framework called the ‘three horizons’ (Sharpe and Hodgson 2007; Curry and Hodgson 2008:
Sharpe, 2013), which co-defines the different qualities of time range that reflexive futures
thinking requires. The framework is described in Figure 3.
7
Figure 3 Looking into the Future with Three Horizons
In Figure 3, the line labelled Horizon 1 represents the viability in a given context or
environment of the current dominant pattern or structure of the system in question. Its
viability is considered to be degrading due to an increasing mismatch with the changing
external conditions. The line labelled Horizon 3 represents a different transformed pattern
which, although seemingly with minor significance early on, turns out to be a much better fit
with the changing external conditions and so becomes the dominant viable pattern. The line
labelled Horizon 2 represents the turbulent, even chaotic, situation of transition from one
pattern to another. An example of this is disruptive technology, which may stimulate both
creative innovation and resistance in equal measure until the technology becomes normal
(Christensen, 1997). It is often not culturally feasible in real world business environments
(due to conservative attitudes, sunk investments, etc.) to jump straight to Horizon 3 early on,
when the strategic fit appears low. For this reason, compromises are made and a turbulent
tension is experienced between competing pressures to conform to Horizons 1 and 3. The
resulting trajectory is represented by Horizon 2.
The three lines in figure 3, taken together, unfolding over time, represent three potential
futures. In particular circumstances, Horizon 1 may prevail by sustaining innovation that
merely reinforces the status quo. An organization may strongly push its ‘success’ formula,
not realizing that it is out of date, and may ultimately overshoot relevance and collapse
(Sterman, 2000). This is what happens when companies resist hearing the ‘bad news’ that
more radical change is needed. In other circumstances Horizon 3 may prevail, when a far-
sighted company spots a change in the environment and moves wholeheartedly to align with
it, even if this means lower profits in the short-to-medium term. This is not the most usual
reaction to accepting the importance of change, however. More often, companies find
8
themselves on a bumpy Horizon 2 trajectory, as they seek to chart a strategic course that
keeps stakeholders on board who are advocating both Horizons 1 and 3. So Horizon 2
represents a turbulent but not sudden transition.
Note the way the lines are drawn is such that they are all present in any given slice across the
time axis. By way of clarification, this can be thought of in three stages. At t1 the H1
‘business as usual’ world is dominant but beginning to fail; H2 innovation is on the
ascendency; H3 is still a fringe possibility often discounted or unnoticed. At t2, H1 is already
in serious decline; H3 is noticed and is beginning to attract resources that displace H1; and
the innovation in H2 is generating a high variety of experimentation. At t3, H3 has become
the dominant paradigm now much more fit for purpose than H1. However, and importantly,
H1 has not gone away. Essential elements are maintained but in a different role, often more
related to basic infrastructure. H2 is calming down as H3 is stabilising the new pattern. Of
course, in the fullness of time and with constant change, H3 will become the new H1.
This three horizon approach offers the prospect that different systems methods might be more
useful and appropriate depending on which aspect of foresight is being attended to.
Alignments cannot be precise, but more a reflection of the predominant orientation of any
given systems method.
Operational research methods that are linear and algorithmic are useful in the more
deterministic and predictive world of forecasts and extrapolation. Systems methods that
account for dynamic feedback, such as system dynamics (e.g., Sterman, 2000), align better
with the world of roadmaps, where performance on the pathway is subject to strong feedback
influences. In the domain of scenarios, where there is high uncertainty, there is also increased
dependency on human judgement, so approaches like critical systems heuristics (Ulrich,
1983) and soft systems methodology (SSM) (Checkland and Poulter, 2006) fit well. Perhaps
the most promising systems approach for the reflexive futures category is the relatively
undeveloped field of anticipatory systems. Miller, Poli and Rossel (2013) are exploring and
developing a research network around what they are calling the ‘discipline of anticipation’.
This is a new field that is opening up and is of great interest to UNESCO.
An alternative approach is to organise systems methods from the perspective of the three
horizons. The qualitative distinctions between the horizons also suggests that different
systems methods may be more suited to problem-solving and decision-making according to
the horizon perspective adopted by the decision maker. A broad representation of this is
given in Figure 4.This is not intended to be comprehensive or finalized, but rather to suggest
an approach for further investigation.
The key question behind more detailed mapping of systems methods onto the horizon chart
relates to the structural pattern of the dominant horizon in a given time phase. This pattern
will to tend to restrict the behavioural range and lead to ‘lock-in’. Assuming our
improvement time horizon is short term in Horizon 1, then we are likely to be dealing with
well-established systems, however complicated they might be. Methods of systems analysis,
such as systems engineering, have been well developed for these situations. If our short-term
9
improvement is more innovative in Horizon 2, then techniques that are strong in exploring
feedback, such as system dynamics, would be more appropriate.
In the middle zone of turbulent transition, where Horizon 2 is dominant and the pattern shift
is disruptive, methods using assumption sets based on the previous history of system
behaviour and assuming continuity into the future will fall out of alignment with their
changed environment. This suggests approaches like SSM for accommodation of diverse
views in the midst of the turbulence. The future viability of a system in the radically different
operating environment of Horizon 3 might be modelled and anticipated by application of the
VSM through a strong system 4 (Beer, 1985). To increase the variety under consideration for
transition, especially with multiple stakeholders or disciplines, Team Syntegrity (Beer 1994)
is another interesting area to explore for hybrid understanding. Providing the group with
agency have truly shifted their perspective towards Horizon 3, then methods like interactive
planning (Ackoff, 1981) will be more appropriate, as these support people in learning into the
future and reinventing their activities.
In the longer term zone, where the strategic interest is in trying to fathom the radically
different nature of Horizon 3, then those methods which allow for more innovative
possibilities will be particularly relevant. Although anticipatory systems as an applied field
has not yet been well developed, there is growing interest in this form of systems thinking as
supportive to foresight work (Poli 2010). An interesting perspective comes from considering
what the qualitatively different attractors for Horizon 3 might be in the context of making a
transition across the fitness landscapes to a new stable state. Finally, since the assumption
sets will be rather different in each horizon mindset, the application of both critical systems
heuristics (Ulrich, 1983) and boundary critique (Midgley, 2000) could help clarify the whole
three horizon pattern and the different assumptions and boundaries between H1, H2 and H3.
This analysis is summarised in Figure 4. It must be stressed that these positions are indicative
and will still depend on how any practitioner decides to use them.
Figure 4 – Orchestrating systems methods for three horizons based foresight
10
A more comprehensive account of the different forms of systems thinking can be found in
Midgley 2003.
Conclusion
The ideas in this paper are based on the first author’s long experience of applying both
scenario planning and systems thinking in strategic management, and the growing sense that
there is an interesting and productive step to be made in stepping back and seeing the two
fields as complementary and even having some common underlying foundations. Poli (2011)
and Hodgson (2013) have explored this from a philosophical and theoretical orientation.
However, in terms of the more practical side of bringing foresight into systems thinking, we
believe it is valuable to think in terms of time boundaries, as seen in the three horizons
model. Table 2 examines the implications for understanding systemic improvement.
Table 2 – Three Modes of Systemic Improvement
Horizon 1
Horizon 2
Horizon 3
Identify and repair self-
defeating feedback.
Identify missing necessary
feedback.
Identify and displace self-
defeating feedback and
feedback that creates lock-in.
Challenge buried
assumptions.
Seek optimisation.
Seek unmet need.
Seek new visions and
patterns of viability.
Be efficient.
Be the disrupter.
Be the future.
The implications for systemic intervention practice can be summarised in five points:
1. Recognise that improvement is entirely dependent on a time horizon to realise that
improvement.
2. Build this consideration into the conduct of boundary critique to clarify the system in
question.
3. Raise awareness of whether the improvement desired is mainly H1, H2 or H3 in its
emphasis.
4. Determine which mode of improvement is most likely to generate the desired results.
5. Match the appropriate systems concepts, systemic problem structuring methods, etc.,
to the desired mode of improvement.
This paper aims to open up one approach for further research and consideration, and provoke
further questioning.
Acknowledgements
The first author is grateful to the Centre for Systems Studies, Hull University Business
School, for the opportunity to try out some of these ideas in a research seminar in March
2014. Useful discussions with, and feedback from, Andrew Curry, Roberto Poli and Bill
Sharpe have helped clarify the ideas in this paper.
11
References
Ackoff, R.L. (1981). Creating the Corporate Future. New York: Wiley.
Beer, S. (1985). Diagnosing the System for Organisations. Chichester: Wiley.
Beer, S. (1994). Beyond Dispute: The Invention of Team Syntegrity. Chichester: Wiley.
Bishop, P., Hines, A., & Collins, T. (2007). The current state of scenario development: an
overview of techniques. Foresight, 9(1). p20
Checkland, P. and Poulter, J. (2006). Learning for Action: A Short Definitive Account of Soft
Systems Methodology and its Use for Practitioners, Teachers and Students. Chichester:
Wiley.
Christensen, C. M. (1997). The Innovator’s Dilemma: When New Technologies Cause Great
Firms to Fail. Cambridge USA: Harvard Business School Press.
Churchman, C. W. (1970). Operations research as a profession. Management Science, (17).
Curry, A., & Hodgson, A. (2008). Seeing in Multiple Horizons. Journal of Futures Studies,
13(1), 1–20.
Galvin, R. (1998). Science Roadmaps. Science, 280.
Hodgson, A. (2007). Appreciating the Future. In B. Sharpe & K. van der Heijden (Eds.),
Scenarios for Success (pp. 291–299). Chichester: Wiley.
Kelly, E., Leyden, P., & and GBN. (2002). What’s next? Exploring the terrain for business.
Chichester: John Wiley and Sons.
Hodgson, A. (2013). Towards an ontology of the present moment. On the Horizon, 21(1), 24–
38.
Kuhn, T. (n.d.). Second Thoughts on Paradigms. Retrieved from
http://eu.pravo.hr/_download/repository/Second_Thoughts_on_Paradigms.pdf
Midgley, G. (2000). Systemic Intervention; Philosophy, Methodology, and Practice. New
York: Kluwer.
Midgley, G. (Ed.). (2003). Systems Thinking (Vols. 1-Volumes 1–1-4). Sage, London.
Midgley, G., & Pinzon, L. A. (2011). Boundary critique and its implications for conflict
prevention. Journal of the Operational Research Society, 62, 1543–1554.
Miller, R. (2007). Futures Literacy: A Hybrid Strategic Scenario Method. Futures, 39(4).
Miller, R., Poli, R., & Rossel, P. (2013). The Discipline of Anticipation: Exploring Key
Issues. UNESCO.
12
Pascale, R. T. (1999). Surfing the Edge of Chaos. Sloan Management Review, Spring.
Retrieved from http://sloanreview.mit.edu/article/surfing-the-edge-of-chaos/
Poli, R. (2010). The Many Aspects of Anticipation. Foresight, 12(3), 7–17.
Poli, R. (2011). Steps Toward an Explicit Ontology of the Future. Journal of Future Studies,
16, 67–78.
Sharpe, B., & Hodgson, A. (2006). Intelligent Infrastructure Futures: Technology Forward
Look - Toward a Cyber-Urban Ecology. London: UK Government Office for Science.
Sharpe, B. (2008). Third Horizon Innovation: finding ourselves in the future. James Martin
Institute for Science and Civilisation, Oxford University.
Sharpe, B. (2013). Three Horizons: the patterning of hope. Axminster: Triarchy Press.
Sterman, J. (2000). Business Dynamics: Thinking and Modeling for a Complex world.
Boston: Irwin McGraw-Hill. p123-127
Van der Heijden, K. (2005). Scenarios: The Art of Strategic Conversation (2nd ed.).
Chichester: Wiley. P283-283
Van Notten, P., Rotmand, J., van Asselt, M., & Rothman, D. (2003). An updated scenario
typology. Futures, 35, 423–443.
Voros, J. (2006). Introducing a classification framework for prospective methods. Foresight,
8(2), 43–56.
Wilkinson, A., & Kupers. (2013). Living in the Futures. Harvard Business Review, (May).
