Category Archives: action learning

Towards a Design for a Virtual Open Source Event

A Virtual Open Source Event (VOSE) aims to hold both sides of the dilemma identified in the previous blog asking what we can learn from going virtual. In the design below, the word ‘matrix’ is used in the Foulkesian sense of “the network of all individual mental processes, the psychological medium in which they meet, communicate and interact”.[1]

The design uses a virtual platform, for example Slack, that can provide synchronous and asynchronous communication among participants over an extended period of weeks. The participants come together because they have direct experience of a problem domain[2] and are wanting to develop new initiatives within it that have the possibility of adding value within that domain. Participants’ direct knowledge of and experience within the problem domain is crucial, whether from the supply-side or the demand-side, because this is necessary to the forms of learning that will arise in the matrices.

The work takes place within four matrices, each with its own moderator and each with its own distinctive focus.[3] Three of these matrices focus on different perspectives on the domain: the Witnessing, Plus-One and Parallel Process matrices. Participants bring these perspectives together in the fourth Development matrix as planned collaborations, in which the focus is to ensure that the spokes of each collaboration’s development wheel can be balanced.

The VOSE ends when a majority of its participants are ready to put their planned collaborations into action.

Plenaries: Two plenaries involve all participants and moderators. The Opening Plenary introduces the event. The Closing Plenary reviews the emergent learning.

Witnessing matrix: Participants form a matrix in which individuals identify and narrate situations in which there is a value deficit and/or injustice that has arisen within the chosen problem domain.  The rule in this matrix is to make no interpretations.

Plus-One matrix: Participants fishbowl around a 3-person plus-one process exploring an individual’s experience of a situation identified in the Witnessing matrix. The individual uses the emergent metaphors to formulate a counter-narrative and underlying dilemma implicit in his or her reading of the situation.

Parallel process matrix: Participants fishbowl around individuals focusing on their understanding of the parallel processes in a situation with its underlying dilemma that has emerged from the Plus-One matrix. The aim is to explore the strategy ceilings constraining the ways in which the dilemma is being responded to.

Development matrix: Participants form collaborations that bring together insights from the other three matrices to form initiatives that can be balanced across all the spokes of the development wheel.

Enaction: The VOSE exists to work through understandings of what is going on in the problem domain and the crises limiting what innovations are possible within the domain. Enaction involves moving to a third moment in which new collaborations can be realised.

Moderators: Each matrix has its own moderator supporting its work.[4] Moderators’ meetings are to provide peer support in sustaining the different kinds of relationship between the participants in each matrix. Participants can observe moderators’ meetings in ‘fishbowl’ and contribute issues for consideration.[5]


[1] The quote is from Foulkes, S.H. and Anthony, E.J. (1965) Group Psychotherapy: The Psychoanalytic Approach: Second Edition Karnac Books: London. The role of each matrix becomes apparent through the fishbowl effect created by the relationship of its participants at any one time to the working of a focal individual or group within it. In effect, the fishbowl turns the work of the focal individual or group ‘inside-out‘.

[2] A problem domain defines the ‘north star’ bringing people together because of their shared concern for what is going on within a domain. This is most apparent in the public and not-for-profit sectors, where issues of health, social services, urban economic growth, housing, transportation etc bring people together with the political will to make something happen. Such ‘north stars’ are not so apparent in the private sector because they take any individual organization beyond its definition of itself to have to consider its role within a wider ecosystem. Examples such as energy, digital communications, insurance, farming and recycling etc nevertheless present the same challenges.

[3] The design is thus a further development of the 12-step witnessed plus-one process. It separates out the identifying of narratives (the Witnessing matrix), the situating of a narrator within a larger ecosystem-of-concern (the Parallel Process matrix) and the development of a dilemma formulation from a plus-one process (the Plus-One matrix). This enables there to be a circulation around these matrices by a larger group of participants within the problem domain. To this it adds the Development matrix, in which the development wheel is used to examine critically the planned interventions emerging from participants’ collaborations in terms of both their balance and their leadership dynamics.

[4] Depending on the numbers of participants involved, the VOSE can be run either by a team of 4 moderators or one moderator taking up the different moderating roles in succession. The key is that the contract with the participants is to be working with their transference to the work of the VOSE. This means that the moderator(s) have to be very clear about the different way of working in each matrix, and to be able to hold these differences. This leaves the participants with the challenge of how they put it all together.

[5] The fourth Development matrix is the place where participants bind together what has emerged for them in the other three matrices in ways that have meaning for them. This binding can be thought of as a Borromean knotting. The challenge for the moderators is to enable the essential incommensurabilities between the other three matrices to be held.

Balancing the development wheel

We are seeing that securing effective testing for COVID-19 is not so easy.  Of course there need to be enough tests manufactured that are effective.  But there are also the questions of where and how to test citizens, who will do the testing, how to protect the testers, where to run the tests once samples are taken, and how to set the priorities for who should be being tested.  Even given leadership, there are huge challenges of training and organization to put all this together in ways that are aligned to and can cohere within different local contexts.

The government agencies in both the USA and in the UK learnt that it was not enough to procure a new piece of equipment.  In order to deploy a capability effectively within an operational theater, the US military took the view that development had to take place along 7 spokes of a development wheel: doctrine, organization, training, materiel, leadership, personnel and facilities. NATO added an interoperability-between-forces spoke to this list, while the UK military removed the leadership spoke from this list, added Information and logistics spokes, and renamed the doctrine spoke as a concepts-and-doctrine spoke and the facilities spoke as an infrastructure spoke.

Putting all of these together in the context of an organization seeking to be edge-driven in the way it creates effects for its clients, we come up with eight spokes for the development wheel, four hierarchically/ vertically-driven because driven by the need to secure economies of scale and scope; and four horizontally/edge-driven because driven by the need to secure economies of alignment within each edge situation and client context-of-use.

  • Why spokes of a wheel? If there is not balance across all eight spokes, the development wheel will not be able to turn.
  • And why does the wheel need to turn?  Being edge-driven requires continuous adaptation and learning by an organization, which, while starting from situational understanding, has to be able to impact on all eight spokes.  Without situational understanding it is not possible to determine what effects need to be created by the organization at its edges in order to sustain its viability, but without balancing the wheel, the organization cannot then be effective.
  • And what does this mean for leadership? Each spoke demands its own form of leadership, and the relationships between these different forms of leadership determine how the wheel is able to be kept in balance. This economy of leadership goes beyond narrower definitions that are identified with just one of the spokes.


Doctrine & Operational Concepts: The principles and framework governing the approach to generating effects for clients within a domain of relevance.

Facilities, Infrastructure & Logistics: The facilities, infrastructure and logistics providing the platform that supports an organization in doing its work.

Leadership and Education: The ability to lead creatively and effectively within the chosen domain of relevance.

Materiel & Technology: The equipment, tools and methods needed to be effective within the chosen domain of relevance.


Edge Organization: The particular orchestration and synchronization of capabilities needed to generate the desired effects within a client situation.

Mission Alignment: The people with the appropriate know-how and ability to work together collaboratively in support of a given edge organization.

Situational Understanding: The way a domain of relevance is defined and its data is fused and interpreted to provide a composite picture and understanding of what is going on in the particular situation.[1]

Personnel & Shared Culture: The people with the socialization, background and mutual knowledge and trust to be able to work together.


[1] A key issue here is not only ‘dark data’ i.e. data that is accessible, but not currently included within the current definition of the domain of relevance (see Hand, D., J., Dark Data – why what you don’t know matters. 2020: Princeton University Press.), but also traces of behaviors that are not yet accessible as data – the kind of traces that require ‘feet-on-the-ground’ and a forensic attention to what-is-going-on (wigo).

What can we learn from going virtual?

Digitalization is changing the competitive demands on every organization across all sectors, at different speeds and in different ways. The covid-19 pandemic is accelerating these changes, driving each organization towards becoming more knowledge-based, so that it finds itself having to compete more by aligning its behaviors dynamically to the differing situations of its clients and by synchronizing its own behaviors with others’ behaviors within the client’s context-of-use. Think wrapping continuing support around the evolving situation of a client.

Something of the enormity of this shift can be grasped in terms my work not being organized primarily around the task boundaries of what I am doing as a provider in terms of my primary task. Rather the work must be organized within the client’s system-of-meaning in terms of the primary risk from leaving too large a gap in relation to what the client wants. Whatever I and others do must be translated and transposed into the client’s context. Think what it takes to perform standup comedy.

This translation and transposition are endemic in the virtual world, in which meaning is endlessly emancipated from our own ways of organizing it. Having to work within others’ organizations of meaning leads us to being between two sides of a dilemma that both need holding, both ‘here’ and ‘over there’.  The virtual world entangles different ways of organizing meaning such that there are always multiple version of the ‘truth’. Think processes for agreeing where to direct investment within a city to impact on the greatest need.

A LinkedIn blog asks: “can group relations work actually be undertaken virtually”? Its five questions can be re-framed in terms of this dilemma arising from the impact of digitalization, leading organizations towards having to hold both sides. The question of whether group relations work can be undertaken virtually thus becomes a question of what innovations are needed in group relations work that will enable us to learn how to hold both sides.


Q1. What happens to ‘the task’ and the capacity to find and hold the task – made more complex without bodies in the room and unidimensional head photos on a screen? Made more challenging to persist with by virtue of the distractions available when working virtually. How are the initial agreements stimulated and supported virtually?

A group relations event is designed to explore transferential and counter-transferential effects arising between people as they emerge in group dynamics.  The event as a whole, however, depends on its participants taking up a transference to the work of the event per se as a work in which all the participants are engaged. Thinking of ‘a work’ in this way is like the way an artist creates ‘a work’, each work revealing to her a bit more of what her work is. When a group relations event goes virtual, what is the work to which the participants have a transference when it can’t be to studying the in-the-room interactions of the group per se? To make sense of this, we need to understand what constitutes a transference to the work:

transference to the personversustransference to the work

In a transference to the person, it is as if the other knows what it is that the individual wants, while in a transference to the work, it is as if the work knows what it is that the individual wants…

Q2. What about ‘time’ as connectivity issues plague beginnings, middles and endings of each GR element – and at every moment in between? Connectivity strength varies across the globe and at various times of the day and night; timezone issues might make for a complex weave of differing biorhythms for members – my morning, your evening? And what about digital security?

The various task systems created by a group relations event have their own bounded timespans of discretion within which participants make the choices implicit in their behaviors. The absence of boundedness created by virtual processes demands of participants that they choose the scope of the complexity that they are attending to, their interactions with each other taking place within these differing spans of complexity. Here we are moving from being dominated by the chronos of clock time to the kairos of what is constitutive of moments that have moment:


The timespan of discretion of a task system is defined by its beginning, middle and end, while its span of complexity is defined by the scope of the networked collaboration needed to to produce a desired outcome…

Q3. Since the ‘territory’ is no longer ‘our’ physical space, how do we contend as consultants with being in the ‘territory’ of the membership – their living rooms, offices, bedrooms?

The territory of a group relations event establishes the boundaries that limit the ways in which participants may be present. Going virtual means that no such boundaries are established while participants’ different ways of organizing meaning are made more apparent.  The resultant experience of edges is in terms of the contexts from which participants speak within their places and local times:


A boundary is marked physically in time and space, while an edge arises wherever there is an experienced need to translate and transpose between different ways of organizing meaning…

Q4. How does ‘authority’ work in the virtual environment? Can useful explorations of transference arise in making sense of what’s really going on?

With transference to the person, questioning the relation to authority involves questioning the relationship to the person. In contrast, transference to the work involves questioning the effects that appear to be being called for in a situation and what then might be necessary to giving rise to those effects. This changes where a participant looks in understanding what is really going on from the other per se to the situation:


Authority questions what participants are being obedient to, while performativity questions the relation between ways of organizing meaning and their effects…

Q5. If these boundaries are fundamental for ‘containment’ – how are we to anticipate, manage and make use of the anxieties that arise and are necessary for deep learning? Surely these virtual anxieties will have a different quality?

The ‘containment’ of a group relations event is derived from the way it holds its boundaries, limiting what participants must deal with. It is also derived from the ways of organizing meaning that emerge within its boundaries in the way participants make sense of what is going on. Going virtual emancipates both the ways of holding and also the ways of containing. Here ‘holding’, as the way in which horizontal complexity is limited, is distinguished from ‘containing’ as the way of organizing meaning. This challenges participants to focus less on how they are working with their anxiety and to focus more on possible innovations in the way they use the organization: 

defenses against anxietyversusdefenses against innovation

Defenses against anxiety defend from what is going on below the surface of a participant’s consciousness, while defenses against innovation defend the way a participant uses an organization to support some aspect of their identification.


Putting all five dilemmas together, placing Group Relations Work on one side and ‘Going Virtual’ on the other, we can see something of what holding both sides might involve:

 Group Relations Work Going Virtual
Q1transference to the personvstransference to the work
Q5defenses against anxietyvsdefenses against innovation

The question, then, of whether group relations work can be undertaken virtually is a questioning of the limits to ‘containing’ as a conflation of holding and containing. Whatever is needed of an event that goes virtual, its design will need to involve holding in such a way as to emancipate containing. Think Open Source Event.

Requisite Authority: when is triple-loop learning *necessary*

by Philip Boxer BSc MBA PhD

What organisation needs asymmetric and tripartite approaches to leadership? Why engage in triple-loop learning driven by dilemmas in sustaining relationships to individual clients’ demands?  What makes action research using plus-one processes so important?

Approached from the drivers of organisational scope, the answer to each question is: any organisation, once responding to the horizontal drivers of performance becomes more important than remaining subject solely to the constraints imposed by its vertical controls, since under these conditions the organisation is in a complex and therefore ‘turbulent‘ environment in which clients must be responded to one-by-one. Health and social care are good examples of such an environment, but all industries are moving towards this condition under the influence of information technologies and the increasing prevalence of multi-sided demands, the necessary corollary being the use of platform strategies.[1] Another kind of answer is: “if competition pushes you into the ‘red zone’ in the diagram below”.

It is easier to understand this ‘red zone’ if we start by considering what makes triple-loop learning not necessary. The diagram approaches this in terms of the way role and task are aligned to each other, requisite authority being whatever role definition is congruent with the task demands on the organisation. Triple-loop learning is  not necessary as long as the ‘red zone’ can be avoided, the ‘double diamond’ providing a diagnostic tool for identifying this condition:


  • Task: Either there are no dynamic cross-boundary relations to demand situations that are driving performance (e.g. providing medical equipment), or, if there are, then they can be responded to solely in terms of a choices defined by the organisation ex ante (e.g. providing a menu of in-home services)
  • Role: Either there is no accountability for performance in the demand situation (e.g. performance of the equipment once sold is down to the purchaser), or, if there is, then the accountability is to the person who signed the contract and not to performance within the situation itself (e.g. “if you are not satisfied with my performance, then take it up with my manager and don’t complain to me”).

Requisite authority involves there being congruence between the role and task sides of this diagram.  Lack of congruence means either too much organisational complexity or inadequate organisational support, depending on which way it goes.

We can add labels to the different parts of this diagram to make it clearer when triple-loop learning does become necessary:

  • Task: There is a dynamic relationship to the client’s situation that demands the dynamic alignment of differentiated behaviors and that involves dynamic linkages across the boundaries of the organisation (e.g. a care pathway has to be configured and continuously adapted to the needs of the individual client).
  • Role: Responsibility for responding appropriately involves bringing together a number of services from different organisations and holding them accountable in ways that are sustainable and that relate explicitly to performance within the client’s context-of-use (e.g. a care manager responsible for through-life management of the client’s condition and accountable directly to the client).

Examining a particular case situation, a hospital group wanted to provide seamless care to patients admitted through their Emergency Department (ED).  The task on the right was therefore to provide a condition-centric episode of care, the episodes being designed one-by-one.  The problem was that the ED was in a matrix relationship to the specialist wards with which it had to negotiate admission after having admitted the patient to ED.  This negotiation was constrained by considerations other than the patient’s condition, such as the receiving ward’s budgets.
Diamond3The proposed solution was to create an ED diagnostic team that had the power to determine where a patient went from ED.  The danger with this was that did not provide requisite authority, simply relocated where power was held without addressing the underlying challenges of designing and aligning care pathways that were sustainable across the hospital group’s ecosystem.  The solution was to set up a forensic process that could track and evaluate the performance of the ecosystem in order to learn what forms of agility were needed beyond the establishment of the diagnostic team.

The outcome from this process was a new organisational capability to backtrack ED admissions and to examine them as symptoms of failure in the primary social and healthcare systems.  This led to new ways of managing patients’ chronic conditions and failures in care funding.

[1] The multi-sided platform strategies of Apple, Google and Amazon are also good examples of this, as are the failures of Nokia and Blackberry through their continuing pursuit of one-sided strategies in environments demanding multi-sideness.

Parallel process and triple-loop learning

by Philip Boxer BSc MBA PhD
What lies at the heart of asymmetric leadership, through which the North-South bias can be balanced off by the relentless pursuit of an East-West imperative? I believe the reflexive process ‘behind the mirror’[1] is at the heart of this form of leadership.  Through reflexive process, the dynamic impact of unresolved dilemmas may be identified and worked through in parallel processes. Triple loop learning is a way of thinking how these parallel processes interact with each other under asymmetric leadership, enabling the organisation as a whole to respond dynamically to the demands arising in a complex environment.

Parallel processes
Parallel process is ‘brought forth’ by a reflexive process(5) engaged in by leadership ‘behind the mirror’.  In this process, dilemmas are looked for in the experience of clients(1). This identification with the experience of clients in their context(1) is crucial. From this reflexive process(5) come questions about the nature of the leadership process(4) that is shaping the nature of interventions(3) on the way the organisation works(2) in relation to its clients(1) within their domains of relevance(0):

The circular nature of the relationships between these layers becomes apparent if we relate them in the form of a learning cycle:

  • 5->4 Reflective Observation: new dilemmas emerge in the experiences of clients in their contexts within their domains of relevance.
  • 4->3 Abstract Conceptualisation: this is the leadership/governance process by which the intent of the organisation is kept sustainable.[2]
  • 3->2 Active Experimentation: this is the approach management takes in intervening on the way the organisation works.
  • 2->1 Concrete Experience: this is the plan of action through the way the organisation works as a theory-in-use.
  • 1->0 Reflective Observation: this is what actually happens for the clients in their contexts.

Looked at as a cycle, three kinds of learning can emerge, depending on the extent to which the cycle is short-circuited.  This short-circuiting arises when a strategy ceiling forecloses certain kinds of questioning of the behaviour of the organisation:

  • single loop: changing how a chosen approach to interacting with a given environment is executed.
  • double loop: changing how the approach to interacting with any one environment is chosen in pursuit of given direct effects.[3]
  • triple loop: changing how the approach to interacting with any one environment is chosen in pursuit of given indirect effects.[4]

Relating this back to the parallel processes, the importance of  the reflexive process becomes apparent – without this capability an organisation does not have the means of questioning its own assumptions as a part of its own ways of creating value for its clients.[5]

[1] ‘Behind the mirror’ is a way of approaching what-is-going-on based on a metaphor with systemic family therapy. While an individual may be working with the family, a team observes what-is-going-on from behind a one-way mirror, trying to make sense of the family dynamics in ways that can support the individual working with the family. It is impossible to get an organisation in front of a one-way mirror (!) but the metaphor is used to indicate a certain kind of critical questioning of the relation between the observer’s interests and what is being observed.
[2] This is the ‘final cause’ of the organization’s behavior. The difference between the ‘reflective observation’ subject to this governance at 1->0 and at 5->4 is whether or not the governance processes are affiliated to an a priori intent, or an intent is able to emerge from an alliance with the client in which the value deficits experienced by clients come first.
[3] The direct effects are defined by a one-sided relationship to demand in which the value is inherent to the direct behaviors of the organisation.
[4] The indirect effects are defined by a multi-sided relationship to demand in which the value is inherent in the indirect behaviors supported by the direct behaviors of the organisation. See the asymmetric nature of multi-sided demand.  The implications of such an approach are that the organisation has to pursue a platform strategy in order to capture indirect value from the relationships between its different kinds of customer rather than solely direct value. See what distinguishes a platform strategy.
[5] This is not to say that this questioning does not happen – only that without a reflexive process, the learning has to take place ‘above the strategy ceiling’. This may work in relatively stable environments, but in complex turbulent environments such an ‘above-the-ceiling’ process does not enable the organisation to be sufficiently dynamically adaptive. See Tempo, Entanglement and East-West dominance.

The Asymmetric Leadership Forum

What’s it like where you are leading at the moment?

  • Is the relationship between your organisation and its customers in balance, or are you having to work out how to handle your customers’ contexts in a more and more ad hoc way – riding your bicycle while re-designing it?
  • Are the outcomes your customers want highly dependent on others’ services as well as your own – do you need to align purpose & activity with other complementary suppliers?
  • What about the challenge & imperative of delegating more leadership and authority to those dealing directly with your customers, moving power to the edge of your organization where your customers interact directly with you…?

If any of this is recognisable to you, then you are at work as a leader in an environment of asymmetric demands, where situational judgements, exceptions, variety, differences – all of these are more like the facts-of-leadership-life than predictability, balance, controls, planning. We call this asymmetric leadership.

As an asymmetric leader you are likely to be working with some combination of

  • Customers’ escalating demands within increasing uncertainty and complex contexts.
  • The challenges of personalisation & individualisation by an increasing number of providers’ networks.
  • Aligning through-life support and condition management for the customer across organisational boundaries.
  • Reducing duplication and eliminating waste, whilst increasing the emphasis on early intervention to secure long-term benefits.
  • Trying to improve outcomes, especially in the case of complex needs.
  • Facing increasing pressure to develop greater resilience and to contain upredictability.

But how do you think and act in a context like this? What are the ways in which you can conceptualise what is happening that can provide some traction, give you a handle on the situation and create opportunities for improving the economy of your leadership effort?

We have some concepts and analytic tools, which we think can help you:

  • Map the ecosystem of organisations, customers and contexts within which you increasingly need to decide how to act.
  • Consider how to strengthen horizontal accountability in ways which hold accountable the individuals who are dealing directly with customers.
  • Develop the fractal resilience of the service systems you design and lead to cope with variation in the scale and scope of individuals’ needs.
  • Establish economies of governance in the way resources can be brought together and combined in individual interventions.
  • Define the indirect value for your customers beyond the immediate value arising from their involvement with your services.

Leading action learning

by Philip Boxer

The leadership of learning from experience involves learning from direct experience of practice – what is sometimes called action learning. This approach depends on establishing a ‘transference to the work’ in the sense that the learning has to be allowed to emerge through engaging with the problematic nature of a situation in practice. It contrasts with consultancy, in which the learning needed is b(r)ought in, based on a transference to the consultant. How this choice is taken up, between consultancy or action learning, will reflect the way power is exercised by the sponsoring system[1] within the client organisation.

The leadership environment
Action learning demands a leadership environment that holds the balance between[2]:

  • is prepared to acknowledge problems that need tackling (E);
  • provides support to tackling them (S);
  • recognizes career paths that make use of action learning (W);
  • uses as process models successful examples of action learning that become taken up by the larger organization (N);

Correspondingly, to be effective, any action learning project has to be able to satisfy four ‘edge’ criteria while keeping them in balance[3]:

  • It should add value (E): the outcome of the project should be to give the organisation a new ‘angle’ and new leadership in the way it creates value for its client.
  • It should be practical (S): the project should be based on data, make effective use of existing capabilities, and the result of the project should produce ground-level consequences for the way these capabilities are used.
  • It should ‘connect’ (W): the project should build on or take account of existing structures and ‘culture’ of vested interests, i.e. it must take notice of what is possible.
  • It should matter (N): there should be an identified sponsor[1] for the project to whom the team can relate and report.

The facilitating role
Crucial to the success of such projects is the person in the facilitating role[4], whose task it is to enable the members of the project to engage in a a ‘plus-one’ process. This type of process is needed to work through each of the following potential sources of error[5] in how they approach their learning:

  • Is the project team placing too much dependence on one account of what is going on? (is it buying in to one person’s version of the story, as though he or she knew and could give a total description of the problem situation?)
  • Is the project team assuming that there will be one right way to interpret the presented problem? (is a particular frame of reference being accorded unquestioned authority?)
  • Where does the project team ‘draw the line’ around the problem? Is the line being drawn in a way which includes or excludes themselves? (Who is part of whose problem in the problem-as-presented, and are the project team able to formulate the problem in a way which includes themselves?)

The relation to the larger organization
One reason why changes introduced by action learning projects do not get taken up by the larger organization is because they are not understood by the larger organization as sources of collaborative learning[2]:

  • mobilizing commitment to change through joint diagnosis of business problems.
  • developing a shared vision of how to organize and manage for effectiveness.
  • fostering consensus for a new vision, competence to enact it, and cohesion to move it along.
  • spreading revitalization to all parts of an organization without pushing from the top.
  • institutionalizing learning through formalizing supporting policies, systems and structures.
  • enabling strategies to be monitored and adjusted to problems encountered in the process of institutionalization.

The importance of networking across boundaries
Another reason is because such projects remain locked inside organizational silos, and are not used to reshape the organization by developing E-W dominance[6]:

  • Using action learning projects as change agents to create a new ‘social architecture’, becoming the basis of networks, determining the intensity, substance, output, and quality of interactions; as well as the frequency and character of dialogue among members of the network.
  • Defining with clarity and specificity the business outputs expected of the action learning project and the timeframe in which it is expected to deliver.
  • Guaranteeing the visibility and free flow of information to all members of a network and promoting simultaneous communication (dialogue) among them.
  • Developing new criteria and processes for performance evaluation and promotion that emphasize horizontal collaboration through networks, openly sharing these performance measurements with all members of the network and adjusting them in response to changing circumstances.

[1] The client system is where the ‘problem’ is, but the sponsoring system is where the power is. So the consultant has to work with the problem within the context defined by the sponsor:
The relation between these two axes parallels the relation between the two axes in the the plus-one process.
[2] The source of this list is ‘Why Change Programs Don’t Produce Change’, by by Michael Beer, Russell A. Eisenstat, and Bert Spector, Harvard Business Review, November–December 1990. Some of the challenges involved in doing this are described in the paper on The Future of Identity
[3] The N-S-E-W refer to the balance that needs to be sustained by the governance system in order to overcome a North-South bias. The use of this approach in one organisation is described in this paper on Marketing Project Groups.
[4] The challenges of this enabling role are described in this paper on  meeting the challenge of the case.
[5] These lead to the three kinds of error in ‘Unintentional’ errors and unconscious valencies:

  • Type I errors of correspondence arising from placing too much dependence on one account of what is going on,
  • Type II errors of coherence/consistency arising from having assumed there will be one right way to interpret the presented problem, and
  • Type III errors of intent arising from leaving the project team’s own interests (and therefore valencies) out of the way the problem is formulated.

[6] The source of this list is ‘How Networks Reshape Organizations – For Results’, by Ram Charan, Harvard Business Review, September-October 1991. Some of the challenges surrounding the formation of network interventions are described in this blog on stratification.

‘Unintentional’ errors and unconscious valencies

by Philip Boxer

How are we to know what constitutes an ‘unintentional’ error? I want to be able to identify ‘unintentional’ errors as errors of intent in order to be able to ask why a diagnostician can be rendered ‘blind’ to the actual nature of the situated condition before him. Is it because no-one can know what is wrong, given the current state of knowledge? Or is it because the diagnostician is blind to the possibility that his mental model is itself inadequate? To clarify this distinction, I start from James Reason’s work on understanding errors – ‘To Err is Human’[1].

Understanding ‘intentional’ errors[2]
James Reason defines an error as the failure of a planned sequence of mental or physical activities to achieve its intended outcome when these failures cannot be attributed to chance. As a cognitive psychologist, he distinguishes three things:

  • the mental model that frames the way the diagnostician makes sense of the situation (‘knowledge-based performance’);
  • some way in which the diagnostician uses this model to establish correspondence between a plan and the situation observed (‘rule-based performance’); and
  • some way in which the diagnostician uses this plan to define actions on the situation (‘execution-based performance’).

Both the “intended outcome” and the “planned sequence of activities” are constructs of this mental model, which corresponds to the diagnostician’s ‘way of making sense’. In these terms, therefore, there are three kinds of error corresponding, in reverse order, to errors of (1) execution, (2) planning and (3) intent:

  1. the way the actions prescribed by the diagnostician’s way of making sense may be put into practice wrongly (‘execution-based performance’),
  2. the way the model is used to ‘make sense’ of the situation may be at fault (‘rule-based performance’), and
  3. the mental model may itself be unable to make the situation tractable (‘knowledge-based performance’).

According to Reason, however, error is not meaningful without the consideration of intent. That is, it has no meaning when applied to unintentional behaviors because errors depend on two kinds of failure, either actions do not go as intended (i.e. an error of execution) or the intended action is not the correct one (i.e. an error of planning). In the first kind of failure, the desired outcome may or may not be achieved; in the second kind of failure, the desired outcome cannot be achieved. For Reason, “intentional” errors thus reduce to errors of planning or execution, so if we follow him in restricting ourselves to ‘intentional’ errors, there are only errors of planning and execution. This is why errors of intent are not addressed in ‘To Err is Human’.[2]

Understanding ‘unintentional errors’
Reason nevertheless differentiates between slips or lapses and mistakes. A slip or lapse occurs when the action conducted is not what was intended.[3] It is an error of execution. In contrast, a mistake occurs when the action proceeds as planned but fails to achieve its intended outcome because the planned action was wrong.[4] It is an error of planning. On the other hand, a mistake in medicine might involve selecting the wrong drug because the diagnosis is wrong. In this case, the situation itself was mis-assessed resulting in the action planned being wrong (unless serendipitously right!).

From the patient’s perspective, not only should a medical intervention proceed properly and safely (i.e. free of errors of planning and execution), but it should also be the correct intervention for their condition in their particular circumstances. We are thus adding that the condition should itself have been correctly identified (i.e. the medical intervention should be free of errors of intent, such errors arising through the adoption of an unwanted aim).

The question here is why the condition itself should have been incorrectly identified, leading to the adoption of an unwanted plan. For Reason, a further distinction is made between mistakes at the level of applying rules and applying knowledge (e.g. ‘Human Error’[1] p56). Reason suggests that this distinction reflects two kinds of error: (i) because the situation was assessed incorrectly; and (ii) because of a lack of knowledge of the situation. In the first case we can agree that this is an error of planning. We would say that the situation was assessed incorrectly in the way the diagnostician ‘read’ the situation using his mental model. But in the second case we must ask what is meant by “lack of knowledge of the situation”. If this were “knowledge” in the terms of knowledge that the mental model uses, then it is the same as (i). But if not this, then it must mean that the existing unconscious mental model leaves the diagnostician ignorant of some aspects of the situation because it has no place for them. In the physical sciences, this would be like using Newtonian mechanics to predict the motion of electrons subject to quantum effects. In order to understand ‘unintentional’ errors, therefore, we need to consider the reasons for the inadequacy of the unconscious mental model being used.

‘Unintentional’ errors as evidencing unconscious valencies[5]
A theory of the subject would have to give an account of the subject’s conscious mental model of his or her world and its trajectories, to which the subject experienced himself or herself as being subjected.[6] But it would also have to give an account of the way in which the subject experienced himself or herself as being subjected to his or her unconscious mental model.[7] These two forms of subjection together constitute a ‘double subjection’, in which the concept of ‘valency’ provides a way of referring to an unconscious investment in particular ways of organising the subject’s conscious mental model of his or her world, i.e. in particular ways in which the conscious and the unconscious are held in relation to each other.

Using the concepts of double subjection and valency, we can now account for all three kinds of error as follows. The subject’s conscious mental model and its trajectories are the diagnostician’s ‘way of making sense’, from which ‘plans’ associated with ‘intended outcomes’ and ‘planned sequences of activities’ may be derived. The third kind of error of intent arises from the nature of the subject’s unconscious valency for particular kinds of ways of making sense because of the ways in which they support the subject’s identifications:

  • Type I: correspondence error, in which the subject’s conscious mental models fail to anticipate the subject’s experience in a given situation (i.e. an error of execution creates a correspondence error);
  • Type II: coherence/inconsistency error, in which an elaboration of the subject’s conscious model with respect to the situation renders the model itself incoherent through its inconsistency with the subject’s unconscious experience. The occurrence of such an error can be said to be a ‘signal’ to the subject of the need to change his or her conscious model but, in general, the error does not determine a suitable elaboration, or even guarantee that one may be found. (i.e. the way the subject’s conscious model is put into relation with the situation creates an error of planning, which reveals the incoherence/inconsistency of the model itself with respect to the situation);
  • Type III: decidability error, in which the subject faces more than one possible elaboration of his or her conscious model of the situation because the model is non-deterministic with respect to the situation, generating mutually inconsistent models from which to choose. This undecidability means that no diagnosis can in fact be made with the available model, making problematic its use in any choosing ‘what to do’.  Under these conditions, for the subject to act ‘as if’ he or she knows ‘what to do’ will be because of his or her valency to that course of action (i.e. acting from unconscious valency to a particular course of action in the face of such undecidability creates in an error of intent).

The implications?
The report following ‘To Err is Human’[2] was on the Quality of Health Care in America project[8]. It considered a range of quality issues related to the overuse, underuse and misuse of treatments.[9] If we are not to question the motives of doctors (i.e. as intentionally following forms of diagnosis that serve their own interests while not serving the interests of their patients), then overuse (e.g. operating on knees unnecessarily) and underuse (e.g. not prescribing a prophylactic orthotic treatment where it could mitigate future conditions such as the need for an operation) should both be considered as arising from errors of intent.

Understood in these terms, it is necessary but not sufficient, in order to address issues of quality, to seek to ensure the use of the best possible conscious models (e.g. through relying on the influence of ‘evidence-based’ forms of research).  The subject must also be prepared to take up the ethical challenge of questioning his or her unconscious investment in his or her particular ways of knowing i.e. questioning being subject to an unconscious model that is inadequate because of what it is structurally ‘blind’ to/lacks[10]. The process by which this ‘blindness’ can be worked with involve triple-loop learning, making use of reflexive processes.

[1] Reason, James, Human Error, Cambridge: Cambridge University Press, 1990.
[2] This section uses extracts from “To Err is Human: Building a Safer Health System”, Institute of Medicine, National Academy Press 1999. Pp46-47. This blog was provoked by the fact that errors of intent were absent from the analysis.
[3] The difference between a slip and a lapse is that a slip is observable and a lapse is not. For example, turning the wrong knob on a piece of equipment would be a slip; not being able to recall something from memory is a lapse. These are errors of commission or omission in the way the plan (of intended actions) is translated into practice.
[4] In medicine, slips, lapses, and mistakes are all serious and can potentially harm patients. For example, in medicine, a slip might be involved if the physician chooses an appropriate medication, writes 10 mg when the intention was to write 1 mg. The original intention is correct (the correct medication was chosen given the patient’s condition), but the action did not proceed as planned. If the terms “slip” and “mistake” are used, it is important not to equate slip with “minor.” Patients can die from slips as well as mistakes.
[5] Extracts drawn from Philip Boxer and Bernie Cohen, ‘Doing Time: The Emergence of Irreversibility’, Annals of the New York Academy of Sciences 901:13-25, 2000; and Boxer, P.J. and Cohen, B. (1997) “Analyzing the lack of Demand Organization”, 1st International Conference on Computing Anticipatory Systems, Liege.
[6] In psychoanalytic terms, this would be subjection to a ‘reality principle’.
[7] In psychoanalytic terms, this would be subjection to a ‘pleasure principle’.
[8] Chassin, Mark R., Galvin, Robert W., and the National Roundtable on Health Care Quality. The Urgent Need to Improve Health Care Quality, JAMA, 280(11):1000–1005, 1998.
[9] The impact of underuse, overuse and misuse on these issues was later published in ‘Crossing the Quality Chasm’: Committee on Quality of Health Care in America (2001). Crossing the Quality Chasm: A New Health System for the 21st Century, National Academy Press.
[10] The process by which the subject ‘conceals’ this structural blindness is through conserving forms of vagueness in in the way he or she applies his or her models.

Consultancy or Action Research?

by Philip Boxer
Consider the relationship between a client and a service supplier in which a problem is being presented. They can each ask “who knows who knows best” what to do about the problem, and there are two possible types of strategy for how they can work together:

The consultancy approach:

  1. The senior management, knowing that they do not know while expecting the consultant to know, asks for consultancy.
  2. The consultant sells senior management a solution that they know how to provide on the basis of the transference that the consultant establishes. (guru)
  3. The senior management asks the consultant to provide people who know how to apply the solution.
  4. The consultant does so by using ‘bright young things’ who can learn how to apply the solution provided by the consultant on the job. (apprentice)

The action research approach:

  1. The senior management, knowing that they do not know while not assuming that anyone else knows better, asks for a technical assessment of the nature of the problem (expert).
  2. The technical assessment establishes what is known and what currently is not.
  3. The senior management sets up an action research process aimed at making tractable what currently is not known (reflexive process).
  4. The action research process creates a way of solving the problem that can be used by senior management.

What is the difference?

  • In the consultant’s approach there is a continuing asymmetry between the knowledge of the client and of the supplier, such that other suppliers are excluded. This approach may provide a quicker initial answer for the client, but leaves the client no better off in understanding the nature of the problem.
  • In contrast, the action research approach focuses all the time on removing asymmetries of knowledge between purchaser and provider such that the learning is joint. This is done by jointly identifying and working through the dilemmas emergent in the behaviours of the purchasing organisation in relation to its client-customers. This approach may generate better learning for the purchaser, but may not provide answers as quickly as senior management would like.

Behind this difference is the question of whether the problem presented by the senior management is itself the problem to be solved, or a symptom of some other underlying ‘real problem’. If the former, then it is reasonable to assume that a supplier can know the answer even if senior management does not. But if the latter, then the problem is going to have ‘wicked’ characteristics[1] and some form of action research process will be necessary, using some form of forensic process.

[1] ‘Wicked’ problems are problems with a circular definition – implementing a solution changes the nature of the problem affects what forms of solution can be effective etc. This concept of ‘wickedness’ was originally articulated by Rittel, H. and M. Webber (1973). “Dilemmas in the General Theory of Planning.” Policy Sciences. A more recent study in the context of Managed Networks in Health Care is Ferlie, E., L. FitzGerald, et al. (2013). Making Wicked Problems Governable? The Case of Managed Networks in Health Care, Oxford University Press.