The Role of Hierarchy in Realizing Collective Leadership in a Self-Managing Organization

This article is about leadership in a self-managing organization. Empirically, it is grounded in a participatory action research study stretching over two years, oriented around the following question: how does instances of hierarchical leadership and collective leadership interact across different dimensions of power in a self-managing organization? The article concludes that although this self-managing organization is largely governed in the day-to-day through collective leadership, this outcome still relies on hierarchical management of discourse and the processes which replace authority. That is, the article concludes that there is a dynamic between hierarchical and collective forms of leadership in this specific self-managing organization that serves to make its mode of self-management viable. In terms of practice, the results suggests that truly radical decentralization of decision making in the day-to-day is possible and can be highly effective. Furthermore, perhaps counterintuitively, results suggests that there can be a role for quite pronounced hierarchical leadership in making collective leadership flourish – at least when transitioning from a managerial hierarchy. Further research into this area might study organizations that are originally started as self-managing organizations and see if the same role for hierarchical management of collective leadership exists also in this type of case.


Introduction
There has been increased interest in academic and practioner literature regarding organizations that seek to do away with formal managerial hierarchies (Lee and Edmondson 2017). The increased interest in self-management has been attributed to companies needing to become more adaptive, internal processes no longer needing to reflect linear production processes, and factors relating to empowerment being increasingly important when trying to secure young talent (ibid.). We can call these organizations e.g., teal organizations (Laloux 2014), collaborative inquiries (Torbert 2004), openly chosen structures (Torbert 1974), holocrachies (Bernstein et al. 2016), or self-managing organizations (SMOs) (Lee and Edmondson 2017) -to name some of the many labels applied to radically flattened organizations. In these organizations people have been found to set wages, fire each other, and solve conflicts -all without falling back on the officially accepted notion that a predefined person (i.e., management) has the right answer to these key points of contention (Lee and Edmondson 2017;Laloux 2014;Dignan 2019). These organizations then seem to challenge the conventional wisdom that formal hierarchical power is necessary if we want to effectively organize people (see e.g., Simon (2013)).
However -as power comes in many guises -just because an organization takes away formal hierarchical authority, it does not necessarily mean that it removes non-formal hierarchical power. We might then seek to understand the extent that these organizations supplant hierarchical leadership (HL) with collective leadership (CL) in practice (Lee and Edmondson 2017). A suitable next topic in the conversation around SMOs might then be to explore the degree that HL still exists within SMOs.
The article will answer the following question: how does instances of HL and CL interact across different dimensions of power in an SMO? First the article describes a theoretical framework which aims to be granular about where and where not HL is replaced by CL when studying SMOs. Secondly, it draws on prior empirical material available on SMOs to note that we should be suspicious of any claim of an organization without any HL. Thirdly, it presents a method for data collection / analysis to distinguish between CL and HL. Fourthly, it presents a case description and a set of findings relating to CL and HL in an SMO. Finally, it presents a set of contributions and implications for further research into SMOs. Perhaps contrary to our modern 'common sense' that a managerial hierarchy is necessary (Lee and Edmondson 2017), the organization studied for this text really was largely successful in replacing HL with CL. However -and this is crucial to understanding the case -this shift entailed the use of HL, both in the short-term work to actualize the CEO's vision against the wishes of many in the upper echelons of the prior managerial hierarchy, and in the medium term to maintain and further develop that vision into a functional alternative to a managerial hierarchy. That is, here CL is found to exist to an unusual degree -but also to depend on HL for its enactment in practice. This has important implications for practioners who often struggle to define the role of HL in enacting an SMO (Dignan 2019;Laloux 2014;Torbert 2004).
The research collaboration with the organization is best captured by the label participatory action research (PAR) (MacDonald 2012) -i.e. the data collector was actively engaged with the firms as a subjective agent helping to shape its transformation.

Theoretical Framework
The following section first discusses power as occurring through three dimensions, here labeled directive, procedural and discursive (Lukes 2004) and then discusses CL and HL as two types of power. Juxtaposed, this results in a two by three matrix that is used to classify processes involving power into one of six boxes. This will allow for a succinct overview of how instances of HL and CL interact across these different dimensions of power in the SMO (Table 1). Lukes (2004) put forward three dimensions which can be used to conceptualize power from a critical standpoint. In the first of Lukes' (2004) dimensions, power is simply the ability to tell another person what they then will do (Lukes 2004;Perrow 1986). In the context of modern organizations, we can call this directive power. In the second dimension we find what we can colloquially refer to as the 'rules of the game'. This is then the process in which those with power define how choices are made, with the result of excluding some outcomes, or making their preferred outcomes more likely (Lukes 2004;Perrow 1986). We can call this procedural power -or the power to shape the process. In the third dimension, we can influence others through control over what appears as real, legitimate, and desired (Lukes 2004;Perrow 1986). We can call this discursive power. We can summarize these dimensions through the following questions: …Who does what? …How is the choice of who will be doing what made? …How does this come to be socially constructed as legitimate?

Power
This is a very short introduction to a large body of literature linked to critical perceptions of power, which in no way does it or the literature surrounding Lukes' (2004) ideas justice. Instead, the article simply takes the three definitions put forward by Lukes' as a starting point for the three vertical categories of the above six-fielder. Now then for the two horizontal categories.

Collective and Hierarchical Leadership
Both leadership (e.g., Palmer and Hardy 1999;Alvesson and Spicer 2012) and CL (Fairhurst et al. 2020;Ospina et al. 2020) are highly ambiguous terms and it is important when using them to be clear about the sense in which they are used.

Healthy Collective Leadership and non-formal Competence Hierarchies
CL is here conceptualized as a type that exists in contrast with HL. In discussing how these two forms of leadership differ, it is important to bring in a sense of nuance in which CL is not reduced to a total rejection of managerial hierarchies (Alvesson and Spicer 2012). In this article, that intention is operationalized through noting that there are different types of hierarchies interacting within a managerial hierarchy. On the one hand, there is the formalized hierarchical structure in which one person is designated as a decision maker in relation A sense maker engages in sense giving to others which e.g., informs their framing of a certain process.
to others (Lee and Edmondson 2017). On the other hand, there is a plethora of competence hierarchies which may or may not overlap with the formal managerial hierarchy. These competence hierarchies are here assumed to originate in healthy informational, experiential, cognitive and developmental asymmetries which coalesce into a task contingent ranking. That is -for each task there are real differences in aptitudes which exists independent of us socially constructing a formal ranking of how individuals objectively perform. If we do not recognize these competence hierarchies as an important aspect of social life in our sensemaking, when removing the managerial hierarchy, we run the risk enacting a totalitarian sense of egalitarianism which run contrary to those parts of us that are healthy and in affinity with competence hierarchies. It can then be considered key to not throw out the baby with the bathwater and not stop socially constructing a recognition of competence hierarchies.
To conceptually separate the managerial hierarchy from competence hierarchies, three different concepts are now introduced in turn: state, CL, and HL. This sense of CL was collaboratively created alongside the practioners, as appropriate in participatory action research (PAR) (MacDonald 2012).

State
The term state can refer to aspects across the three dimensions of power as it applies in a specific situation (e.g., who fills the dishwasher? …or what the role of dialogue should be in getting the mandate to tell someone to fill it? …or why is dialogue about dishwashers good or bad?). If answers to these questions are preserved in an individual throughout their interactions with others, their state is preserved. If not, their state is adapted. Leadership, or influence, is then the process through which these states come to be and be transformed.

Collective Leadership
For leadership to here be called CL all individuals within the collective have states that are both potentially self-preserving and self-adapting through their relating to one another. This definition of CL as a type is anything but new -and is sometimes said to exists in relation to a type of lens (Spiller et al. 2020;Torbert 1974;Gebser 2020). This position resonates e.g., with the practice turn within leadership in general which puts emphasis on that which occur beyond any one individual (Carroll et al. 2008) or "his or her singular interpretation" (Raelin 2011: 201). These perspectives also resonate with e.g., the contemporary philosophical position of what Freinacht (2017) calls 'metamodernism', andwhat Gebser (2020) called 'integral aperspectivalism' (Johnson 2019). Within these philosophical positions there is a subtle shift in how we think about ourselves, from being independent individuals, to 'dividuals' that actualize our various yet shared potentials, e.g., our potentials for leadership or followership, in relation to others (Freinacht 2017). That is, CL as objective behavior (as a type) is here constructed to exist in relation to CL as a subjective and intersubjective lens held by practioners, which considers the potential of mutual transformation as a core aspect of life (e.g., Freinacht 2017;Fletcher 2004;Spiller et al. 2020).
In this it is also worth underlining that CL is sometimes said to be closer to an embodied skillset than just a set of precepts -requiring conscientization, self-inquiry and reflexivity to come about (Spiller et al. 2020;Torbert 1974;Bohm et al. 2004). It is also sometimes constructed as existing beyond what is normally considered professional, relying on emo-tional support (Torbert 1974) and ties of affection (Spiller et al. 2020) to face and transcend divisions and potentially realize the energy stored up in the tension between superficially opposing viewpoints (Spiller et al. 2020;Torbert 1974Torbert , 2004Bohm et al. 2004). It is also often described as existing beyond purely cerebral cognition (Fletcher 2004) and to be intrinsically bound up with embodied awareness of emotions (Torbert 1974(Torbert , 2004Bohm et al. 2004;Spiller et al. 2020;Fletcher 2004;Joiner and Josephs 2007). The interconnected notions of leader/followership, interdependence and relationally oriented skills is summarized well by Fletcher when writing that "leading and following are two sides of the same set of relational skills that everyone in an organization needs in order to work in a context of interdependence" (Fletcher 2004: 648).

Collective Leadership and Competence Hierarchies
Referring to the importance of keeping the baby when throwing away dirty bathwater, please note the repeated use of the word potential in the prior sections on CL. The term CL here resonates with the many descriptions that hold space for the possibility of mutual transformation and not the absolute imposition of mutual state transformation in every contingency (e.g., Spiller et al. (2020) and Torbert (1974Torbert ( , 2004). In some contingencies, all the information, all the expertise, all the relevant developmental history might be actualized asymmetrically in one person rather than being equally distributed in a group. We can imagine this as a spectrum where interdependent individuals on one extreme end of the spectrum are fully egalitarian in their mutual transformation, and everyone's state is equally transformed through a dialogue. This might represent the ideal outcome in circumstances in which there is no clear basis for the group to sense into and name an underlying competence hierarchy. On the other end of the spectrum this same group might find themselves in contingencies in which the competence hierarchy is fully apparent to everyone and centered around one person. In this case the same group might engage in a dialogue with a very asymmetrical outcome, in which one person state is fully preserved and everyone else is transformed. We can call these two poles polyphonic and monophonic and note that the whole spectrum between these poles is a healthy part of both individual and collective thriving.

Hierarchical Leadership
In contrast to this, in what is here referred to as HL, this latter monophonic outcome is not the result of fluid dialogue which recognizes an underlying competence hierarchy but rather the result of a unilateral imposition of one individual's state over others. HL is then oriented towards (a) a zero-sum enactment on power, in which one persons' ability to influence others reduces other people's ability to self-preserve in their current form and (b) where this is done without the other nodes' active and ongoing consent.

Synthetizing this Section: The Difference Between Collective Leadership and Hierarchical Leadership
In HL then, the mode of framing is in terms of "either / or"; it is either this way, or that way. CL meanwhile is oriented towards a "both / and" mode of framing; your ability to influence me does not stand in opposition to my ability to influence you. Instead, both of us, and the collective, stands to gain from the partial mutual preservation and the partial mutual adaptation of our individual original states. A shift from CL to HL is then to "[reenvision] the very notion of power from power over to power with" (Fletcher 2004: 650). We can say it represents a shift from a monophonic conventionality to the potential for transconventionality (Torbert 1974(Torbert , 2004) -i.e., the potential for transcending two or more conventions that are embodied by two or more people. We can also note that CL is rather difficult compared to HL -and we should, based on extant theory (see e.g., Torbert (2004)), see it as skillset that must be developed. We are also well served to acknowledge that healthy CL sometimes includes monophonic outcomes; the greater the asymmetry in skills given a specific task the more one person should be heard at the expense of the airtime of others. What separates this conception of CL from HL is that monophonic outcomes in CL is subject to ongoing consent and a social construction of the natural task contingent hierarchy. That is, the hierarchy is fluid across time and space and depends on who's input is intersubjectively understood as most relevant in the moment rather than who has the most authority delegated to them by the arrows and boxes of a formally defined managerial hierarchy.
This concludes the initial exploration of the 2 × 3 six fielder.

Comparing Prior Depictions of Power in SMOs
To be clear about the definition of an SMO, the paper builds on Lee and Edmondson's (2017) three-part definition. It is an organization which (a) removes the managerial -subordinate relationship for some, or all, processes and (b) replaces this social construct with clearly formalized means to make decisions that do not include a managerial hierarchy in such a way that (c) it applies to everyone in the same way.
To understand the current state of research relating to the different dimension of power in SMOs, twelve sources of prior case studies were analyzed to get a sense of how this somewhat fragmented area of research relates to power dynamics (Barker 1993;Puranam and Hakonsson 2015;Baldwin 2015;Foss 2003;Lovas and Ghosak 2000;Ashforth and Reingen 2014;Ravasi and Verona 2001;Jha 2020;Felin 2015;Gray et al. 2015;Ackermann et al. 2021;Torbert 1974;. The article will now use the 2 × 3 typology to explore similarities and differences in the contributions of these cases. Firstly, it will discuss the link between directive and procedural power in SMOs. Secondly, it will explore the nature and importance of discursive power in SMOs. Finally, it will discuss an important controversy regarding CL and HL in SMOs. In each of the reviewed cases, procedures that supplanted HL with CL was put in the foreground. From this we can learn e.g. that CL augmenting procedures exist as complementarities to one another (Foss 2003;Torbert 1974Torbert , 2004. We can also learn that a lack of top-down mandated HL does not mean that a more taxing form of HL cannot appear among supposed peers (Barker 1993). We have also learned that endemic power struggles within an SMO can be a boon if the organization is able to use predefined procedures to oscillate between different framings -and hence become able to make use of two different but complementary ways of seeing the world to enact an organizational level reflexivity (Ashforth and Reingen 2014). Furthermore, we have seen that a core benefit of SMOs can be found in these procedures ability to enact loose coupling as an organization level design principle -specifically because teams are not organized in a hierarchical manner through HL but can seek out those synergies that they themselves are capable of discerning through CL (Ravasi and Verona 2001). To summarize this theme in the research field we might say that procedural elements which are not dependent on one formal person, but rather part of the commons, is the primary tool which SMOs use to decentralize power in an effective and long-lasting way -and many of the benefits / drawbacks of SMOs are understood in the literature in relation to the procedural dimension of power. Crucially in relation to power we can see that these procedures do not always work to remove power differentials but can also end up reinforcing them (Barker 1993) and that some procedures are there to manage or mitigate the result of ongoing attempts to enact influence over others, rather than completely remove these attempts (Ashforth and Reingen 2014).
Beyond this clear focus on the procedural aspect of SMOs, several cases highlight the importance of discursive power to provide a shared frame of reference in the absence of HL in the directive dimension (Puranam and Hakonsson 2015; Baldwin 2015; Lovas and Ghosak 2000;Ravasi and Verona 2001;Jha 2020;Felin 2015;Ashforth and Reingen 2014;Torbert 1974Torbert , 2004. We can note that some commentators view monophonic HL in relation to the discursive dimension of power as beneficial -see e.g., case descriptions regarding Valve. Others (e.g., Torbert (1974;, conclude in sharp contrast to the discussions based on Valve, that a shared discourse arising through polyphonic CL is a criterion for an SMO to work well. When it comes to understanding the role of HL in enacting SMOs there is no consensus in the case literature. For example, some consider the CEO to likely be an important stabilizing factor and view ad hoc HL engagement in decision making in the directive dimension as potentially important (Jha 2020;Puranam and Hakonsson 2015;Baldwin 2015). At the same time practioners like Dignan (2019) argues the opposite -claiming that intervention is the opposite of what is required for creating a well-functioning SMO. This point of contention does not just exist between researchers studying different cases. Foss (2003) and Lovas and Ghosak (2000) come down on opposite sides on this debate regarding the same firm (Oticon); the latter views the selective HL based interventions of Oticon's CEO as highly functional aspects of its operations, while the other sees this as the cause of the firm's eventual reversal from being an SMO into becoming a more traditional matrix structure. That is, there is a notable controversy in the literature on if and how the CEO should make use of HL.
In summary, we can (1) conclude that the procedural layer of power is important to supplant HL with CL in the directive dimension of power. We can also note that (2) discourse has a special role to play in the decentralization of power in SMOs. We can note that (3) there is no agreement in the literature of when and if HL has a role to play in enacting an SMO. We can lastly note that (4) we should be careful to buy into surface level depictions of SMOs if these do not sufficiently problematize the likely existence of at least some HL alongside CL.

Method
As noted above, there is ample reason to be suspicious of the notion that there would not be a degree of HL in the SMO if viewed in situ -regardless of what the organization espoused on paper. For this reason, the research design needed to go beyond what could establish through interviews and official documents and ideally view the organization from within (Bell and Willmott 2020).
To gain insight into the inner workings of the organization participatory action research (PAR) was selected as a suitable method (MacDonald 2012). That is, the researcher actively influenced the organization (in e.g., helping them set their strategy), and the organization influenced this text (helping the researcher iterate on the above theoretical framework, conclusions, and discussion).

Case Selection
Ten different SMOs (fitting Lee and Edmondson's (2017) criteria) was approached to gauge which organization was most keen to co-create both the research process and ideally also their own journey towards becoming a more fully articulated SMO. Several CEOs were excited at this prospect, but one stood out as the most transparent, engaged, and keen to learn about the theory.

Data Collection
A great aspect of this co-creational approach was that the researcher could participate in meeting, as well as view the organization from within by helping to co-create a two-day strategy setting workshop involving around 150 people. The access also included having planned interviews, which took around two hours, with six staff members -three senior and three more junior. In terms of written documents, the research process focused on the various iterations of their employee handbook for working as an SMO (what they called 'their spine'), as well as other documentation (reading material from trainings, the literature used to define the content of trainings and different versions of the CEO's purpose statement). Lastly it included a survey regarding organizational challenges and what people appreciated with one another.
Overall, it is estimated that the engagement lasted 180-200 h within the organization over at 2-year period. Around 100 h were related to meetings regarding the strategy days, and the facilitation and harvest from this event. Around 40-50 h were informal conversations with the CEO where we discussed theory and his challenges. These were evenly spread over the two years. 12 h were formal planned interviews. The remaining time (around 30-40 h) involved meeting observations and informal ad hoc interviews related to the meetings after and / or before they started.

Data Analysis
When considering CL as a research lens, temporality (Maupin et al. 2020;Fairhurst et al. 2020), contextual boundary conditions, multiple levels of analysis (Maupin et al. 2020), and the heterogeneity of individual experience (Fairhurst et al. 2020) has been put forward as especially important issues to consider as design constraints for CL oriented methods. Furthermore, given the suspicion that there is potentially more HL than the organization's espoused theory, it is also worth noting that it is important to separate the macro level espoused theory from the theory-in-use on the individual level. That is, we must go beyond understanding the heterogeneity of individual stories and also become privy to the hetero-geneity of individual behavior to the extent that we understand if and how it involves ambiguity and breakdowns in the subjects' espoused theory of leadership (Alvesson and Spicer 2012), a point which has also been made in relationship to methods for surfacing Luke's (2004) discursive dimension of power (Hathaway 2016) and regarding methods for studying CL in general (Fairhurst et al. 2020).
Given these design constraints, it can be considered suitable to select a process perspective involving in part organizational discourse analysis (ODA) (Alvesson and Karreman 2000), allowing for the surfacing of both espoused theory on the macro and micro levels, as well as event and longitudinal sub-case analysis (Yin 2011) which will allow us to document the theory-in-use on the micro level (Maupin et al. 2020;Fairhurst et al. 2020).
In terms of ODA, discourse with a large D (i.e., that which focuses on cultural assumptions in the context (Alvesson and Karreman 2000)) is understood through the evolution of the companies' handbook, informal and formal interviews with the CEO, the former head of HR and a co-founder and the chief architect of the handbook. The survey data was used to understand the heterogeneity of answers to what matters when it comes to staff attributes and organizational challenges, allowing for an analysis of where the organization agreed with the formal espoused culture and where there was considerable disagreement.
Meanwhile, for each sub-case the timing of events is significant (Maupin et al. 2020;Fairhurst et al. 2020), meaning that sub-cases were analyzed as processes, regardless of if these traces decision-making processes that unfold across the timespan of events in meetings or if it traces more substantial longitudinal shifts that affected many people -such as a new strategy or ways to train people.
The primary raster on each sub-case is, as noted above, the six-fielder introduced in the theoretical framing. Each sub-case was understood as a process that shifted states of people, either in terms of CL or HL, and in terms of if it fitted in the directive, procedural or discursive dimension of power. In some sub-cases more than one dimension was relevant, and it was more or less ambiguous whether it fitted a monophonic CL or a monophonic HL.

Findings
The findings section is built on the following logic. First it gives a brief introduction to the case. Secondly, it considers the link between the directive and procedural dimensions of power by comparing the official narrative to observations through a summary of sub-cases. Thirdly it considers the link between the procedural and discursive dimensions of power, again juxtaposing the official narrative with observations -but this time by going deeper into one sub-case. Finally, the overall case is summarized in relation to the 2 × 3 six-fielder.

Case Overview
The case in question is a consultancy with about 150 people and about 20 MEUR in annual turnover. It is headquartered in Stockholm, Sweden, with an additional office in Gothenburg, also in Sweden. It consults large and medium size firms and is slowly branching out into complementary services through, e.g., an acquisition relating to IT and one conference / event management firm.
The firm is presently formally organized as an SMO (as will be further disclosed in the findings). That means that per Lee and Edmondson's (2017) definition not even the CEO can tell another person what to do. Instead, the organization relies on formal decisions making rules which does not depend on managerial authority (these rules will be thoroughly explored later in the findings). The organization makes heavy use of multilateral communication tools (as opposed to emails) and non-mandatory weekly meetings in which people can challenge and support each other, as well as make use of the decision-making processes to make decisions that don't rely on managerial authority. As will be further discussed late in the findings, tailor made self-leadership development programs was deemed essential to make this SMO function well and was attended on a voluntary basis of around 95% of staff.
Founded in 2005, it started its journey into becoming an SMO in March 2016 when its members started to examine the differences between what they said and what they did in a workshop. This led to what the CEO termed a 'state of urgency', where everyone in the organization created an individual action plan, and the leadership group started to meet with a view to organize a new vision for becoming an SMO. Already nine months after this initial meeting the strategy of 2017 was defined through an "open space" (a bottom-up process) rather than a top-down plan. In situ data collection began in 2018 and continued for two years with varying intensity.
It is worth noting that the shift to becoming an SMO represents a sharp turn in the organization's developmental history. One day it was guided by a CEO with an "either / or" HL way to engage with the world and the next day this same CEO was dead set on enacting a "both / and" CL mode of framing in the business. Though considerably successful in some regards, this did not happen overnight nor without considerable pushback.
Due to the highly personal details divulged in the text, all names, including the name of the organization, has been disguised in the case description.

The link Between the Directive Dimension and the Procedural Dimension
In any organization that fully qualifies as an SMO, directive hierarchical power cannot formally be considered a valid way to organize people. We therefore start the analysis by detailing how the organization formally relates to directive power through the procedural dimension, first in terms of its espoused theory and then in terms of its theory-in-use. No Formal Hierarchical Leadership in the Directive Dimension In this case we can see Lee and Edmondson's (2017) criteria play out across many different formalized processes. For example, work execution is fully decentralized, with all managers having been given other formal roles and everyone being explicitly "responsible for their own personal management of time and priorities", as put by their employee handbook. In removing all formal management, this handbook also puts forward the process of how monitoring of work and quality assurance through performance reviews should occur on a peer-to-peer basis within self-managed teams and/or self-managed mentor relationships. The organizational design is also decentralized, with a certain number of open forums handling different parts of coordination (e.g., different forms of competency development, sales, strategy, and health) with "membership being open to all that find value in participating", as framed by the CEO and the handbook. Budgeting is decentralized in the sense that people are responsible for allocating time across internal and external projects in ways that makes sense to them and through dialogue with their peers. Furthermore, the budget is decentralized in that any deci-sion, including e.g., "the acquisitions of other companies [the largest possible investment the CEO envisioned for the company] can originate anywhere in the organization and come to be realized through e.g., the strategy forum", as put by the CEO. Finally, the strategy of the organization can be said to be partially decentralized, with the two co-founders always participating in the strategy forum, while "everyone was free and encouraged to join its sessions".
We can summarize this by noting that there is formally little scope for anyone, including the CEO to give orders and expect these to be followed. The Formal Processes which Support CL in the Directive Dimension Instead, peers are meant to help inform and collaboratively coordinate each other's direction through predefined formal processes intended to catalyze CL. This is largely intended to occur through two different dialogue-based processes. The first is called the "advice process", and through it someone comes up with "a proposal that they then refine in relation to advice given by others that are either experts or affected stakeholders", as put by the employee handbook. When they have gotten this advice, "they are free to make a decision that affects others in a way that is firmly grounded in care for the relationships", as put by the former head of HR. The second CL based process is called "integrated decision making", and here everyone that is affected by something gathers and is heard individually before a dialogue is initiated. The aim of the integrated decision-making process is to eventually reach a point where everyone can give their consent to the decision.
The previously mentioned forums (like e.g., the strategic forum) are also intended to work in a "coaching manner", meaning in this context that "participants help each other set goals, and expand their framing of those goals", as put by one of the key architects of their employee handbook. This occurs "through a give and take process, which aims to be as cocreating as possible" -meaning that "problems and solutions are sourced from everyone, and then taken action on by those best suited for the task".
In other instances, more specific processes were in place which supplanted these two generic processes for facilitating dialogue. For example, performance evaluation was decentralized through a detailed procedure (called the "conscious developmental dialogue") for getting feedback and setting and following up on goals with a mentor. Everyone is here meant to enact CL in the form of "mutual feedback, appreciation and confirmation in order to grow, create value and feel good as a person", as framed by the guide. This process starts with "sensing into the trust and safety" which the handbook considers "to be key in creating space in the recipient for truly hearing those that have something to share". Theory-in-use in the Directive and the Procedural Dimension How do these intended processes pan out in practice? The vast majority of the decisions making processes in the directive dimension that was witnessed really did work according to the handbook. That is, it was only outliners in which the two patterns laid out by the handbook did not largely reflect the observed reality. This was true regarding small things, like e.g., whether to use this or that piece of academic evidence when selling a certain consultation service to a specific client, who should join a sales meeting, or where they should host an after party to their conference days -and in big things -like how they should formulate their value proposition for their new flagship offering.
It is also worth noting that CL as a type came to be in relation to a shift in the lens the practioners used to understand leadership. For example, when discussing a suggested format for an upcoming meeting two junior staffers and a senior staff member references the importance of building trust, relaxing, slowing down and generally "caring for the atmosphere" to be able to lower defenses in order to "really take each other in and make good decisions together". In the discussion, the humanizing aspect of being less formal was considered a key strategic resource by a junior staff member, who quoted a mindfulness role model that he had as having said something like "when you take off your mask, it's easier for everyone else to do it". This conception of leadership was also visible when a junior staffer joined a chat group about sourcing a new consultancy service… after getting up to speed and expressing her own view she thanks the thread for "everyone leading our way in this process" -indicating that her sense of leadership holds that it resides in the collective as opposed to e.g., in the senior members of the chat.
When engaging in short ad hoc interviews about how they experienced the meetings, overall, there was a broad consensus that matched observations -people valued the two decision making processes and used them.
Instances that are worth noting, in which HL did feature in the directive dimension were centered around a few people that had been formal leaders in the prior managerial hierarchy and did not always appear to buy into the spirit of the transformation. As reported in interviews by themselves and others, these people had initially strongly resisted the move towards an SMO. Occasionally, when observed, they clearly found the loss of hierarchical power difficult to come to terms with. Sometimes they fell back on a supposed authority which no longer was there in the same way as the followership was now either lacking or as best existing in an ambiguous state making HL appear as highly ineffective. For example, in one meeting in which around a dozen stakeholders had convened to discuss a new service offering one such person which had been less pleased with the idea of the transformation dictated "the right way to proceed" in a way that clearly went against both the spirit and the practice of integrated decision making and which "landed poorly", as phrased by one of the other participants; "[he] simply does not always get it". The same the day the CEO also criticized the output of the process and appeared to be exasperated with both the HL used by the person in question and the "small perspective on value generation it seemed to illustrate".
Overall, there were few attempted instances of HL this first of Lukes (2004) dimensions in this organization's daily life -and none which seemed to be that successful. That is, seldom did anyone try to use authority to get someone else to do something they did not consent with, instead the directive dimension of power was mostly negotiated on the spectrum of CL as specified by the procedural dimension of power. That is, here the organization's espoused theory was congruent with its theory-in-use across most sub-cases. Summarizing the Implications of the link Between the First Two Dimension of Power From this we can conclude that this organization satisfies the criteria of being an SMO in that it (1) removes the managerial-subordinate relationship for either specific or all processes and (2) replaces this with a formal set of rules which (3) applies to everyone in the same way -regardless of title (Lee and Edmondson 2017), noting that the CEO used these same processes when interacting in the directive dimension.
In this we can also conclude that the directive dimension is organized through the procedural dimension both in terms of the official narrative and the in situ ongoing organizational life.
As CL defines the directive dimension, an interesting divergence from how power is usually discussed through Lukes (2004) dimensions is worth noting. Usually, power in the secondary procedural dimension is used to limit the possible future states of the system in the directive dimension -making those states prefeed by elite certain or at least more likely. For example, the way that representative democracy in the US works seems indifferent to outcomes that are preferred by most people while making those that benefit the elite highly likely (Gilens and Page 2014). In this case the opposite can be said to be true; the way that the CEO has reshaped the procedural dimension allows everyone to partake in defining states in the directive dimension. The article will now describe how this outcome came to be through HL in the procedural and discursive dimensions.

The link Between the Procedural Dimension and the Discursive Dimension
The need for Discursive work In starting this discussion, it is important to emphasize that the internal climate of the organization used to be oriented around a HL lens. Before 2016 the consultancy measured everything, and decisions came to be through a competitive meritocracy. To the CEO, and the managers, this had made perfect sense and matched the how they had always worked.
This mode of operation then required relatively little internal discursive work as HL matched their larger societal context's HL lens, as generally accepted at the time by the leadership and staff alike. Then, when the CEO had woken up one day and decided that he wanted act out a new and CL oriented lens "he had his work cut out for him", as he put it. Suddenly, there was an enormous need to do work in the discursive dimension of power.
For example, perhaps unsurprisingly, when explaining how he wanted to transform this very lucrative business from the core on out, his co-founder, many of his mid-level managers and those high performers (who depended on a HL congruent bonus procedure) was not in any way convinced of the merits of CL or the move towards becoming an SMO. Given that this position initially permeated a large portion of the prior power structure, the process of redefining the procedural layer and breaking up a managerial hierarchy which was interwoven "with both the identity and individual earnings of key stakeholders" and was "clearly at odds with a locally influential [discourse] in our industry", as framed by the CEO, "this did not happen overnight". The Discourse which Shapes the Procedural Dimension One way in which the CEO's discourse shaped and enabled the new CL based decision-making processes was through optional personal development courses that most people (≈ 95%) took. An interesting aspect of these trainings, as well as the CL process that they mean to premier, is that they "only in part depend on easily discussed concepts such as relationships, dialogue, and mutuality"i.e., terms are somewhat straight forward to capture with words in formal descriptions, as framed by a key sense maker. They also in part depend on more visceral and experiential concepts such as "being mindful", "being one's full self" and "giving one's consent"which are concept that "words do little justice" as put by one of the chief architects of the employee handbook in which they are used.
The prior head of HR noted in relation to this, that those formal procedures in some cases require a shift in perception which she believed "could only come about through personal development". The CEO put it like this… "maybe it is about a shift when your locus of attention is primarily rational and outward directed and you [really care] about how people view the address of your apartment… of if you have the right 'shabby-chique' design for your kitchen or whatever… to one where your attention is more inwards directed and more experiential … how does this decision feel to me, in this moment, does it open me to oth-ers? Does it close me to others? What is most important here for us to open to each other?". Another member of the board was even more esoteric when he said that "it is about attuning to one's soulfulness". The handbook also touches on this type of aspect already in the introduction, putting forward that "the power of the organization is sourced from within" and that the intention of the organization is to be "a space in which people can self-actualize".
The handbook also constructs a how to this why when it puts forward that conscious presence, or mindfulness, is "a form of subjectivity that is a way to know our selves and to develop".
To facilitate this shift in perspective and behavior, the organization held ongoing trainings in mindfulness, value-based living, and later also reflexive / empathetic communication skills. They further created an in-house collective leadership development program together with external experts in these types of techniques.
Overall, it can be noted that the SMOs formal processes are internally constructed as both enhancing and depending on "embodied mutuality", and that this is a property which is understood to exists within and between people. In this we can note that the discourse which puts forward that these qualities are important in turn legitimizes the procedural layer of power in the organization as these processes are then constructed as a way to live something important in practice.

A Recurring Pattern
But of course, these training programs, or the processes which they were designed to support, did not come into being overnight. Instead, they were the result of intensive discursive and procedural work by the CEO in legitimizing these states over others. There was a recurring pattern for how this occurred across important instances in which the procedural layer or trainings were defined.
First there was a (a) mostly private sensemaking process, in which the CEO identified a need in the organization and privately read, personally practiced, and consulted with outside experts on e.g., mindfulness, personal development, adult development theory, value congruent living, self-discovery, deconditioning -as well as the procedural layer or discourse within other SMOs. The CEO then became (b) a dominant sense giver in a group he formed to act on this need, in which other people's states were transformed and his was largely intact through him doing discursive work. If (c) a conclusion which stemmed from the either / or mode of framing that this text associates with HL as a type, this perspective was simply not allowed to help define the framing of the issue at hand. The result of this was to clearly limit scope of the CEO state transformation when it came to (re)forming the procedural layer, as he did not consider procedural dimension states not corresponding with his own both / and state in the discursive dimension to be legitimate. To the extent that his own state was transformed it was largely in relation to outside experts rather than internal stakeholders. In the end, it was usually (d) the same few people who finalized the advice process with him, and in this process mostly helped shape how his state was communicated -rather than substantially transforming the state itself.
For example, the above pattern (a through d) matches how CL as a concept came to be defined within the SMO, how the processes for seeking advice on the owners' directive was defined, how the process in which an annually occurring companywide two-day bottom-up strategy conference came to be redefined and how the metric for measuring company success was redefined.
There were not any sub-cases in which important procedural aspects were defined or redefined in a way that did not match this pattern. Partial exceptions to this statement can be found in more ambiguous sub-cases: e.g., does relatively minor changes in the format of a weekly meeting reflect a shift in the procedural dimension of power or is that more the directive dimension? What is clear is there was nothing that affected the whole organization in a meaningful way which did not broadly fit the above process. Is this Pattern CL or HL? To answer the research question then, which involves describing how CL and HL interacts in the organization, it important to first discuss whether this pattern is an example of CL or HL. Let us dive deeper into this by exemplifying the above (a) through (d) steps by going deeper in a sub-case detailing when the organization reformulated its metric for measuring its degree of success. Now, before the process of redefining the metric even got going on the group level, the CEO had (a) clearly done a great deal of thinking around a clear metric(s) for measuring different aspects of success. To him, the ideal solution should both capture the inherent complexity of various relationships between actors, while at the same time be simple enough to use in daily life. It should also be relevant on the individual and collective level. Furthermore, it needed to guide the organization towards financial viability as a hygiene factor while at the same time not reducing its mission to only being about fiduciary responsibilities to shareholders. One solution he had tried was a "sort of balanced score card" process, but in his mind, these BSCs were not always successful in guiding decisions towards synergies between different types of stakeholders, but rather often pitted different metrics and priorities against one another. In contrast to this, he wanted the organization to explore how many different stakeholders could be satisfied at once (e.g., customers, nature, society, partners, other team members, owners, and the future development of the organization through learning). In this reasoning we can note, in relation to the 2 × 3 six fielder, that the CEOs state is about how the content of the procedural level needed to guide the organization to transcend various potential trade-offs in the directive dimension (between e.g., complexity and simplicity, between individual and collective relevance, between now and the future, between financial viability and a more transcendental purpose). We could say that his state is then very much about a 'both / and' sensibility.
This issue then came to fore of the organization when it became necessary to set new goals as part of an annual strategy setting conference. This conference then became the spark which caused a group to meet with the intention of arriving at a decision through the integrated decision-making process. Once in the meeting however it was clear that (b) the CEO was the dominant sense maker in the group, doing discursive work to frame the issue as per his initial private sensemaking.
When this framing did not resonate with someone in the ensuing initial round of the integrated decision-making process (in which everyone is supposed to be heard in turn), and this person quite clearly only considered the profit motive, the CEO (c) broke with the prescribed (integrating decision-making) process. His tone, his body language and the content of his words was authoritarian and completely discarded the perspective being offered as simply not relevant to the complexity which others had offered.
In doing this, the CEO constructed a line between legitimate and illegitimate processes using power in the discursive dimension, shaping the procedural layer through which the directive dimension is then understood to be successful or not. He does this with a view to enact his both / and notion -yet does this through either / or HL. As evidence for this designation, we can firstly note there was no 'collectively explicated yet fluidly agreed on competence hierarchy' in that room. That is, the power to shape the discourse in this way did not become legitimized through a CL sensibility in which reflexivity in participants lead the group to intersubjectively recognize the CEO's expertise. Secondly, in doing this, he ignored the values professed by the other person, and enforced his values, as if a diverse stakeholder perspective, which includes a shareholder perspective, is value neutral and naturally superior to a narrower shareholder perspective.
This same thing happened later in the week. In this meeting the explicit topic was (d) very much oriented to how to communicate the CEOs state onto others, i.e., how is the 'correct way' to use the metric most effectively taught -is it through case workshops? Is it through theory? Is having clearly outlined and shared basic assumptions perhaps most important? How can it be integrated into a then nascent CL leadership program? In this meeting again the CEO unilaterally ranked the competency of the group saying, "you get this thing", pointing at someone, "you kind of get it" pointing at someone else and "you really don't get in" pointing at another external person. This last person later explained that he felt "dumbstruck and closed through this experience" -and highly unlikely to voice the type of "either / or" sensibility which preceded the ranking.
Again, this is an example of HL. Firstly, the overall intention of the meeting is confined to understanding how one state should be copied, i.e., it is clearly a fully monophonic relationship to state in which mutual transformation is not possible. Furthermore, the allusion to competence hierarchies is not fluidly agreed on in dialogue, but rather unilaterally imposed with the result of cementing one state as superior to others. That is, in summary, given our definition of instances of HL, there is HL in the link between the discursive and procedural dimensions of power -and it is specifically through this discursive HL which the procedural layer is shaped to be oriented toward CL 1 .

Conclusion
Based on the sub-cases discussed above; how does instances of HL and CL interact across different dimensions of power in this SMO (Table 2)?
Overall, CL in the directive dimension has over time become the everyday reality of the organization. For example, in the beginning getting the integrated and decision-making processes to work was reported to have been much more difficult than two to four years in, when this process appeared to the obvious way to make decisions. In conjunction with this type of shift, another lens on what leadership is has come to permeate the directive dimension of the organization, as exemplified by the above sub-case in which three people emphasize what occurs between people as important for bringing out that which is within people as key to enacting sensible CL in practice.
1 This pattern seemed to hold, with some ambiguity in order (e.g., the (a) part comes and goes many times in relation to defining what CL is for the organization). There are also more or less significant attempts by the CEO in inviting others into co-create what goes on in (a) and (b) processes which he nevertheless remains a dominant sense maker, but not to the extent where the effect of (c) (i.e., curtailing the potential state shift in relation to the procedural dimension to ones which fit the CEOs state in the discursive dimension) would instead suggest that the processes are CL instead of HL.
One might have then expected that CL in the procedural dimension and discursive dimensions would also become more prevalent over time. This was not the case, at least not over the two years of data collection. As noted above -instances of collective leadership in the procedural and discursive dimension were ambiguous at best, and perhaps best understood as non-existent.

Discussion
These finding challenges several aspects of extant theory on SMOs.

Little to no HL in the Directive Dimension of Power
There is a widespread assumption that we need vertical differentiation (Simon 2013). That is, many perspectives assume that we need other people to direct other people -that we need a managerial hierarchy in terms of the day-to-day (Lee and Edmondson 2017). Prior in situ research into SMOs, by e.g., Barker (1993) and Ashforth and Reingen (2014), seemed to confirm this when they found dysfunction and power struggle to ensue in the absence of a formal hierarchy. That is, one might have assumed that accounts put forward in the popular literature, by e.g., Laloux (2014) and Dignan (2019) regarding SMOs would be hyperbolicand that the depictions in the handbook of this particular SMO were unlikely to be matched by observations from within the organization.
However, it turned out that the accounts in initial interviews and their handbook seemed in this case to hold. That is, decisions that were witnessed in the directive dimension followed either the integrated decision-making process or the advice process as described in the findings. That is not to say that competence hierarchies did not exist, or that these sometimes did not lead to outcomes closer to monophony rather than the polyphonic pole of the spectra -but rather that these outcomes were the consensual result of dialogue and that There were outliers which occurs in this box, e.g., a person who was previously a senior manager was no longer able to lead in this way as the followership was now lacking. Procedural Instances of collective leadership in defining the content of the procedural dimension was not found regarding any question that had significant weight.
The content of the procedural dimension is limited through HL to states that makes CL more likely in the directive dimension. Discursive Two to four years in there was no ongoing concerted discursive work in the organization to mount an ideological counterpoint to the CEO's both / and sensibility. When interacting in this dimension, there was an enforced dynamic where the CEO was cast as a leader and other were followers 2 .
The state of the procedural dimension is meanwhile legitimized through HL in the discursive dimension -both when it comes to defining decision making processes and in terms defining the contents of trainings which support them. See the (c) part process description that detail the link between the procedural and discursive dimensions of power in the case.
people 'danced' between being leaders and followers in the sensemaking process as was productive given the task at hand.
An implication for this on positions like the one put forward by Simon (2013), is that we do not need a managerial hierarchy to manage the directive dimension of power -it is possible to set strategy, handle hiring, resource allocation, conflict resolution etc. etc. without a formal managerial hierarchy. This is not just true within the espoused theory of SMOs, but also is borne out by this longitudinal highly intimate study of how both the smallest and most complex decisions were made.

The Links Between Directive, Procedural and Discursive Power
In prior research, as cited in the initial overview of research into power in SMOs, the importance of alternative procedures to HL has been highlighted again and again. Several of these authors also include a discussion on how a different discursive lens is important but rarely (but for Torbert (1974Torbert ( , 2004) seek to create either a theoretical or empirical link between the discursive and procedural dimensions of power.
The in situ empirical findings in this case agrees with non in situ research on e.g., Valve and Oticon, in that people in this organization work in a different way which involves radically less hierarchical authority in the day-to-day than is common to most businesses. However, this unexpected outcome (in the directive dimension) came to be, in this case (nota bene), because the CEO used his authority to break down old ways of working and create new ways of working (through a HL reformulation of the procedural dimension). Crucially this entailed maintaining these procedures through HL (in the discursive dimension).
What this paper does then, that is new, is that it draws an empirical link between how the directive dimension comes to be CL oriented through CL augmenting procedures that are formulated though HL and then justified through intensive HL based discursive work to redefine basic assumption about what is legitimate. More studies of how this works in organizations that are started as SMOs rather than transformed into SMOs would be an interesting empirical focus for future research.

There might be a Healthy role for HL in Healthy CL Structures
It was noted in the introduction that there are many ambiguous answers in the SMO literature regarding the role of the CEO and his or her use of HL. Some believe that is important (Lovas and Ghosak 2000;Jha 2020;Puranam and Hakonsson 2015;Baldwin 2015) and some find that it is detrimental (Dignan 2019;Foss 2003;Laloux 2014). In Barkers (1993) case, we saw just how truly terrible things can get in dysfunctional CL oriented structures where management abdicates responsibility in all three domains of power without the CL oriented competence or structure first being in place. Meanwhile, in Foss (2003) case on Oticon, we noted how selective intervention by management in the directive dimension was potentially fatal for the SMO.
In this case, using a more granular approach to understanding dimension of selective intervention by the CEO, we saw an alternative to those two cases of dysfunction, with people co-managing the directive layer in a highly functional CL oriented way, allowing for the type of guided evolution discussed by Lovas and Ghosak (2000) to occur in practice through CL rather than HL. However, this occurred through selective interventions by the CEO in the procedural and discursive dimensions.
Could it be then that there is a role for selective intervention in the two higher of Lukes (2004) dimensions of power, which allows for a functioning CL in the directive dimension? Clearly more cases need to be juxtaposed for such a conclusion to be found to be generalizable, but this case could be the start of such a practice and performativity-oriented exploration around the role of HL in SMOs. It would also be highly fruitful to contrast the analysis of this case with cases in which the directive dimension functions well without selective intervention in the higher dimensions -if these indeed are found to exist in practice through in situ observations. Its important to note here again just how far into the organization data collection had to get before it became clear that HL still dominated the procedural and directive dimension. Interviews, surveys, document analysis would not have shown this. It was through action research that the nuances of how HL shaped the discursive and procedural dimensions became visible.

The role of Paradox Within SMOs
In Ashforth and Reingen's (2014) discussion, it is noted that the organization was divided between those that wanted more focus on profitability and those that wanted more focus on purpose. That SMO let a partisan conflict endure, meaning that CL did occur in the procedural and discursive dimensions in their case, albeit in a dysfunctional way.
In this case, this same divergence between profitability and purpose existed within the organization, but in this case the CEO forces the both/and solution through HL as discussed in the last of the above sub-cases. Similarly, in this case, the construct of collective leadership was understood to be a skill, which meant that it was understood as being an example of a competence hierarchy -something which someone was better at than others. That is, in both these sub-cases there is a need to relate to surface level contractions between ideologies / priorities and either accept them and oscillate between poles or find a solution that transcend them through both / and thinking (e.g., purpose driven profitability, or a hierarchy of abilities to relate non-hierarchically to others).
Understanding if and how such tensions occur in other cases, as suggested by Torbert (1974Torbert ( , 2004, might be a fruitful key area to focus on for further research into contemporary SMOs. Perhaps it is the resolutions of paradoxical tension (such as the tension between profit and purpose, or the notion of managing self-management as visible in this case) that defines that heart of this phenomena? That is, in the absence of hierarchical monological sensemaking, perhaps it is both the potential difficulty and promise of CL in SMOs to create institutions that can oscillate between, or transcend, otherwise polarized monological perspectives?

Implication for Practice
The findings in this article indicate that procedures can replace authority as the primary mode of coordination and decision-making in complex interdependent organizations. Taking action based on the recommendations of e.g., Dignan (2019) or Laloux (2014) might then be fruitful if you are looking to create a more agile organization.
In doing so, you can use your formal and non-formal power to define and legitimize new non-hierarchical ways of coordinating work and training staff. You should however not use it to interfere with the work itself.
You should also be careful to remain reflexive (i.e., be able to consider more than one perspective, and ideally combine the best of several) in your leadership; the ability to juxtapose and integrate divergent viewpoints -even when these appear to be paradoxical -might a core aspect of leading an SMO.
These finding are tentative, and we need more research based on in situ observation of SMOs. We also need more brave leaders to let go of preconceived notion about the role of hierarchy in realizing effective economic processes. Ideally, we can build bridges between practical and theoretical perspectives of this verdant empirical terrain.
Funding Open access funding provided by Stockholm School of Economics.

Data Availability
The raw data is available in Swedish in a non-redacted format. To respect privacy concerns, redactions are necessary depending on what aspects of the source material are interesting to researchers. Given this qualification, the findings of this study are available from the corresponding author upon request.

Declarations
Disclosure of Potential Conflicts of Interest Chris Kummelstedt worked as a consultant during the strategy setting workshop, as discussed in the paper. Chris also invoiced for some of the interviews and reflection with the organization. In total Chris invoiced about 70,000 SEK over the course of two years.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.