Deconstructing the Voice of Technology: Looking through the Actor Network Theory (ANT) Lens


In 2013, Raghu Ananthanarayanan and I published our book on the Tensegrity Mandala, proposing an alternative organization design and philosophy for firms preparing to make a leap into the world of networks. We believed that the paradigm of traditional hierarchical structures and associated control mindset can only be transformed if the firm is able to nurture and resonate networks that mirror and respond to a dynamic interplay of multiple voices – of wealth, of markets, of communities, and of technology. Our claim was that this dynamic dance or interplay reaches a state of equilibrium defined as a set of ‘creative tensions’, when key roles within each network get enlivened. It is then the organization may be perceived as aligned and innovative.

This blog seeks to build on further, and add new perspectives to the network lens of an organization. It seeks to decipher and interpret the voice of technology – taking a deep dive into how technology shapes not just culture of the firm but also its identity, its structural design, and its perpetuity in a chaotic world. This blog raises a perspective as well as ethical questions on how Artificial Intelligence (AI) would get integrated into organization design, leave alone throughputs and value adding processes.

I would be deploying the Actor Network Theory (ANT) as a lens of exploring the nuances of organization design and transformation by focusing on technology. I would like to thank Prof. Mukul Vasavada of IIMA, who was the first to introduce me to the works of Bruno Latour – Latour’s influence on ANT has been extremely significant.


Introducing Actor Network Theory (ANT)

ANT is a unique approach to interpreting networks. Traditionally we have looked at human beings as representative nodes in a network. However ANT is a creative lens where one can identify non-human actors, and how such non-human actors, co-create and define networks as well as organizations.

As per Wikipedia, Actor–network theory (ANT) is a theoretical and methodological approach to social theory, where everything in the social and natural worlds exists in constantly shifting networks of relationships. ANT posits that nothing exists outside those relationships. Thus, objects, artefacts of technology, ideas, and processes are seen as just as important in creating systems as humans.

When deployed in a firm’s context, it introduces the notion of an ‘ACTANT’ – an ‘Actant’ is seen as an amalgam of man and machine, that impacts how value gets created. Latour says that an Actant emerges when ‘humans cast relevant components of their agency and knowledge into artefacts to which action programs and capabilities are delegated’ – as a result of this delegation, artefacts become holders and dynamic vehicles of human agency, thereby replacing humans in doing things.

Let me give you a very simple example to make sense of an Actant.

As a reader of this blog, you would be using emails in your correspondence to others. Whether you choose Outlook or Mail or Gmail, there is a feature of organizing your mails as a ‘conversation’. The conversation keeps track of your correspondence across days, weeks, and sometimes months, with a host of stakeholders who may have joined in a dialogue around the theme referred to as the ‘subject’ of the mail. The mailbox feature also books your calendar, reminds you of meetings and deadlines, opens up apps that lets you know more about the people who you are likely to meet etc. If four people are a part of this conversation, every subsequent mail would alert all four and add to the momentum or slowness of dialogue within. It is almost as if this feature (ACTANT) has its own power in grabbing your attention, managing your responsiveness and making you perform.

Three decades ago, the very absence of this actant would have impacted your ways of networking, sharing information and data, collectively reviewing – in fact collective processes may have been near impossible if the four of you were in different geographies. Thus, the conversation strand in your mail box, is an actant, with its potency and its agency.

The more you look around, you would discover how technological artefacts are determining our choices and our sense making – how the very same also determine the nature of social networks that we become part of, and of our effectiveness and impact in systems. Take for example the term – PMO (Project Management Office) – despite the numerous approaches to project management, you would be quick to discern how a project management methodology contains routines that shape human behavior.


The Voice of Technology

Dimension 1

Ostensive versus Performative

In our book, we had very briefly referred to ‘routines’, focusing more on the dynamic and operational capabilities of the firm, and how these create a vector / momentum. In this blog, I would like to spend a lot more time on the notion of routines and how ANT impacts these.

Routines have generated multiple meanings in organization theory – my favorite or preferred lens would be to see ‘Routines as a set of preferences that seem to energize conditional patterns of behavior within an organization – these soon become a gene, or a memory, or a ritual for replication and imitation’. Routines are not simple but complex meta-habits in an organization.

Organisational routines may be conceived as ‘generative systems’ that produce recognisable or discernible, repetitive patterns of interdependent actions, carried out by multiple actors as per Feldman & Pentland in their paper in 2003[1], within a pre-existing social context. Feldman and Pentland brought in tremendous insights on the nature of routines, by defining and deploying two categories – Ostensive and Performative.

  • Ostensive routines embody structure, largely through rules, policies, and procedures – often the reference is to possible or potential behavioural pattern. Taking on the example of PMO – the routines unleashed through a PMO offer an understanding of the ideal way of deploying a project.


  • Performative routines stress of specific actions by specific people at specific times and places. Unlike Ostensive routines that are driven around structure, rules and policies, performative routines are ‘agency’ driven.

A lot of research has focused more on designing  ‘ostensive routines’ – especially when one is looking at an organization level phenomena. This also reinforces my belief that most organizations seem to design and leverage ‘ostensive routines’ to align and structure the dynamics between the human being and the technology.

Take for example – the process of documentation.

Whether you are a detective on the police force, or a key account manager on a business development trip or a research scientist in a laboratory – the documentation process (or an actant) becomes not just a methodology but a meme for the organization. If it isn’t documented – it is not real.

Most managers seek to co-hold the two routines – ostensive versus performative; the latter perhaps never as much given its due in managerial decision making, and often embedded as intuitive or the exception to the rule. Thus ostensible and performative routines often seem to be a split, where the latter is akin to the underbelly of the system, including the associated rituals of learning from performative routines.

If I was to relate to our book, the roles of the Craftsperson, the Throughput Developer and the Innovator (3 roles that straddle the Voice of Technology), it is imperative to examine the following:

  1. The nature of Actants that merge the craft, the technology with the human being. The way actants have emerged over the years, it becomes difficult to envisage what belongs to the human being and what belongs to the inscribed artefact that governs him or her.

For example, the throughput developer from an ANT lens would have to discern, institutionalize, and design the new actants that translate technology into creating value. One such actant that engineers of my age would relate to is ‘Poka Yoke’ techniques for human behaviour modification.

2. The nature of routines that energize the role – ostensive versus performative, when it comes to diagnosing or sensing the context, when it comes to making the decision, and when it comes to reviewing and learning from it.


Dimension 2

Decentralization versus Control


One of classic quests of the modern age is towards accomplishing complex tasks within networks without investing too much energy into Control and Coordination. In most decentralized networks, the fragile balance between control and decentralization is achieved through the design of ‘Artefacts’ / actants – the artefact itself becomes the hidden ruler and or manager.

One of my clients was exhibiting its cutting-edge work on robotics for knee and hip surgeries. The ecology of technology solutions that this global giant has created included (a) mapping software that could look at the patient in a pre-operative stage and then its outputs including angle of surgery, bone size, incision etc., would be transferred into (b) a robotic arm that would ‘guide’ the surgeon into making cuts and incisions that is otherwise humanly impossible. The robotic arm is another example of an actant that allows the surgeon to be more precise, more creative, and more effective as it becomes the holder of human agency, perhaps at the cost of replacing many routines of the surgeon.

Working with technology, and looking at the crude example above, most designers of technology straddle the second dilemma – that of decentralization (agentic) versus control (stabilization). Most technologists would refer to three processes of innovation that actants and ANT design dwell on:

  • Variation: To what extent is the Actor network allowing for unbalancing and agentic choices? How much variation should a system be able to handle. Continuing with the robotic arm example – when does the arm allow for the surgeon to go against the parameters given to him or her by the machine, and when does the arm block him or her.

In one of the researches on the development of open source code, it was quickly realized that variation had to be allowed within the interface of the human being (the open source member) and the technology – for it was crucial for new innovations.

  • Selection: The process of selection is around prioritization and keeping some ideas on the back-burner.
  • Stabilization: The process of stabilization is around artefacts and actants assuming control.

As you would have guessed, ANT and artefacts capable of control come handy in the second and third processes – which are influenced by the need for Control.

If this becomes the trend and as AI catches on, this conundrum between decentralization and control becomes more complex – for what we are now saying is that the agentic maverick (who creates variation) has to learn to live with the Control of the Artefact and the Actant.

So I guess the routines around radical innovation and incremental improvement get immediately destined to be split and this may fragment the firm further – or at least it builds another hierarchy of human beings. There are some human beings that get to be more agentic and there are some that comply with the actant.



The set of assumptions, Raghu and I began with in our model on tensegrity mandala, have been challenged by the actor-network theory, and the notion of actants and artefacts. Thus, the three critical roles that we had designated that would own and resonate with voice of technology – that of the Craftsperson, the Throughput developer, and the Innovator, need more thinking through as the gap between the human actor and the technology dramatically blurs.

There was a presentation by a professor from INSEAD, who spoke of AI impacting five routines (generic processes) of management:

  1. Planning and Analyzing the Tasks
  2. Delegating the tasks to various members of the organization
  3. Monitoring and reviewing task accomplishment
  4. Performance management including training and learning
  5. Motivating

All these generic processes or routines, he claimed, had core algorithms that could be deployed in any organization and in any industry vertical. He also claimed that AI could do a better job than the subjective human being on the first four processes.

I am still not convinced of his claims, but the categorization of management processes were around ‘instrumental processes’ and these arguably are strengthened by ANT.

I am still thinking whether sentient processes can be designed around artefacts and actants. After all one of the influences on ANT has been socio-technical systems of the last century.

[1] Feldman, S. M., & Pentland, B. T. (2003). Reconceptualizing organizational routines as a source of flexibility and change. Administrative Science Quarterly, 47(4), 94–118.