I had the opportunity to along to the Responsive Organisation Unconference in London, where I lead a session and attended many others – more about those in a later post. The discussions during the day started a chain of thoughts about organisation structure, specifically, how structure has been disrupted by technology and how technology might help to reconstruct it.
Beyond Hierarchy Again
Hierarchy remains the predominant paradigm in business, with tree-like organograms dominating the landscape. The underlying reality is that organisations have and are shifting away from that model, and that shift from hierarchy to process to commitment and network management is an ongoing journey. Hierarchies simply can’t adapt fast enough for the rate of change that technology has unleashed on the world, and if your business isn’t fast enough, it isn’t going to be interesting. Strict hierarchies inhibit cross-functional communication, filter out critical pieces of information in the cascade process, and slow communication down. They can, however, provide stability, clarity and consistency. There are contexts where hierarchy is an appropriate structure, don’t write it off completely!
Networks Not Works And Real Work
So what is the alternative? Networks, you cry, of course. But networks are a curious things. An organogram represents a form of network, but it is a hierarchical network. If we are going to talk about “networked organisations” then we need to be a little more nuanced. When I started working in computer networking, IBM ruled the roost, with its mainframes and hierarchical networks (yes, I am that old) “Strong, stable, dependable.” Then the Internet, and with it Cisco, came along, with a distributed form of networking that turned out to be even more robust and dependable, and significantly more adaptive. The resilience of the Internet comes from its highly distributed form of control. No central point of control means no central point of failure (and my, how CEOs have become a central point of failure in recent decades).
The protocol that you have probably never heard of that allows the Internet to work is something called BGP. It uses the concept of Autonomous Systems to construct the network. I mention it, at the risk of giving myself flashbacks to the days when I daily risked bringing the whole thing down with a single key stroke, because it evolved into an incredibly stable and robust system that has survived an incredible amount of change. The Internet survived the arrival of The Web, online video, mobile phones and social networks, none of which were even conceived of before it was initially designed. The Internet exists as a collection of distributed autonomous systems, and the term was established to mean “a network of networks.” So do distributed autonomous systems hold the answer to the company structure of the future? Is their independant-interdependence the model for the leaderless organisation?
In short, no. In conversations with Philip Sheldrake it became apparent that while systems like The Internet, bitcoin, wikipedia and git hub are interesting studies, they don’t provide answers to the key challenge of distributed organisations: How do you deal with change, and steer the organisation through it? Open source software development is often held up as the ultimate in distributed, leaderless, networked execution, but experience of it says that it is anything but. Yes, it is a great model, but it doesn’t work in the way the most people perceive it to work. The successful open source projects all exhibit an exponential contribution profile (few contribute much, and many contribute little) and depend on a benevolent dictator to keep things on track. They are anything but egalitarian and efficient.
Flat networks require stronger leadership than hierarchies, and without that leadership, the network rapidly fragments in to separate competing networks (or forks in software terms). The network needs a control plane to manage its structure during times of change. Many start ups are fantastically functional flat networks, but fail spectacularly when confronted with either rapid growth (restructuring) or fundamental change (directional uncertainty).
One of the fundamental concepts we had in mind when we built Milestone Planner was the idea of distributing control out to the furthest parts of the network. Let project team members make autonomous decisions, then communicate them “at the point of impact.” – by default every team member can update every part of the project. Teams still have leaders, and there is clear accountability. We borrowed heavily from the concepts of social network platforms, but it turns out that there were more lessons there than we had realised. I strongly believe that, at their heart, all business problems are communication problems, or rather, they are communication failures.
Social platforms introduced us to the idea of ‘Fans’ and ‘Followers’ that enabled conversations to scale. Users could select who to follow (or unfollow) and build their own networks – distributed networks, with distributed control. But, and this is a big ‘but’, in the limit (as the network grows increasingly large) friend/fan/follower models turn into another form of broadcast. We follow more and more people, and rapidly become overwhelmed by communication. Facebook came up with the now defunct “EdgeRank” algorithm filtering the level of communication to be manageable. EdgeRank was optimised towards interest and attention (Facebook needs eyeballs for ads), leaving the intellectually hungry, well… Hungry. Twitter filled the void by allowing users to build networks based on serendipity and value creation, rather than historical relationships, and thus it thrived, for a time.
The challenge with distributed control is that the distributed points need enough information to make the right decisions, but not so much information that they become overwhelmed by the process of communication. Email has become the control plane in the modern organisation, and people are overwhelmed by it, being constantly pushed irrelevant information and missing out on the detail. The friend/follower model can help here, but it needs a little help itself. The friend/follower network needs effective discovery mechanisms to enable it to scale. In short, it needs structure (or meta-structure).
To build an organisation that is responsive, and agile, neither the traditional hierarchy, nor the friend/follower model works. We talk a lot about networked organisations, but having been on the leadership teams of networked organisations for over a decade, they too have their weakness. Legal and regulatory structures means that someone needs to take ultimate responsibility (or to ‘be in charge’). There needs to be directional steering and shepherding to ensure that the network remains cohesive. There is a structure that addresses these challenges. It turns out that the answer was in our heads all the time. Literally, in our heads.
The brain is a network that adapts to change with incredible agility, even if it does have some quirks. The brain solves the design problem using heterarchy – adaptive networks that structure and restructure according to need. Heterarchy isn’t a new concept in organisational design, and many of us have worked in them for many years. Virtual Teams, in their original form, were organisational structures that rapidly established purpose-specific teams, leadership and authority. The teams were fluid and adaptive – the very model of a heterarchy, where roles were context specific rather than permanent for an individual.
Don’t Rest When You Design for Flow
When virtual teams become static, they are no long a heterarchy. They become the aching grid lock that is the modern matrixed organisation. The recent turbulent economic times have sent organisations mixed signals. When change stops, hierarchy sets in, or rather rivalling hierarchy sets in. It explains much of the organisation angst that is happening in larger businesses today. As economic growth, hopefully, re-establishes itself, change will flow through the organisation. The challenge will be which businesses can create the control plane to support heterarchy. We have the technologies to communicate and adapt to change like never before. What we need now is operating models that enable us to use them to build organisations that are both responsive and robust enough to deal with the challenges ahead.