Alignment without bosses

An introduction to information hierarchies in complex adaptive systems

“I would tell my staff about the ‘dinosaur’s tail’: As a leader grows more senior, his bulk and tail become huge, but like the brontosaurus, his brain remains modestly small. When plans are changed and the huge beast turns, its tail often thoughtlessly knocks over people and things. That the destruction was unintentional doesn’t make it any better.” ― General Stanley McChrystal


A common concern of de-bureaucratisation is how to align autonomous teams to higher-level organisational goals. So entrenched is the paradigm of managerialism that many executives do not believe that disparate teams and individuals can successfully coordinate strategies without having a boss or manager providing top-down control. However, as General Stanley McChrystal’s witty apologue suggests, dictatorial management often causes more problems than it solves.

It’s interesting to note that most complex adaptive systems have no problem at all aligning their sub-components to a common purpose. Termite colonies, for example, divide the work to be done amongst differentiated castes such as workers, soldiers and alates. Notably, there is no ‘managerial’ caste bossing them around, yet they are able to coordinate their activities to construct highly sophisticated ‘cities’ and maintain them for thousands of years1. Another example of self-organisation is our own bodies. Our various organs and systems work together seamlessly, via integrated regulatory mechanisms, without the need for any central controller2.

How?

Recent scientific advances suggest that the concept of information might hold the key. It may be that complex adaptive systems self-organise by using information hierarchies, a theory put forward by Jessica Flack and collaborators in the Collective Computation Group at the Santa Fe Institute.

In very basic terms, lower-level components compress and percolate information up to higher-level information structures, which in turn are used for fine-tuning the behaviour of components.3

The theory is too dense to unpack in this short transmission, but we can get a sense of how it applies to organisation design via some case studies.

Alignment Teams

The construction of alignment teams is an effective mechanism for optimising collaboration between different groups. They work by creating information hierarchies.

For example, when scaling Scrum, Nexus Integration Teams can be used as a focal point to address constraints across multiple Scrum teams. Membership includes individuals from each Scrum team who have the necessary skills and knowledge to help resolve issues. Through the mechanism of the integration team, bottom-up intelligence is aggregated at a higher level to fine-tune the trajectories of each Scrum team.

Another example is Fusion Cells. These were originally designed to combine US military, intelligence and law enforcement resources into a unified network to beat Al Qaeda.

A Fusion Cell is a small, ad hoc group filled with representatives of larger departments, agencies or organisations. They are empowered by their home organisations to collaborate, innovate, and quickly share their insights. They’re expected to leverage their combined capabilities and perspectives to fuse raw data into actionable insights, then push those insights to frontline elements as quickly as possible.4

Both alignment teams are examples of non-dictatorial information hierarchies. Information is aggregated and used for alignment but no top-down control is exerted.

The hypothetical example below shows how they work in a little more detail.

Example

In the diagram above, the alignment team comprises of appropriate delegates from each of the teams 1, 2, and 3; say a product development team, a customer service team and a marketing team. It also includes key stakeholders from different parts of the organisation, say strategy and compliance. The teams are collaborating on a new product launch. The alignment team formed organically and will dissipate after launch.

The alignment team meets regularly enough to enable trajectory correction of the individual teams. Meeting frequency depends on how fast work evolves at the individual team level. Meeting too frequently introduces unnecessary communication costs while meeting too infrequently introduces misalignment risks. Some optimal point needs to be found.

Each team allows compressed information to percolate up to the alignment team. Flack refers to compressed information as being ‘coarse-grained.’5 Examples of it may be the status of a critical dependency, or a schematic of the latest design iteration. Fine-grained information, details the whole does not need to know about, remains at the team level.

Coarse-grained information is generated via the collective computation (‘C’ ) of the work teams, as well as by other parts of the organisation.

The alignment team aggregates this information to form an impression of the macrostate, a gestalt view of the various dynamics in play.

The teams then tune their trajectories to the macrostate view through downward causation (‘D’), meaning that the higher-level information drives behaviour in lower-level components.

In this example, behaviour-tuning may include adjustments of schedules, refining team-level strategies, reacting to an innovation, etcetera.

Why it’s important

One might argue that such an intricate analysis of simple teaming mechanisms is overkill. And for the simple examples presented it certainly is. However, understanding nested dynamical processes is key to designing self-organising structures.

Alignment teams are just one form of adaptive mechanism. They provide a useful focus for understanding the role of information hierarchies in organisations.

Leave a comment

Learn more

For an introduction to information hierarchies, start with:

Life’s information hierarchy by Jessica Flack

Flack JC, Erwin D, Elliot T, Krakauer DC. 2011 Timescales, symmetry, and uncertainty reduction in the origins of hierarchy in biological systems. In Cooperation and complexity (eds K Sterelny, B Calcott, R Joyce), Boston, MA: MIT Press.

Jessica C. Flack, ‘Coarse-Graining as a Downward Causation Mechanism’, Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 375, no. 2109 (28 December 2017): 20160338, doi:10.1098/rsta.2016.0338.

Walker, S., Davies, P., & Ellis, G. (Eds.). (2017). From Matter to Life: Information and Causality. Cambridge: Cambridge University Press. doi:10.1017/9781316584200

Image credits

Formation, by Antranias

Alignment teams, from my training slides

1

Stephen J. Martin, Roy R. Funch, Paul R. Hanson, Eun-Hye Yoo. ‘A vast 4,000-year-old spatial pattern of termite mounds.’ Current Biology, 2018; 28 (22): R1292 DOI: 10.1016/j.cub.2018.09.061

2

Although the brain is a central regulating terminus for many bodily processes, it itself is a network without a controller. Lisa Feldman Barrett offers some accessible literature on this topic.

3

see Learn More section for readings about information hierarchies

4

McChrystal Group, Fusion Cell Playbook

5

Jessica C. Flack, ‘Coarse-Graining as a Downward Causation Mechanism’, Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 375, no. 2109 (28 December 2017): 20160338, doi:10.1098/rsta.2016.0338