Inspired post by Dave Pollard today on Â the challenge of scale and the confusion of control. Â Complicated systems require few connections in order to be manageable:
It is because business and government systems are wedded to the orthodoxy of hierarchy that as they become larger and larger (which such systems tend to do) they become more and more dysfunctional. Simply put, complicated hierarchical systems donâ€™t scale. That is why we have runaway bureaucracy, governments that everyone hates, and the massive, bloated and inept Department of Homeland Security.
But, you say, what about â€œeconomies of scaleâ€? Why are we constantly merging municipalities and countries and corporations together into larger and ever-more-efficient megaliths? Why is the mantra of business â€œbigger is betterâ€?
The simple answer is that there are no economies of scale. In fact, there are inherent diseconomies of scale in complicated systems. When you double the number of nodes (people, departments, companies, locations or whatever) in a complicated system you quadruple the number of connections between them that have to be managed. And each â€œconnectionâ€ between people in an organization has a number of â€˜costlyâ€™ attributes: information exchange (â€œknow-whatâ€), training (â€œknow-howâ€), relationships (â€œknow-whoâ€), collaboration/coordination, and decision-making. That is why large corporations have to establish command-and-control structures that discourage or prohibit connection between people working at the same level of the hierarchy, and between people working in different departments.
Why do we continue to believe such economies of scale exist? The illustration above shows what appears to happen when an organization becomes a hierarchy. In the top drawing, two 5-person organizations with 10 people between them have a total of 20 connections between them. But if they go hierarchical, the total number of connections to be â€˜managedâ€™ drops from 20 to 8. Similarly, a 10-person co-op has a total of 45 connections to â€˜manageâ€™, but if it goes hierarchical, this number drops to just 9.
This is clearly â€˜efficientâ€™, but it is highly ineffective. The drop in connections means less exchange of useful information peer-to-peer and cross-department, less peer and cross-functional learning, less knowledge of who does what well, less trust, less collaboration, less informed decision-making, less creative improvisation, and, as the number of layers in the hierarchy increases, more chance of communication errors and gaps.
But, what about complex systems?
So back to the purpose of this post, to answer these questions: 1. What is it about the â€˜organizationâ€™ of the Internet that has allowed it to thrive despite its massive size and lack of hierarchy? And: 2. What if we allowed everything to be run as a â€˜wirearchyâ€™?
To answer the first question, the Internet is a â€œworld of endsâ€œ, where the important things happen at the edges â€” and everything is an edge. â€œThe Internet isnâ€™t a thing, itâ€™s an agreementâ€. And that agreement is constantly being renegotiated peer-to-peer along the edges. If you look at the diagram above of the co-op with the 45 connections, youâ€™ll notice that the nodes are all at the circumference â€” around the edges. There is no â€˜centreâ€™, no â€˜topâ€™. And the reason the organization isnâ€™t weighed down by all those connections is that theyâ€™re self-managed, not hierarchically managed. The work of identifying which relationships and connections to build and grow and maintain is dispersed to the nodes themselves â€” and theyâ€™re the ones who know which ones to focus on. Thatâ€™s why the Internet can be so massive, and get infinitely larger, without falling apart. No one is in control; no one needs to hold it together. Itâ€™s a model of complexity. And, like nature, like an ecosystem, it is much more resilient than a complicated system, more effective, and boundary-less. And, like nature, that resilience and effectiveness comes at a price â€” it is less â€˜efficientâ€™ than a complicated system, full of redundancy and evolution and failure and learning. But thatâ€™s exactly why it works.