Organising People through Distributed Networks

I’ve written a 3 page feature for HR Magazine this month (this edition also has me on the cover as one of their ‘Movers and Shakers’).

The feature focuses on the creative design of people centric organisations and summarises some of ‘The Social Organization’. Therefore, if you’ve not read and don’t want to read the book (you should!) you can at least read the summary of the section on organisation models in HR Magazine.

However, I’ll also be outlining a lot of the main points from the book and the article here. I’ve already reviewed the advantages and disadvantages of traditional organisation models (vertical functions and horizontal teams) and the first more people centric model: communities, and especially communities of performance. If you're not read this article, I recommend you do so before reading this post.

In my next article, I'm going to explain how I think organisations can use networks of performance in their organisation architecture. However, before I do this, I want to explain what I mean by exploring other types of networks operating in organisations.

In 'The Social Organization', I suggest that one way of doing this is to look at the science of networks. At a very simple level there are three main network topologies, shown below:

Baran, P (1964) On distributed communications: I. Introduction to distributed communications networks, RAND Corporation, RM-3420-PR, as of December 27, 2016. Available from: https://www.rand.org/pubs/research_memoranda/RM3420.html [accessed 12.1.2017]. Published with permission in 'The Social Organization'.

The three models in Figure 4.1 consist of nodes or vertices, which can be the people or groups within an organization; and links, edges or arcs, which are the connections between them. Links can be undirected / symmetrical / mutual or directed / asymmetrical, ie they can point in one or both ways. They can also be seen as single / uniplex or multiple / multiplex types (eg about learning and innovation) and can be identified at single or multiple value levels (eg rated 1–10 by importance). Degrees are the number of links into a person or between two people. Social networks, featuring the relationships between people, tend to be much more complex than these three models, which were originally drawn to represent different types of computer networks (which is why the diagram labels the nodes as stations). The relative simplicity of these computer networks is the reason I am using them here to explain the basic working of human networks.

It might also be interesting to note that the distributed network in Figure 4.1 was the basis for packet switching, which enabled the creation of the internet. However, nothing in this chapter has anything to do with online social networks such as enterprise social networks or ‘the social network’, ie Facebook. These are simply technology-based platforms for developing the real social networks I am writing about in this chapter.

The network labelled (A) on the left-hand side of Figure 4.1 is based around a centralized, single ‘hub and spoke’. Social networks often form like this as a result of weaving together a small number of previously scattered relationships. This form of network is simple and efficient but the person or people at the centre may also become bottlenecks and obstacles to communication, performance and innovation. These networks are also not very robust: eg if something happens to the person or people in the centre the whole network fails.

A single hub-and-spoke network can also be represented as one parent with multiple child relationships. This is shown by unfolding the network and moving the central node to the top, with what were the spokes hanging beneath it. This is called a tree diagram and is the basic format of functional organisation shown in a traditional organization chart.

A centralized network will become stronger and more effective when there are direct, cross-cutting links between people/groups who are not in the hub of the network – eg between a person or group in the top left and one in the top right of network (A) rather than all of these links going to the centre. Increasing the extent of linkages across a network is called ‘broking’. This involves reaching across structural holes, which are empty paths between two people who were previously unconnected (Burt, 1995). This creates higher levels of bonding, or closure, within the network. Closure speeds up communication and decision making, which facilitates the emergence of norms and higher levels of trust, and offers more opportunity for change.

As the number of cross-links increase, two particular roles may develop. These are, first, connectors – people who have many links within the network. And second, energizers, who have less links but whose enthusiasm helps new ideas to take hold. All of the links in this network are most likely what are termed ‘strong ties’. These are relationships between people who know each other and cooperate together. Strong ties are useful for execution as they bring people together to get work done and support each other.

However, there is generally little diversity in this sort of network as every- one tends to have the same experiences and perspectives. Everyone already knows what everyone else knows.

The most adaptive and resilient network form is (C), a distributed network, which is why the internet uses it. People or groups in these networks have multiple parent–child, or peer-to-peer relationships. However, human networks never look quite like this distributed network. Instead of this, as networks grow, they tend to become clumpy. They form decentralized hubs, clusters or subgroups consisting of many dense connections surrounded by more loosely coupled areas of sparser linkages. Professor of network science Albert-László Barabasi explains that this clumpiness develops because people tend to attach to others who are already well connected. Just as the rich get richer, the well connected get even better networked. This is called a scale-free network, and it follows a power law or 80/20 pareto distribution rather than a normal distribution curve (Barabasi, 2002). That is, there is a long tail of very dense hubs and another of empty network spaces.

These multi-hub networks can be either more or less decentralized. Network (B) is still fairly centralized as hubs still connect into one main centre and there is just one lateral connection between hubs shown at the top of the diagram. This is the type of network you might see in a horizontal organisation with a programme management office (PMO) structure. Programmes and projects manage themselves but still report to the PMO. Community based organisation also looks fairly similar to this from a network perspective, though the hubs can often have slightly higher closure and the links between and across communities (see below) are more personal rather than role oriented (and hence not necessarily between different hubs).

A more decentralized version of a decentralized network would typically have more of the hubs connected laterally to each other. In addition, these links would often form between people outside the centre hubs, eg between a person in the top left and someone in the top right of network (B) rather than, or in addition to, the two more central people shown below them. This means that the network starts to look like more of a cross between (B) and (C). From now on, I refer to this as a distributed network, whilst recognizing that social networks are inherently clumpy and therefore partly decentralized as opposed to truly distributed.

Additional key roles in this type of network are the people with links across the network, between one hub and another, bridging between different parts of the network. These cross-boundary brokers are known as boundary spanners.

All these lateral connections are most likely to be what are termed weak ties (Granovetter, 2001) – links between people who do not know each other well. These people spread knowledge of what is going on in the different subgroups and provide a basis for sharing ideas across the network, which is important in sparking innovation. However, too high a proportion of intergroup broking or bridging compared to bonding relationships can lead to reduced trust and cohesion (Burt, 2007).

It is lateral ties that provide the small-world effect (Watts, 2003) that recognizes the common scenario where strangers can often identify a common acquaintance. This is based upon being able to reach other people through a small number of steps, or a short path length, popularised through the idea of six degrees of separation (Milgram, 1967). This means that networks tend to ‘fold back’ on themselves, ie our friends’ friends tend to be friends (Baker, 2012).

These network chains do more than pass information though. The earlier sections of this chapter on mindreading and harmonizing explained that we are influenced by our connections. We are also influenced quite strongly by our connections’ connections and their connections too. For example, research has found that if a person’s friends are happy then that person is 15 per cent more likely to be happy too. If the friends’ friends are happy they are 10 per cent likelier. And if the friends of those friends’ friends are happy the person is 6 per cent more likely to be happy (Christakis, 2009). The chain of influence drops off after three degrees, although these days social media means we can also be influenced, eg through online social rewards such as Facebook likes, by people who can have high degrees of separation. Focusing on social networks and online social networks therefore makes ensuring high levels of individual engagement more important too.

Valdis Krebs describes how networks may also evolve into a further form that is a core/periphery network (Krebs and Holley, 2002). This will be reviewed in Chapter 13.

So, all organisational groups are networks, and all organisational networks are groups, though the groups in a distributed network are of a fairly nominal kind. And it is these distributed networks / nominal groups which I'm going to focus on when I write about organisation networks.

Also, once again, as with communities of performance, it's worth adding that these networks don't need to be seen as parts of the organisation architecture. Most organisational networks are informal / social and perfectly effective in doing what they want to do like this. However, if there is the potential for them producing something important for the organisation (which there often is) then I think it's important they are recognised as this - hence networks of performance. I'll be posting on these shortly...

(But before anyone shoots me, I do recognise that organisational networks as well as communities need to be managed in a much looser, softer, bottom-up way compared to traditional functions and horizontal teams. In fact, managing is probably the wrong word to use, and the title of this article should also be something more like Enabling people to organise...)

Here are my previous posts on organisation models:

https://www.dhirubhai.net/pulse/hr-magazine-new-organisation-models-jon-ingham/

https://www.dhirubhai.net/pulse/whats-wrong-traditional-functions-why-do-most-still-use-jon-ingham/

https://www.dhirubhai.net/pulse/from-vertically-focused-functions-horizontal-cross-teams-jon-ingham/

https://www.dhirubhai.net/pulse/making-matrices-work-jon-ingham

https://www.dhirubhai.net/pulse/communities-performance-jon-ingham

https://www.dhirubhai.net/pulse/buurtzorg-communities-performance-jon-ingham

You'll find more in the HR Magazine article, or 'The Social Organization'.

Jon Ingham, @joningham, https://linkedin.com/in/joningham, [email protected], +44 7904 185134.

Top 100 HR Tech Influencer - Human Resources Executive

Mover and Shaker - HR magazine