Blog for July 2016 Seminar: Understanding Networks

Conrad Taylor writes:

The 80th meeting of the Network for Information and Knowledge Exchange (NetIKX) took place on 14 July 2016 on the topic ‘Understanding Networks’ and was addressed by Drew Mackie and David Wilcox, who also took us through some short exercises. The meeting was chaired by Steve Dale, who has worked with Drew and David on a number of projects.

Drew has researched around network analysis. David’s background is as a journalist (Evening Standard) and he has tried to give people a voice within regeneration and urban development issues. They exercise their joined skills typically in projects for community development and social service strengthening.

In my account of the meeting, I do not exactly follow the order in which the points were made. I also offer my own observations. Where those deviate significantly from the narrative, I’ll signal that in indented italics, as here.

The idea of networks
The concept of a network has many possible applications, such as computer networks, but Drew said our focus would be networks in general and how they can be represented visually and thus analysed. Visualisations appeal greatly to Drew, who is by background an architect and illustrator.

There are various ways in which the nature of an organisation or a community can be expressed, e.g. through stories. Network thinking is a more structural approach. In network representation, one typically has some form of blob which represents an entity (such as a person, department, organisation), and lines are drawn between blobs to show that a relationship exists between the entities on either end.

Mindmap diagrams and ‘organograms’ are forms of network diagrams representing hierarchical set-ups, designed to limit the number and kind of connections possible. Others networks are more freeform.

Hierarchical organisation is just one way in which networks can be constrained. Other examples of constrained networks: connections between components in an electronic circuit are anything but random. You cannot travel on the Underground between King’s Cross and Seven Sisters without passing through Finsbury Park. Connections may also be strongly typed: the connectors in a genealogy diagram may indicate ‘was married to’ or ‘was the child of’, and some connections are not possible – you can’t be the mother of your uncle, for example.

‘Anything that can be drawn as a set of nodes and connections is a network,’ said Drew. The nodes could be people – they could be ideas. For the purposes of this workshop, we considered networks where the nodes are people, organisations and institutions: while not being accidental or random, such networks are not particularly constrained.

People who work with networks
Drew identified four kinds of people who may work with networks. These roles are not mutually exclusive and can overlap.

Network Thinkers understand the power of thinking in terms of networks and promote that view, usually applying it to their particular field, such as management or urban design. In economics, he mentioned author Paul Ormerod, who is a visiting professor at the UCL Centre for Decision Making Uncertainty.

Network Thinkers recognise that networks may have been designed for a purpose (‘intentional networks’), or may emerge from a variety of connections and purposes ‘unintentional networks’; the latter have patterns which evolve and change over time.

Network Analysts are probably those most likely to work with formally diagrammed representations of networks. They survey networks to figure out which nodes are more central, which are more on the periphery. For example, someone may not themselves have many links, but they may link key clusters within the overall network and thus play a central role.

For a simple network with up to about 20 nodes it isn’t too difficult to spot these characteristics in a network diagram, but when the diagrams get more complex it is a good idea to use software which not only draws a representation, but can also perform mathematical analyses (as described more below).

Network Builders help networks grow by creating and strengthening connections between other people, not just their own. Often these connections are between people (or organisations) already connected to the ‘builder’, who might also be described as a broker or bridge-builder. In the kind of community building work that David and Drew do, these people are out there in the community and serve a valuable function.

Networkers, Drew defined as people who are trying to build their own network. They may call their contacts ‘a network’, but more properly it is a list of their direct contacts.

Uses of network theory and analysis
Drew mentioned a number of applications for network thinking.

Organisations, partnerships. A prominent use is in management of organisations, e.g. creating networks to optimise the flows of knowledge and information. A more expanded but similar use is to facilitate partnership working between organisations, communities and individuals: this is a major focus of the work which Drew and David do, and Drew promised to give us examples.

Life transitions. Within a project for the Centre for Ageing Better, they are deploying network analysis with a time dimension, showing how a person’s networks of support and friendship and engagement can change as they age. In the example he showed us later, a fictitious aggregated persona had her network connections changed as her husband retired, then died; she compensated for this by joining activity groups in the community, but later her ill health prevented her from attending them. Also changing over time was her relationship to agencies and individuals in the health service.

Space design. As an architect, Drew notes that network theory can be used in urban design, to identify those places that are most central to the structure of and life in a city. Epidemiology uses network theory to understand how infectious diseases spread, and behaviours which have positive or negative health consequences (from jogging to alcoholism).

Military doctrine is defined by NATO as ‘fundamental principles by which military forces guide their actions in support of objectives… authoritative but [requiring] judgment in application.’ Drew said that the US military now talks about ‘fighting networks with networks’. In the US Military Academy at West Point, Virginia there is now a Network Science Center, a multidisciplinary research project for representing and understanding physical, biological and social phenomena through network-analytical approaches (see Security services, police forces and of course intelligence services also use network analysis.

Some network concepts
Link maxima. There is quite a bit of maths in network theory, but some levels are easy to understand. Consider, for example, the relationship between the number of nodes, and the number of connections possible between them.

My explanation: suppose I have two friends: in network terms we are three nodes (ignoring our other friends for the sake of argument!). I’m friends with Jim, also Anna, but they don’t yet know each other. So Jim has one connection within the network, Anna has one, and I have two. I introduce Jim to Anna; now each of us has two connections, and the maximum number of possible connections between three nodes (three connections) has been reached.

Try drawing a series of simple circle-and-line diagrams, and count the connections possible. With four nodes, each can have up to three connections and the maximum number of connections is six. In a network of six nodes, each can form up to five connections; the total possible number of connections is 15.

There is a general formula; where n = the number of nodes, the maximum number of connections is n times (n minus one), and the total divided by two. With ten nodes, the maximum number of connections is 45. Double the size of the network to 20 nodes, and now 190 links are possible.

Of course, in a real live network, not everyone is connected directly to everyone else; in any case we just wouldn’t have the cognitive ability to maintain so many links. In networks which Drew and David have mapped, the most links any one node directly makes is about 15. But everyone is connected to everyone indirectly through intermediate nodes in the network.

Centrality. Network theory identifies several forms of ‘centrality’, which broadly stated is a measure of which are the most important nodes in a network system. Today, said Drew, we would look at closeness centrality and betweenness centrality.

As I understand it, the most basic kind of centrality measure is ‘degree centrality’, which simply means the number of links each node has. A person with links to 2 others in the network has less degree centrality than someone with 10. But this can be complicated if the links have some directionality. Consider, on Facebook or Twitter someone may have two million incoming links (‘likes’ or ‘follows’) and so is popular, but has few outgoing links and so is not particularly gregarious.

More complex centrality indices use the idea of the ‘length’ of a path between nodes. This is potentially confusing, because the spread of nodes on a network diagram is bound to mean that some connecting lines appear longer than others, but this is not what is meant. Length here is measured as the number of hops it takes to get from one node to another. If A is linked directly to D, and D directly to M, and M directly to Y, and that is the only way to get from A to Y, then the length of the path from A to Y is three hops.

A simple definition of closeness centrality is centrality to the network as a whole. Nodes which have a high closeness score are best placed to spread information across a network, and they also have a good overview of what is happening across the network.

Suppose you have 26 nodes in a network labelled A to Z and you want to calculate the closeness centrality of node M, add up the number of hops it takes to get from M to A, from M to B, from M to C and so on. The sum of all those lengths for M, divided by the total number of nodes, has been called its index of ‘farness’, and its index of ‘closeness’ is simply the inverse of this.

Betweenness centrality notes that some individual nodes are central to different bits of the network: this is common in networks made up of people. We can identify clusters of nodes that hang together, versus clusters weakly linked to the others. You might do this to identify who to lobby, to whom to feed information, to have the most effect on the network. Nodes with high betweenness centrality act as important bridges within the network, but may also be potential single points of failure.

Betweenness centrality is more difficult to compute. Repeating the above example, we would ask how often does node M act as a ‘stepping stone’ on the path between any two other nodes? This concept was introduced by Linton Freeman in 1977 to help identify, in human networks, who in a network has the most influence or control on communication between other people.

Eigenvector centrality measures how well connected a node is to other well-connected nodes, and such nodes generally play a leadership role within the network.

As I understand it, this is a kind of ‘metameasure’ based on already computed centrality indices for the nodes. Connecting to a node with a high closeness or betweenness centrality (a well connected and influential node) counts for more than connecting to one with a low score. You raise your eigenvector centrality score by connecting to as many well-connected people as you can.

Network density. There are various definitions of this metric. Drew thinks the most useful one is, the average number of connections per node within the network. This works for any network size.

Our imagined ‘A to Z’ network has a theoretical maximum of 325 connections, and if they were all active, each node would have 25 links and we could call that situation ‘100% density’. But polling the network, we may find that A actually has 5 connections, B has 7, C has 3, D has 11 and so on.

Clusters and communities. Network analysis software can identify clusters of nodes which tend to hang together. This is not because they share a common characteristic, but because of the place they occupy in the network. The software can then auto-colour those nodes in groups to help you to notice them. Usually these network clusters turn out to have a basis in the nature of real functional links within the community.

One method for detecting hierarchical sets of communities and sub-communities in large networks was developed at the Catholic University of Louvain in Belgium and is called the Louvain Method. It’s available as C++ or Matlab code and is used in social network analysis tools such as NetworkX and Gephi. (See a pretty thorough if dense explanation on Quora at

Another approach to cluster detection is able to notice clusters that overlap, and that it would seem is what the Kumu web-based network analysis tool uses.

Types and uses of networks
David now took over the meeting. As a journalist he had noticed how in a community affected by some proposed urban development project, a ‘helicopter view’ might reveal disconnected initiatives across the community; how to join them up, how to overcome the silo mentalities which crystallise around different professions and cliques? Thus he became interested in network thinking.

David showed us a couple of diagrammatic slides originated by Harold Jarche. One, labelled ‘the network learning model’, creates a space between two axes. The vertical axis indicates ‘diversity’ and ranges from ‘structured and hierarchical’ at the bottom to ‘informal and networked’ at the top. The other axis has ‘goal-oriented and collaborative’ at the left and ‘opportunity-driven and cooperative’ at the right. Ranged up the diagram from bottom-left to top-right are three slightly overlapped balloons representing three levels of networks for sharing and learning:

  • Work Teams (structured, goal oriented): based inside a formal organisational structure, sharing complex knowledge, driven by deadlines, strong social ties, co-creating learning.
  • Communities of Practice (half-way along both axes): spanning shared concerns across organisations, a trusted space to test ideas, people don’t know each other personally, but integrating work and learning.
  • Social Networks (informally co-operative, opportunity-driven): high diversity of ideas and opinions such that you might find stuff you hadn’t considered in your task group; weak social ties.

Visualisation and analysis software
I have already mentioned the use of specialised software to help represent networks and to analyse them. After the exercise and a break, Drew returned to this topic. Most network analysis software, he said, has an analytical and heavily mathematical flavour: examples are UciNet and Gephi. But recently, easier-to-use software has appeared and he described three that he and David have used.

  • yEd is a free , open-source diagramming package for Linux, Mac and Windows. I have used this myself, but for drawing a particular kind of non-social network diagram: Entity-Relationship Diagrams (ERD) used in database design. According to Drew, yEd also has some ability to analyse network maps.
  • Kumu is their current favourite and main recommendation. It is a web-based system, and you can sign up for a basic free account at You can draw network diagrams with Kumu or it will make them from data and do the analysis; it can also hold stacks of attribute information attached to the nodes and the connections, which enables clever searches and filters on the diagram. Drew and David have been combining network analysis with Asset-Based Community Development methodology (of which, more later), and being able to annotate the notes with what assets they bring to the table has been very useful.
  • Polinode. Drew described this as ‘a very slick program’, also web-based, and business oriented. It has a built-in survey mechanism which is useful for collecting information about people in your business network and automatically populating the network diagram accordingly.

Example network maps

Readers may want to look at the PDF file of the slides to see the examples described here. The slides can be found online by clicking here unless you are reading this off paper, in which case the URL is: 7HUxJTYr0o/edit?ts=5784d539#slide=id.g115d229400_2_10

The first example was created through a survey conducted for the Irish Crafts Council, polling designers and makers, suppliers, retailers and agencies in Ballyhoura, South Tipperary, Wexford, Kilkenny and West Cork. It presents as quite a dense diagram with over 400 nodes and an overlapping mesh of connection lines which in places all run into each other so it is hard to distinguish them. The software discovered three major clusters, based on the link patterns. Interestingly, the clusters were based strongly on geographical proximity – it wasn’t the case that jewellers would network with other like craftspeople across the region, for example, but across the crafts, people networked locally and helped each other out.

The study also revealed that economic development agencies had lots of connections; in West Cork in particular, the agency played a leading role in the network. Meanwhile, though the Wexford and Kilkenny cluster showed a very dense pattern of connections, they were mostly connections within cliques of craft workers, and as such were not very influential across the area.

A second example was for a regeneration partnership programme for Berwick upon Tweed; in this diagram, all 50 or so nodes were organisations. Seven, highlighted on the diagram by the software, were major ‘hubs’ with multiple linkages, with the Borough Council as the most central, playing a ‘brokering’ role between the more strategic organisations at the top of the diagram, and the tightly focused local organisations at the bottom.

Within this project, they then compared the network graph with the results of a survey in which each organisation within the network was asked to rate their perception of (a) the skills held by the other organisations, across five categories and (b) resources those organisations also had to offer, across the same five categories. Dramatically, the Borough Council which the network analysis had identified as being ‘most central’ scores spectacularly the worst on both counts! This leads to interesting discussions. So do you pump money and training and resources into the Council as the centre of that network, or bypass them with a new project? (What actually happened was that all Borough Councils in Northumberland were disbanded.)

Kumu again, in detail
Drew was keen to point out that whereas in the past different software tools would have been needed to work on the different phases of the Berwick upon Tweed project, they were able to do it all simultaneously in Kumu. If anyone is interesting in pursuing this, after this session, he suggests that we get a free Kumu account. That will give each of us a Kumu ‘name’, and he suggested he could put up a Kumu site where we could discuss this stuff and experiment.

There are various ways of getting network data into Kumu. You can draw directly to the screen; Drew likes drawing so he appreciates this. Or, you can type commands into a screen terminal. Comma-separated database files (.csv) can also be uploaded. Kumu can ingest spreadsheet files from Google Sheets, and these in turn can be fed from Google Forms, Google’s form-interface web-based input software. Drew also noted that Kumu is planning to introduce its own integrated survey module soon.

Kumu, as already explained, lets you add extra data to nodes, add node attributes, and tag nodes, which makes search and filtering more powerful. Drew also believes that the developers are very responsive and they listen to how people want it to develop.

More examples, fictitious and applied
Drew showed a network map for ‘Slipham’ – a fictitious community which they use for testing ideas and policies. It is populated by the kinds of local people, organisations and services which would be typical for most communities: there’s the General Hospital and a group GP practice, a branch of Age UK, a number of local councillors, a Somali Association, the Rotary Club, the Police, several sports clubs, etc… and a number of individuals who provide bridging functions through their multiple engagements.

The ‘centrality’ measures for the nodes are emphasised on the map by having the more central, better connected nodes displayed as a proportionately larger circle. The circles are coloured – automatically by the software, on the basis of attribute data that has

been entered for each node. Nodes around education are coloured yellow, red signifies health and social care, and blue is socio-political. (How Kumu displays nodes and links can be customised by bits of Cascading Style Sheet coding, as used in Web site design.)

Drew switched to the Slipham map in Kumu itself, online, and demonstrated how each node can have attributes stored ‘within’ it. He also showed some of the ways that a map can be probed, for example by clicking on one node and having the map display only those other nodes directly linked to it as working contacts – or perhaps within two ‘hops’ rather than one. Selecting two nodes at opposite sides of the map, he got Kumu to show the immediate links of each, helping to identify a couple of nodes shared between them, which could be used as conduits for contact or liaison.

Drew demonstrated a network map produced for NHS Education Scotland (NES). The connector lines on here were interesting in three ways. Firstly, they displayed as curved rather than straight; they displayed with three grades of thickness; and they also seemed to indicate directionality, as each line had an arrowhead at just one end.

The purpose of this investigation was to identify sources and flows of information. The thickness of the line indicates the ‘volume’ of flow (an attribute which you can control by adding a value to the data behind the connection), and although Drew did not explain this, the arrowheads clearly indicate the direction of information flow.

Drew used this diagram to warn of an effect in mapping in real life, which is that one often works for a single client within the network (in this case, the NES), who readily provide their own links, and those initially dominate the map. If you want a more thorough picture, you will have to make contact with the organisations they have identified, and survey them to try to ascertain their links too. It may take a few iterations of this process to get the wider picture.

A second exercise (or game)
We had already had a simple discussion exercise about networks in our table groups. Drew now offered something more playful. He used Kumu to present an abstract network map where the nodes were identified by numeral only. Displayed next to this map on the left was a listing of the ‘top ten’ nodes ranked by betweenness centrality. Each table was to ‘adopt’ one node and then try to promote it up the centrality score table, either by adding a link, or deleting one. (The ‘adopted’ node did not have to be a terminus for the link added or deleted.) Based on table choices, Drew input the changes into the Kumu map and re-displayed the rankings. We did this a couple of times.

The game was competitive and fun, and less confusing than it might have been, because three tables adopted the same node, and two another. The biggest effects came from making or breaking links between nodes which were already well connected. This was a good exercise in learning how to ‘read’ a network diagram.

Collecting information for mapping
After the exercise and a coffee break, Drew gave us guidance about how to prepare for network mapping by gathering information. Where a community is dispersed or hard to collect together, you might prepare an online survey; they’d used Google Form, but Survey Monkey also works. The data may have to be fed in via Excel or Google Sheets, and as Comma Separated Values (.csv). London Voluntary Service Council is currently doing a network mapping exercise using online forms.

If you are holding an event where participants are present, you could get people to input data straight into Kumu, or create a drawn-up paper sheet or questionnaire. Drew showed a model: an Organisational Mapping Sheet they had prepared to collect data for a project on tobacco reduction. Each organisation notes their name at the top of the sheet, and adds some ‘interest keywords’ – I guess this is used to sift the nodes into categories, and so would work best with a predetermined tag vocabulary.

The sample illustrated then had a number of small repeated tables, the first of which was for ‘your organisation’. One box asked ‘Sharing?’ – if you think your organisation is good at sharing, you tick it; if not, you leave it blank.

The sample form then listed five rows of activity: Online communications, Technical, Management, Financial and Community, and next to each of this was a box for ‘Skills’ and another for ‘Resources’. If you have technical skills, you tick that box, and if you have financial resources, you tick that. Otherwise, you leave them blank.

The sample sheet shown had seven other mini-tables identical to the first, except that rather than being about ‘your organisation’ this was for your private opinion about the sharing abilities, skills and resources about the other organisations with which yours was most in contact. There was a note to assure people that ‘individual contributions will be confidential and unattributable’.

This is just one example. Depending on the theme and the nature of the community, the skill and resources sets may be different. Instead of a tick or the absence of one, you could calibrate the data with a numerical score, for example a plus or minus figure, or a number between 1 and 5, or a number of ticks. A calibrated assessment seems to have been used in the Berwick upon Tweed case study described earlier.

Other supplementary means for collecting data could be face to face or telephone interviews. If you are ‘iterating’ the investigation by contacting other organisations named in the linking, telephone interviews make a lot of sense unless you have an online form resource and can invite the second-round participants to fill that out too.

Not that difficult!
Drew showed one example of a fairly complex network map encompassing about 70 organisations, which had been created by the manager of the a Children’s Centre in Croydon to indicate those involved in some way in Croydon’s ‘Best Start’ programme, for children under 5 in families at risk in some way. Following a workshop, she went home, created a Kumu account and without previous experience of network mapping created the network map in two hours.

Another advantage of developing a network map in an online environment like Kumu is that Drew was able, as it were, to ‘look over her shoulder’ and help her remotely to develop her network map further.

One problem with a network map created thus by an individual is that, although she thinks those links exist, she doesn’t know for sure, and the links are unqualified in other ways. A maxim in the network mapping community is ‘the node knows’ – best not to speculate but ask people and organisations in a prospective network what their connections really are.

Time-base networks: the CFAB example
Networks can have a time dimension, and Kumu can cope with these too. An example might be a flow-chart, or a process-mapping chart.

Family Maps. As mentioned briefly above, a recent project which Drew and David have tackled is for the Centre for Ageing Better (CFAB), which wanted to investigate how technology can be used to assist people in later life. They started by developing six ‘personas’. A persona is a fictitious person who embodies a set of characteristics typically found together, so can represent a sector of the population in a simulation game such as one that CFAB ran with 50 people at one of their conferences.

These personas were based on research conducted by the polling company IpsosMORI, who have a database of population characteristics from polling, plus focus groups. IpsosMORI had already concluded that three factors dominate in securing well being in later life: financial security, health, and social connections.

Based on IpsosMORI cluster work, the CFAB project created ‘Mary’, whose tagline was ‘Can Do and Connected’. The character was represented by a cartoon portrait by Drew, in which she says ‘I want to remain independent as long as possible!’ Five sentences explain that she is 73, owns her home outright, but feels she has to watch her spending. She recently lost her husband, but stays positive with support from friends and family, and engages in local activities. She has long-term health issues, but hopes things will improve and stays optimistic. She uses an old mobile phone for calls and texts, but her attitude indicates she would explore technology further if she thought it would help…

Mary lives in ‘Slipham’ (of course) and has connections with various agencies there such as the General Hospital, a community nurse and a bowling club, plus several individuals who are friends, children and grandchildren, etc.

For this project (perhaps through the gaming process?) they also developed time-based maps which showed how Mary’s network might evolve over time: she compensated for the loss of her husband by joining community social activities such as ‘PowerAgers’ (a walking group), but later had to give them up due to advancing ill health, which also changed her needs and her network of support from health and social care agencies. For the network mapping in Kumu, this evolution of her network was coded by tagging each node in it for inclusion in various year bands. You can then advance year by year (in this case, in five-year steps) to see how the persona’s network develops.

The purpose of the exercise was to explore how support services might be better co­ordinated to help people as they get older – and the role which technology might play in that. This was allied to investigation of how vulnerable people are to social isolation.

Drew spoke of the phenomenon of ‘social ageing’ – how our social connections change as a result of ageing. A related concept is ‘network risk’, which spots which kinds of contact network are vulnerable to sudden collapse. They tend to be the ones dependent on physical activity – but could also be impacted by poor public transport provision, or financial hardship, meaning that you can no longer afford to participate in activities.

Multi-level maps. Drew showed how for the CFAB exercise they created a custom version of ‘Slipham’. So long as the node entities reside in the same Kumu project, their links and other attributes will be inherited by other maps created within the project. Drew pointed out that a couple of the nodes on Mary’s personal map are also on the wider Slipham map – others, which might be relevant to Mary’s future happiness and well-being (such as Age UK Slipshire and the University of the Third Age), were not.

Linking to Asset Based Community Development projects

Drew’s final slide showed a very complex network map developed around several projects in Croydon, on which he and David are currently working.

I was interested to note that in the Croydon work, the network mapping is part of a larger programme using ABCD methodology – Asset Based Community Development. My awareness of ABCD has come from another community development practitioner, Ron Donaldson – who spoke at the NetIKX#78 event – and who uses ABCD in some of his own work.

Asset Based Community Development is an approach to developing activities and services within communities which focuses not so much on community needs as on the skills, resources and capabilities of individuals and groupings and organisations within a community. The approach was developed in the 1990 by John L McKnight and John P Kretzmann at the Institute for Policy Research at Northwestern University in Illinois, USA. The website of the ABCD Institute which they founded, anchored within the university’s Center for Civic Engagement, is at

For example, we may find out that the Methodist Church has a meeting hall, the school has a grassy area suitable for a neighbourhood fair, Darren is a whiz at Web sites, Ant is a cartoonist, Charmaine and Sue make Jamaican patties, Nguyen is a videographer and videomaker with his own kit, three of the gents from Men In Sheds would like to teach basic woodwork, Sarah has a pillar drill and lathe and can weld, Pushpinder creates theatrical costumes, three Green Party activists want to encourage materials recycling, Jordan can drive a minibus… If these assets can be put together in inventive ways, the community can start to help itself rather than waiting on help from on high.

A key tool in ABCD is the Capacity Inventory which gathers data about who has or can do what, and also finds out how they are connected to projects in the community. To me it has now become obvious that rather than a static card index or other capacity inventory database, an interactive network map with data behind it such as Drew had shown us using Kumu fits beautifully with Asset Based Community Development.

Audience feedback
Many people likes the example of ‘Mary’. Steve Dale said that it is not uncommon for our networks to shrink as we get older. Rob Rosset wondered if that is something we accept, or struggle against. Steve felt that maybe the ageing brain is not as able to cope with lots of connections, but thinking about people he knew years ago, such as in his Navy career – well, their paths have diverged from his anyway, and he would rather stay close to a smaller circle of family and friends who mean a lot to him.

Someone else affected by the story of ‘Mary’ thought exclusion and isolation are the other side of networking. She added that many people are not confident and outgoing networkers, so as well as thinking about how to strengthen our networks by building the links between those who link readily, we should also think about those who stand on the sidelines only for lack of encouragement.

Conrad reflected on the game we had played. Some people with try to strengthen their local dominance in a network, whereas if they were less egotistical and were prepared to connect on an equal basis with people at the heart of other clusters, more could be achieved. Drew commented that a network map can identify someone who occupies a strong network position – but that doesn’t guarantee the right constructive attitude!

David Wilcox reported that a concern of the London Voluntary Service Council is that in the current austerity climate, voluntary action is being crippled as the agencies and associations which used to serve as hubs are taken out. How can the existing groups become better at using network thinking and technology? But those organisations rarely have those skills and capabilities.

David has begun thinking it would be good to develop a range of personas which might represent Londoners – as a starting point for examining what kinds of connections they tend to have, and what they might benefit from in the future – either on their own, or assisted by ‘Network Builders’. Because it looks as if increasingly we are going to have to create social infrastructure from the bottom up. He’d be interested to know if anybody else would be interested in that, to get a project going.

Clare Parry thought that people may share a neighbourhood, but the communities don’t connect – the example she offered was of traveller communities not connecting to settled ones (different ethnic and cultural communities would be another case in point). David said that this was a feature of the ABCD work in Croydon which has been going for about four years. They have volunteer ‘community connectors’ and Drew has been using network mapping to help them identify useful points of inter-community connection.

Finally, Martin expressed concerns about the ability of network maps to misrepresent situations if the data input is wrong or insufficient. Drew said that network maps can give you insights – but if you really want to know what’s going on, you have to investigate that in the field.

This was an interesting and well-attended NetIKX meeting. It’s nice when we have use of the Upper Mews Room at the British Dental Association – it accommodates 50 comfortably around tables and is well lit, very suitable for the round-table syndicate groups which are a hallmark of NetIKX meetings. (To learn more about NetIKX, see

As one of the early slides said, there are various ways of getting a picture of how things and people are organised, such as through stories. Network analysis is a more structural and structured method. But I think more people are comfortable with stories and I suspect some of my NetIKX colleagues felt they had waded beyond their depth when we tackled network theory! This may be why the story of ‘Mary’ resonated so well – however fictitious, there was a story in there. The stories around other projects such as the Berwick upon Tweed one also helped bring these to life for me.

I was intrigued enough by betweenness metrics and other abstract aspects of network theory to do more reading around them, if only to help explain them. I hope my expanded explanations of how these things work with reference to my fictitious abstract ‘A to Z’ network are (a) helpful and (b) not misleading!

Obviously there are subtleties of social network analysis and visualisation which we didn’t cover (and which could have led to rapid cognitive overload has it been attempted). For example: the directionality of links; the strength of links; how if at all to weight the value of a particular node’s contribution to the network. I look forward to playing with Kumu to discover more and I have signed up, as suggested by Drew.

On the CFAB example and social ageing —The story of Mary made me think too. Many people of my age (early 60s) and even decades older find that the Internet and social media, even Facebook plus digital photography, help us keep in contact with friends dispersed across the globe, people whom we meet face to face but rarely, and even make new friends by being introduced online to friends of friends. Quite cost effectively, and even if we are housebound.

Writing and reading – perhaps falling out of fashion? – can network us with others in great depth. Frequent emails are important to my mother and me. Text can link us to ideas across time as well as space, for example by reading books – or accounts like this of interesting meetings we may have missed…

In Mary’s story of ageing, illness decreased her access to wider networks, but that is not the only factor. There are many activities I cannot join for lack of funds. I cannot begin to express how grateful I am to have a 60+ Oystercard and therefore free travel across London!



0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published.