“Connected: The amazing power of social networks and how they shape our lives” was the title of the talk given by Dr Nicholas Christakis at the RSA earlier. Due to rather poor time management, I didn’t make it to the event itself, but followed it online. This link should eventually have the video and downloadable audio of the event.
I’d recommend checking out the full talk, as Christakis is an engaging speaker and his theories make a lot of sense. Rather than recap the full session here, I’ll instead focus on a few areas.
The hypothesis of the talk (and book) is that social context plays an important part in our behaviour and attitudes, and our ties tend to form groups of likeminds. Things ultimately spread in networks.
In his data visualisations, he displayed his theories by using nodes to represent people, with lines acting as connectors.
The number three was a dominant theme throughout the talk.
Christakis noted that there are three theories in how things cluster.
- Induction – Person A’s behaviour directly affects Person B, who then mimics Person A
- Homophily – Person A and Person B both have the pre-existing condition independently, and group together because of this
- Confounding – Person A and Person B are proximate, and share an exposure to an external factor
The confounding theory refutes the idea of network effects. Yet for network effects to be proven, the nature of the connections need to be understood:
- Mutual friendship – where both person A and B are friends
- Ego-perceived friendship – Person A befriends person B, but Person B ignores them
- Alter-perceived friendship – Person B befriends Person A, who ignores Person B
Christakis argues that different relationships will have different effects. He notes that if we were to map our relationships, they wouldn’t form a uniform pattern like a regular lattice but instead vary across three dimensions
- The number of friends/connectors per person/node
- The interconnectedness of friends – are the nodes I am connected to also connected to one another?
- The position within the overall network – is my node in the centre or towards one of the edge?
The final three of his talk is in degrees of influence. Christakis posits that we are not only influenced by our friends but also their friends and their friends’ friends. Three degrees of influence.
He believes that we should look at the networks, rather than the individuals, when formulating policies and strategies, because properties aren’t understandable when just looking at individual components. He used the (excellent) example of carbon. When carbon atoms are linked together in one way, they form graphite. When linked in another way, they form diamond. Two very different structures, with very different properties (And the one with more connections is more valuable).
Thus, he believes we live connected lives (even though he talked about part of it being a genetic trait) because the benefits outweigh the costs. We break off bad connections, and strengthen good ones. We create networks to spread and sustain good and desirable things – things we couldn’t as individuals.
I enjoyed the talk immensely, and would recommend people watch/listen to the full 75 minutes. I appreciated the depth he went into when attempting to determine causation, rather than just correlation. His argument was quite persuasive and of course it has repercussions on how we would be framing our objectives.
It’s got me thinking about whether the value of people within a network differs. Christakis claimed a network could shed its bad apples – I’m not convinced since a breaking of a first order tie doesn’t necessarily break the second order tie, where influence can still pertain. If we were able to break our ties and influence our networks, then surely only good things would spread, and not things like viruses or unhappiness. But notwithstanding, are some apples “better” than others?
Whether through Berry & Keller’s Influentials, Gladwell’s Tipping Point typologies or another example, people have attempted to segment the population in attempts to harness the spread of messages. But does the number, strength and position of connections impact on the value of that person, or is a person only as valuable as his or her network?
Instead could it be analogous to Belbin’s team role functions? A balanced team needs the whole range of roles and contributions in order to be successful. Would a network comprising purely of influentials become less valuable, due to the absence of other types of people to influence?
And so when we devising sampling structures or STP strategies based on their attitudes or behaviour, should we be attempting to create a proxy of individual positioning within a relevant network in order to predict the dynamic interplay of ideas and actions? I’m not even sure if this would be possible, but it would certainly aid our predictions on whether something is sustainable or not.