‘How Stuff Spreads’: Experiments in social networking

Posted on September 27, 2013 by

1


Spread_nu

Brett Cherry, Philip Garnett and John Bissell relate their experience in teaching young people about social networks through a unique workshop that encourages learning about human behaviour, innovation and modelling.

The young people’s workshop for the 2013 British Science Festival in Newcastle organised by the Tipping Points project was an experiment in itself and the results are worth exploring. Our workshop for students (KS4) was called ‘How Stuff Spreads’, a title borrowed from Tipping Points researcher now based at University of Bristol, Professor Alex Bentley. The challenge was to develop an engaging activity that students would enjoy while learning about the diffusion of information within social networks. Not only did the game Colour Spread (based on an idea by TP statistician Dr Camila Caiado who helped organise the workshop) spark interest in the bright, young minds of the people who attended, they actually REALLY liked it.

threecolours

Start with three colours

Scientists model the spread of information (diffusion) in different ways, one of which is to look at how information diffuses neutrally, meaning that the probability of a trend or product being copied is proportional to the frequency in which that trend or product appears in the population. Known as the neutral model it is thought that a surprising amount of human behaviour at the population level can be thought of (modelled) as individuals simply copying each other. Everything from the fashions we follow and the products we buy. In neutral copying no difference is assumed between the intrinsic value of an idea or product. In our colour game `Colour Spread’ we assume such a neutral model: choosing a colour doesn’t really put you at an advantage or disadvantage because it’s simply a colour, there is no incentive behind it.

At the start of the workshop we passed out numbered labels that were clearly visible, which allowed us to track which ‘agents’ within the model spread particular colours, all the way back to the original seeds (the first people given colour cards). During the experiment we attempted to control for gender, e.g. males would receive odd numbers and females even, or control for different schools (if we happened to have more than one school attending the same session).  We would ‘seed’ three colour cards amongst students, sometimes asking for volunteers or else choosing them at random.

Spread_nu

Neutral model from experiment showing spread of different colours between students (agents).

After the seeds received their colour (green, orange or pink) they would then have to find another person to persuade to take that colour from the designated colour station, both agents (the persuader and persuaded) would then go to the network station to have their colour and numbers recorded. Of course no one had to pick a certain colour if they didn’t want to, in fact, no one really had to pick a colour at all, but everyone usually did, including teachers. Most participants were quite taken with the game and almost immediately got involved, with some students even starting their own mini-campaigns to get their peers to choose their colour. So despite it being ‘neutral’, students did compete (perhaps highlighting our innate desire to belong to a successful group), but knowing that the specific colour they chose made little difference in itself. However, when providing students with an incentive (sweet) and the ability to ‘cheat’ (swap their colour for another), things started to get a lot more interesting.

lolly

Give an incentive

We ran the same experiment again, but this time we told students that if they held the most popular colour card at the game finish, meaning the one that spreads the most, then they’d get a lolly. For example, if people with green cards managed to convince most of the population to ‘go green’ they would win the prize in the end. In this experiment they could trade in their own card and encourage others to do so openly.

This is what happens to the network model when you provide an incentive:

Spread_Ins

When an incentive was provided (lolly) nearly everyone chose the same colour.

Sometimes we told students they could trade in their card for another colour, but often they would figure it out for themselves. After all, the experiment was about demonstrating to them that behaviours can be learned socially, spreading widely and rapidly through a network. For example, if one person traded in their card then another would and so on. The person at the colour station would see everyone trading their card in for green, but during some sessions it could change suddenly to pink or orange. The growing fervour amongst students resembled that of people trading at a busy city stock exchange. Students would monitor the developing network and change their behaviour accordingly. Most people wanted to buy into the colour that would win.

Spread_Ins_Cheats

Example of model where students were allowed to ‘cheat’ making it possible to trade in their colour for the same as everyone else.

In some cases prior to running the incentive experiment we would encourage students to think carefully about how they could collaborate so that everyone would win. Naturally, students would get all of their peers to convert to one colour so they could each win a lolly. But not all students would convert.  Some would keep their colour regardless and in fact attempted to encourage others to take their colour even though it was clear that a certain colour, often pink or green, would eventually dominate. What we wanted to teach students with this experiment is that incentives influence human behaviour, and in this case it was quite clear that giving students a lolly if they won caused them to pay much more attention to what colour they chose and who they persuaded.

‘Connectors’ and ‘Super Spreaders’

In Malcolm Gladwell’s book The Tipping Point he provides examples of individual agents that he calls ‘connectors’ — people who have a knack for social networking and are able to spread their influence easily. Social scientists call these agents ‘super spreaders’ as they are capable of spreading information quickly and efficiently throughout a population. The same idea could be applied to the spread of sickness or disease as well, from the common cold to the HIV virus. There are also historical examples of connectors in social history. Throughout our experiment sessions we would occasionally give historical examples of important connectors such as Dr Martin Luther King when he gave his ‘I Have A Dream’ speech in the US which was copied by different populations throughout the world, influencing millions of people.

In the experiment we were able to identify the connectors because (as noted earlier) each had a number and as you can see connectors were essential to spreading colours. Usually the colour with the most connectors would win. The connectors’ influence allowed certain colours to spread effectively. The people who were good at influencing their peers to choose their colour were also great ‘salespeople’, a term also used by Gladwell in his book. Those good at selling ideas are (no surprise) great at advertising and marketing, which is essentially the business of spreading influence, whether for example through corporate behemoths like Coca Cola, governments or even environmental groups. They all have a product to sell whether its material or not. And it is often the immaterial ones that can have the most influence in society.

Spread_Prod

In this model notice how connectors 35, 12 and 16 were decisive in spreading the colour green. Or 32, 14 and 8 in the case of pink and 30 for orange.

Now what’s interesting about this is that you can think of a range of parallel examples in politics, culture, science and religion to name only several broad topics where social copying applies. In fashion, for example, certain styles or brands of clothing catch on because of the influence someone has modelling it, such as a pop music celebrity. Other influential people may copy them and then others perhaps less influential copy them also. Gladwell uses the example of Hush Puppy shoes in New York City. Social networking online is also obviously a brilliant example of how things spread through Facebook and Twitter. Think also of political campaigns. The number of connections a politician has within certain regions are crucial to the success of their campaign overall: without the right connectors politicians will have less influence and receive fewer votes.

Drawing Diffusion

Besides Colour Spread we had another similar activity for students to try that was devised by computer scientist and biologist Dr Philip Garnett. This was similar to Colour Spread in that it involved encouraging students to persuade one another, but was more subtle than the previous activity. In this experiment students would get others to copy their drawing once they had copied from another. Seeds were used as in Colour Spread, though this time – rather than being handed coloured cards – they copied one of our `seed’ drawings.

These are the original images that the first school copied from:drawing2

drawing1

drawing3drawing4

We then chose one copied drawing of each original image to give to the next school to copy and then used drawings from the second school for the third and so on. Students would copy the drawing then get one of their friends to copy their drawing. As you can see there is a space for the student to fill in the number of the card they copied from, name, gender etc.

Template

Students were allowed to copy any drawing they liked and could also copy in any way they would like. What would happen is that session after session, school after school, the drawings would diffuse in unique ways. Compare for example the original drawing of the house above with two drawings by students during one of the last sessions.

house2

house1

Psychologists say that an image of a house is something unique to the subject. That a person’s idea of a house is unique to them because of the social and cultural meanings attached to it. So when asked to reproduce a drawing of a house students would often add to the drawing that they were copying things that illustrate their internal idea of a house, rather than faithfully reproduce the house in front of them. Over time we spotted innovations and the house started to look significantly different from the original. For example, sometimes students would add window frames or birds, a rainbow, or a fence etc. Understanding human interpretation is key here, how our understanding of something as basic as a house is shaped by our own ideas, so that even the simplest everyday objects we encounter are impregnated with meaning. It also says something unique about people as individuals; they create the world around them just as they are shaped by it.

shapes

The other images that we chose to be copied are more abstract (lines and shapes) and therefore it is unlikely that an individual would have their own internal model of what they are like. These images are therefore more faithfully reproduced; the students would attempt to copy what they have seen rather than draw what is in their head. However, no human copying process is perfect so these images evolve through time, just as information does as it spreads through social networks. We hope to produce an animation that shows how these drawings changed over time and make it available to the schools who participated in How Stuff Spreads.

As a form of public engagement we found that How Stuff Spreads worked rather well, in fact, even better than we had originally thought. It is always difficult to predict how young people will take to an activity they know will likely be didactic in some way, in this case teaching them about social science. But the medium is the message after all. Many students were so involved with the activities themselves that learning and having fun were indistinguishable. Perhaps this also had something to do with core lessons that speak to something fundamental about our humanity: that we love to copy each other, and in doing so find new ways to create, innovate and evolve over time.