In this presentation we demonstrate how groups of people adopt a certain trend or fashion, how rumours propagate through networks, how to communicate information to a large group of people in the most efficient way, and how the framework of network analysis can be used to better understand how we communicate.
The full research is available on http://noduslabs.com
2. Watch
the video on
vimeo.com/2035117
SOCIAL NETWORK
The nodes are the people, the connections are interactions
between them (visualization by Gephi - www.gephi.org)
3. TYPES OF NETWORK
Scale-free - degrees distributed following power-law
(a few, but significant # of well-connected and disconnected)
4. TYPES OF NETWORK
Small world - special case of scale free – tightly-knit loosely
connected communities with short distance between the nodes
5. TYPES OF NETWORK
Random - degrees distributed “normally” across the nodes
(most have an average number of connections)
6. S I R
S I S
S I R S
EPIDEMIC MODELS
S: Susceptible, I: Infected, R: Removed/Recovered
(Ball 1997; Newman 2002; Newman et al 2006; Watts 2002)
7. Watch
the video on
vimeo.com/36958670
CONTAGION DYNAMICS
Message = Virus
8. “healthy” “infected”
HOW DOES IT HAPPEN?
Ideology, Trends, Collective action, Protest, Meme...
9. most “friends”
adopted a
trend, so the
blue node does
the same finally
1. INFORMATION CASCADES
Herd-like behavior, influenced by the others. Only when
“conversion threshold“ is exceeded (Watts 2002; Hui et al 2010; Young 2002)
10. no connections between the nodes many connections between the nodes
= cascades not possible = cascades can occur
2. GIANT COMPONENT
Most nodes must belong to the same component
for the global epidemics to occur (Watts 2002; Newman et al 2006)
11. 3. START WITH A GROUP
Rapid spread of disease within tightly connected communities
can lead to an epidemic outbreak even if the links are loose
12. WHY?
Because once the contagion is spread within the group, it will
spread across super-network to the other groups (Ball 1997).
13. Better than random nodes, but still not Optimal - leave the same number
perfect - immunize random groups of susceptibles in each group
STRATEGIES OF RESISTANCE
Leave the number of susceptibles the same in each group, thus
preventing the virus from spreading within and throughout.
14. 99%? 10% IS ENOUGH.
Committed 10% can change the opinion of the majority as long
as they persistently broadcast their message (Xie et al 2011)
15. * not too many!
4. BUILD SHORTCUTS
Scale-free networks with shortcuts are better in propagating,
dense networks are better for cascades. (Kuperman 2001; Yan et al 2008)
16. 1. Amplitude of contagion increases with the higher number of random shortcuts
(Cummings 2005; Kuperman 2001)
2. Small-world wirings (links between different communities) enhance network
synchronization (Barahora & Pecora 2002).
3. Synchronization (simultaneous information cascades) are boosted if the links are
made between the nodes of varying degree (Boccaletti 2006)
4. Assortative networks (well-connected nodes attract each other) are good in
percolating (spreading the message further and maintaining the endemic contagion
for a longer term period). Disassortative networks (nodes with varying degree
connect together) are better in sync, but the contagion is periodic and short lived
(Bragard 2007; Newman 2002)
THE ETHICS OF PROMISCUITY
Or how to make random connections,
without driving your network crazy.
17. 5. FOCUS ON BROKERS
The nodes that connect different communities, are the best one
to target when spreading a message. (Stonedahl 2010; Freeman 1997)
18. Image: CC Laura Billings @ FlickR
6. MESSAGE = VIRUS
The message should have the capacity to replicate itself
across the network.
19. Watch
the video on
vimeo.com/33742762
Rumours started on Twitter during the London riots were much more long-lived when started with a
query, which in turn produced statements in support and opposition of the original statement.
START WITH A QUESTION
@someone “Is it true what BILD wrote that Angela Merkel
disappeared?” #weird #politics #germany #shithappens
21. Against Putin Facebook group The viral message should imply
“against Putin”,
not “protect animal rights”
THE SAME PURPOSE
The message should reiterate the purpose that brings
the target network together.
24. 1. Information Cascades
(people should be talking to each other)
2. Giant Component
(most of the people should be connected to each other, bring the “loners” in)
3. Focus on Groups
(better the more densely connected ones, 10% can be enough)
4. Make Random Shortcuts
(communication outside of one’s community, diversity of links)
5. Information Brokers
(people who connect different communities together)
6. Message = Virus
(the message should have the capacity to replicate itself)
SUMMARY
Information Epidemics and Viral Contagion
25. Ball, F. (1997). Epidemics with two levels of mixing. The Annals of Applied Probability, 7(1), 46–89. Institute of Mathematical Statistics.
Retrieved from http://projecteuclid.org/euclid.aoap/1034625252
Ball, F., Neal, P., & Lyne, O. (2010). Epidemics with two levels of mixing. MOdelling Complex Systems, University of Manchester. Institute of
Mathematical Statistics. Retrieved from http://projecteuclid.org/euclid.aoap/1034625252
Bastian, M.; Heymann, S.; Jacomy, M.; (2009). Gephi: An Open Source Software for Exploring and Manipulating Networks. Association for the
Advancement of Artificial Intelligence
Freeman, L. (1977). A Set of Measures of Centrality Based on Betweenness. Sociometry
Vol. 40, No. 1 (Mar., 1977): 35-41
Hui, C., Goldberg, M., Magdon-Ismail, M., & Wallace, W. A. (2010). Simulating the diffusion of information: An agent-based modeling
approach. International Journal of Agent Technologies and Systems (IJATS), 2(3), 31–46. IGI Global.
Kuperman, M., & Abramson, G. (2001). Small World Effect in an Epidemiological Model. Physical Review Letters, 86(13), 2909-2912.
Newman, M. E. J. (2002a). The spread of epidemic disease on networks.
Newman, M. E. J. (2002b). Assortative mixing in networks. Physical Review Letters, 89(20), 5. American Physical Society. Retrieved from
http://arxiv.org/abs/cond-mat/0205405
Newman, M. E. J., Barabasi, A.-L., & Watts, D. J. (2006). The structure and dynamics of networks. Princeton University Press. doi:10.1073/pnas.
0912671107
Stonedahl, F., Rand, W., & Wilensky, U. (2010). Evolving Viral Marketing Strategies. Learning.
Watts, D. (2002). A simple model of global cascades on random networks. Proceedings of the National Academy of, 99(9), 5766-71.
Xie, J., Sreenivasan, S., Korniss, G., Zhang, W., Lim, C., & Szymanski, B. K. (2011). Social consensus through the influence of committed
minorities. Physical Review E, 84(1), 1-9.
Yan, G., Fu, Z.-qian, Ren, J., & Wang, W.-xu. (2008). Collective Synchronization Induced by Epidemic Dynamics on Complex Networks with
Communities. Science And Technology, 0, 3-7.
Young, H. P. (2002). The Diffusion of Innovations in Social Networks. Economy as an evolving complex system 3, 3(1966), 1-19. Oxford
University Press, USA.
REFERENCES
26. INFORMATION EPIDEMICS
AND VIRAL CONTAGION
We used Gephi for network analysis and visualization –
download it on www.gephi.org
We used NetVizz app by Bernhard Rieder to get
Facebook data.
More on www.noduslabs.com -
Contact: Dmitry Paranyushkin | dmitry@noduslabs.com
Twitter: @thisislikecom | Facebook: Nodus Labs