Information Epidemics and Collective Action


Formal Metadata

Information Epidemics and Collective Action
Network Analysis Perspective
Title of Series
Part Number
Number of Parts
Paranyushkin, Dmitry
CC Attribution - ShareAlike 3.0 Germany:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor and the work or content is shared also in adapted form only under the conditions of this license.
Release Date

Content Metadata

Subject Area
Large groups of people can drastically change their opinion, adopt a completely unexpected trend, come out to protest on a square, adopt a certain ideology, have an amazing time at a party, or start using a certain product on mass scale. While all these social phenomena are diverse, one thing in common is that they involve information dissemination that happens in a synchronized way, evoking a certain response from the population at once. In this lecture I will demonstrate how epidemic theories from network science can be used to study information contagion and trend/rumor propagation (so-called information cascades). We will use real examples from Facebook and Twitter (Russian protest movements and UK riots), as well as Gephi software to visualise the sample data. We will learn how successful campaigns (both in marketing, politics, as well as the social sphere) manage to become viral and to provoke a collective action on the side of participants. We will also see how trends, rumours and ideologies are generated and proliferated through social networks. We will also find out how information becomes viral and what one can do in order to increase the message's contagious potential. The session will be held by Dmitry Paranyushkin from Berlin-based Nodus Labs, a research organization specialized in using network analysis and complexity science to enhance our understanding of cognitive and social processes.
Trail Context awareness Group action Process (computing) Presentation of a group Multiplication sign Shared memory Control flow Field (computer science) Frame problem Cognition Message passing Software Meeting/Interview Internetworking Hypermedia Order (biology) Self-organization Software framework
Group action Mathematics Graph (mathematics) Software Computer network Order (biology) Interactive television Vertex (graph theory) Mathematical analysis Software framework Connected space
Revision control Observational study Information Scientific modelling Multiplication sign Computer network 1 (number) Dynamical system
Point (geometry) Group action Multiplication sign Scientific modelling Dynamical system Thresholding (image processing) Videoconferencing Energy level Data conversion Data structure Subtraction Physical system Graph (mathematics) Process (computing) Information Artificial neural network Interactive television Mathematical analysis Cognition Computer animation Software Personal digital assistant Computer network Vertex (graph theory) Differential equation Pressure Family Data type
Standard deviation Group action Artificial neural network Twitter Connected space Number Random graph Degree (graph theory) Facebook Computer animation Software Internetworking Computer network Universe (mathematics) Vertex (graph theory) Website Data structure Data type Family
Point (geometry) Group action Existence Randomization Multiplication sign Connectivity (graph theory) Twitter Impulse response Goodness of fit Data structure Subtraction Scaling (geometry) Information Channel capacity Keyboard shortcut Oscillation Extreme programming Oscillation Random graph Message passing Computer animation Software Personal digital assistant Computer network Order (biology) Vertex (graph theory) Normal (geometry) output Freeware Family
Web page Group action Channel capacity Information Connectivity (graph theory) Connected space Facebook Broadcasting (networking) Message passing Computer animation Order (biology) Computer network Vertex (graph theory) output
Computer virus Group action Multiplication Information Observational study Channel capacity Multiplication sign Robot Basis (linear algebra) Shared memory Extreme programming Message passing Computer animation Strategy game Personal digital assistant Telecommunication Computer network Core dump Order (biology) Mathematical singularity Right angle Service-oriented architecture Subtraction God
Message passing Group action Mapping Informationstheorie
Facebook Group action Message passing Mathematics Computer animation Observational study Computer network Vertex (graph theory) Right angle Data structure Subtraction Connected space
State observer Frequency Group action Mathematics Multiplication sign Computer network Mereology Distance
Point (geometry) Computer virus Group action Presentation of a group Multiplication sign View (database) 1 (number) Mereology Field (computer science) Number Facebook Mathematics Strategy game Circle Data structure Subtraction Email Process (computing) Scaling (geometry) Information Interactive television System call Message passing Computer animation Blog Telecommunication Order (biology) Computer network Service-oriented architecture
how you each show W and how hello my name is Dmitry and and
so here it is said that I work another slabs which is a research organization that studies how the frame of the framework of networks can be used to understand cognition and social processes and 1st of all an allergist say what I'm going to dull just make the like have an hour presentation of some of the main ideas in the network analysis that can be used to understand protease movements and to because there is not so much time if you have any questions we there can have them after the session on the outside the uh during the breaks and so on so I'm very open to speak about it after and the 2nd thing is that I just want to addressed in the context where the stock is actually happening it's the revolt track and do most of the discussions of in this track are focusing on protease movements and somehow we all share a similar understanding of what the protest movement is it's when a group of people uh shares a certain idea becomes fascinated with strongly enough that the unified for a collective action in order to spread this idea to the rest of the people who might not always be uh sharing the same values and we've seen that happening during the last years in the occupy Wall Street movement and In the process in Africa Russia and so on and sometimes worked well sometimes it didn't we also know that social media plays an important role in this process we also know that the government's they started to also do some kind of crackdown on the freedom of the Internet in order to stop his movement and sometimes even use the same techniques in order to counteract protest movements so we kind of know what's what's happening in the field generally but what I want to speak about today and it's something that I feel is often missing from this discourse of the technicalities of how protest movements actually work and function how does it happen that people can get into a certain idea at the same time how does unified for collective action can we ensure that the message
spreads through network in the most efficient way to a large group of people how can we structure our interactions in the ways that would enable the network of be robust against external influence and at the same time open and adaptable enough so that it could be responsive to whatever challenges that encounters in the outside world and all these questions are very complex because we're dealing with a lot of individuals that interact and the framework of network analysis can be very useful to think about this phenomenon so come in network analysis that but people are represented as nodes and their interactions the connections between them once we start seeing communities as networks we can use various tools from uh graph analysis in mathematics in order to see how this communities come together proliferate information and fall apart and the 1 of
the basic models that's used when we talk about information contagion and the dissemination of information networks are of the epidemic models which are usually used to study diseases and I'll just go through this quickly because we don't have so much time and 1 of the basic ones is that sigh are so that means that the knowledge can be in several different stages so it can be susceptible S uh I infected and are recovered by or the more of a whole complex version of the same model is susceptible infected recovered susceptible again back to infected so this is very basic epidemic models can then be applied to any dynamics that involves people getting into a certain ideas so for example they can be susceptible to something let's say like buying pneumonitis sneakers something like this then become infected with the AIDS so they buy them
than the fashion wears off they recover then the new fashion comes again they go back to being infected with a certain idea so this is a basic
epidemic model how it works and then 1 other important concept here are information cascades it's hard like behavior influenced by the others when the conversion threshold is exceeded so let's say that the node is you and this is all the friends 0 family that you have a certain time passes by and a small proportion of your surrounding becomes interested in a certain idea let's say they think that the government system in Germany stocks and they want to change it so they will tell you that the case something has to be done but because it's only let's say 10 per cent of the people that surround you it's not enough for you to change their opinion so some time passes even more and you have already 70 per cent of your surroundings interested in the same idea or or maybe believing in a certain thing or adopting a certain chance to become much more susceptible to accepting and also because of the social pressure because you hear it all the time and so on so at some point you also become infected with a certain idea that the surrounding you so this is a basic model of how informational cascades work and so on the global level this is how the contagion occurs and enough I can play this video here no but we have different groups inside the networks that are susceptible to a certain piece of information they become infected then they spread information further to other groups some of the group's recovered during the process and the this kind of dynamics happens globally but when we model interaction processes in this way then we can use the tools from graph analysis also differential equations to see how exactly does it happen that information spreads through networks what network structures are the most conducive to to the spread information um what network
structures that are better to make sure that information spreads simultaneously and in a synchronized way and so on so I'll just go quickly through the different types of networks that
exist so 1 of the most common 1 is scale-free network it's the when that most of the nodes inside the network have very few connections and few with significant number of nodes of very well connected so that the number of connections among the knowledge is distributed unequally they have a very Vriend degree of the influence that they have in the network Internet is structured for example like this you have websites like Twitter and Google and Facebook where most of the people go and most of the websites on the internet and maybe only visited by the people who actually created them so this is scale-free network structure then we have another type of network random network it's where most of the
nodes have more or less the same number of connections and and there's a few deviations of the very few nodes are very well connected and very few notes have very few connections but most of the knowledge they share more or less the same number of connections this is a situation that you normally encounter at the workplace or in university where everyone knows each other so everyone has more or less the same number of connections in the group and then we have also small world networks which are very typical for whatever social structures were involved in of our friends family and so on so this
means that we still have a very well connected network almost as well connected as the random network but the community structure inside the network is quite prominent so we have different groups of nodes which are more densely connected together than with the rest of the network so we have all these different communities that interact together globally as well this is a very typical structure for good normal social networks that we have your friends maybe family it's
all the different groups that have weak ties together and form different communities so when we talk about uh and dissemination of information in networks the structure of the network is very important because if you different independent researchers both in epidemics and viral marketing they found out that for example if we talk about random network where most of the people know each other and to a certain disease or information enters this network it's much more susceptible to high amplitude oscillations so the lifetime off infection is shortly but at the same time such a random networks are much more likely to synchronize globally so basically if you have a group of people where everyone knows each other and everyone interacts all the time and you communicate a certain message to a large enough proportion that informational cascades occur then that uh you will ensure that there is a certain chance happening in the network that a lot of people become interested in the same idea at the same time however it will be quite short-lived so it will be a short-lived trend that disappears after a while on the other extreme we have also networks that are more like scale-free with a prominent community structure such networks are very good and proliferating information for longer distances ensure that almost everyone in the network receives the message but if they're much less likely to synchronize so the global cascades are much less likely to occur and then we have something in the middle which is mostly the case of scale free networks where there's different prominent communities that there also connected together through random shortcuts such networks can synchronized easily and at the same time they can also proliferate information globally very efficiently so when we talk about disseminating information networks the introduction of random shortcuts in the community structure that's quite distinct is very important and this enables the network to maintain uh the capacity to synchronize globally but at the same time to also have enough resistance against this kind of oscillatory movement where it can change opinion very quickly or where a certain piece of information or a trend in rural so introducing random shortcuts in the community structures is very important for each disseminating information networks then we have also another very important point it's the existence of the giant component so most of the nodes in the network should belong to the same interconnected structure a very good examples for example right now here I don't know how many people know each other but after we finish this stock if you all knew each other and you're interacting with each other it would be much easier to continue this discussion in this group however because everyone is disconnected it makes it very difficult for this network that we also attempt temporarily formed here to actually generate some information on its own it always needs external input and impulse in order to kind of get involved in certain things so when you
communicate information to large groups of people it's important to make sure 1st that they are interconnected so that they have the actual capacity to interact together and then it will make
it much easier for you to communicate a certain message to this group of people and what happens on facebook quite often with pages that are established for brands of all 4 broadcast movements is that lots of people that belong to the page of all that belonged to the group are actually not connected to each other so they don't have the friends belonging to the same group that means that you communicate a certain message to them but they don't spread it on their own so they always need external input in order to continue acting within the constellation so forming giant component and connections between the nodes is very important for information to spread through then also 1 of the important thing is when we communicate to a large group of people it's important to start with a group of interconnected nodes and then this group will spread information to the rest of the network through the connections that it has to other groups and at the
same time this also tells us what the possible strategy of resistance could be if we have a network that we don't want to be influenced to easily so in this case and studies in that the tumor immunology they found that the least efficient immunization strategy was to immunize a certain proportion of groups in the population so for example the households and the most efficient immunization strategy was to immunize a certain proportion within each group so that means that if we want to be able to resist uh being influenced easily the network that we're operating it should be comprised of groups that don't share the same opinion so there's maybe some confrontation even within the groups that means they cannot easily synchronized together and that means they wouldn't also spread information globally so these are the 2 extremes that we can have on 1 side groups and that share a certain consensus and that makes it easier to communicate certain piece of information to them that they spread further in network and on another side as a strategy of resistance to have groups that slightly disagree so it makes it harder for them to synchronize globally and so basically what I want to emphasize here is that if we're talking about a group of people which is able to unify for a collective action and share the same idea certain things it's important that every member of this group acts on the basis of policy singularity where they belong simultaneously to different communities that are distinct from each other that don't relate together but that enables them to have a multiplicity of different opinions and to choose at each time which group they belong to and at the same time be open to any other information from other groups and another important idea when they spread information is to focus on brokers on the people that connect different groups together because then they will disseminate information through the rest of the network and of course the message itself it should be structured like a virus should have the capacity to replicate
itself and there was a very interesting study done of London rise by the gods where they found that the rumors that started with the question were much longer lead them the rumors that started with a certain assertions so for example if you want to spread the rumor it's much easier if you say uh not just like kind of American disappeared but you're right is it true that build robots that are American disappeared so you start with the question and there's a lot of layers of truth that have to be actually unfolded in order to get to the core of the message and to confirm it or not so Starting with the question is always good to make sure that information goes
viral but it's also important to recontextualize so
to know that kind of their mental map of the group that you addressed so you can propose something that relates to the information theory to have but that's still also carries some kind of novelty in this message
and of course it's important that the purpose of the group is related to the message that you communicate so it should redirect the purpose that brings people together if you try to communicate to the group that's against putting that the animal rights are important it's not going to work so well but if you say that they should be leaving the fact that animal rights important because it will have to to put it in down maybe it will work better so this should be always connection to the purpose of the group together and that 1 of the important study found that once the 10 per cent of
people is committed enough to certain idea they can spread easily through the rest of the network and then I'll just go quickly uh in a couple of minutes and then I will be done through the Russian protease movement we analyse different facebook groups that were established to actually be able to lose protease that happened in December and January in Moscow and 1 of the groups was putting must leave we found that we analyze it actually once in december and once in January 2012 after a big demonstration that happened in Moscow and we found that while the group was able to attract new members of the community structure inside the group didn't change and also the communities were not very well connected together and more than half of the nodes didn't have any friends that belong to the same network so it made it very hard for this
network to actually introducing a change because it was not connected enough and that it was not old enough to external challenges that it faced and also when we analyze the which people actually belong to each group we found that it was a highly politicized group that
consistent of people who worked with the opposition leaders there is the group here where it's a Georgian activists who are against putting because of that war that happened before so so so it was a highly politicized group with a lot of ideology insight on the other hand another group that fought in a much more efficient way and that was much more successful in implementing change was the volunteer and activist group that was just gathering people to be observers of the next presidential elections over the same period of time this group managed to attract 50 % more members so it's rose almost twice and then the members of the already part of the network they actually connected together much more but it had much less notes that didn't have a new other friends in the network so only 1 3rd of the people did know anyone else but two-thirds were actually connected to each other in the network and this group was much more interconnected it had lower average path and the distance you have to travel in between the
nodes and that the community is actually what was most interesting they change the structure so the Hobbes which were important they lost their importance and they gave way to the new newcomers from the periphery who were able to actually also influence the kind of processes that happen within the group and when we analyzed who these are the members of this network we found that it was mainly journalists and bloggers who were not politically active before soldiers group that was formed for a very practical purpose was much more efficient in evolving in something that could be used a full and that actually work much better now that we look in retrospective political change didn't really happen but what happened is that people they for the 1st time in Russia got a taste for extra actively participating in very practical important prompted processes of election of observing election and of being active in this field so successful groups are the ones where people talk to each other with the peripheries integrated where Hobbes are leaving the center to form new groups when a small number of orphans of people who are not connected exists and when the unifying slogan is not something the political and general and abstract but more more a call to action than people are constantly reminded about the purpose that brings them together so just to go through quickly the the main points that are made in this presentation it's important for information dissemination that there is a prominent community structure that has random shortcuts in between the communities it's important to focus on densely connected homogenous group in order to spread information and information brokers are important because they connect different communities together so they can be addressed in order to proliferate information inside the network and message should be structured as a virus in order to be able to replicate itself of course I looked into so all these details from the point of view of how to make it easy for information to spread through the we can use exactly the same strategies to also be able to resist any external influence that might be encountered through social networks and for example the way how Facebook makes people
interact locks people into informational bubbles where we're surrounded by quite homogenous community actually makes it much easier for external agents to influence people and to influence the behavior so it's always important to resist I think this kind of drive that exists in society right now to filter out your circle of communication through 1 group that's always still be able to go into different groups and be part of different communities maybe even have different lives and that would enable you to have an overview of what's happening on a global scale while also being able to be active in a certain field as long as you feel is important so these are my contact details if you want to speak about it after you very welcome to send me an e-mail and we can also speak after or if you have some questions right now I think I have maybe a few minutes you're you're welcome to to ask them you you you you you you you you on the church


  486 ms - page object


AV-Portal 3.9.2 (c7d7a940c57b22d0bc6d7f70d6f13fde2ef2d4b8)