Facebook in Decline

Facebook has seen consistent growth over the past few years, but it has recently seen this growth peak. The number of users is now declining, down nine million monthly users in the United States. The number of minutes that Americans spend on Facebook has also declined; however, growth in India and Brazil has increased. It is possible that growth has stopped in America due to Facebook reaching its peak and that other countries are still catching up.

Another explanation offered is the adoption of smart phones in Brazil and India which in turn has led to adoption of Facebook as a mobile app. This pattern is similar to the idea of information cascades and adoption of technology in networks: as more people buy smart phones and begin using Facebook, it becomes more beneficial for others to do so as well, especially if the initial users are friends of later adopters. Adopting Facebook on new smart phones has a higher benefit as more friends join and can derive value from being able to interact with each other on the site.


FBI Soon to Be Watching Your Online Communications the Moment they Happen

A government task force is preparing legislation that would pressure companies such as Face­book and Google to enable law enforcement officials to intercept online communications as they occur, according to current and former U.S. officials familiar with the effort.

The FBI believes it does not have the capabilities it requires to tap and trace internet communications of potential terrorists, and is proposing to fine companies that fail to comply with wiretap orders. They believe that suspect’s activities can be missed as critical evidence.

The draft of the proposal states that a court can levy a series of escalating fines in firms that fail to comply with wiretap orders. Companies that do not comply with an order within a certain period would face an automatic judicial inquiry. After 90 days, fines that remain unpaid could lead to fines.

The proposal would not dictate how the wiretapping capability must built and leaves the development to the companies. Small companies are exempt from fine.

As I read this article and many other on CISPA and the previous SOPA bill have me curious as to the capabilities of the federal government when they have control of all the information pertaining to the nodes of the network and the direction in which the edges between them travel. The private companies have the capability to do that, but only for their services. What happens when both graphs overlap and a more detailed and defined graph is formed? The information that could be learned from this overlapping of graphs would critical to understanding the human network.

Yours Truly,



Finding the Easy Way to Do Hard Things

This past week I came across a research paper by Nicholas Christakis and James Fowler in which they discuss their testing of a means for detecting disease outbreaks early in their development.  They start from the idea that individuals who are more central to the network are more like likely to be close (as measured in terms of the number of edges) to the individual who introduces a disease into the network, therefore will catch the disease earlier.

Here’s the super-clever part.  Looking at previous research on and attempts at heading off outbreaks, they concluded that actually mapping a social network is too slow and onerous a task to be useful.  Epidemics can move quickly, and large social networks can be very difficult to map.  Instead, they pursued a much easier process for determining who in a network is more central.  Taking advantage of the “friendship paradox” (as they describe it, the strange fact that “your friends have more friends than you do”), they took a random sample of Harvard undergraduates and asked them to name a friend.  This second sample, not the original, randomly chosen students but the friends of the randomly chosen students, should theoretically exhibit greater centrality than the random selection, meaning that this group will generally be closer to the source of the disease and therefore catch it sooner.  They tested this using students at Harvard during a flu outbreak and found that they were correct; the epidemic peaked earlier in the second group than in the first.

What I find intriguing and important about this work is that they manage to reveal information about the network structure without actually having to map the network.  As we’ve learned during this past semester, there are many common, stable properties of networks that could be used to make predictions about network behavior.  However, this is often useful only so long as one has a map of the network.  By teasing out an easily discoverable group that exhibits higher than average centrality, the researchers were able to use this group as a “sensor” for coming outbreaks of the flu.  This reminds of me of what Dr. D when we were studying the adoption of new technologies that have network effects; our abstract quantitative understanding of the phenomenon, which is dependent on having information we would never have in the real world, points us in the direction of a qualitative understanding that is easily applied in the real world.  Similarly, Christakis and Fowler’s approach allows for one to make potentially powerful predictions about the spreading of a contagious disease within a network without having to do the difficult-to-impossible work of precisely mapping the network.

(Christakis happens to also have given this related TEDTalk)

Golden Age for Science of Social Networks: Bountiful Applications


Clever social scientists are coming up with new ways to put the burgeoning amounts of data on social networks to work for societies they are supposed to reflect. By employing the help of inquisitive physicists these psychohistorians are analyzing patterns in the network dynamics we’ve discussed in general models, and refining models to study at far more precise levels. A researcher at Northeastern University has managed to devise an algorithm that analyzes a person’s mobile-phone records and then with 93% accuracy predict where that person will be at any point during the day. In another interesting application Dr. Vesipigani along with his team have created a program named GLEAM (Global Epidemic and Mobility Model) which splinters the world into hundreds of thousands of squares. Using these squares as hubs the program models travel patterns between other squares (busy roads, flight paths, etc.,) which demonstrated potential in 2009 when it successfully mimicked an outbreak of a strain of influenza known as H1N1 as it made its way across different countries with commendable accuracy. Within the guts of these models there are apparent network components such as transiency and network flows at work. The idea of using the rich new sources of data from mobile technology to model social network dynamics aspires to achieve such lofty goals as to predict potential tipping points where riots might break out, or model the breakdowns of trust between banks and customers that cause financial crises. I myself am interested to see how well we could use similar social network analysis to predict the business cycles of industry and performances within the market, perhaps measures of a firms social network could be revealing of sudden upticks in its market value.

Any thoughts are very welcome.


The Economist: Dr Seldon, I Pressume

Cancer Epidemic?

We’ve been learning about epidemics this week, focusing on the way diseases cause epidemics through networks. We also learned ways in which these epidemics can be prevented. However, an article I found states that prosperity is what is causing a particular cancer epidemic in Latin America: “Latin America Threatened by Mounting Cancer Epidemic: Study” (source: http://health.yahoo.net/news/s/nm/latin-america-threatened-by-mounting-cancer-epidemic-study).

The article states:

A multinational team of researchers found the current state of cancer care and prevention in Latin America incompatible with the socioeconomic changes taking place in the region, where an increasingly urban populace faces mounting lifestyle-related cancer risks.

 Although cancer is not considered contagious, there are certain confirmed risk factors that will increase one’s chance of being diagnosed with cancer. In this case, prosperity is causing this region to imbibe in many cancer causing behaviors, such as drinking alcohol, smoking, eating until obese, and leading sedentary lives. Although this is extremely similar to the behaviors of Americans (who lead the world in obesity), and Latin Americans are less likely to contract cancer, they are twice as likely to die from it. This is due to their lack of preventative measures, lack of health care, and the patients seeking treatment too late.

The study recommended Latin American nations make major changes to their healthcare policies, such as dedicating more funds to public health, widening healthcare access so cancer patients can be treated earlier and developing better national cancer plans. It also envisions shifting funds away from costly end-stage cancer treatment toward palliative care.

 These are some of the ways that may help curb this ‘epidemic’, although this epidemic does not travel the way most diseases and epidemics do. The study calls for immediate change, otherwise it will be almost too expensive and deadly to deal with within a short span of just 10 to 15 years.

WoWconomics: Epidemics in the World… of Warcraft

Here we are friends, the final installment of your favorite weekly blog post: WoWconomics.

This article is not NEW, however, it is too relevant to ignore…

This week in class we learned about predicting the spread of epidemics through networks, and we have learned about several models that economists and scientists have developed to help model future outbreaks. However, we may need to look no further than the internet to find the best ways to model potential outbreaks, and with the help of the developers of World of Warcraft, Blizzard Entertainment, researchers have been able to see what an epidemic looks like in a real population. A population made out of fictional people.

World of Warcraft, in 2007, had an accidental glitch that mirrored a global epidemic more realistically than any model science had ever predicted, because it actually affected real people. When a virus called Corrupted Blood, which was supposed to be isolated to a separate part of the game, was accidentally introduced to the main population of an online server via an accidental teleportation by a player, it caused rampant destruction across tens of thousands of players in a matter of hours. Scientists have since looked back and attempted to model the viruses path across the server.

World of Warcraft offers a surprisingly close replica to our society in several ways that made this study particularly relevant. First, some players had access to abilities that allowed them to travel incredibly far in a short amount of time (similar to those few nodes in a network given long links) which allowed the disease to spread quickly across long distances. Additionally, characters were able to be re-infected, which some diseases allow for in reality as well. Also, some players acquired the disease and logged off immediately, and when they re-logged on at a later time, they still had the disease, allowing it to have longer gestational period than the game would allow. This mirrors real diseases in that some allow for people to be “carriers” and not succumb to the disease as quickly.

In the end, Blizzard fixed the problem by removing the disease from the game, claiming it was too dangerous for their players, but the real life implications of studying that disease have helped scientists see not only what a disease could do to a population, but also learn that potentially MMORPG games could be a suitable solution for scientific experimentation.


Link: http://www.time.com/time/health/article/0,8599,1655109,00.html


Reverse Epidemic Diffusion

In Wired magazines “Finding the Sources of Epidemics,” Samuel Arbesman discusses how networks can be used to find the source of an epidemic using as few as 20 percent of that network’s nodes. According to Mr. Arbesman “they (Pedro Pinto and his team) explore how the leader of a terrorist organization can be identified and even show how this methodology could be used in finding contamination sources in a subway system. They found that they could determine the source of a contamination to within a single subway stop by monitoring the behavior fewer than 20 percent of the stations.”

In this case, reverse diffusion was used to find the source of a cholera outbreak in S. Africa after the fact. However, models are one thing and their applications are quite another. In all of these examples except one, reverse diffusion is useful for finding a static hub. But no evidence backs the claim that the leader of a terrorist organization could be caught by working backwards with statistics. For one thing, the leader might move, making he or she harder to pinpoint than the proverbial hot zone subway stop this author mentions. Secondly, the leader might operate through proxy. Finally, reverse diffusion requires that 20 percent of the nodes in the network be analyzed. The last requirement is the most debilitating, because with sick people, the researcher can deal in absolutes – this person is ill and this one definitely is not. But how could this method be applied to something as complex as terrorism, where the researcher can’t be sure if their nodes are the right nodes? In terrorism, the researcher has to diagnose degrees of ideological infection, which is a far more abstract concept than a viral epidemic. http://www.wired.com/wiredscience/2012/08/finding-the-sources-of-epidemics/