We’ve had yet another interesting and eye-opening week of posts on network structure and dynamics. One prevalent theme was that of cascades through a network. In one example, gblogger explained the link between cascades and political primaries, specifically the currently ongoing Democratic race for the Presidential nomination. The author explained how certain demographics of voters tend to be more influenced by external signals than others, that the staunchly partisan voters within the Democratic Party are not easily swayed. In further blog posts, other authors dived into the way the media can be influenced by cascades, and how this can subsequently induce or intensify cascades within the public.

Another interesting example of cascades that a few authors discussed were emotional rather than information-based. In his post, the engine driver, went into the biology behind emotional cascades. As it turns out, we as humans are biologically hard-wired to feel empathy for others through what are known as “mirror neurons” that trigger the emotions within ourselves that observe of those around us. In another post on a similar topic, Guitar Guy, discusses laughter cascades. Recent research in the field of emotion has yielded results that confirm the conventional wisdom that laughter is in fact contagious.

Lastly, one poster, llenroc, examined Metcalfe’s Law, in terms of its original context, Ethernet. Today, Metcalfe’s Law extends far beyond this one example and is commonly cited in the study of network dynamics.

By an overwhelming margin, this past week’s posts have been about the potential irrationality of information cascades.  Moviegoers, home shoppers, hush puppies, flash mobs, stamp collectors, and wall street (see a great post by ehp1459, and a post about the Indian capital markets by randomvoice) all have a bit of blind-following-the-blind in them.  Posters were exceptionally keen to point out that the collapse of Bear Stearns may have followed some herding behavior: the value of the stock dropped to $2 after bad news and fear over investment banks, as posted by sl1201.  Lehman followed similar behavior even when management assured the company was on sound footing.  Outside of mortgage backed securities, eep writes about how Beanie Babies may have experienced a bubble of their own.  Let’s not forget pokemon trading cards too!  Thus, cascades potentially causing products to be mispriced.  On a further extreme, ninjaspleen looks at Asian Art — which may have decreased in quality as prices went up.  As prices went up, artists may have lowered creativity to output more and fullfill demand.  I’d like to see more work into how information cascades in a product market affect the underlying value of the product.

Although we’d like to be species-centric, Amcoops245 reminds us that not only humans follow cascading behavior.  Army ants can apparently follow each other until they form what’s called a “circular mill”: “Some mills can reach up to a 1,200 foot circumference (with a lap time of 2 1/2 hours) and will continue until the ants collapse and die or a group accidentally stray from the circle.”  Let’s hope humans don’t learn too much from those particular ants!  On a hopefully unrelated note, jeb369 points out that political elections exhibit cascade behavior when executed sequentially.  This type of behavior occurs less when elections are executed in a blind fashion, and each individual uses only their own information to judge.  Thus the wisdom of crowds loses legitimacy when the crowd follows the minority; any witness of proper hooliganism can attest to that.  Just like the ants, humans follow humans, and according to shoes14’s post (on a paper by Bikhchandani, Hirshleifer and Welch), humans don’t usually realize they’re part of an information cascade even when it occurs.

A few brave souls defend information cascades as a force of good.  sls14 writes about cell phones and positive externalities.  In a situation where it’s better to have all users choose one cell phone carrier than many carriers, it’s good for everybody to follow an information cascade.  For example, myusername asserts that information cascades were critical to the success of Blu-Ray over HD DVD.  However, even in a world with positive externalities, herding can lead to noncompetitive behavior.  Not long ago Microsoft Windows and Internet Explorer made up the majority of the OS & Browser market.  Many would contribute the success of these two pieces of software to the simple fact that they had a big base to start off with.  Sometimes information cascades can also be used to propogate potentially important information.  After all, much of the news that we get about “hot issues” come through the grapevine — through friends and family.  Matt fingers the topic of Global Warming, but surely many issues of the time have benefited in visibility through cascade effects.  However, sometimes cascades can take a long time to truly propagate information.  knight2612 brings up the case of LASIK eye surgery.  According to knight2612, Negative side effects only started cropping up a decade after the initial batch of individuals were operated on.  Thus for the first 10 years, there was a positive cascade based off no side effects, and the negative cascade settled in after that.

Outside of Information Cascades, jian2587 posts an interesting post about the applications of PageRank within a social context.  How can we use similar graph algorithms to learn about social settings?  bes36 writes about Visualizing Wikipedia and other large graphs — a good post for those interested in seeing some implementations of graph visualization.  Both irishlass and sammy make informative posts about auctions.  Finally, kudos to lepidoptera for posting xkcd comics.

This week, many postings covered the recent class topics of information cascades and web search/keyword-based advertising. In fact, one could almost say that the class blog experienced an information cascade, as specific examples within each topic were repeated and “hot” this week.

Some of the popular subjects…

1. Search engines and the use of link analysis: We have learned about the basics techniques behind ranking websites, which are constantly being updated and changed. Many posts this week talked about developments in Search Engine Optimization.

solee426 wrote about Yahoo’s new task-based approach. Pomegrenade discusses Twitter, a new site that combines Q&A with social networking. Milktea introduced Searchme.com, a new and interesting visual search engine. elashish14 tells us about a study that shows that the position of a link on the Google search result page has more influence over college students than the abstract that is written with each link. whosnitsuj and asr96 talk about ways of cheating the Google PageRank system, questioning if some relevant sites should be allowed to break the guidelines and how Google should deal with “Google bombs”, respectively. alisa shows how PageRank technology can go beyond web searching and is being applied to prioritizing genes to analyze in cells. Interesting!

2. Online ads: Techniques for pricing and distribution of these are also constantly evolving as the internet and the market evolve.

uniqueNewYork talks about demographic bidding and leprechaun speculates on what a Do Not Track bill would mean for the online ad industry. apb37 tells us about Google’s new Pay Per Action product that attempts to counter click fraud while cw125 talks about how Google is dealing with click-through rate trouble. melodiadorata introduces interactive advertising that we might expect to see on Facebook soon.

3. gsr8, jgr29, Guitar Guy and 3dza0micron give different perspectives on the issue brought up in John Tierny’s article in the New York Times. The article talked about diet researcher Ancel Key’s idea that a fatty diet is related to heart disease and how this was a prime example of an information cascade that was later disproved.

4. bpnet, Yates, Nodes of Steel and Hy95 talk about how information cascades can be blamed for the housing bubble, and each get into different details, opinions, and related issues.

In the case of these blog postings, an information cascade (of sorts) of popular subjects was a positive one; it allowed deeper analysis and the thorough coverage of some of the recent major topics of concern. There were some unique ideas brought up this week as well.

zqjl introduces to us the recently-proved (and quite interesting) road coloring theorem.

Gbblogger writes about how various auction formats are used on online travel websites.

ngf2 talks about a study that showed that subconscious brain activity representing enjoyment increases when a person (even a wine expert) drinks a wine that is more expensive.

This past week we saw a number of interesting applications of network theory on graphs ranging from social to business to information networks. Two especially insightful posts discussed voter networks in the United States and the importance of social networks in the music industry. In his article on voter networks, pastafarian1859 examined the different phenomenon of “blue states” and “red states” in the US over the last several election cycles. Another compelling perspective on this issue would be that of Schelling’s segregation model in which nodes voluntarily segregate themselves in order not to be a minority. Could it be that Democrats and Republicans voluntarily segregate themselves into certain parts of the country? How might this play into the urban and rural dynamics that the author discussed?

In a very interesting article on the potential Yahoo-Microsoft merger, elashish14 explored the situation from the perspective of an unstable three-node network. Since Yahoo, Microsoft, and Google all compete in the online search market, it models easily to one of the simple examples in which a triad can be unstable: three negative edges joining each node. Since we know that this situation cannot last due to its structural imbalance, one could have used the underlying economics to predict Microsoft’s bid. In a purely logical world, when this sort of structure arises, it must come to pass that two of the nodes will ultimately team up so that we have two clusters of all positive edges connected to each other by all negative edges, a situation very close to what we see unraveling.

Another intriguing post, this time about an information network, looked at how YouTube became blocked in much of the world due to the single action of a Pakistani ISP. The author, mdc53, considered the incident a failure of the network and raised a number of compelling questions about network vulnerability and security.  “If a country as relatively small and undeveloped as Pakistan could very easily accidentally disrupt YouTube’s traffic and block the site for users worldwide, then what sorts of cyber-chaos and confusion could potentially be unleashed by a malevolent country or group? Furthermore, what’s to prevent something like this from happening again, and do certain internet protocols need to be revised or reworked?”  These are important questions that are probably being examined in the light of the recent incident.

Several posts this week about game theory as applied to various circumstances. There were some very creative applications, ranging from loolooloocy’s analysis of the dating games in TV show HOUSE, to several political issues: the Israel and Hezbollah conflict, democratization in Russia, presidential elections, and the Cold War.

Many of these posts give examples of situations of reaching a Nash Equilibrium that doesn’t maximize social welfare. In fact many real-life situations mirror the Prisoner’s Dilemma, where different sides may have a more favorable option to choose given all possibilities from the other side, but where this N.E. leads to a non socially-optimal result. How unfortunate that this is the way that so many issues pan out in the world today. If only game theory experts who are interested in maximizing social welfare ruled the world, eh?

Irishlass talks about how the Prisoner’s Dilemma applies to the law as well, where a pro-plaintiff product liability rule is the best choice for two states, but where this reduces social welfare.

Also touched upon this week: the Blu-ray/HD battle. Mltstar4 gets ahead in the lecture material to talk about how the popularity of Blu-ray is attributable to network effects, such as those covered in Tipping Point. Jad346 goes on to show some game theory scenarios between Toshiba/HD and Sony/Blu-ray.

Keep up the good work, everyone! Definitely read posts from the other students because they might spark an interest that could help you with your own postings, or just help you learn something new or think about it in a new way.

There are several interesting topics covered in recent posts on the course blog. For starters, it’s looking like we can find examples pretty much everywhere of networks that behave according to the rules we learn in class. Students have found examples in everything from investing within one’s network in the financial industry to Republican party factions and presidential candidates.

I think this class is exciting primarily because it brings up so many interesting subjects. A few topics from the postings inspired me to look for further related readings. Students can feel free to use these for future blog post ideas.

On the matter of cell phone networks and how analog cell phone service is disappearing, read this article from BusinessWeek, The Dawn of Green Mobile Networks. According to the article, the number of cell phone subscribers in 2015 is expected to be 5 billion, up from the 3 billion users today. This will require new infrastructure in less developed, poorer areas. Using renewable sources of energy can allow a successful (and more sustainable) expansion of the mobile network.

One post talks about the Silicon Valley network and how and why it works. This article discusses the uniqueness of technology clusters within the global network of technology industries, and Seattle’s potential to be a new cluster.

A few students discussed mergers between existing online networks (MySpace/BBC and MonsterTrak/CSO). More about connecting over the web with a post about Friend-Of-A-Friend and one about other professional networking.

Check out this New York Times Freakonomics column about the affects of MySpace and social networking on society, with comments from quite knowledgeable people with interesting comments.

Finally, in the spirit of Valentines Day perhaps, a posting on monogamy compared to the Prisoner’s Dilemma. Check out this article from the Guardian to read more about fidelity versus infidelity as related to developing social intelligence and natural selection. Best quote? “animals that engage in monogamous pair-bonding… tend to have larger brains than promiscuous animals.” So if you’re smart, be faithful.

In the past week there have been several additional interesting posts to the class blog. At this point the number of new posts is small enough that we can briefly comment on each of them. There were two posts on game theory. amcoops245 discussed the need to use mixed strategies in playing poker. Poker is a more complex game than those we have analyzed in class, but the ideas of game theory can be applied to it. A good poker player does not play a pure strategy in choosing how to bet as this would eventually allow the other players to infer too much about the quality of the player’s hand from bets. Instead, good players bluff occasionally. klevine discusses how to interpret climate change negotiations as play of a repeated prisoner’s dilemma. Again this is a complex game. If the game had a definite ending time, then sustaining cooperative behavior would be difficult, or impossible, just as it is in the one-shot prisoner’s dilemma. But as the end of the negotiation is not specified in advance, cooperation is possible.

Several other posts discuss the use of network structure in various settings. jamesww noted how referrals in the real estate industry form triadic closure between the friend of client and the client’s realestate agent. sqi2 noted how medical researchers are using properties of networks of interactions between biochemical agents to better understand how to target drugs. Finally, pomegrenade used a network analysis to discuss protests against the Church of Scientology and church’s apparent understanding of influence in social networks.

As a way to summarize some of the activity on the Networks class blog, the course staff will be posting to a “digest blog” that will run in parallel with the main class blog. Once the blog schedule gets up to full speed next week, there will be more content than we can possibly hope to summarize, but the idea in this digest blog will be to point out a few of the posts and link to further reading on some of them.

To get this underway, here are a few of the ideas that have been discussed on the class blog in the past couple of weeks. A number of posts have concerned the ways in which information, ideas, and behaviors can spread through social networks; this will be a large topic in the second half of the class. In particular, pmd8 talks about a widely-read study from this past summer on how obesity is “socially contagious” — although you don’t “catch” obesity from your friends in an epidemiological sense, the probability that you become obese is significantly higher if your neighbors in the social network are obese. jld263, in the first post of the semester, talks about the Genocide Intervention Network, which aims to spread information and awareness about situations leading to possible genocide, and to facilitate fund-raising in support of the prevention of genocide. And ninjaspleen talks about recent work of Duncan Watts that calls into question whether people who are typically deemed “highly influential” in a social network actually have the impact that is usually attributed to them.

A number of posts give nice applications of some of the themes that we’ve discussed in the past few weeks. notamoose considers how structural balance can be used to interpret some of the rivalries and alliances in the Democratic and Republican Presidential primaries that have emerged over the past month. irishlass discusses current developments in kidney exchange, which is an emerging application of matching markets for a medically important goal. dagreg discusses how triadic closure and structural balance naturally operate in the world of massively multi-player on-line games; as we’ve discussed at various points in class and in the book, on-line domains such as these provide us with fascinating sources of data to study such effects at a quantitative level.

Several recent posts discuss the impact of network effects. sithswine182 discusses how Zune, Microsoft’s new mp3 player, relies on network effects and why its strategy has not been successful. timo writes about Dell’s endorsement of Ubuntu Linux and what kind of implications this might have for future operating systems that wish to leverage network effects.

rorschachex also discusses Dell’s decision to support Ubuntu Linux, but from the perspective of voting and information cascades. Dell’s decision was a direct result of its new IdeaStorm site, which allows the public to vote on new product design concepts.

Several posts discuss the internet’s role in the small-world phenomenon. Both starvinmarvin and md6 argue that the internet helps create random edges (i.e., those defined in the Watts-Strogatz model), and thus is responsible for the small diameter of the global social network. someone cites a study by Duncan Watts and colleagues at Columbia University that confirms the six degrees of separation phenomenon using an email experiment.

On the other hand, ocj2 discusses a paper written by Judith Kleinfeld which discredits Stanley Milgram’s findings in support of the small-world phenomenon. Kleinfeld argues that although mathematicians have developed computer models of how the small world phenomenon could logically work, empirical evidence suggests that we actually live in a world deeply divided by social barriers such as race and class. She claims that the popular belief of the small-world phenomenon is mostly due to psychological reasons.

portmanteau, jmholloway and frozenatcornell all discuss Digg’s DVD-decoder fiasco, which is an incident covered by our class lecture on intellectual property rights. Digg deleted several posts which linked to articles revealing a code which can allegedly be used to circumvent the digital rights management on HD-DVDs and Blu-ray Discs. In response, angry Digg users flooded the site with references to the code and criticized the site’s decision to censor content. Digg shut down at one point, and then returned online with a message from Kevin Rose, Digg’s founder. In the message, he declared that he would not delete any more articles that contain the code. The Digital Rights Management (DRM) group has threatened the bloggers with legal action.

A number of posts this past week discussed various positive and negative aspects of information cascades. On one hand, companies can use social networks to generate additional social valueforf their products (e.g., FireFox, Nintendo, MySpace). If successful, such strategies can also serve as an efficient advertisement tool.

As mentioned in class, efforts in computer science have been made to measure diffusion through networks. In particular, progress in this area can lead to new insights on how one might best select a group of early adopters to maximize coverage of a social network. Both mrjeets and cuecon204 discuss a recent paper by researchers at IBM which study information diffusion through Blogspace. The model presented in this paper is actually derived from a diffusion model proposed in part by Professor Kleinberg in 2003.

However, information cascades can also generate negative effects on society. Some examples include perpetration of false information. One recent post by dijkstra discussed the recent news about Marilee Jones after her academic credentials were finally revealed to be false after nearly 29 years of being accepted at face value. Other potentially more harmful effects include how political and corporate decisions might be adversely affected by information cascades.