In the March 4th, 2008 issue of the Cornell Daily Sun, columnist Ben Birnbaum wrote an article entitled The Audacity of Hype. In the article, he discusses the trend, especially of the younger generation, of supporters of Senator Barack Obama for the Democratic nomination to lack any substantial reason to support his nomination. The author even quotes an interview with a Texas senator. When asked why the senator supported Barack Obama, he could not give a valid reason. So why do so many people support a candidate if they have not done the research to find out if he is, in fact, the right person for the job?

To what this all boils down is an information cascade. When you look at the long list of Obama supporters, which include John Kerry, Chris Dodd, as well as Oprah Winfrey, it is hard not to think that they know something about him that you do not. So, like many young voters, you disregard the small amount of information you’ve researched and believe that since so many have supported Obama, they cannot all be wrong. This happens because the individuals who support a candidate see the choices of supporters who have made decisions before them. As an example, say Oprah Winfrey and John Kerry are the first two Barack Obama supporters. If Oprah supports Obama, one can deduce that she must have received a high signal. If Kerry supports Obama, he must also have received a high signal. Then, the information cascade starts, and there is a higher probability that Obama is a good candidate regardless of what signal the third person receives. To be fair, this occurs with any candidate, but the article speaks specifically about Barack Obama.

The Cornell Daily Sun article can be found here: http://cornellsun.com/node/28481

Over Spring Break, you may have noticed that Pizza prices have become higher in some pizza parlor around the country. Are these arbitrary events, or are they the results of interaction of a complex economic network?

Given what we have learned in ECON 204, it is not surprising that the rise of pizza prices are not an arbitrary event. Indeed, the owners of pizza restaurants have experienced a drastic increase in the cost of basic ingredients that they need for everyday operation, such as flour and cheese. These have resulted in large increases in the cost of running a pizzeria in many locations in the United States.

As unrelated as it sounds, activism in sustainability shares part of the blame for the increased pizza prices. Such activism has led to devoting an increased number of acres of farmland to corn, one of the components used to manufacture biofuels. Due to the complex economic network that underlies the nation’s limited farmland, this means that less wheat is produced, and in turn, there is less flour to sell to pizza-makers. To make matters worse, cold and wet weather have destroyed much of the remaining wheat crop this year.

The result might be devastating to the operators of smaller pizza shops: doubling of the cost of flower in a month. Other vital components for pizza have also become much more expensive than in the past: cheese cost have multiplied by two, cooking oil prices have tripled, and motor fuel costs are rising rapidly, all within the last year.

This leaves the owners of such businesses with an interesting question: how to maintain profitability, the loyalty of customers, and the quality of their product in the face of these significant changes. There have been different solutions:

- Some shops have raised prices

- Some shops have cuts costs not related directly to their food products

- Some shops have chosen to decrease portion sizes

It is unclear how many shops will be forced out of business by this situation. What is clear is that the complex interactions in networks cannot be ignored in any business line or field.

References:

http://findarticles.com/p/articles/mi_qn4176/is_20080301/ai_n24373538

http://www.timesunion.com/AspStories/story.asp?storyID=674157

Diet and Fat: A Severe Case of Mistaken Consensus

By John Tierny

New York Times Article: October 9, 2007

In class we have been discussing this notion of information cascades and how easily we can be influence by the choice of others. One interesting figure that was mentioned in the article points to research done by a group of economists. If, say, 60 percent of a group’s members have been given information pointing them to the right answer (while the rest have information pointing to the wrong answer), there is still about a one-in-three chance that the group will cascade to a mistaken consensus. So how does this relate to the medicine field? “Cascades are especially common in medicine as doctors take their cues from others, leading them to over-diagnose some faddish ailments and over-prescribe certain treatments. Unable to keep up with the volume of research, doctors look for guidance from an expert — or at least someone who sounds confident.

What this article describes is how information cascade brought about the idea that low-fat diets will reduce heart disease. The argument is that there has been no scientific proof that the former is true and as a result of this recommended diet, people have been switching to diets rich with carbohydrates, which is believed to obesity and disease. So if this diet hasn’t been scientifically been proven to be effective, how does it have the endorsement of “92 percent of the world’s leading doctors.?”

It can all be explained with the idea of information cascade. It all goes back to Ancel Keys, a prominent diet researcher a half-century ago. He became convinced in the 1950s that Americans were suffering from a new epidemic of heart disease because they were eating more fat than their ancestors. However, upon making this argument, two glaring problems were made apparent. First, according to the best estimate possible, ancient hunter-gatherers ate just as high or higher than the diets we practice today. Second, there wasn’t really a new epidemic of heart disease. Yes, more cases were being reported, but not because people were in worse health. It was mainly because they were living longer and were more likely to see a doctor who diagnosed the symptoms.

In 1957, the American Heart Association declared that dietary fat did not correlate directly with heart disease, however, three years later, the association changed position, largely because Dr. Keys was on the committee issuing the report. This report led to an acceleration of the cascade in the 70s when Senator George McGovern issued a report advising Americans to lower their risk of heart disease by eating less fatty foods. Again, this report wasn’t based on hard scientific data, instead, much of it was due to the cascade effect. This report then led to the creation of the food pyramid in the 1980s. So basically, the food pyramid that we all learned about in elementary school was largely due to this phenomenon of information cascade.

In an effort to help control this cascade effect, a group of scientists at the National Academy of Sciences released a report stating that there still wasn’t enough evidence to recommend a low fat diet to all Americans. However, they were promptly excoriated publicly for denying the danger that had already been proclaimed by many reliable sources.

This is a matter of such enormous social, economic and medical importance that it must be evaluated with our eyes completely open. Thus I would hate to see this issue settled by a cascade. Tests continue to refute this notion of non-fat diets, although it makes sense, its not one hundred percent valid. Even though it’s an interesting topic, it’s even more interesting how information cascades plays a role in even in subjects like this.

http://www.starcitygames.com/php/news/article/12201.html

The preceeding link is an interesting look at information cascading gone wrong. The presented example is about Magic: The Gathering (a popular trading-card game). MTG is popular enough to receive scrutinous examination from a set of hard-core fans. Some of these fans are renowned enough to encourage followers – i.e. many players will play the same decks as the “pros” constructed at tournaments.

As it turns out, deciding what deck to use is a special case of the accept/reject scenarios we’ve gone over in class. A deck is “good” if there is a high probability of your winning with that deck. The article above examines a specific deck whose value was embarrassingly overestimated for a long time. The deck “Ghost Dad” achieved notable success towards the beginning of its popular use. While there were many reasons to suspect this deck was bad (see above article) it got played over and over again in tournaments. The author of the article explains this as an occurrence of an information cascade.

There was something about the deck that got the cascade started, though. The deck has some “surprise value” that is hard to play against if you haven’t seen its strategy before. Because of this surprise value, the deck was acclaimed as good by many players. So other players started second-guessing their own doubts because, as in all information cascades, the seemingly rational thing to do is to give a lot of priority to the crowd’s opinion (since the crowd is composed of more people than an individual is).

Eventually, though, players caught on to this new strategy. The few players who weren’t swayed by the information cascade played different decks. And now that Ghost Dad’s strategy was no longer a surprise, it was very often defeated by these few players. Eventually, players of the game started noticing this and the illusion of intelligent design was removed. The actual process of this discovery is analyzed more in the article.

All in all, this article is a telling real-world example of some theory we’ve gone over in class.

http://www.shmula.com/197/digg-as-a-game

In this article, the author analyzes why and how users artificially inflate their digg scores on articles. Digg, for those not familiar with it, is a site that people join and post articles, pictures or videos that they found interesting and the other members of the digg community vote on whether or not the article was interesting. The main problem they discuss is that there are a few users or user groups that have been continually pushing their articles into the most “digged” spots, they discuss the underlying economics of why a user tries to achieve relevant digg posts, and explores the information flow within a larger network (how a few well connected nodes can really influence the whole rest of the network). This article was very interesting as it showed a large number of amateur economists trying to apply many of the economic concepts they learned to discuss problem of hijacked ratings on digg. They debate on the Game Theory nature of posts, as well as Information Cascades and the power of a small set of well connected nodes in a larger network (ie network bridges).

The article discussed the basics of game theory, such as strategies, payoffs, information, and rationality, and the prisoner’s dilemma outcome which we discussed in class. While the game theory analysis was interesting, what more caught my attention was when they started talking about information cascades. “The Digg system of measuring a story’s importance is based on a semi-random set of people voting for or against a semi-random set of news stories.” So the factors at hand are a couple of networks, one for people and the other for news stories, and people trying to post relevant articles to gain esteem within the digg community. This article discusses without explicitly mentioning it how if a well connected node receives a piece of information, how it can effectively control its dissemination, which is graph theory. Also, as the topic becomes more relevant or gains more attention, there are Information Cascades, where users who usually wouldn’t be interested in a topic pick up articles because if many other people think it is relevant, they might have better information. It is quite interesting to see how complex human interaction is and also how relevant so many of the topics network theory are to being able to describe a seemingly unrelated problem.

PS I was going to post a different article that discussed another topic, but when I saw everyone else had been posting on Information Cascades, I decided to post on it as well. (Can someone say theory in action?)

http://www.cato.org/pubs/regulation/regv26n4/v26n4-2.pdf

Given the discussion of information cascades in class and my interest in politics, I came across an article that highlighted the idea of majoritarianism.  In our democratic government, the idea of “majority rules” runs public policy.  So long as this type of ruling continues to exist, information cascades will continue to play a huge role in state intervention.  This way of running public policy proceeds smoothly so long as the cascades have an objective and is based on sufficient scientific evidence however, information cascades are not always “correct”.  At times, poor public policies arise from the state’s decision acknowledging “false” cascades.   

We cannot assume that the election results, bureaucratic processes, and the political decisions meaningfully voice the public opinion.  Public choice economists would largely agree that public policy often does not meaningfully voice the public opinion, especially in the short run. This is due to the actions of special interest groups.  These groups organize to defend a certain interest while taxpayers and consumers are too scattered and unorganized to fight every government intervention.   The small interest groups then exploit these individuals and continue to spread their opinion through an opinion cascade.  Some examples from the past and present are the actions of soft-drink manufacturers financially backing up the drive for Prohibition and nicotine patch manufacturers that have had influence in the anti-smoking cascade.  According to public choice economist, Kuran, “politics consist of controlling society’s choice through manipulation of public opinion”.    

We often assume that majoritarianism always reflects the public’s best interest, but this article has proved otherwise.  The majority of consumers are vulnerable to the concentrated interest of these small interest groups and what began as the opinion of this one group may proliferate as the majority’s opinion.   This is why when these groups have the wrong interest, incorrect cascades can develop followed by bad public policies.   

Herd Beliefs

“If a hundred people believe it, it’s a cult,

if a thousand people believe it, it’s a philosophy,

if a million people believe it, it’s a religion,

if a billion people believe it, it’s the truth.”

http://theuniverseas.com/human-herds

http://blog.seattletimes.nwsource.com/techtracks/2008/03/microsoft_yahoo_make_social_network_plays.html

http://sites.google.com/a/opensocial.org/opensocial/Home

This article shows how important online social networks have become. Some of the largest software/tech companies found it necessary to join together in an effort to minimize the differences between using most of the more popular social networking websites. Yahoo joined in the OpenSocial campaign, created by Google. OpenSocial “defines a common API for social applications across multiple websites,” in an effort to have similar structure and applications that work across websites, thus allowing its users greater ease of use and portability. The hope is that a user will be able to use an application designed to interact with one website and will still be able to use that application with another without having to re-do everything all over again.

What seems to be happening here is that a group of social networks will have common applications between them, that will allow their users to have a single application that does the same task for both websites. In effect, it is an information network joining social networks (which are websites, and thus part of an information network). Thus, we see a blurring of the lines between social and information networks, which allow these websites to act as both at the same time, affording its users the benefits of both worlds. Users are able to interact socially with each other, and because of the websites they can interact in new ways, such as browsing for friends based on similar interests, in the more “train-of-thought” pattern found in information networks.

Also, because there is not one giant all-encompassing social network available on the internet, the advent of OpenSocial means that users will have the ability to use the same applications across multiple social networks without the need to redo things multiple times, thus bringing the websites together more tightly. This will hopefully allow users the choice of multiple social networking websites so that they can use the same website their friends use, without reduced functionality or multiple copies of the same thing.

The cover story in USA Today’s Money Section investigated the Botnet Scams that are spreading across the internet. This botnet is comprised of thousands of bots that spread viruses and spam throughout personal emails and webpages. An example used in the article is a virus that was spread following Heath ledgers death. This email supposedly had a link divulging “the real reason” behind his death, when really it was just a part of a botnet scam. If an individual were to click on this link they would have been infected with the fast spreading Mega D Botnet, spreading spam for male enhancement pills. Studies and reports have shown that 40% of the 800 million computers connected to the internet in a day are involved in the botnet network of spreading email spam. Not only are email systems being attacked, but sensitive data is being stolen from banking and shopping websites, as well as spreading fresh infections. Everyday internet users come across spam mail or viruses, as they have become such a big part of the internet network.

The manner in which botnets operate directly relates to many topics we have covered in this class. The internet itself is a giant network connecting people to information they desire. Email has developed into a means of reaching individuals, and therefore has become a key way for these scammers to work, spreading their viruses and spam. Webpages themselves have recently been targeted as well, as botnets strive to extract important data. In class we have discussed the inner workings of hubs and how they work. The web tends to be a very difficult concept to grasp for many, as its comprised of so many networks. Searching the web alone is a very difficult task, as those botnet scammers control certain pages and areas. A lot of the difficulty surrounding searching the web has to do a lot with the reliability of the source, as any individual can be an author of a page. Information is geographically dispersed, enabling any of these scammers to create their own page. The game theory principle is based on the world reacting to your actions, or in this case, searches on the web. The scammers want their spam pages to score high for any search, whether that means writing the phrase over and over again, anything to attract these innocent web seekers.

Bot Net Scams.

Given our recent in class discussion of social network effects and the similarities of these topics with my interests as an ISST major, I absolutely have to comment on the post made by my classmate earlier today entitled “Information Cascades and Pop Culture” and more specifically the referenced NY Times article from Dunken J. Watt’s entitled “Is Justin Timberlake a Product of Cumulative Advantage?” My interest in this topic begins with the belief that the concept of “Cumulative Advantage” and many other social concepts discussed in this article can be explained more precisely using the ideas and conceptual framework we have been laying out in our Networks lectures. As I began reading Dunken’s article I was became increasing anxious to here the ‘network’ aspect of this social commentary. What I eventually realized is that the sociologist Dunken J. Watt’s is giving a very insightful description of what we would call “network effects” but from the perspective of traditional sociology. In fact the author doesn’t even discuss the concept of a “network” until near the articles end and when he finally started to form conceptual ties to economics, information theory, and network theory that I was so anxiously awaiting I honestly felt like it was not only “off” but at the very least incomplete. (I only became aware of the author’s prestige after coming to this general sentiment so please humor me if you can).

Dunken begins a paragraph on the second page by paralleling the idea of “cumulative advantage” with emerging network theories in economics.

“Economists like Brian Arthur and Paul David have long argued that similar mechanisms affect the competition between technologies (like operating systems or fax machines) that display what are called “network effects,” meaning that the attractiveness of a technology increases with the number of people using it.”

The economist that Dunken is referring to are responsible for the modern theory of increasing returns, also called positive feedback depending on your department/textbook. Positive Feedback is known as a network “Externality” and is defined by wikipedia an “An impact (positive or negative) on any party not involved in a given economic transaction.”

In the sense of economics, we have seen that positive & negative externalities can arise when individuals are not making their value assumptions independently. From a network perspective we interpret the concept of externalities slightly differently by stating that the value of a connecting to a network depends on the number of other people already connected to it (Varian, 180). However this is not a parallel to the “cumulative advantage” concept as Dunken states but rather fundamentally the same mechanism or “network effect”. To make this point more concrete I’ll use the fax machine example previously mentioned. The nodes in this network are the fax machines and the edges represent the connections (and hence the ability to exchange information) between them. The “total value” of the network will then be increased with the addition of a each new fax machine to the network- as one would generally expect, but more importantly the marginal value of this additional fax is also increasing! This increase in marginal value property is another form of the the externality known as positive feedback and is often called “increasing returns to scale” in most economic settings. What I have attempted to show with this comparison of network and economic “externalities” is that if we think of information exchange as an economic transaction and vice versa, we can develop more than just “parallels” between these network mechanism but also fundamental governing principles and mathematical formulations. What I feel this author failed to do was recognize that the seemingly complex and ambiguous mechanisms driving our cultural development & societal evolution can be accredited to the same positive feedback externalities that took America’s old industrial economy from an “economies of scale” to “economies of networks.” we have in the information industry (Varian, Ch.7). But dunken seemed to downplay the importance of networks in his social analysis by implying that network effects were themselves not enough to explain the unforeseen shifts in consumer demand (what I believe modern economics may call a “tippy” market.)

“even in markets that don’t exhibit obvious network effects (like markets for low-carb or organically produced food, fuel-efficient vehicles or alternative energy technologies), sudden shifts in consumer demand can still arise, persist and then shift again. These shifts often come as surprises but are soon explained away as mere reflections of changing public sentiments. Yet while in some sense these markets do reflect what people want, that is true only of what they want right now. If markets not only reveal our preferences but also modify them, then the relation between what we want now and what we wanted before — or what we will want in the future — becomes deeply ambiguous.”-Dunken

I would say there are most definitely network effects in the form of externalities exhibited in the markets for low-carb food and alternative energy. And although you may not find these network effects obvious, that really doesn’t change the fact that the theories that we have applied to our information and economic networks has also predicted many of the empirical observations of our social network and explains why stock prices are just as hard to predict as the next Justin Timberlake or Tickle Me Elmo—which are in ways “cultural stock” if we think about those products using the same conceptual framework we use to define economic markets.

Many subtle and non-intuitive events can happen in these network markets and I believe this is what Dunken meant by saying that these markets are “deeply ambiguous.” Perhaps a better understanding of this market structure is given by Varian in Chapter 7 of Information Rules, “The beautiful if frightening implication [is that] success and failure are driven as much by consumer expectations and luck as by the underlying value of the product. A nudge in the right direction, at the right time, can make all the difference.” This is also what I believe to be one of the most compelling arguments/insights from Malcom Gladwell’s “The Tipping Point.” However Dunken seems to misinterpret these new market principles when he makes the statement, “Yet while in some sense these markets do reflect what people want, that is true only of what they want right now.” –But isn’t this the very assumption of the most basic markets we study in economics? Given what we’ve learned about our society and the social *networks* that arise across all humanity, shouldn’t we expect this “cultural” marketplace to operate under many of the same principles as a network of transactions on wallstreet?

I believe the missing conceptual component that merges Dunken’s sociology perspective of social networks with that of modern economic & network theories is that information exchange is as much an economic transaction in the market for consumer products as the exchanging of cash for corporate stocks is on the stock market. The ‘two-way’ nature of economy that has grown out of the information age is such that markets not only reveal our preferences but modify them as well. And although Dunken seems to be focused on the ambiguity that arises from this fact, the end result of this so called ‘ambiguity’ is a culture that can rapidly evolve and progress, where value can be created not only intrinsically but extrinsically as well. This hard to quantify “extrinsic value” is what I believe contributes to the volatility and “tippyness” of network markets and is the reason for the deserved skepticism Duncan calls for at the end of his editorial:

“That doesn’t mean we should stop trying to anticipate the future, any more than we should stop trying to make sense of the past. But it does mean that we should treat both the predictions and the explanations we are served — whether about the next hit single, the next great company or even the next war — with the skepticism they deserve.”- Dunken

However in this giant “conglomeration network market” of economics, society, and information that I envision, our dear friend skepticism is merely a semiotic resource for expressing the value of the imperfect information associated with randomness that is inherent to the market. What I believe this means is that the need for skepticism is controlled by the same “ambiguous” demand functions that initially sparked Duncan’s call for skepticism which may turn his closing argument into a rather moot point. What do you think?

In 1995, the two management consultants who wrote the book Discipline ordered 10,000 plus copies of their book from many small bookstores scattered across the country [1]. Likewise in 2002, David A. Vice a reporter for the Washington Post ordered 20,000 copies of his book from BarnesandNoble.com and later returned 17,500 of them[2]. The obvious questions that arise is why would someone want so many copies of ones own book and secondly why didn’t the authors order the books directly from their publisher. These questions can be best answered by the theory of information cascades which was discussed in class this week.

What are Information Cascades?

Information cascades occur when individuals modify their own choices based on the choices of others. A misconception about herding or information cascades is that it is irrational. Herding behavior can propagate both valid or invalid information but it reflects individuals making decisions rationally based on limited information. Individuals may choose to follow the crowd because they believe the crowd knows more than their do. For example, perhaps an individual lacks the time and expertise to research which candidate to vote for, so they decide to vote the same as their friends.

Cascading decisions can result for two reasons. First people may imitate the decisions of others for informational reasons. They may believe that a choice is popular because more people have found it to be better. Secondly people may imitate others because of a direct payoff due to the popularity of a product. For example, reading a popular book allows one to converse at a party or seem current. Such phenomena are sometimes are called network effects.

To become an information cascade a few criteria have to be met. First sequential choices where people can observe the actions of all previous choices is required. Second, decisions or assessments have to be binary with only two choices possible. Why binary choices are required for cascades and examples of the blind leading the blind are explained in a New York Times Blog [3]. Lastly, the payoff must be uncertain otherwise people would always pick one of the choices.

Cascades and Bestseller Lists:

The authors described above realized the importance of hitting the New York Times bestseller lists and hence their strange purchases. These lists can become a self-fulfilling phenomena with the benefits of increased prominence and further sales resulting for listed books. Thus by channeling their purchases for the book Discipline to many small bookstores tracked for the bestsellers lists, the authors described were able to artificially boost their rankings.

With the advent of the web, gaming these such rankings has become much more commonplace. A Wall Street Journal article talks about how one PR firm, Ruder Finn, claims it can boost the rankings of books on Amazon.com and BarneandNoble.com for a fee of $10,000 [4]. The fact that most books sell less than one copy a day on Amazon.com combined with the short term nature of these rankings(refreshed every hour) makes them particularly volatile and vulnerable to manipulation. One author for example convinced his fans to purchase his book on one day and at the same hour to maximize the jump in the rankings hoping to gain prominence. Ruder Finn for its fee has popular authors to recommend the new book on many email lists and offers free goodies along with purchases to customers as enticements.

This probably brings up the question, shouldn’t the best books get to the top by themselves? Many arguments show why this may not necessarily be the case. One reason is the difficulty in judging the quality of a work. This is illustrated by the many classics that were turned down by many prominent publishers before getting published. One reason measuring the quality of books can be so difficult is because people don’t know precisely what they like [5]. Instead they are often searching for the best which by definition is what other people like and think is the best.

The randomness involved in popular products was illustrated by an online experiment described in the article “Is Justin Timberlake a product of Cumulative Advantage?” The experiment showed that on different runs of the experiment, different songs became the most popular. It was found that people tended to pick their favorites based on the most popular songs. Thus some good luck early on could be amplified by herding behavior resulting in a low quality song becoming most popular. For example, a song found to have middling quality placed 1st in one run and then 40th in the next run. Thus the experimenters argue, in the cultural marketplace there is much uncertainty about which products will turn out to be blockbusters and which will not.

The randomness inherent in the selection of successful products by consumers in the marketplace means it makes sense for authors to push their product. Jump starting the momentum behind a book, having favorable early reviews or increasing its visibility by pushing it in an influential top-seller list may give it the momentum necessary to become a cultural phenomenon and a blockbuster. The recognition may in turn make the author more famous and future books more visible. This explains why many authors and literary agents spend so much time, money and effort trying to build a buzz for their books.

1: Stern, William. “Did Dirty Tricks Create A Best-Seller?” BusinessWeek. August 7, 1995

2: Kirkpatrick, David D. “Author’s Attempt to Promote Book Backfires”, New York Times, March 6, 2002

3: Tierney, John. “How the Low-Fat, Low-Fact Cascade Just Keeps Rolling Along.” New York Times, October 9, 2007

4: Bialik, Carl. “A Few Sales Tricks Can Launch a Book To Top of Online Lists.” Wall Street Journal Blog, March 23, 2007

5: Watts, Duncan J. “Is Justin Timberlake a Product of Cumulative Advantage?” New York Times, April 15, 2007