There is a very deep ocean of information relating to game theory.  In sticking with my theme of relating classic examples back to class topics, I found an amazing game theory website.  gametheory.net has an amazing depth of information and is a great resource for all game theory topics.  If one were interested in combinatorics, they could even play Nim through the site.  However, more importantly is a set of links to very many game theory outlines at various universities throughout the country.  If any students in the class are interested in hearing the material from class in a different perspective, follow the student links to the “Lecture Notes” section to hear what a variety of leaders in the field are teaching.  One thing I would like to address in this blog, is the popular press section. 

 http://www.gametheory.net/News/Items/117.html 

That is one article from Discover Magazine which covered the research of a CalTech economist researching irrationality of behavior in game theory.  While in class, we assume that everyone behaves rationally and follows our basic mathematical models; often real-world results are confounded by the irrationality of people.  Information cascades may not form where they should because people might choose to go against the signals received purely to go against their signals.  This is not rational, but it happens.   Another thing I would like to point out, is from the Schelling book when he discusses equilibrium and such in the first few chapters.  While the economic perspective of a beach may dictate that people go until the beach is so crowded that noone else wants to go, this is not socially optimal in many situations.  Situations often rely on the rationality and the irrationality of people equally.  Like with the opening story about the students filling in the back of the auditorium, we see that the students were achieving an optimum, an irrational optimum.  Often, it helps to look (as the researcher from the article above did) at the data of irrationality as much as the data which confirms the rational behavior.   There are many other resources on this site, and I highly recommend it for students in this class and anyone interested in Game Theory.

http://www.ecommercetimes.com/story/34422.html

In Online Social Networks and the Profit Motive, Elizabeth Millard explains entrepreneurs’ wariness in investing in social networks. How exactly does one advertise in an online social network? While financial analysts do see online social networks as promising markets, the question remains how to tactfully and efficiently advertise on such venues. “[Advertising on online social networks is tricky] because network members give personal information, they might flinch when that data results in targeted advertising,” Millard explains. Since targeted ads clearly result in some privacy concerns, online social networks began attempting to acquire revenue elsewhere. Additions of premium memberships (extra features for a monthly fee) became a profitable and ethical way to acquire revenue through social networks. Such a scheme works very well as these social networking sites can initially attract users with their free, basic features and then offer members more services at a cost. Additionally, free trial offers can be extended to the core members in hope that a taste of these new services will attract them to acquire premium memberships.

A different issue arises with online business networks. Business networking isn’t used to meet people with similar interests but rather to facilitate industry, and Millard explains how such networks must be treated differently then purely social ones. LinkedIn spokesperson Konstantin Guericke, when interviewing with E-Commerce Times told Millard, “The way our members look at the site is not as a community, but as a tool. Hammers don’t come with advertising, so why should we? You pay for your tools, and that includes business tools like databases and applications.” This brings up a very interesting point. Online networks seem to be categorized into those whose services, as deemed by the public, should be free (like MySpace or Facebook), and those that naturally should require payment (like LinkedIn and other technology tool sites). However, certain “social” online networks, like relationship finders, require a fee for their services. The fact that activities like sharing picture with friends on Facebook is deemed to be a free service, while matchmaking is not, elucidates a lot about the relative social importance of different networking venues and their underlying purposes.

There is one issue on which virtually every member of the baseball punditry agrees: steroids ruined baseball in the 1990s. Players took the illegal drugs because they made them stronger, and this strength enabled them to hit more home runs, which in turn gave them greater leverage in contract negotiations. Major League Baseball did not regulate the issue, refusing to punish its players for steroid use, and so it is generally agreed upon that steroid use was widespread on most teams for much of the 90s and the beginning of the 21st century.

Hand in hand with this belief that steroid use was widespread is the idea that steroid use leads directly to gaudier statistics. As the above chain of logic states, take steroids and you’ll get stronger and hit more home runs–on the surface, it seems to make sense. And if one looks at the league’s statistics from the past 15 years, they seem to support this theory–the number of home runs climbed higher than anyone thought it could, a feat best personified by Mark McGwire and Sammy Sosa’s pursuit of Roger Maris’ single-season home run record of 61 in the summer of 1998. Both of those players have since been implicated in steroid rumors, although nothing has ever been proven.

So is this logic true? Does steroid use lead to better individual performance? In today’s New York Times, J.C. Bradbury, and economist and professor at Kennesaw State University, argues that the steroid explanation is faulty. His article, which can be found at the following link:

“What Really Ruined Baseball”

attributes the spike in statistics to a different factor: league expansion. Bradbury’s article alternates between dismantling the steroids explanation and supporting his own theory.

How exactly would league expansion affect individual performance? Bradbury uses a theory of the evolutionary biologist Stephen Jay Gould in his explanation:

“…in competitive environments, as the variance of the quality of participants shrinks, opportunities for great performance diminish.”

So when MLB expanded from 26 teams to its current number, 30, over the course of five years in the 1990s, its talent pool was significantly diminished. There were more roster spots to fill, and so players, on average, became less talented. By using Gould’s hypothesis, Bradbury concludes that with less talented players (he refers to them as riffraff) populating the league, great hitters were provided with many more opportunities to have great performances.

This league-wide expansion then, and the resulting drop in talent, is the explanation for baseball’s recent statistical aberrations, according to Bradbury. It’s not just limited to hitters, though; he cites numerous figures in which the game’s other half, pitchers, have also been able to exploit less talented players–for example, a recent spike in numbers of strikeouts.

Bradbury’s explanation is essentially a tipping point–something that may seem insignificant at the time, like the expansion of an already large league by a mere four teams, can have large and unintended consequences in the future. It is interesting to consider how decisions like this could impact organizations in many different fields–these kinds of situations are not limited to baseball. Is expanding your restaurant franchise a good idea? How about building that additional wing on your church–have you considered the costs of maintaining it, especially if your membership declines?

I also enjoyed Bradbury’s explanation because it flies in the face of so much conventional wisdom regarding increases in home runs and other statistics. Baseball is, more so than any other American sport, a game of history, with its numerous instances of individual outcomes inspiring people young and old to compare players from past generations to those of today. So when that comparison becomes muddled by something like this huge increase in home runs, people search for an explanation, and many seem to have latched on to steroid use as a catch-all for the game’s current state. While I don’t agree completely with Bradbury’s column–there is no single explanation for what happened to baseball in the 1990s–I respect him for approaching the debate from a different angle and offering a convincing alternative explanation, much like Gladwell does for other situations in his book.

Political Information Cascade

(http://select.nytimes.com/gst/abstract.html?res=F30F1FFF3E540C708EDDAA0894DF404482)

The Presidential political race is almost upon us as most notably in the news democratic contenders begin to set their platform agendas. Before either party can rally behind their presidential candidate, the party members through the presidential primary must elect a candidate. The ‘Primary’ as it is known has two phases. First, party members from each state will vote in their state’s primary election. Later in the year after every state has voted, a general convention is held where the states vote again and the candidate who emerges victorious wins the party’s nomination for the presidency. While the general convention may seem like the more important part, the race is normally decided long before the general convention is held. In fact, the most important part are the votes of the first couple states because their outcome creates an information cascade that reasonably predicts the outcome of later voting states. That is why the news makes such a big deal about the outcome of early states like Iowa, New Hampshire, and Nevada.

However, this schedule is about to change. Over the last couple of months, many powerful states like New York, Texas, and California have moved up their presidential primaries so that they can become the most important battleground states for presidential hopefuls. Since their votes count for so much, candidates will have to focus winning those states instead of the initial small fries. One March 23rd, The New York Times published an article entitled To increase Florida’s Influence, Lawmakers may set a January Date for Presidential Primary. The article is about how Florida wants to move their primary ahead of all the other “megastates” so that they can become the most crucial state for candidates to win since they will start the information cascade. Larry Sabato, a political scientist at the University of Virginia says, “It’s the smartest move I have seen…they [Florida] are the only ones that has figured out that if they go first, they could be the megastate to tip all megastates, and Florida could decide the nomination.” In essence, this is a fight to see who will be the first state to trigger the information cascade on arguably the most powerful level and decide the nominees for the 2008 presidential election.

“Obesity runs in families - and friends, too”, by Alvin Powell of the Harvard News Office, can be found here:

http://www.news.harvard.edu/gazette/2007/03.08/09-obesity.html

            This article summarizes some of the conclusions made from the Framingham Heart study by Nicholas Christakis, professor of medical sociology at Harvard Medical School and professor of sociology in the Faculty of Arts and Sciences. The Study collected information about health, diet, and exercise, along with information about the family and friends of each of the subjects of the study. Taking periodic measurements every two to four years beginning from 1948, the Study was able to cover the spread of three generations: grandparents, parents, and children.

            Data from the Study show that Americans are indeed gaining weight: “The percentage of adults aged 20 to 74 who are overweight has increased from 44.8 percent in 1960 to 65.2 percent in 2002. Those who are seriously overweight or obese increased from 13.3 percent to 30.5 percent over the same period.” Christakis, after studying the case of 5,124 children and their families and friends, was able to determine a correlation between mutual friendship and weight gain. He noticed the most significant chance of weight gain occurred when two people each considered the other a friend. If a friendship were directed in only one way, the weight gain of a person named as a friend was often not affected by the weight gain of the person doing the naming. Thus the directionality of the friendship-weight gain connection is concluded to be two-way.

            Christakis also found that gender also affected weight gain between friends. In his words, “Men are much more influenced by weight gain in men and women influenced by weight gain in women.” This is an important distinction to make. Gender introduces an additional commonality between members of the study, thus creating a kind of focal point as described by Kossinets and Watts. As discussed in their analysis, students who shared a class were more likely to exchange email. This situation can be transferred to weight gain in that people of the same gender were more likely to affect each other’s weight gain.

            By determining the correlation between friendship and weight gain, the directionality of such friendships, and the focal point around which most weight gain is associated, Christakis managed to draw a few conclusions about the network effects of obesity. While these trends were too complicated to be described exactly (“more akin to the ripples and interference patterns generated by throwing a handful of stones into a still pond than that of dropping in a single rock”), he did notice a change in accepted societal norms of weight. He believes that such data of weight gain can be explained in part by “the acceptability of gaining weight”, thus, the influence of someone’s opinion on weight gain directly affecting another person. Such a relationship can be directly applied to cascading effects that we learned in class. For example, person A’s friend B buys a tasty, not-so-healthy brand of ice cream. B tells A how great it is. A, dismayed by the lack of nutritional value of the ice cream, is hesitant to buy it. A’s friend C does not quite have A’s will-power, and buys some of the ice cream also. C also tells A how wonderful this ice cream is. Now A, with two positive signals about the ice cream, is likely to want to find out about this awesome ice cream, despite its health drawbacks. Cascading is now likely to happen: A’s other friends may hear about how A, B, and C all liked this ice cream, and are then compelled to find out the truth for themselves. So we see that the cascading effect indeed influenced these people to make a choice detrimental to their health.

The academic paper entitled The Authority of Supreme Court Precedent: A Network Analysis can be found using the following link:

http://jhfowler.ucsd.edu/authority_of_supreme_court_precedent.pdf

This academic paper provides an in-depth explanation of how network analysis can be applied to study Supreme Court precedents. A network was constructed that consists of 30,288 majority opinions written by the U.S. Supreme Court from 1754 to 2002. The paper describes a method for creating hub and authority scores that can be used to identify the most important and most influential cases in history. The method used is a direct application of what we learned in class about hub and authority scores before Spring Break.

The following describes how hub and authority scores are used in the study:

“The authority score of a case depends on the number of times it is cited and the quality of the cases that cite it. Symmetrically, the hub score of a case depends on the number of cases it cites and the quality of the cases cited. Thus, authority scores indicate the degree to which a case is thought to be important for resolving other important issues that come before the Court, while hub scores indicate the degree to which a case is well-grounded in previous important rulings” (3). “A case that is a good hub cites many good authorities, and a case that is a good authority is cited by many good hubs” (12).

At a basic level, one could just use the number of inward citations to measure a case’s importance. This is the method of degree centrality. The problem with this method is that it does not use all the information given in the network to arrive at a conclusion because it treats all citations with the same weight. Another method is eigenvector centrality which indicates important cases using a computed vector of importance scores. However, this method biases downward the importance of recent cases and also assumes that only inward citations contain information about a case’s importance. Using hub and authority scores allows both inward and outward citations to be considered when assessing importance of cases. The study of cases using the hub and authority scores method facilitates the most accurate identification of key precedents in the network.

The above comparison relates to the idea of distinguishing between nodes in a network that have multiple reinforcing endorsements and those that simply have a high in-degree. This was touched on in Homework 4 from last week and was also mentioned in class. The method of using hub and authority scores is a much more comprehensive and accurate method than the others described.

In addition to identifying importance, authority scores are also used in the study to detect the rise and fall of precedent, to analyze changes in the issues that the Court prioritizes, and to study changes in the importance of competing legal rules within a certain area of law. The paper is able to verify its findings by successfully matching the results of important cases found using network analysis with evaluations by legal experts.

http://www.orgnet.com/orgchart.html

http://waderoush.typepad.com/twr/2005/03/james_surowieck.html

In an article called Organizational Hierarchy: Adapting Old Structures to New Challenges, Vladis Krebs says the structure of government is not efficient as it could be because it has not adapted to the agile and creative structures like those seen in the business world.  Krebs uses the example of the hierarchy of the

US intelligence community to discuss how an alternative structure could benefit the government.  The government wanted to improve its accuracy in the war and its agility to prevent terrorist attacks. In Krebs example, the president sits at the top of the hierarchy and has direct connections to the head of each intelligence agency.  In an effort to improve the efficiency of this network, it was proposed that a new position be created that would act as an intermediary between the heads of each agency and the president.  This intermediary would be responsible for analyzing and synthesizing the information from each agency.  

Krebs argues that a better modification would be to simply create connections among each of the heads of the agencies.  A new position, he says would be better only if the goal was to have “accountability and budget responsibility;” however, he believes that an interconnected structure would be better if the goal is “a smart, agile learning organization — able to adapt to a changing enemy.” 

Reading this article made me first wonder how the social connections function in information hierarchies such as that of the government.  The dynamics are much more complex than Krebs portrays.  I couldn’t help but think of a blog post by “Travels with Rhody,” that outlines (almost word for word) a lecture given by James Surowiecki on the Unwisdom of Crowds.   Surowiecki would argue that simply adding connections among the heads of each agency would not necessarily yield better decisions.  He warns that organizations can be “too connected,” even when other group members are really smart.  

When we were learning about information cascades in class, we discussed how the order of speakers could influence the final decision because the first few people could set in motion an information cascade.  Once this happens, it will take someone with more power (either in position or the information they possess) to stop this cascade.  So how does an organization structure itself so that it can efficiently integrate information while maintaining the most accurate information?   While Krebs believes that an interconnected structure would maximize the quality and efficiency of decision-making, Surowiecki emphasizes that the best organizational structure would “range across hierarchies” and include individuals with diverse opinions and information.  Krebs’ idea of information sharing is a good one; however, it is important to keep in mind the subtle (but tremendous) power that information cascades can have on groups trying to make a decision.  More information is needed to truly understand the dynamics of group decision-making, in order to develop the best organizational structure for both business and government.

http://waderoush.typepad.com/twr/2005/03/james_surowieck.html

Rhody writes about the arguements that James Surowiecki revealed in a conference he gave. Surowiecki talks about all the fuss about collective action and collaboration over the last decade with all the new websites. But in a way he looks at the wisdom and use of these type of situations with scepticism. Rhody briefly brings up his book, “Wisdom of Crowds” and some examples that he used like a group of people only being a pound off from guessing the weight of an ox after it had been slaughtered and dressed. He says the wisdom of crowds works well when there is a true answer, and as long as some choices are better than others. He says the key to coming up with the right information is the use of private information some may be wrong but the diversity of each individuals answer is a good thing. Surowiecki believes that the best way is not to have everyone to use their own information and mix it in with others but for each individual to come to their own conclusion and then share it. The less personal interaction the better. He then contrasts it to a centralized type of collaboration that linux uses with a large group working but one or a few people decide on the final decision.Then he shows how ants are able to figure things out but shows that humans don’t have the same tools as ants do. He says the more humans interact the dumber we seem. He gives two reasons for why this is; One being that human beings herd, and second being that we imitate. People like the comfort of a crowd if you want to look legit then do what everyone around you is doing. He says that we are imitation machines and shows this by an example that he used in his book when he had scientists just looking up into the sky on the street corner and people began to do the same thing. He then brings up “The Tipping Point” and its explanation of what a information cascade is. Collective decisions may not be tied with quality. He quotes Pascal as the problems of the world arise because a man cannot sit in a room and think quietly by himself. He then says that that isn’t what he thinks we should do since some information from others can be useful but how can we have group intelligence without the influence of others in order to keep the independence.

In this article we look at information cascades and network effects. Although we see what information other people have we can’t rely on that since it alters our own opinions the best thing is for others to develop their own information and answer and then use what other people have decided on as well because as Rhody and Surowiecki have shown that humans are too easily influenced and this ruins the information that each person brings to the group decision. In information cascades after the first two people decide on the same option (accept/reject) then all others “should” follow despite ones own opinion in the matter but what if the opinions of the ones before were right or wrong. That is why in some cases the group can be very smart but in other times they can appear less intelligent than they would seem than if they were individually making a decision.

Topix Reinvents itself as citizen journalist site

It seems as though user interactive sites, such as Wikipedia or even UrbanDictionary.com, have longer lifetimes than read only sites. Following this trend, Topix has transformed itself from a software-based news aggregator site to a citizen journalist hub where site visitors can post whatever news they please.  Prior to Topix’s reinvention, users were not sticking around the site for very long, visiting only a few pages before becoming disinterested and leaving the site.  This change was brought about in hopes to get users to read more pages of the website, creating a higher click-thru-rate which will attract more advertisers.

This is a good example of what we discussed in class today - how some sites or programs won’t have payoffs unless other people are participating in the network.  These include most social networks and media sharing programs like Kazaa or Limewire.   In this case, Topix can exist without users contributing, but it wouldn’t thrive for very long.  It needs advertisers to keep the site running and Topix has choosen to allow more user interactivity to achieve this.

In last Sunday’s Times Magazine, Rachel Louise Snyder writes with news of “a network theory for building a savings account.” When John Caskey, an economics professor at Swarthmore, conducted interviews in two low-income communities, he found that many “didn’t save not because they actually couldn’t [i.e. they couldn’t afford to], but because they believed they couldn’t.” The problem was intense social pressure to avoid saving money from friends, family, and other members of people’s social networks: “the minute people got a little surplus, friends and family would start asking for loans.”

What the Consumer Federation of America is doing to reverse this situation is to promote saving using the same network effects that currently impede it by “creating a network of support for saving” by promoting saving to members of churches and community groups. The theory appears to be that if many members of a community save, someone with a surplus will be able to save that money instead of loaning it directly to others. Furthermore, non-savers experience a sort of information cascade as their friends and family start saving, encouraging them to start saving for themselves. When people are able to save their savings victories with each other, it reinforces the accomplishment and promotes even more to save.