Traffic over the internet is becoming more and more congested. When routing one point on the world wide web to another, internet service providers often want to find the shortest path between them in an effort to optimize data transfer rates. However this often results in many signals being routed through the same path and causes congestion. In an effort to ease the congestion, sometimes users are routed through a different, longer path but with less congestion. This trade off is the process of optical routing and flow control and often will result in faster data transfer rates.

The process of routing and rerouting data based on transfer rates that depend on traffic is similar to network theory on finding Nash Equilibria between multiple nodes in a graph. Applying basic network theory to complicated things such as the internet and web traffic routing can often result in more efficient and well connected graphs.

http://en.wikipedia.org/wiki/Optical_IP_Switching

Today we learned in class examples of positive externalities, some examples being: vaccines, fax machines, ebay etc.   But I wanted to focus on cell phones, more precisely how they are considered a positive externality.  For one, people are not going to want to be a part of a plan that is not widely known.  This is because it is not wide spread, so you do not know who else has the phone and it might not seem like a reliable service.  The most well known carriers are: At&t (Cingular), T-mobile, Verizon, Sprint and Nextel.  But there are also many that most people, including myself, have never heard of.   After looking at all the carriers I learned these carriers exist: CellularOne, SunCom wireless, U.S Cellular, Centennial Wireless, Edge Wireless, nTelos, Qwest and Surewest.  There are also many other carriers that are not on this list, but the fact is that they are unknown.  So do they also fit into the category of a positive externality?

For a positive externality to occur an individual action must create a positive and uncompensated effect on others payoff.  So, in the instance of cell phones, the well known carriers are surely positive externality because when you purchase a cell phone, you create a network between the people you know, and their ability to reach you.  This is especially the case when, for example, you both have At&t; therefore whenever a call is placed between you two, it is always free because you both have the same carrier and that is one of the benefits of At&t.  This would be a positive externality because another is benefited by your purchase of the cell phone. 

But what about those carriers that are not well known?  You would think that they could not be considered a positive externality because no one has heard of them, so most people would not trust to purchase a plan from them. But to an extent, it could be a positive externality if you purchased a plan with one of these cell phone services, because then you are a form of advertisement.  If you like the carrier, you tell others about it, they then purchase the phone, and then you and the others benefit.  But as some say, sometimes this negative outlook on the specific service is ignored because of a positive benefit.  “While a case could be made that cell phone service does not measure up to this level of service, the fact remains that we seem to be willing to accept that trade off because the advantages that mobility provide outweigh the added expense and inconvenience inherent with today’s cell phone service” (http://www.ispplanet.com/perspectives/2005/ wireless_rant_p3.html).  In conclusion, I would agree that even if the cell phone carrier has never been heard of, it is still considered a positive externality because if you were to purchase a plan, then you would be sort of ‘advertising’ that phone carrier, and also others would be willing to purchase it more.  Also the fact that cell phones in general are a positive externality because once you get the phone, you open up a new network of people that can contact you.

Before spring break I attended a presentation for HEI hospitality, a company which buys and sells hotel properties. During the presentation we were addressed by one employee whose job requirements included not simply web design, but web page modification. The purpose of her modifications is to increase the rankings of the web pages for their hotels so they come up high on web based searches. She brought up a few web pages of real estate properties she has done work on as examples. We were shown the page before and after work was done. Part of the business is making sure that the company’s web pages are easily accessible by basic web searches via google, yahoo and other readily used search engines and that they come up above competing hotel sites. Some methods by which this is done is by adding links to the pages so they come up higher on searches. Links from their corporate site to their specific hotel web sites and other various links are added based on rank. As learned in class this is a commonly employed method of raising web page rankings.

Another even more basic method by which the ranking of a web page can be increased is by enlarging the font size of commonly searched phrases. For example if someone were to search a hotel brand such as Sheraton and a city name, several hits will come up. By enlarging the font of “Sheraton” and raising the geographic location of the name of the hotel on the site, it may come up higher in searches. Just by increasing the font or placing the name higher on the page the rank of a page increases, and this means increased traffic and increased room bookings for the hotel.

Her job includes searching these pages regularly to see how high they come up in both general and advanced searches and making the changes necessary to improve their rankings. Just as the structure of the financial market dictates job availability, the nature of the internet creates jobs within certain web based industries and this is a perfect example.

Link: To the HEI hospitality web site http://www.heihotels.com/

To the site for Sheraton, Music City http://www.sheratonmusiccity.com/

Rumors: Classic information cascades

 People deal with rumors everyday in their life. From a high school student to a company CEO, there is always some sort of a rumor circulating about in their social network. In a high school for example, a competitive student, say A, might have spite against another student, say B, who is getting better grades than him. If that is so, the kid A might want to get back at the kid B and start a rumor about him by saying that he was cheating. How would it be possible for him to do so? What is the tipping point that would lead other students to believe that? One way would be to go to an official and tell them about it. But if the lie is caught, then A would be in a lot of trouble. Another possibility is that A could tell his friends about this. But they could easily see through his lie and call him jealous and deceitful. Hence, A has to look for that one person in his class that is a source of all rumors, that person who is trusted by other students and is a center of gossip in the network. Acknowledging the information by that person would be the tipping point in the cascade of that rumor and would result in its spreading throughout the school. Even though other students might not know whether the student A is innocent or not, they would go with the crowd and side with the more influential person.

 According to the above explanation, 2 major factors can be identified leading to a successful rumor. The first major factor is the spread of the rumor. If the student in the above example tried to spread the rumor about student B in his family and relatives, then that would not have been successful because it was not in the right network. Spreading of the rumor in his family would not have affected the student B as it is not the correct network and the spread would not be efficient as the signals received by the nodes in this network would not have significance among them. The second most important factor is the type of node that receives the rumor. As I explained in the above example, if A tells the rumor to another student whose word is not as significant as some of the other important students, then that would not lead to an efficient spreading of the rumor. Hence, it is not only the spread but also the nodes that spread the rumors further. Such kind of strategy of spreading a rumor efficiently is evident in our daily lives. We see these everyday in financial markets and real estate networks.

 Let me explain about a real estate network. Suppose prices in an area for good houses or industrial plots are reasonable. If daily newspapers and television channels don’t talk about real estates, most people don’t consider investing their money on property and lands. However, a real estate salesperson can think of stirring up the activity in the market by spreading the rumor among people that land prices are going to rise rapidly very soon and it would be a very good idea to invest in them as it would lead to 100% profits. Spreading such a rumor through a couple of people would be inefficient and would possibly result in a failure. However, if that salesperson goes to a central node, a real estate enthusiast who actively deals with sale purchase of lands and who further spreads it around, then that rumor could result in a wildfire among people. Such a rumor would result in people actively thinking about buying real estate and this would increase the activity in the real estate market. People would ignore their own signals and intuition and would go with the crowd. The frequency of sale purchase of real estates would drastically increase leading to an increase in the demand of land, hence ultimately increasing the real estate prices. At this point, where the rumor actually comes out to be true, the information cascade comes crashing down because now everyone is aware of the fact that the prices are rising and would start investing in real estates. Hence, even though the rumor started by this salesperson was incorrect to begin with, it ultimately led to desired result that the person wanted and the rumor turned out to be true. A similar kind of pattern can be observed in stock markets where a company’s share price can increase if someone leaks a rumor to a magazine that the company is releasing a new product. People will think that the company’s share price would go up and will start buying their shares. As long as the company does not reveal its intentions and the rumor keeps on circulating, the information cascade will proceed and people would tend to go along with the crowd. However, as soon as the company reveals the truth about the rumor, the information cascade crashes as everyone will be aware of the outcome, whether the company really has a product or not.

 Another example that I would like to mention here is celebrity networks and company rumors. Celebrities and famous multi-billionaire companies deal with a ton of rumors everyday. People who might be against a company’s prosperity or a celebrity could try to spread rumors about them that might be untrue. I came across one such article about Google named:  Is the Google magic coming to an end?

This article talks about the fall of Google and an end to Google’s era. It says that Google share prices have been seen falling recently and a number of key members of the Google community, such as Ethan Beard and Sheryl Sandberg, have left Google to join Facebook, one of the largest growing social networks in the world. All this information about Google could be completely random and unrelated, that is, the above mentioned people could have left Google for their own reasons which could be completely unrelated to each other. However, if this information linking Google to Facebook were to gain stead and if people started believing in this, then it could pose a potential problem for Google. Its employees could think that if Google has started to fall, then it would be best for them to search for jobs elsewhere. They could ignore the hard facts that Google is rated the best employer of the year and is one of the best companies to work with and they might go with the crowd. This could lead to more employees leaving Google which could potentially make this rumor true. Hence, these examples perfectly depict how rumors are a classic form of information cascades and how the tipping point in the network could change the rumor into a truth.

Article link:

http://www.rediff.com/money/2008/mar/29forbes1.htm

Rumors: Classic information cascades

 People deal with rumors everyday in their life. From a high school student to a company CEO, there is always some sort of a rumor circulating about in their social network. In a high school for example, a competitive student, say A, might have spite against another student, say B, who is getting better grades than him. If that is so, the kid A might want to get back at the kid B and start a rumor about him by saying that he was cheating. How would it be possible for him to do so? What is the tipping point that would lead other students to believe that? One way would be to go to an official and tell them about it. But if the lie is caught, then A would be in a lot of trouble. Another possibility is that A could tell his friends about this. But they could easily see through his lie and call him jealous and deceitful. Hence, A has to look for that one person in his class that is a source of all rumors, that person who is trusted by other students and is a center of gossip in the network. Acknowledging the information by that person would be the tipping point in the cascade of that rumor and would result in its spreading throughout the school. Even though other students might not know whether the student A is innocent or not, they would go with the crowd and side with the more influential person.

 According to the above explanation, 2 major factors can be identified leading to a successful rumor. The first major factor is the spread of the rumor. If the student in the above example tried to spread the rumor about student B in his family and relatives, then that would not have been successful because it was not in the right network. Spreading of the rumor in his family would not have affected the student B as it is not the correct network and the spread would not be efficient as the signals received by the nodes in this network would not have significance among them. The second most important factor is the type of node that receives the rumor. As I explained in the above example, if A tells the rumor to another student whose word is not as significant as some of the other important students, then that would not lead to an efficient spreading of the rumor. Hence, it is not only the spread but also the nodes that spread the rumors further. Such kind of strategy of spreading a rumor efficiently is evident in our daily lives. We see these everyday in financial markets and real estate networks.

 Let me explain about a real estate network. Suppose prices in an area for good houses or industrial plots are reasonable. If daily newspapers and television channels don’t talk about real estates, most people don’t consider investing their money on property and lands. However, a real estate salesperson can think of stirring up the activity in the market by spreading the rumor among people that land prices are going to rise rapidly very soon and it would be a very good idea to invest in them as it would lead to 100% profits. Spreading such a rumor through a couple of people would be inefficient and would possibly result in a failure. However, if that salesperson goes to a central node, a real estate enthusiast who actively deals with sale purchase of lands and who further spreads it around, then that rumor could result in a wildfire among people. Such a rumor would result in people actively thinking about buying real estate and this would increase the activity in the real estate market. People would ignore their own signals and intuition and would go with the crowd. The frequency of sale purchase of real estates would drastically increase leading to an increase in the demand of land, hence ultimately increasing the real estate prices. At this point, where the rumor actually comes out to be true, the information cascade comes crashing down because now everyone is aware of the fact that the prices are rising and would start investing in real estates. Hence, even though the rumor started by this salesperson was incorrect to begin with, it ultimately led to desired result that the person wanted and the rumor turned out to be true. A similar kind of pattern can be observed in stock markets where a company’s share price can increase if someone leaks a rumor to a magazine that the company is releasing a new product. People will think that the company’s share price would go up and will start buying their shares. As long as the company does not reveal its intentions and the rumor keeps on circulating, the information cascade will proceed and people would tend to go along with the crowd. However, as soon as the company reveals the truth about the rumor, the information cascade crashes as everyone will be aware of the outcome, whether the company really has a product or not.

 Another example that I would like to mention here is celebrity networks and company rumors. Celebrities and famous multi-billionaire companies deal with a ton of rumors everyday. People who might be against a company’s prosperity or a celebrity could try to spread rumors about them that might be untrue. I came across one such article about Google named:  Is the Google magic coming to an end?

This article talks about the fall of Google and an end to Google’s era. It says that Google share prices have been seen falling recently and a number of key members of the Google community, such as Ethan Beard and Sheryl Sandberg, have left Google to join Facebook, one of the largest growing social networks in the world. All this information about Google could be completely random and unrelated, that is, the above mentioned people could have left Google for their own reasons which could be completely unrelated to each other. However, if this information linking Google to Facebook were to gain stead and if people started believing in this, then it could pose a potential problem for Google. Its employees could think that if Google has started to fall, then it would be best for them to search for jobs elsewhere. They could ignore the hard facts that Google is rated the best employer of the year and is one of the best companies to work with and they might go with the crowd. This could lead to more employees leaving Google which could potentially make this rumor true. Hence, these examples perfectly depict how rumors are a classic form of information cascades and how the tipping point in the network could change the rumor into a truth.

Article link:

http://www.rediff.com/money/2008/mar/29forbes1.htm

Robert Hansen, a business professor at Dartmouth University, has drawn some interesting conclusions about the role of investigative reporting in the 21st century mainstream media. His argument uses some important aspects of information cascade theory.

The catalyst for his observations stems from the Israeli/Hezbollah conflict we all remember from two summers ago. He believes that even though Israel gained the strategic victory, Hezbollah “won the communications war” due to their savvy control of the media. This is all he says about the issue, and moves on to a more general analysis of investigative journalism.

With the increase in use of online news sources, mainstream media are cutting their budgets, which equates to dropping reporters, and thus decreasing the amount of investigative journalism. In his view, this paves the way for cascades and herding. He gives the example of a news story. As always, you have private information that tells you whether a story is true or not, and you also observe how many people pass the story on to others. If you assume people only pass of true stories, then after the first two people have passed the story on, you must assume that it is true and pass it on, leading to a cascade. The problem with this, he says, is that we don’t know anything about the first two people who passed the story on; they may have had bad information. As a result, many untrue stories will be passed around and taken as fact (leading to the reason why he believes Hezbollah won the communications war).

Hansen also talks about herding in the media. With the lack of independent investigative journalism, when a network sees a story other networks are running, they are more inclined to run it too even though they may have done no research on its validity.

Towards the end, a quote from Hansen summarizes the problems with the media and information cascading perfectly: “Amazingly, in this new world of technology, we are getting less information being produced, yet more (false) consensus in the world on what is truth, and therefore worse decisions being made.”

Link: http://robertghansen.blogspot.com/2006/09/decline-of-mainstream-media-and-theory.html

We have spent much of our time in class discussing how the web is an information network. An evolving extension of the world wide web exists called the semantic web, originally conceived by World Wide Web Consortium (W3C) director Tim Berners-Lee. Just like the web, the semantic web is an information network. Unlike the regular web though, the semantic web aims to allow computers to perform similar operations that humans do regularly. Actions such as buying items off eBay, or checking out books from a library currently requires human interaction to be achieved. The semantic web aims to allow a computer to perform similar operations without the need for human direction.

 Anyone familiar with Hypertext Markup Language (HTML), the language used primarily to write web pages, knows that webpages frequently have meta tags at the beginning of the document to specify to the computer the content of the web page. The semantic web takes this idea a step further by publishing data in a Resource Description Format (RDF) that allows computers to understand the web page in similar terms to human perception. Extensible Markup Language (XML) is primarily used to implement this data format on the semantic web.

 In his article Semantic Networks and Social Networks Stephen Downes makes the argument that the information used to represent people or information in RDF should include the use of social network analysis, represented as social meta-data, in order to increase the accuracy of semantic web searches. So if you could represent a person’s social behavior and attributes on the web as data, this information, used in conjunction with the semantic web can result in web searches that will return what the user is most likely looking for. In essence we are searching for a resource based on the contents of the resource and the social properties of the author. Of course the key here is to define the identity of the author and user in such a way that reflects the web content that is related to them. With properly defined identities I could theoretically run a search on articles on a certain subject (say video gaming) authored by people who are similar to me.

 As is easily seen the above discussion is simply a combination of the social networking we see on sites such as Facebook, Flickr and Match.com applied to web searching.

 Source: http://www.downes.ca/post/31624

In class, it was mentioned that in aggregate, people tend to make good decisions. This is because while there is variability amongst individuals, their decisions generally tend to fluctuate around the optimal decision. While this does occur in many situations, there are also other conditions that can cause a group to make poor decisions. This phenomenon has been dubbed “groupthink.” Groupthink is a model of how individuals can behave in groups to come up with poor decisions. While there are several factors that can cause poor decisions to be made, the most central to the model is the concept of group cohesion. The idea is that individuals will highly value the cohesion of the group, possibly more than they value making an optimal decision. As such, an individual may not speak out against the initial leaning of the group for fear of threatening the cohesion of the group. This stifles dissent and causes hastier, poorly thought out decisions. Groupthink is an old idea (dating back to 1972). The article that I read (see below) gives a good description of groupthink: how it works, its predictions, the major empirical work done on it, and the general state of the theory today.

Group decision making can be modeled in a way that we have modeled other decision making processes. For simplicity, the decision should be a binary (“yay” or “nay”) decision. Suppose that, like in our information cascade models, everyone has their own private signal. Also suppose that it is an open discourse, and everyone can either choose to respond “yay” or “nay.” A crucial variable in this particular model will be group cohesion. If group cohesion is high, an individual’s payoff will not be determined on the outcome of the decision, rather his/her payoff will be determined on whether or not cohesion was maintained (and therefore the individual should elect to not dissent from the direction the group is leaning towards). If this condition is met, the individual’s decisions are likely to resemble an information cascade, not because (or not fully because) people infer that the others had good signals, but because the payoff of the group making a good decision is less (to the individual) than maintenance of group cohesion. Therefore, provided group cohesion is high, once an individual perceives group consensus is arising he/she will conform to this decision regardless of his/her personal signal.

The above is of course a simplification, but the overarching point is that groups are fallible. While sometimes groups decision making results in better decisions than an individual, this is qualified by the nature of the group and the individuals comprising the group.

Article: “Groupthink: an examination of theoretical issues, implications, and future research suggestions” by Marc D. Street.

http://sgr.sagepub.com.proxy.library.cornell.edu/cgi/reprint/28/1/72

With the success of the One Laptop Per Child (OLPC) program, interest in one of the key features of the project, wireless mesh networking technologies has dramatically increased. This technologies radically changes norms of network topography and enables increased internet accessibility in inaccessible areas. In a typical end-user network, individuals access the internet via a gateway at their internet service provider (ISP). Their connection to the ISP is direct and there is no connection to other users directly. This typical topography is known as a “Star” network To access the internet then, end users would need a direct connection to the ISP gateway so, for the wireless internet example, they would have to be within the range of a wireless access point. With wireless meshing, internet users are able to connect not only to the gateway wireless access points, but also to other wireless users within signal range. This means that user A need only be close to another user B who has access to a wireless access point to gain access to the internet and that user A need not actually be in range of the gateway himself. This has tremendous applications in rural areas like those designed for the OLPC, however, there are clear disadvantages to the technology.

There are clear advantages and disadvantages to each network topology. Performance is one of these major considerations. In a “Star” network, the Internet Service Provider (which usually ahs a high bandwidth connection to the internet backbone) is only a distance of 1 away from each user. In the mesh network, the ISP may be several hops away and the data would have to pass through many low-bandwidth users to reach the destination. There is also the problem of network routing. Determining the best possible route through the mesh to the internet becomes challenging. Network security also comes into question. On a “Star” network, all data is passed immediately to the, usually trust-worthy, ISP, whereas in a mesh environment data must pass through potentially untrustworthy users along the way. These considerations weigh heavily on the future of wireless mesh technology.

It is clear that, in third-world environments and applications designed for the OLPC, incorporating wireless mesh technologies is a major advantage. Until these technologies can be spread into the mainstream and incorporated into existing infrastructure, certain design limitations must be addressed. Though most of these limitations result from the network topography itself, more sophisticated mesh routing algorithms and increased network size could potentially benefit the success of the technology.

There have been many examples of information cascades discussed in class, including restaurant choices, routes of travel, fashion/fads and rides at an amusement park. One of the more significant debates of the 20^th century and the early part of this century has been on the climate, namely the issue of global warming. This article discusses the media coverage on this debated topic. It refers to the information cascade that has occurred, due to the majority of scientists who support the mainstream explanation of the increased global temperature. The author of the linked column accepts global warming, just not the effects that it has on particular climate changes. The proposition described is that the media provides more emphasis on the evidence supporting the majority explanation of global warming, rather than equitably distributing coverage on alternative reasons.

What we have also learned is that in an information cascade, everyone involved is behaving rationally, but that may not always be the right action to take. As for the everyday person, many of them are not entirely informed, or even adequately informed to make a knowledgeable judgment on the global warming debate. Putting any political affiliation aside, I have found that many realize this and they tend to put their trust in the scientists that are experts in the field. There is also a strong possibility that they are influenced by the alleged media coverage that the article argues about. One of the results of this wide acceptance of the mainstream argument of global warming has produced a beneficial event worldwide. ‘Earth Hour’ as it is called, occurred this weekend in 35 countries, which led to 60 minutes of black-out from major buildings, businesses and homes. Any proponent of any current global warming explanation, or even one who doesn’t believe in it at all, must see this event as a positive result of an information cascade, due to the reduction (though small) of greenhouse gas emissions. This contribution demonstrates how people all over the world can come together and make a difference.

http://www.opiniojuris.org/posts/1199289350.shtml

http://news.yahoo.com/s/afp/20080329/sc_afp/climatewarmingaustraliaearthhour_080329115530