This post is inspired by my Spring Break trip to Europe. Anyone who flew out of the northeast the Friday before break probably experienced the same headaches I did due to canceled flights, long delays, and generally unhappy and unhelpful airline service reps. I figured that storms were a common problem for airlines and was surprised that they didn’t have a better plan for handling the backups and delays. It made me think of the network infrastructure of airlines. Here is a study by the National Academy of Sciences on the worldwide air transportation network, its structure, and the roles of different types of cities.

To briefly summarize their findings, they saw that this network is a small-world network and “the most-connected cities are not necessarily the most ‘central’… this surprising result can be explained by the existence of several distinct communities.” These communities seemed to be based on a few factors, namely population density, economic development, and state and geographical borders. A few ideas stick out in this paper as being relevant to our coursework. For instance, the paper discusses the clustering coefficient, “the probability that two cities that are directly connected to a third city also are directly connected to each other.” This is similar to the phenomenon of social closure. They also come to the conclusion that connectedness and centrality do not necessarily imply one another by analyzing the betweenness of cities in the network and noticing that cities with low degree (not hubs) sometimes have large betweenness values. “Although the most connected cities are located mostly in Western Europe and North America, the most central cities are distributed uniformly across all of the continents.”

This could explain the huge delays that occur when a site that connects two major communities shuts down. These sites are usually very central and without them the entire network experiences delays costing enormous amounts of money each year. This article is more pertinent to the material from earlier in the course, but it still is a good example of an application of the material we’ve studied.

Today, professor Easley talked about information cascade, which was really interesting to me because sometimes I just followed my friends or other groups of people even though it was not exactly what I wanted to do. After the class, I was looking for other examples of ‘information cascade’ and I found many related topics to it. Since they were not covered in the class, I want to introduce some of them to you. What I found interesting is: bandwagon effect and groupthink. (there is many other related topics, but I only chose two.)

1) Bandwagon effect: 

The bandwagon effect is the observation that people often do (or believe) things because many other people do (or believe) the same. The effect is often pejoratively referred to as herding instinct, particularly as applied to adolescents. Without examining the merits of the particular thing, people tend to “follow the crowd”. The bandwagon effect is the reason for the bandwagon fallacy’s success. (from wikipedia http://en.wikipedia.org/wiki/Bandwagon_effect)

How ‘following the crowd’ can be a problem? Well, as some of you learn the theory of supply and demand in microeconomics course - I learned it in ECON 101. The theory assumes consumers buy products based on their prefrences and they decide to buy or not based on the price. However, if people’s decisions depend on others’ decision, the story is different. For example, suppose my value of some book is originally $50 and its price is $55. Since I would be better off not buying it, I do not want to buy it. However, after a few days, I get to know that many people around me have already bought the book and they like it. Moreover, I hear that some people who have not bought yet are going to buy it. Then, I would probably buy it with this price before it increases. If I did not get the information that many people bought it, I would not buy it. The crowd’s behavior affects my decision.

2) Groupthink:

Groupthink is a type of thought exhibited by group members who try to minimize conflict and reach consensus without critically testing, analyzing, and evaluating ideas. Groupthink may cause groups to make hasty, irrational decisions, where individual doubts are set aside, for fear of upsetting the group’s balance. The term is usually used as a derogatory term after the results of a bad decision. (from wikipedia http://en.wikipedia.org/wiki/Groupthink)

I’ve experienced how grouthink can make a serious problem. Two years ago, I had a sight-seeing trip to Mo-hang (in Korea) with some friends of mine. It was our first time to have a trip without parents so that we were so excited. We did not discuss how much money we would need to cover travel expenses; we thought ‘a hundred of thousand won’ would be fine. Even though it was enough money, we bought a lot of alcohols right after we arrived at the inn. Everyone was pretty sure we could go back home because we remembered the price of but tickets from Seoul to Mo-hang. Few of us were thinking about making ‘a nest egg.’ We thought saying any kind of concerns showed one was scared or not brave. Anyway, We used all money except for some for tickets. Next day, the owner of the inn told us that he could not provide a ride to the terminal because the car had a problem. We got nervous because if we missed the bus, which would leave in a couple of hours, we could not go back home and even we did not have money to stay longer or to ride a cab. It was the outcome of our groupthink. Fortunately, we could get a ride from other groups of people in the same inn to the terminal, but we could have not been able to return home safely. It was a really horrible experience for me.

If you read the information on wiki, you can find out that there is many ’serious’ problems caused by groupthink such as Space Shuttle Challenger disaster.

 There is many other related topics, but these two are easiest to understand and analyze for me; it seems that it happens everyday! I believe that to know about bandwagon effects and groupthink could help you not easily give up your own opinion or change your decision. You sometimes need to think one more time before you decide based on other’s decision or behavior. In that way, you could avoid some serious problems.

New York Poised to Join Move to Earlier Primary

Today’s discussion on information cascades brought to mind a growing trend in the upcoming 2008 election. Presidential nominees in both major political parties are traditionally decided during the primary season, and for the past few decades, the first major proving grounds for a candidate’s success (or lack thereof) have been the Iowa caucus and the New Hampshire primary.

However, one of the major criticisms of this process is that states with later primaries lose their relevance. By the time those primaries occur, the bandwagon effect has already taken hold: most of the candidates have dropped out and both parties’ remaining delegates and resources are devoted to the front-runner.

The article above focuses on New York’s motion to change its primary date to February 5, joining about 20 other states that have done the same or are considering doing so. An article examining the new development also appeared in the Boston Globe (requires subscription):

The simplest explanation for electoral momentum is that people like to vote for winners. “There are a lot of people out there who don’t follow politics all the time,” says Michael McDonald, a political scientist at George Mason University. When a candidate wins an early primary, “that’s a signal that this might be a candidate you need to look at.”

…

A study by Stephen Farnsworth of the University of Mary Washington and S. Robert Lichter of George Mason University’s Center for Media and Public Affairs showed a sharp shift in the tone of the network news coverage of Dean and Kerry after the 2004 Iowa caucuses and New Hampshire primaries: Newscasts dealing with Dean began to fill up with negative statements about him, and Kerry coverage became distinctly more positive.

And then there’s the money. Donors, perhaps even more than voters, don’t want to back a loser. It’s a more recent phenomenon, says Mayer, “but one of the most powerful things you get out of doing well in the early going is a big infusion of funds.”

As of now, it is unclear what effect this will have on the primary season. On one hand, it could counter the information cascade scenario, in which the results from relatively few primaries decide the course of both political parties. Because of the potentially large number of states that will hold simultaneous primaries, a mixed result could keep multiple candidates in the race, many of whom would otherwise withdraw. On the other hand, and for the same reason, a strong showing by any one candidate would make that Feb. 5 one of the most decisive dates of the 2008 campaign, and the primary season would be shorter than ever.

Summary | Full Article (JSTOR)

This article shows how Information Cascades can be demonstrated in a very controlled setting. The example is in a classroom with 3 urns, 3 blue marbles and 3 red marbles. The teacher puts 2 blue marbles and 1 red marble in the first urn (urn A) and 1 blue marble and 2 red marbles in the second urn (urn B). Then the teacher randomly selects one of the first 2 urns (by flip of a coin, for instance) and dumps it into the third urn without the students knowing which urn was chosen. Students are then allowed, one by one, to draw a marble from the third urn with replacement, and then predict which of the urns A or B was chosen and announce it to the rest of the class. Though the students knew the contents of urns A and B, they don’t know which urn was chosen to be dumped into the urn they are choosing from. Nor do the other students in the class know what color the marble is that each student draws, they only know the prediction that student makes.

After following these rules, it tends to be the case that Information Cascading occurs and after the first 2 students, and the subsequent predictions tend to agree with the initial predictions regardless of the color of the marble drawn. Without information sharing (students don’t announce their predictions), this result would be surprising, as if the student picks a blue marble it is twice as likely that the chosen urn was urn A. With information sharing (students do announce their predictions), Information Cascading begins to come into play when additional predictions add no new information to the subsequent drawers. For example, if the first 2 predictions are for urn A, no matter which color of marble the third student draws, he will choose urn A because the first 2 predictions will outweigh the new information from drawing the marble. Therefore, subsequent students do not gain any new information because they cannot infer whether the third student drew a red or blue marble. Information Cascading does not always arise. According to the article, cascades happen only about 80% of the time.

While Google has successfully made a media empire selling text ads, Adwords certainly is not the final word on network-based advertisement delivery innovation. Their auction system provides a price based on pseudo-semantic value, rather than the value of the audience viewing the ad. Expanding upon the earlier post on social capital, the web could provide media vendors with a way to price their ads not solely based on the demand for the specific outlet (or classification of venue) by advertisers, but by the value of the actual audience members. Never before has media been so freely distributed and users so easily tracked that this hypothetical situation could exist.

Since the cost of media delivery is so low, the value of publication is overshadowed by the value in garnering a user’s attention. The attention economy is a framework from which we can value this attention as a basic commodity within social media. Seth Goldstein, an entrepreneur behind AttentionTrust, believes that this metadata, which can take the form of a user’s clickstream history is not only valuable, but belongs to the user.

However, putting aside the issue of metadata ownership, AttentionTrust allows for the data to be shared with other web applications and web services. For instance, one company, BuzzLogic, allows clients to track “influence” in the world of social media. Similarly to theoretical valuations of social capital, BuzzLogic claims:

“The BuzzLogic system uses data gathered by its monitoring tools to deliver its social maps. Based on criteria determined by marketers themselves, BuzzLogic social maps chart the relevant relationships that evolve during blogging conversations to show influence and how it changes over time.”

As both clickstream and social relation data is accumulated for many users on the web, it is feasible to improve systems such as Adwords with far more advanced methods for advertisement valuations, based on the very ideals which make collaborative media valuable to the user.

Having made it to the final round of interviews with Google last semester for the position of Online Sales and Operations Coordinator, I have found the past several lectures on keyword-based advertising particularly interesting.  Before flying to the offices in Mountain View, Ca I tried to get thoroughly acquainted with AdWords program.  That was when I discovered Google’s “AdWords Learning Center“, a lessons catalog that anyone with questions about the AdWords program can reference.  

Basic Definitions introduces users to some of the terms that have come up again and again in class (e.g., “Cost-per-click (CPC)” , “Clickthrough Rate (CTR)”, “Quality Score (QS)” ) and seeks to clarify the relationship between them.  The class handout on keyword based advertising assumes that the position of the ad is the primary determining factor in CTR, and since CTR is highest for the ads at the top of the search page, these ads will be the most expensive to place (Price= CPC*CTR).  It must be noted however that in practice the price is more complicated that this to calculate, since Google places heavy emphasis on Quality Score.  As noted on the definitions page, QS is determined largely by CTR and the relevance of the ad text; the higher the CTR and the more relevant the text, the lower the minimum bid and price  per click.  It is counterintuitive that a higher CTR could result in a lower price, yet it makes sense: a higher CTR means more people are clicking to view the ad. Since advertisers pay Google on a Cost-Per-Click basis, this means increased revenue for the company.  At the same time it creates synergies for the advertisers, since a higher CTR means that more potential customers are choosing to view the ad.  

Although the process of selling AdWords slots is highly automated, Google also has a large army of AdWords Coordinators, who work with advertisers to help them meet their advertising goals by providing excellent customer service,  reviewing ads and keywords for quality and accuracy, troubleshooting, and optimizing advertiser keyword lists and ads in order to maximize their return on investment.  If you are curious, check out this page of Google Jobs to see the positions for new graduates.  The interviewing process itself was quite an experience!

Google’s New Frontier: Print Ads from BusinessWeek.com

Last November Google, the renowned profiteer of Internet keyword-based ad search, launched a new advertising initiative for printed ads. In this test project, Google would allow advertisers to bid on ad space from many of the top newspapers in the country, including the New York Times, the Washington Post, and the Chicago Tribune. Although the program is very similar to Google’s keyword-based ad search, it differs in that it allows the newspapers to set minimum prices and Google itself does not actually purchase any space; rather, it just provides the tools for advertisers to directly communicate with the newspaper. Many advertisers, such as eHealth, values this program in its ability to provide ROI (return on investment) metrics, which means advertisers can track how much profit is gained or lost on the amount of money intially invested into the advertisement (an extension of Google AdWords technology, which allowed advertisers to see exactly how many clicks a particular ad is receiving). The enormousness of Google’s advertiser network is hugely appealing to newspapers although some clients on both sides worry that Google will eventually consume and dominate the print-advertisement business and leave very little of the bidding to newspapers and advertisers.

Consequently, it seems that one of the key features of Google advertisement sales is its ability to return large amounts of information back to the advertisers about their investments in different advertisements. This is vastly important in the long run, as both the NYT and the Seattle Times have announced (as of March 12) huge revenue gains in advertisement space ever since the launching of the Print Ads program test. This could become a huge resource for Google to tap into, since newspaper advertisement generally still produces almost twice as much revenue as online ads. However, I feel like one of the main features of online keyword-based advertisement is missing, and that is the fact that it is keyword-based. Because online ads are geared specifically, using Google’s search algorithms, towards consumers that have blatantly expressed interest in that direction, there is a much higher chance that an ad returned based on a query will be relevant to the customer’s needs. Print ads could not possibly aimed as specifically as online ads and arguably would not produce as high rate of return as online ads. However, I imagine Google’s expansion into smaller and more specialized journals and magazines would potentially offset the disadvantage of generality.

See also Can Google Go Glossy? by BusinessWeek.com written at the end of 2005 after some of Google’s earlier attempts at print-based ads.

Article Link

Danny Hillis and his start-up company, Metaweb Technologies, are trying to change the Internet. Hillis is trying to create a new way that the world’s digital information is stored. In his vision, the internet would be a public centralized database that would allow computers to automate many of the services that are now performed manually via web browsers. The database, called Freebase, would store its information in such a way that computers would be able to make relationships between the information.

Freebase and the other efforts of Hillis and Metaweb Technologies focuses in on the same question that was brought up in class: exactly what is the size and scope of the internet? Furthermore, it begs to examine how information is able to stored despite an ever-changing digital world that is increasingly dynamic. As the NY Times article points out, “In its ambitions, Freebase has some similarities to Google — which has asserted that its mission is to organize the world’s information and make it universally accessible and useful. But its approach sets it apart.”

Critically analyzing Freebase using our discussion from class of the size and scope of the web, I am left with key concerns that cast shadows of doubts on Freebase. I am curious to know exactly how Freebase will expand and handle dynamic features. I want to know what Freebase will do with the “dark matter” of the internet that we’ve estimated to store enormous quantities of sensitive information. If Freebase is able to address these and other concerns, then maybe this metaweb approach to storing digital information can work. But would it be worth it? In other words, Freebase features would really have to dazzle users to switch from using the web as they know it.

http://www.google.com/ads/georgewashington.html

The George Washington University Medical Faculty Associates is a health care practice in Washington D.C. One of the more popular services is their fertility care.  The media and marketing manager, David Garofalo, had previously used print and radio advertising to get perspective clients. Specifically he searches for potential egg donors. With print and radio advertising he got about two or three inquiries per month.  Garofalo then decided to use Google AdWords as his main outlet and began to get three or four inquiries per day. This 2000% increase helps the clinic because it gives a deeper pool of egg donors to choose from making it easier and faster to pick a donor. Garofalo believes that AdWords gives him more control over his advertising budget because he can see which keywords are generating the most hits and which ones aren’t. He is able to change his ads at any time unlike print ads which run for a set period of time. 

This article describes an instance of using keyword based advertising. Google AdWords prides itself on being a cost-effective way to advertise. This is because it uses the “cost-per-click” method, giving advertisers broad options of keywords and then being able to pick the most popular ones. Using a form of Overture’s method (the “kind of” second price auction), Google AdWords sells keywords, charges by number of clicks and sells multiple slots.