http://seekingalpha.com/article/67381-cpc-advertising-takes-its-toll-on-google?source=financialcontent

In class, we have recently been discussing how Google auctions off its advertising slots. In short, Google uses a General Second Price bid auction (GSP) where the advertisers report some value per click on their advertisement. This value is taken as a bid which then is used to establish prices for the slot. In the end, the advertiser will pay the cumulative price of the price per click multiplied by the click-through rate of that certain slot. The interesting thing about the general second price auction is that even though there will always be a Nash equilibrium among bidders, it is not a dominant strategy for advertisers to bid their true value. Furthermore, even though there may be a Nash equilibrium that maximizes social welfare, this does not always occur. It seems interesting that Google would use a bidding procedure that does not maximize social welfare whereby an auction procedure that did could produce greater revenue. However, there are constraints to consider in Google’s bidding market and it is those constraints that Terence Channon considers in his article “CPC Advertising Takes its Toll on Google.” The model of the General Second Price auction presented in the book had to strip down many of the complexities of Google’s ad market in order to create an understanding of the GSP. Channon reintroduces some of these complexities and how the complex market that Google faces has lowered profits which might make the cost per click auction not a dominant strategy for Google in terms of revenue maximization.

The first challenge Google faces is that the actual companies in the search term have taken actions controlling the advertising market. For example, companies that have trademarks such as Direct TV and Comcast restrict who can advertise on the page when their trademark is used as the search term. These companies are ultimately controlling who can advertise on the page and filtering out major competitors. Such actions have greatly reduced Google’s revenues both in terms of the price per click it can charge and the actual number of clicks that the ads generate. The trademarked companies are reducing the demand for Google’s ad market for that specific search term and reducing the price per click. Furthermore, with fewer relevant advertisers, the number of clicks has reduced greatly. Therefore, such restrictions by the organic search company has generated a double reduction in Google’s revenues. Strategic behavior between parent and subsidiary companies has also had revenue reducing effects for Google. In short, the parent company will tell the subsidiary company not to bid for the top slot. The parent tells the subsidiary that the parent will get the top slot, the subsidiary the second slot and they can share the surplus from the cheaper price per click rate. Though this price ceiling benefits the advertising companies, it is at the cost of Google’s revenue. Finally, Advertising companies are starting their own price per click campaigns which further reduce Google’s supply and revenue.

Another topic brought up in class shortly was the idea of “click fraud” or causing companies to pay large sums of money to Google by clicking on their Ad multiple times. Google has technology to detect fraudulent clicks to protect advertisers and it was posited that 3-15% of clicks are in fact fraudulent. Though the programs installed are not perfect in identifying all fraudulent clicks, even a low percentage of fraud detection could cost Google millions or even billions in revenue because they guarantee reimbursement for detected fraudulent clicks. So, I guess we can’t raise the price of Google’s stocks by clicking more on their Ads and generating more revenue. The costs of any such fraudulent click campaigns goes right back to the advertisers themselves. These complicating issues have led the author to proffer that Google may increase revenue by utilizing a CPA or a cost per acquisition auction rather than a CPC (cost per click auction). It’s interesting to consider if such a program might also lead to bidding your true value to once again become a dominant strategy. This interesting review of Google’s advertising market shows how restrictions on trade and collusion can greatly distort an otherwise fairly efficient market mechanism.

The United States government employed (directly or indirectly) many of the pioneers of game theory, and much of their time was spent considering the distinctly game-theoretic problems the Cold War presented. The Cold War was in some senses a game like any other in game theory, albeit on a much larger scale and with much more immediate consequences for billions of people. In the grandest scheme of things, each side essentially had two strategies: spend or not spend. Undoubtedly, each would have been better off choosing not to commit huge sums of money to defense and intelligence. The unfortunate catch was that each player’s dominant strategy was to spend, irrespective of the opponent’s choice. This led to a profusion of advanced technology and weaponry, but since each side’s forces were roughly balanced against each other there was no net gain from these massive outlays of capital. [More discussion here]

Embedded within the larger game was another game, just as serious as the first. The playing of this game was a cause of great concern for the governments and citizenry of each of the players (not to mention much of the world), and both sought to understand the mechanics and outcomes of it. The nuclear standoff with the U.S.S.R. and the notion of “Mutually Assured Destruction” had all the hallmarks of game-theory. [Lecture on this subject]

Both nations had a policy of commensurate response to a nuclear attack, and to this end each maintained two classes of weapons platforms. The primary ones were the first-strike mechanisms, which were designed to initiate a strike and completely eliminate the target’s ability to retaliate. These were usually hardened missile silos on the territory of one of the powers or its allies. To prevent a first-strike, each country invested heavily in second-strike weapons to be employed afterwards. Owing to the fact that much of the command-chain of the military might not survive the initial bombing, “fail-deadly” systems were created. These served to ensure that isolated military agents be able to retaliate without explicit authorization. Former Defense Secretary Robert McNamara stated that the arms race could be reduced to the objective of preventing the other side from attaining first-strike capability. This desire to present a credible second-strike capability as a first-strike deterrent spurred the development of independently operating nuclear weapons centers. Among these were ballistic missile submarines and continuously-flying bombers tracing routes over preselected targets. These were designed to retain the ability to retaliate even if a significant portion of either country’s defense infrastructure was destroyed, and typically would have several authorization procedures by which they could deploy their weapons. Increasingly, these and other strike-survivable technologies blurred the distinction between what would be considered a first-strike and second-strike platform. [Nuclear Chicken]

A scenario is convenient to illustrate the next point: imagine that that U.S.S.R. had struck the United States, its principal allies and overseas bases with a coordinated atomic attack. The President, deep below Washington in his emergency bunker, now faces a dilemma. His country is destroyed, and no amount of nuclear attrition with the Soviets will undo that. He might ask himself, or his advisors, what incentive does the United States have to respond in kind? Indeed, he might conclude that the best response is no response at all, and elect not to inflict further damage to the planet and the citizenry of the U.S.S.R.

This sort of situation raised another issue that was considered during the Cold War – if one of the sides believed that the other would not reciprocate even if it retained the ability to do so, that side might choose to preemptively strike and end the standoff. To this end, both the United States’ intelligence agencies and the Soviet’s KGB expended some effort to convince the other that retaliation was a certainty (in the context of games, these are known as commitment devices). Some narratives of the conflict detail CIA efforts to represent President Nixon as a demented drunk who would be so out of his head that he wouldn’t hesitate to respond. Correspondingly, attempts were made by the Soviets to represent Soviet Secretary-General Brezhnev as a doddering, senile old man, who likewise would retaliate.

Thankfully, the game was never played to conclusion.

In “What was I thinking?” Elizabeth Kolbert reviews a new book by MIT economist Dan Ariely, “Predictably Irrational: The Hidden Forces That Shape Our Decisions” (Harper, 2008). Kolbert begins her piece by describing a seemly irrational act of purchasing an additional book from Amazon simply to get free shipping. Throughout her review, she describes the history of behavioral economics – a branch of economics that began nearly twenty-five years ago which demonstrated that humans were not simple calculating machines as classical economics would have them.

In class, we covered the Ultimatum Game, a situation in which an individual is given a sum of money to be shared with another player. Traditional economics would predict that the receiver should accept any offer, since even the smallest offering is better than no deal at all. And yet, few individuals are willing to engage in an unfair deal, and are willing to punish their colleague for not behaving fairly.

Is this irrational behavior? It would be better to think of behavior as being predictable or not. If we imagine that for most of our human history we lived in small communities, what may be considered irrational behavior makes a whole lot of sense. One-off games of sharing money simply do not describe the complex social and historical nature of how we live and interact with others. Make a bad deal in a small community and you are likely to be shunned by other potential business partners. Fair trading, as a strategy, becomes a completely predictable behavior.

Instead of looking at humans in the classical economic view of resource maximizers, it may more informative to look at humans as risk minimizers. Minimizing risk does not mean the same as maximizing payoff. For instance, a subsistence farmer may resist planting a different crop (even though it may potentially yield larger payoff) simply because of the risk. A bad year for a subsistence farmer may mean losing his land, moving to a refugee camp, or other highly-undesirable outcomes. I could potentially make more money if I invested my money in higher-risk annuities, and yet, I plan to retire someday and would rather minimize the chance that I may lose my savings over attempting to maximizing earnings.

In sum, economics attempts to model human social behavior, and yet many of the central assumptions about how we behave are not entirely accurate.

The lingering threat of war between two nations is unsettling, especially if both sides understand the consequences of their actions.  A particular example of this is the Cold War, the power struggle between the US and the Soviet Union over nuclear arms, among other things.  In the article(see link at bottom of post), game theory was applied to the handling of the tension between both nations.  During the Nixon presidency, Henry Kissinger believed in applying the madman theory, or Giant Lance, to dissuade the Soviets from using their weapons.  The madman theory postulates that if you convince your opponent of committing an act of insanity, such as dropping an atomic bomb, the opposing side will back out of the conflict.  This approach is similar to the prisoner’s dilemma.  Consider two people chained together by the ankle at the edge of a cliff.  There is considerable risk that one person will want to jump off.  If one person yells to try to prevent the other from jumping, then both are released, and the one who remained silent receives a payoff.  The theory is that if you convince the other person you want to jump, then he or she will yell, in order to save both lives, and be released.

Nixon played the role of “out of control” president, and told his chief of staff, H.R. Haldeman, “I want the North Vietnamese to believe that I’ve reached the point that I might do anything to stop the war.  We’ll just slip the word to them for God’s sake, you know Nixon is obsessed about communism.  We can’t restrain him when he is angry-and he has his hand on the nuclear button.” Nixon’s insanity even lead him to multiple meetings with the Soviet ambassador to the US, Anatoly Dobrynin, who was quite alarmed by the situation.  Dobrynin claimed that military force doesn’t accomplish much good, and that Nixon should reconsider less futile options.

link - http://www.wired.com/politics/security/magazine/16-03/ff_nuclearwar

Google’s Wireless Auction Play

The Federal Communications Commission (FCC) is currently auctioning the rights to build a new wireless network. The current frontrunners in the bidding, which began on January 24th and will end soon, are AT&T and Verizon. Google was previously in contention, but withdrew after its bid of 4.7 billion was surpassed. While the bidding is kept anonymous, the FCC anounces the latest offers several times a day.

The auction has already raised a record 19.4 billion dollars; however, the efficiency of this method has still been brought into question. The FCC has blocked the licenses into various regions and implicated the rule that once a corporation bids for a license, it cannot bid for a larger license in later stages of the auction.  This has caused a lack of competition for some of the larger blocks and an abundance of competition for the smaller ones.  As a result the larger blocks are selling at rockbottom pices while some of the smaller blocks are grossly over priced.  It is estimated that if different rules the licenses could sell for billions more.

 This relates to our studies because  not only is Google (which we are currently studying) involve, but an auction is used to sell the licenses for the new wireless networks.  If no restrictions were placed on this auction then the FCC would undoubtedly make more revenue and social welfare would be increased.  Therefore, because this auction has a restriction it cannot be profit maximizing.

New Technology Protects Internet Advertisers From Click Fraud

In class this week we learned about search engines such as Google and Yahoo which hold auctions for keyword search results and have advertisement placement based on a pay-per-click process. Google says that “You’re charged only if someone clicks your ad, not when your ad is displayed”. Google charges about $2.66 per click if you’re one of the top three advertisements on their site. What would stop a person from going on theses website and clicking on a competitors advertisements, driving up the number of hits? This would charge the company being advertised more money hurting the company. This also alters the keyword search auction hurting the search engines. This crime is called “click fraud”. What would stop companies from tainting online auctions or running false auctions? For example what would happen if someone hacked a server while an online auction was in progress? They could change the winner, price and placement of an advertisement.

There is click fraud prevention technology in the works for websites such as Google and Yahoo to keep click count accurate. Yong Guan an electrical and computer engineer from Iowa State “is developing technologies to track down cyber criminals and make Web transactions more secure.” Guan is focused on such problems as network threats, click frauds and auction frauds. He plans to attack this problem by reinforcing the wireless infrastructure and beefing-up privacy protection. For the wireless infrastructure Guan plans to limit the access to networks by the use of managed accounts; by granting permission to access information. Some search engines already use an IP address as an indicator to eliminate multiple clicks. However hackers have already been known to network many computers together to circumvent this safety precaution through the use of viruses. That is why a managed account would fix this problem basing the click on the person signed in rather than the computer location or IP address. Also, for the keyword auction people could spam the internet with multiple copies of the same webpage moving there site up on the list. Filters which could scan sites and eliminate duplicates could solve this problem. These safety precautions once in place will help search engines and companies for a short while, however overtime hackers will find away to outwit these measures as the internet expands.

Watching Monty Python the other day, I found my thoughts wandering once again back to Networks 204 and Gladwell’s description of a syphilis epidemic in Baltimore: how it spread from the projects along local highways during summer months and contracted during the winter. Explicitly drawing a connection between contagious viral diseases and transportation networks is essential to understanding the ideas behind the book, especially given the current era of globalization. Six-degrees of separation and small world phenomena is only impressive because of the context of the network population’s massive and ever growing size (6 billion plus people), increasingly connected by aviation, marine, rail and motor-vehicle transport infrastructure.

A recent academic article from the School of Informatics and Center for Biocomplexity at Indiana University developed statistics referring to the spread of contagious diseases like the flu as it passed through links (i.e. flights) to various nodes (i.e. airports). Transportation networks not only diffuse sickness, but also concentrate various people onto a small area, such as airports, seaports, train stations, subway cars, ferries, and so on. This gives the virus ample time and proximity to spread from the host carrier to victims, particularly given the unhygienic nature of these facilities. The interconnectivity not only allows people and (animal) products to move far and quickly, but also transit any viruses that they may be carrying with them.

Natural functions such as migratory patterns are also attributable to the spread of diseases, like the Avian Flu. The Siberian-Caspian-Nile flyway, used as a superhighway by millions of birds, is one such suspected culprit, though others argue that viruses like the Avian Flu will more readily spread through the human engineered transportation network. After all, one would expect diffusion concentrated mainly along the path of the flyway, though empirical evidence has shown otherwise. In other words, the Tipping Point has not (yet) been activated. However, if the disease spread through migratory patterns theory was realized, places like Alaska, where New and Old World birds meet, would create an “international viral transfer center.” Alaska could then be interpreted as a gatekeeper role—one that could be culled in order to stop the spread of a disease by re-dividing (or quarantining) the networks.

Diseases provide interesting network implications in other ways, like power exchange theory and link analysis. Those who are at the center of a network, with more links embedding them in the overall structure, are generally considered to have more power over other nodes. Conversely, they become extremely weak when bombarded with many virus-carrying links during a flu pandemic. On the other hand, those who are at the edges of a network appear less vulnerable through their own self-quarantining habits. Additionally, there is one remaining twist in that those who occupy centralized nodes on a network may become immune to viral diseases due to their proximity and constant contact. During the discovery of the New World, Native Americans were especially vulnerable to European diseases because they were now the outside edges of a large Euro-centric network, whose occupants had millennia to build genetic immunities.

In any case, I’m not dead yet.

See http://www.checkit.nl/pdf/eyetracking_research.pdf 

The emerging field of analyzing internet search behavior is important in many disciplines, but it becomes especially relevant for advertisers wishing to sell products to consumers.  In constructing websites that would potentially be attractive to consumers, one may assume that advertisers would seek to maximize the directness and ease of finding information while still presenting information in a convincing and manner.  Most consumers begin by utilizing search engines to navigate and narrow the options of purchasing goods.  As mentioned in class, search engines themselves offer a noteworthy scheme as to how consumers may evaluate the prospective advertisers of a good.  Thus the process of exploiting the results of the search algorithm, with consideration of how consumers appraise these results themselves, is the topic of a new study that may shed light on how advertisers may increase their attention from consumers.

This study, conducted by the search engine media agency Checkit and the market research agency De Vos & Jansen, sought to map how one perceives search results on a website, and to score the characteristics that draw the eye to a particular result.  The researchers monitored the viewing patterns of internet searchers using an infrared camera system.  Each of the searchers was assigned to search for various items, behaving as would a consumer.  After analyzing the pattern of eye movements across a search results page, the researchers found that those consumers who were looking to purchase a particular item viewed more search results than those consumers who were merely looking to find out more information about their product.  Moreover they found that the average viewing time for a particular result, regardless of whether a consumer was looking to make a purchase or for information, lasted for about 1.1 seconds.  Thus it is clear that advertiser have very little time to make an impression on consumers, and these impressions must be imparted at the level of the search results page in order to be effective.

While the implications of this study are understandable, that advertisers have a short time to captivate consumers, this approach to consumer behavior is important because it begins to take into account how we think and process information.  Indeed the notion that advertisers may begin to tailor their offerings according to the consumers needs lends itself to the idea that the process of searching and deciding on a product is an ever-streamlining experience.  Furthermore the best way for one to design a search result description would be to include precisely the concise information that will relay that the product is adequate to fit the consumer’s purposes (the informational aspect) and that it stands up well to competitor’s products (the transaction aspect).

http://news.bbc.co.uk/1/hi/technology/7153637.stm

Social networking websites such as Facebook and MySpace make 20 millions dollars a month selling advertising space on their website. Yet, Facebook’s 50 million members do not benefit from any of the company’s earnings. Now, there is a social networking website, called Yuwie, that shares more than half of its advertising revenue with its members. Yuwie was launched in July, 2007 and already has over a half a million members. It is completely free to join. You earn money when people view any of your pages within the site, a pay-per-click payment scheme. Also, if you invite other people to join and they mention you as their referral, you earn a percentage of their earnings.

Here is how earning money through Yuwie works. If you refer some friends, who refer some more friends and you take this 10 levels down, you get a certain percentage of their earnings. Here are the percentages that you would earn:

Level          Your %

You             10%

Level 1       10%

Level 2       10%

Level 3         4%

Level 4         4%

Level 5         4%

Level 6         4%

Level 7         4%

Level 8       10%

Level 9       10%

Level 10     30%

These values are the percentages of the RSR (revenue sharing rate) that you would earn at each level. Lets assume that the month’s RSR is $0.50 per 1000 clicks. (The RSR is variable from month to month.) This would imply that you would earn 10% of that, or $0.05 for every 1000 clicks. But, you also earn a percentage of the RSR on your referrals’ clicks. Therefore, if you had a full referral tree of just three people referring three people all the way down 10 levels, you would have over 88,000 people in your referral tree and you could have an earning potential of over $10,000 for that month.

Levels     Referrals     Earnings

1               3                   $0.15

2               9                   $0.45

3               27                 $0.54

4               81                 $1.62

5               243               $4.86

6               729               $14.58

7               2,187            $43.74

8               6,561           $328

9               19,683         $984

10             59,049        $8,857

Totals       88,572        $10,235

To learn more about Yuwie in greater detail, you can watch this introduction video here: http://ww3.yuwie.com/yuwie.asp?r=353535

I thought the concept of earning money for social networking was a novel and ingenious idea. We mentioned in class that websites make a killing on advertising, yet none of it is shared with the users of the web pages. Millions of people spend a lot of time on their Facebook profiles and writing blogs and only ever get self-satisfaction out of it. It is a very lucrative business to charge advertisers for space, especially on a social networking website that gets millions of hits per day. If the website owners are willing to share the wealth, they gain additional profit from the surge in popularity, and everybody wins.

http://domino.research.ibm.com/comm/research_projects.nsf/pages/kittyhawk.index.html/$FILE/Kittyhawk%20OSR%2008.pdf

Once upon a time there was a computer. Then there were many. Eventually it seemed like a good idea to hook them up. And so the Internet was born. So far we’ve considered the growth of the Internet in close relationship with the number of computers connected to it. This may all be about to change. On Feburary, 2008 IBM announced a new research endeavor: “to explore the construction and implications of a global-scale shared computer capable of hosting the entire Internet as an application.” According to the white paper linked above this monster will, at hardware limit, house 67.1 million cores running on 32 Petabytes (~ 32,000 Gigabytes) of memory.  IBM hopes this undertaking will change the way the Internet currently exists, from poorly planned clusters of commodity computers to a more efficient SMP (Symmetric MultiProcessing) design.

I will, however, steer away from the technical complications that Kittyhawk must overcome to achieve it’s ultimate goal, and analyze a world where Kittyhawk does exist, where the entirety of the Internet is being run off a single computer. Some thoughts and observations. IBM headquarters would require security rivaling (maybe exceeding) nuclear silos. Moreover, a centralized implementation of the Internet puts redundancy measures into question: what if Kittyhawk fails? Granted it will probably have an innumerable number of fail safes, but reality still dictates for some chance of failure. IBM would probably, eventually stand trial for monopoly (particularly from the European Union). But ultimately we have lost the unending scalability that so characterized the Internet in the first place. Internet growth would now happen in steps, as Kittyhawk2 and it’s successors make it out of the lab. There is undoubtedly much to gain from this research for the sake knowledge itself. But if the decision to implement this replacement is ever on the table, would it be worth it?