Predicting Popularity
Justin Bieber, Taylor Swift, Lady Gaga, David Guetta, Lil Wayne, John Mayer. Coldplay. What do they all have in common? For starters, they are some of the most prominent music artists today, selling millions of albums worldwide and regularly releasing singles that place on the Billboard Hot 100. What is the secret to their success? This is a question that many in the industry would like to know the answer to, for obvious reasons. If success could be boiled down to a few telling traits, budding bands would know what to strive for, and talent managers would know what to search for. Unfortunately, the answer isn’t quite so simple, as fame seems to be the only characteristic shared by all of these artists. Whether it is in public image, musical genre, age, personality, or style, all of these performers are vastly different. What is it then that makes certain artists and songs international hits, while others never get to see the light of day? The simple answer may just be their popularity.
At first this may hardly seem like an answer. We already know that these artists are popular, what we don’t know is why. Surely there must be a reason why an individual would prefer one song over another: perhaps the song is simply better, or maybe the song fits better with the listener’s tastes. While we can’t deny the importance of quality and the musical tastes of the listener, what we often overlook is how heavily we depend on others in making our decisions. Since we usually look at a song’s popularity in deciding which songs we want to listen to, even small differences in popularity are subject to what is called “cumulative advantage”, known to us more familiarly as “information cascading”, where a song that happens to be slightly more popular than another will tend to become more popular still. As a result, even tiny random fluctuations can compound over time, leading to enormous differences among initially indistinguishable competitors. This means that if history were to somehow repeat itself, even with the same musicians and musical tastes held constant, the names at the beginning of the article might be very different.
Unfortunately, we are not able to turn back time, but thanks to the power of the internet we are able to simulate it on a smaller scale. In a study in which more than 14,000 people participated, participants were asked to listen to, rate, and possibly download songs by bands they had never heard of. Participants were divided into two groups: those who were only able to see the names of songs and bands, and those who were also able to see the number of times a song had been downloaded by previous participants. This second “social influence” group was further divided into eight parallel worlds, where participants could only see the downloads of people in their own respective worlds. All eight worlds started out identically with the same set of songs with zero downloads each, but since they were kept separate they subsequently evolved independently of one another.
If the subjects really did choose music independently of one another, the highest rated songs should also be the most popular in both the independent and social-influence worlds, while drawing roughly the same amount of market share. What resulted, however, was quite different. Popularity in all the social-influence worlds was found to be much more polarized – popular songs were much more popular, while unpopular songs were also much less known. In addition, the individual songs that became hits in each world were different, showing that introducing social influence not only exaggerated hits, but also made them much harder to predict. This is not to say the listener’s own impressions counted for nothing; songs that were ranked highly in the independent condition still stood a better chance of ranking highly in social influence worlds than songs that were judged to be worse. However, the listener’s opinion was easily overruled by what he found to be the general consensus of those who went before him. For example, the song “Lockdown” by 52metro, which was ranked 26th out of 48 in quality, wound up placing first in one social influence world while ranking 40th in another. Overall, a song that was ranked in the top 5 in the independent world only had a 50 percent chance of finishing in the top 5 in the social influence worlds.
The results of this study show us that social influence was just as important in determining a song’s popularity as the inherent quality of the song itself, a conclusion with far-reaching implications. The long run success of a song is essentially determined by the decisions a few early-arriving individuals. The choices made by these first comers are subsequently amplified and cemented in place through information cascades. Since these first comers and their musical tastes are essentially selected at random, no amount of research or algorithms can predict the next big hit more accurately than simply throwing a dart at a list of names.
Cooperation in Prisoner’s Dilemma Situations
In class and on this blog, many real-life examples of Prisoner’s Dilemma situations have come up. These include arms races, performance enhancing drug use, corruption, unstable markets, and female shaving (3). For all of these, it is clear that each player’s dominant strategy is the “defect” choice and, so, the only Nash Equilibrium is “defect-defect”. In the actual Prisoner’s Dilemma, this corresponds to both prisoners confessing.
The option where everyone cooperates is always best, but it is unstable. For example, it would be best if no country possessed weapons of mass destruction. In this situation, though, any given country has a huge incentive to acquire such weapons because it would then be more powerful than any other. Once one country has these weapons, all others become vulnerable and need to arm themselves. So, stability is only achieved when every country has the weapons – the worst option overall.
Let’s now imagine a game of this type that can be played repeatedly and indefinitely – the Iterated Prisoner’s Dilemma. Now, all previous game outcomes are known to both players during each round. The goal for each player is, of course, to maximize the sum of the payoffs from each round. So, is the best strategy as simple as always defecting or accepting? Or is the best strategy more complicated and based on previous information? In the 80s, Robert Axelrod asked this same question and held a competition to find out. Competitors submitted their strategies in the form of computer programs. Then Axelrod ran a round-robin matching of each program with every other. The game that each played had a simple symmetric Prisoner’s Dilemma payoff matrix.
The results (1) clearly showed that a relatively simple algorithm called “Tit for Tat” was the best strategy submitted. The rules for this strategy are as follows: a) on the first round cooperate b) on every subsequent round, do whatever your opponent did on the last round. For an in-depth overview of why Tit for Tat is so successful, see the linked notes (1). There are, however, two quick things we can say about the strategy. First, it promotes cooperation by reciprocating any defect quickly and fairly, as the name (coming from “This for That”) suggests. Secondly, Tit for Tat is “the equivalent of an [Evolutionary Stable Strategy] for prisoner’s dilemma” (2) because it holds up to a wide variety of mutations. There are, however, a small set of strategies that can overtake Tit for Tat (1). What’s important when comparing strategies, though, is how they hold up on average over many different strategies. In this sense, Tit for Tat is the best.
One of the most famous examples of putting Tit for Tat to work was the Mutual Assured Destruction (4) understanding during the Cold War. Every day, both sides (the US and USSR) made the choice to either launch an attack on the other or not. This is an iterated Prisoner’s Dilemma situation because of the payoffs (either superpower would have been better off if the other didn’t exist) and the fact that the game is played over and over again. Now, by treating this as an iterated game and have both sides use Tit for Tat, we can avoid the bleak Nash Equilibrium that the single iteration of the game has. Each nation now knows that a nuclear attack on the other would be quickly and equally reciprocated – Tit for Tat. In this situation, the only rational choice for either nation was (thankfully) to not attack. This shows the power of Tit for Tat in promoting cooperative behavior.
So, while playing a Prisoner’s Dilemma game once has the unfortunate rational outcome of both players defecting, there are actually ways to maintain cooperation when the game is played over and over again. Of these, Tit for Tat is the best known strategy.
Opinion Poll Madness!
It’s no secret that opinion polls get a little out of control around political elections. Sure – depending on who publishes the poll, the accuracy of the sample, and the questions used, poll results can be all over the place. But there seems to be an information effect on polls as well. We saw one such effect arise in problem set 5 (question 7), which showed a huge difference in poll results once votes were publicized in real time; options that already had a lot of votes tended to get even more, as is often the case with anything popular. After all, voting for an unpopular option might lead to a “wasted” vote. Are poll results and media coverage causing an information cascade?
In the past few weeks, the front-runner in the endless stream of GOP debates has changed several times. Following a flurry of transitions between Rick Perry, Michelle Bachmann, and Herman Cain, Newt Gingrich is pegged as the current leader, though he seemed to be all but out of the race until recently. Have things really been that dynamic in the debates that we should expect the front-runner to change on a weekly basis? Or is some other force at work? This article sheds some light on the political factors that might be at play here – I agree that with such a wide field of candidates and no obvious heir to the GOP throne, we should expect a more dynamic debate than usual.
But what about the fact that there have been so many debates, all televised, and all receiving a fair amount of media coverage, and the frequent polls measuring the candidates’ performance? Governor Perry’s blunders easily ring out as some of the most important news stories of the day. Even a pizza mogul’s tax plan has no trouble making its way into the international media, and perhaps even bringing him into the #1 spot in the polls. I think the frequency of the debates, the huge media buzz surrounding each one, and ultimately the polls themselves, seem to skew public opinion so drastically that the front-runner one week may be trailing far behind just a week later.
So what are opinion polls really doing? Do they measure what people think, or are they shaping our thoughts themselves?
Braess’s Paradox in the NBA
Braess’s Paradox states that adding extra capacity to a network when the moving entities selfishly choose their route can reduce the overall efficiency of the network. The reason why has to do with Nash Equilibrium. In Nash Equilibrium, drivers will have no reason to switch routes; however, in a system not in Nash Equilibrium, drivers (assuming everyone is going to do what is in their best interest without consideration for other drivers) will change to the route they believe to be the fastest.
Braess’s Paradox has many parallels to a theory in sports known as “The Ewing Theory.” The Ewing Theory states that a team will perform better if they lose their star player. More specifically, if one player is used significantly more than the other members of the team, the team is not maximizing its potential. When the star player is removed, efficiency goes up. Just as we saw in Seoul, South Korea when a speeding up in traffic occurred as a result of the removal of a major highway. And, just as we saw in 1990 when overall congestion in midtown Manhattan decreased by shutting down 42nd street.
I think the Ewing Theory was somewhat of an anomaly. For example, lets look at Michael Jordan and LeBron James, two first ballot hall of famers. Would the Bulls have won 6 championships without Michael Jordan? Would the Cavaliers have won 60 games each season without LeBron? Absolutely not. When Jordan left the Bulls in 91-92 the team made it only to the second round of the Eastern Conference playoffs. When LeBron left the Cavs to join the Heat, the Cavs won the least games in the NBA.
Facebook and Information Cascades
http://gigaom.com/2011/11/22/six-degrees-what-does-it-mean-to-be-facebook-friends/
http://gigaom.com/2011/11/02/social-media-tipping-points-and-revolutions/
The phenomenal social trend that is Facebook has been continuing to grow, now boasting over 700 million active users. Recently, the company conducted some social research similar to the study that Stanley Milgram performed in the 1960s regarding his theory of the “six degrees of separation”. The new research on Facebook’s user and their connections to each other actually proved that, in many cases, it did not even take those 6 degrees for any one user to be connected to another, no matter what part of the world they are in. In fact, the average “distance” between users was actually only 4.7. And that is a decreasing value – in 2008, it was 5.28.
A trend that researchers have found accompanying this increasing number of users and decreasing distance is the speed at which information can spread. Matthew Ingram, the author of the article analyzing this new research, states “status updates and group memberships and photos and videos [Sociologist Zeynep Tufekci says] can create an “information cascade” that overcomes the typical information vacuum people often experience in such situations.” Tufecki studied the use of Facebook and said information cascades in the context of the Arab Spring uprisings in Tunisia and Egypt and believes that, due to the small distance between users on Facebook, information cascades can happen more easily and in a way that can “actually influence the real world”. Particularly, Tufekci is interested in the use of information cascades via Facebook and how they affect revolutionary efforts. She states “before the internet and the social web came along, people had no way of knowing whether their own dissatisfaction or revolutionary fervor was shared my bothers, apart from a small group that they might know personally… and in order for a movement to break out and become a significant force, the members of that movement have to know that others are also willing to fight — and possibly die — for that cause. Social media, Tufekci says, makes it possible to see this happening in real time, and that helps create momentum.”
Kindle Fire vs. iPad
This article is about how the Kindle Fire is lagging behind the iPad in sales. Several different companies have attempted to come up with products to rival those made by Apple; yet, most products seem to constantly fall short when it comes to sales. Why is this so? It’s certainly not because their features aren’t as developed. Every company has almost the same resources and technologies available to them, but the reason Apple is so successful is because it is a brand name. Apple is like the Gucci or Prada of the technology world. Saying you have a Kindle Fire is like telling somebody you shop at Costco instead of at Burberry. This idea that Apple is an elite brand gives its users a positive payoff, and thus causes a cascade, compelling more and more people to buy Apple products.
I can personally attest to this. The last time I got to choose a new phone, I was torn between the iPhone and the Droid. I actually liked the Droid interface much more than the iPhone’s, and liked several features of other phones that iPhones did not include. However when the time came to make a decision, I ended up picking the iPhone simply because it was a brand name. I will be the first to admit that pulling out my iPhone gives me just the smallest sense of social approval. This will then cause a cascade – after I got an iPhone, my brothers got iPhones as well.
http://www.ohio.com/business/new-kindle-fire-delivers-but-still-disappoints-1.246605
Five Keys for Creating Viral YouTube Titles
http://searchengineland.com/five-keys-for-creating-viral-youtube-titles-37989
There is a lot of information available online referencing SEO, or search engine optimization. Individuals, amateur and professional, are trying to get the marketing edge on their product by simply altering the way they name their videos. The goal of a YouTube advertiser is to get the audience to watch the target video to connect them to the target brand. Google’s AdWords keyword tool allows the user to gage the popularity of certain keywords. Using such techniques, advertisers are capable of working with their titles to get their videos to the top of the search list for specific search queries. The optimization is popular because the optimization requires no additional capital investment by the user unlike buying advertisement space on a Google results page, and can have astounding effects on a YouTube video’s popularity.
Search Engine Optimization, or SEO, is the YouTube user’s attempt to predict YouTube’s page ranking engine to effectively place their product, in this case a viral video, onto the first page of the results page. The first step to optimizing the search for the video is the title. Online tip sites, like this article, recommend the user to determine the frequency of specific keywords being typed into the search queue. The title acts as the keywords that establish the video to the keyword oriented results pages. The more effective titles are those that connect the video to the popular search pages. The higher visit rates the result pages have, and effectively the keyword, the more relative value the video has. The concept here is fractionally similar to the page ranking theory earlier explored in lecture. The relevancy of the video is correlated to how many results pages it can be linked to via simply its title. This concept is strictly isolating the optimization of the name of the video and effectively disregards paid advertisement section.
Expanding from the results page of the YouTube site, videos from YouTube can also show up on the relevant results pages on search engines like Google, Yahoo, and Bing.com. Making titles that are relevant for these search engine results pages is also another mean of marketing the specific video. One specific example in the article was the simple inclusion of the word “video” in the title to target non video oriented search engines significantly boosting the relevancy of the video to the search engine.
Simply by looking at title optimization, there is so much that a user can do to optimize their views. The difference between the next viral video and the video with 1,200 views can be as simple as the word “video”. Boosting relevancy and increasing clicks can be done completely cost free in this case and brings networks into everyday life of the common media consumer.
Why didn’t Google Plus take off?
In this situation I consider two technologies, Google Plus and Facebook. As we learned in class, with two technologies like these there are two possible Nash equilibrium with regards to the adoption of the technology. The equilibrium are attained when people use the same technology. The scenario turns into a game that looks like this:
| Google Plus | ||
| Google Plus | g,g | 0,0 |
| 0,0 | f,f |
Google Plus entered the market of social networking sites over the summer. At the time it first began accepting members from the general public, it was very difficult to get an account. Google claimed that it was not ready to be released to the public and thus was not allowing just anyone to sign up. Instead, getting an account meant having to put in a fair amount of effort. This may have been an attempt by Google to build hype around their new product. Unfortunately for Google, this tactic did not work as within a couple months it became clear that Google Plus was not going to overtake the stronghold that Facebook has on social networking.
We learned in class that in order for a new technology to be adopted, it must be the case that given the percentage of users of the new product p, the payoff of using the new product g, and the payoff of using the old product f, p>(f/(f+g)). In the case of Google Plus, it is clear that either the payoff of using the product was not great enough or the percentage of people using the product was not high enough. Since the product was being released for the first time, there was no percentage of the population using the product yet. It is possible that in order to help spark the original burst of population growth, Google made membership somewhat exclusive. I think this strategy worked at least somewhat because I remember a lot of hype around getting a Google Plus account. However, whether or not this campaign was successful, it is clear that the payoff of using Google Plus was not high enough as it never caught on the way it could have.
At this point, Google Plus no longer has the “fresh” “new” appeal. Thus, in order for the product to be successful it either needs an incredibly awesome marketing campaign, or some highly original functionality that would push its payoff far and beyond that of Facebook.
The Motion Control Cascade
Motion control took the technological and gaming world by storm when the Nintendo Wii was first released in 2006. Initially seen as a gimmick and wasn’t anything revolutionary, the Wii captivated children, adults, and elders with their simplistic remote control approach to their controller with motion control capability. It was to usher in a new wave of people who played video games that were previously turned off by the complexity of Xbox and Playstation controllers. Once families experienced it at their friends’ houses and in stores, they immediately jumped on the bandwagon to also get Wiis, and soon the Wii was the must have item for 4 years.
Since then, the entire industry has been taken over by the idea of motion control with Microsoft introducing Kinect, and Sony introducing PlayStation Move, their own ideas on motion control. The information cascade initiated by Nintendo had all consumers only buying their motion control technology. For example, if a parent went to a store to choose between one video game system or the other, 9/10 parents would find themselves buying a Wii because they had heard of it and seen it in action elsewhere. They made their purchasing decision based on what others had done and rather the information they themselves knew about the product. Nintendo dominated the market for quite some time because of this, before Microsoft or Sony had brought out their own motion control. When there was a decision to be made, the consumers would adopt this new technology from their private information and from others’ choices. And because many people owned it and had online communications, it was a direct benefit to new consumers to purchase the product so their Wii experience could connect them to their friends and family.
Nintendo herded their consumers into the world of motion control and their competitors were forced to follow suit. Though many still perceive motion control as a gimmick, it is apparent that gimmick or not, people will buy the product if it catches on and if its popular. There’s no kind of advertisement like the word of mouth.
http://onlinegamerkid.com/462/how-did-nintendo-wii-become-so-popular
Information Cascade
Information cascade refers to a strange phenomenon where people tend to follow other people’s previous actions and decisions. In contemporary world, where social media like Facebook and Twitter prosper, information cascade plays even more significant roles by affecting larger and larger population. For instance, if a person makes a blog post on a particular product or even political idea, any person who reads about the post is more
likely to buy the product or support the same political ideology, respectively. Such phenomenon is rather rational, just like most people fall for effective advertisements by big companies. However, information cascade can sometimes cause arbitrary and dangerous collective action, like revolution.
According to the article, Riots and Revolutions in the Digital Age, information cascade can explain peoples’ collective economic, social and political actions. The article examines a number of current riots such as Arab Spring and summer riots in London. Information cascade makes revolutions and riots take place more easily because it lowers crime equilibrium, meaning the workers will fear less about the following punishments when they realize there are many other people on the same boat.
Source:
http://www.isn.ethz.ch/isn/Current-Affairs/ISN-Insights/Detail?lng=en&id=133669&contextid734=133669&contextid735=133668&tabid=133668&dynrel=4888caa0-b3db-1461-98b9-e20e7b9c13d4,0c54e3b3-1e9c-be1e-2c24-a6a8c7060233,40db1b50-7439-887d-706e-8ec00590bdb9
keep looking »