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.