5
time, and definitely not guaranteed, even for the best performers. Tastes are fickle, and herd
behavior often takes over.
Even the experts, with much at stake, have difficulty picking winners. Columbia Records turned
down Elvis Presley in 1955 and the Beatles in 1963. They turned down Bob Dylan in 1963, and
almost rejected “Like a Rolling Stone” in 1965, which was later named the greatest rock ‘n roll
song ever by Rolling Stone magazine.
8
Or consider Sixto Rodriguez, the subject of the documentary movie Searching for Sugar Man.
Rodriguez recorded two-and-a-half albums from 1970 to 1975, which were commercial flops.
But he was a huge success in South Africa, and his music became the battle hymn of the anti-
Apartheid movement. And – amazingly – he was unaware of his fame and influence.
Both good and bad luck play a huge role in the rock ‘n roll industry. And the impact of luck is
amplified in a superstar economy.
This was clearly demonstrated in a fascinating experiment conducted by the sociologists Matt
Salganik and Duncan Watts.
9
With the musicians’ permission, the researchers posted 48 songs
in an online music library. Subjects were invited to log on to the library and sample the songs,
with the opportunity to download the songs for free (slide 4). Participants could see the list of
songs, ranked by the number of times each one had been downloaded up to that point. They
could also see the exact download counts, so they were aware of the popularity of each song
based on the collective opinions of other participants (slide 5). From there, the subjects could
click on a song to play it, and then were given the option to download the song for free (slide 6).
For the first 750 participants, the researchers faithfully tallied and displayed the number of
downloads. However, the subsequent 6,000 participants were randomly – and unknowingly –
assigned to one of two alternative universes. In one universe, they continued to see the true
download counts (slide 7).
For the other participants, the researchers deviously created an alternative universe where the
download counts had been flipped, so that the 48
th
song was listed as the most popular song, the
47
th
song was listed as the number two song, and so on.
10
After this one-time inversion in the
ranking, the researchers let the download tallies grow on their own. They wanted to see if the
cream would rise to the top, or if the boost in ranking that the worst song received would lead it
to become popular.
Here’s what happened in the world where the download counts were presented accurately (slide
8). By the end of the experiment, the top song (“She Said” by Parker Theory) had been
downloaded over 500 times, while the least popular song (“Florence” by Post Break Tragedy)
8
“The Hit We Almost Missed,” The New York Times, December 3, 2004.
9
See Matthew Salganik and Duncan Watts, “Leading the Herd Astray: An Experimental Study of Self-Fulfilling
Prophecies in an Artificial Cultural Market,” Social Psychology Quarterly, 2008, Vol. 71, No. 4: 338-355. My
description of their work combines two of their alternative universes to provide a simplified overview of their
experiment, but is faithful to their main findings.
10
All participants provided informed consent before taking part in the experiment.