The Science Behind What Determines Popularity

Can popularity be manipulated?

Happy Thursday, y’all! So today I wanted to talk about a fascinating research experiment into popularity that I came across a few years ago via NPR. Princeton professor Matthew Salganik attempted to learn why works of art become popular. In short, Salganik wanted to learn more about what factors determine if a piece of art is popular. He wanted to focus on two areas: The ‘quality’ of the art, the role (if any) that social influence played. Basically, he wanted to know if art was popular because it was good, or simply because other people liked it.

To conduct his experiment, Salganik decided to focus on music. He took 48 unknown songs from 48 unknown bands. Salganik purposely picked songs that he knew had not been heard by the people he would be surveying. This way, the people being surveyed would not have heard the songs before or even the artist. So they would have no preconceived notion of how popular or unpopular the songs would be. At least in theory.

Salganik surveyed 30,000 teenagers. What Salganik did was divide these teenagers into one of two groups;  The first group would listen to the music and then rate the songs from one to five stars.  Then after listening to the songs and rating them, the person would then have the opportunity to download the song for free.  This was the ‘independent’ group.

The second group was called the ‘social’ group, and it was divided into eight smaller groups.  Each person in each of these eight groups follows the same process as the independent group.  They listen to the songs, then rate them and finally are presented with the option to download the song for free, or not.  The big difference is that with the social group, every member can see how many times every song has been downloaded by members of their group.  In short, they can see which songs are popular within their group and which ones are not, and they have this information available to them before they rate each song.  But they are only able to see the popularity of the songs within their group (of eight groups within the larger social group).  They can’t see the popularity levels for the songs in the other 7 groups of the larger social group.  Also, in some cases the songs are ordered based on popularity (most popular listed first) and in other groups the popularity of each song is shown, but the list isn’t sorted by popularity.

What Salganik found was that when participants were made aware of the popularity of the songs (but the songs were not sorted based on popularity) that the more popular songs were rated more highly.  When the songs were actually sorted according to popularity, this affect was magnified.  So the popular songs became much more popular and the songs that were lower ranked became even less popular.

Salganik appeared at the Thought Leader Forum in 2011 and explained in more detail some of his findings from this study:

There’s this idea that the more people can see what other people are doing, the more they’re going to find the best thing. But in fact, what we see is that when people can see what other people are doing, they start following people, who are actually following other people who are following other people. And this process of following can become decoupled from the underlying reality.

To give a concrete example from these experiments, there is one song, “Lockdown” by 52 Metro, again a song no one has heard of by a band no one has heard of. In one world, this song came in first. It was the most downloaded
song. In another world, this exact same song came in 40th out of 48. This exact same song competing against the exact same other songs.

But you can see to the extent that when we have these kinds of feedback processes, when people are following what other people are doing, slight initial fluctuations at the beginning can become locked in, and then that leads to
very different outcomes, even for the exact same song.

Isn’t that fascinating?  All of this points to a fundamental truth: We as human beings gravitate to that which other human beings have identified as being ‘popular’.  We trust each other and seek out input when we are choosing, especially when given a wide variety to choose from, as the participants in Salganik’s study were given.

And this works on us, as well. Case in point: I was scrolling Notes here on Substack yesterday, and I noticed like 3 updates in a row were from people I follow, who were all sharing content from another Substack user. My immediate thought was ‘Whoa, who is this? Everyone is talking about them, I need to check them out!” So I immediately clicked their profile and saw who it was. All because I saw 3 updates in a row from people I am following, who mentioned this same person.

And ultimately, the algorithm here at Substack determined I would see those updates in that order. If they had been spread out, I may not have noticed that all 3 were about the same person.

I think this also speaks to another fundamental truth: We are simply lazy. We have become used to letting the algorithms do the work of finding interesting content for us.

For reference, I found the leaderboard for Business substacks here. These are supposedly the top Business substacks based on number of paid subscribers, I believe. I had never seen 90% of them shared in any of my feeds.

What does this mean for you?

The obvious takeaway is that we want to highlight when others promote us, our products, writings, whatever. But perhaps more than that, we want to see more customers/readers/clients promoting us organically. We want to see more promotion happening ‘in the wild’.

This is one reason why I talked last week about positioning your offerings or products so that they create a benefit for others. I bought your e-course, and you taught me how to do X. Well guess what? I am going to tell others about that! You don’t have to encourage me to promote the fact that you taught me how to do X, I will WANT to tell others!

The reality is, it’s easier to encourage people to engage in existing behavior than it is to ask people to change their behavior. For example, if someone is already promoting you, it’s easier to get them to continue to engage in that behavior, than it is to ask a customer who has never promoted you, to do so for the first time.

I hope you found that study as interesting as I did! Please have a great weekend, and I will see you again on Monday!

Mack