Viral Video – Spikes and long tails

| May 7, 2010

In the introduction to the viral video part of this case study, we posed this question: Over how much time must views of a video be spread out if the video is to be considered “viral”?

Over the years, views pile up. A few of ours are now at a couple of thousand views. The videos with the most views tend to have been linked to at some point by a larger site, but plenty of the views build up steadily over the years.

Take a look at these YouTube Insight graphs from our Evil Interfaces and Online Predators videos. They show immense spikes of activity when they were linked to by major sites (BoingBoing, Slashdot, and the EFF’s website).

At some point, Gus realized that the “spike” pattern looked familiar: it looked a lot like what economists and statisticians may refer to as a “power law” graph with what has come to be known as a “long tail.” One of the main places we see graphs like this is in charts of which books, movies, and music are popular. The reason people talk about the “long tail” today is it’s seen as presenting a new business model for these markets. Chris Anderson of Wired has pointed out that Amazon and other online companies today are successful not because they can deliver volume when it comes to popular “hits,” but rather because they can deliver a tremendous breadth of “niche” books, movies, and albums to the small numbers of people who are interested in them.

This raises a question about viral video. In the grand scheme of things, what matters more — a brief, weeklong burst of hits on a video from a major site, or the long tail of the few people a day who find the video and view it over the course of the years it’s online?

Part of the question may be, How do people react to a video when they stumble across it on their own, and it’s not being buzzed about by everyone they know? Is their reception of it different than it would be if everyone in their circle of friends linked to it on Facebook? Do people assume that if everyone is linking to it today, it’s “hot” now and will be played out and boring by next week? If they watch it again a week later, will they therefore not link to it, even if they think their friends will enjoy it?

How does viewers’ reception of a video differ if they reach it through search results — starting with a search they planned out themselves — as opposed to finding it linked to another video on YouTube as they click through out of boredom, as opposed to hearing about it from a friend? All of these are possible ways to find a video. All of these questions remain to be answered; the medium is still quite new.

But what if people never link to, or tweet about, or post the videos they watch? Has a video still gone “viral”? Can the video still be a “success” if people watch it over time and don’t refer their friends to it? Do some people or some communities just not post and share links?

Let’s take a look at the views from the Snopes videos. Out of all the viewing patterns we’ve seen on our show, these are the most remarkable: there are just about no other videos which have been viewed so consistently by so many people over a period of a couple of weeks. In other words, they don’t appear to have a power law pattern.

Almost all of those views come from a single related video (which we’d posted this video as a “response” to, some time before this “mountain” pattern appeared). Why do you think views on this video picked up — and then dropped off again? Is this pattern really different from a “spike”? Could this pattern be re-created, and if so, how?

Finally, make a case for which patterns of traffic you think are more valuable to the show, in the grand scheme of things. Is the spike more valuable, or the tail? Or should we hope for more “mountainous” patterns?