A Pseudoscientific Digression on Internet Virality

“The publication in this issue of the research paper…marks the end of more than 8 months of widely reported controversy over whether some of the data now freely accessible should be withheld in the public interest. As a result, people worldwide are now much more aware of the potential threat that this virus, commonly known as “bird flu,” poses to humanity. And the open publication of new data concerning the potential of H5N1 to convert directly to a form that can be transferred through the air between ferrets will motivate many more policy-makers and scientists to work to reduce the likelihood that this virus will evolve to cause a pandemic.”  — Bruce Alberts, Editor-in-Chief, Science 22 June 2012: Vol. 336 no. 6088 p. 1521

 

Five months after it was uploaded on July 15, 2012, Psy’s Gangnam Style became the most watched video in human history. It reached a billion views on YouTube before year’s end, and as of the writing of this sentence (7/25/2014), it had 2,046,723,177 views. That’s over a quarter of the world’s population, and almost half the world’s internet users saw the video in 2012. If internet access has not been dramatically expanded since then (which admittedly is one of Larry Page’s top priorities), its current views would mean that 85% (or more by the time you read this) of the internet world has seen it.

Consider for a moment that the next closest video right now is Justin Bieber’s February 2010 upload, Baby, with just over 1 billion views. Not only will it be tremendously difficult to break Psy’s record, it likely won’t be done by any existing videos in the Top 30, most of which were uploaded before 2012 and all of which (besides the aforementioned) still have well below a billion views.

Amazingly, the only non-music video member of the Top 30 sits at #4: Charlie Bit My Finger – Again!, uploaded by Harry and Charlie Davies-Carr in May 2007. Despite being uploaded 7 years ago, there are comments on this video every few hours, the most amusing ones referencing baby Charlie as the inspiration for Luis Suárez’s recent behavior on the soccer pitch. On that note, the most-watched Tyson vs. Holyfield clip has only garnered about 6.3 million views, which speaks a lot to the nature of a viral video – namely, that “viral” may be synonymous with “current,” since even something so newsworthy in 1997 doesn’t translate well to 21st Century viewers. More on that later.

 

What Makes a Video “Go Viral”

Many things can make a video go viral, but I contend that the most important is the content itself and the effect that the content has on its viewers. This content is comparable to the genetic material contained within a virus (aka virions), and its effects on viewers are comparable to a virus’s effects on its hosts.

According to a New York Times article a couple of months ago, the viral nature of a video seems to be directly proportional to how much warmth (for themselves) and hope (for the world around them) the video instills in its viewers. The article goes on to state that even videos that end sadly can go viral if the video’s overall message is positive, such as a cancer story ending in death, but with a moral for the living.

One study they referenced measured the influence that gloomy Facebook posts by friends in rainy cities had on the posts by their friends in sunny cities. The result was that gloomy Facebook trending, while it did occur, did not gain nearly as much traction among users as happier trends – the evidence being that friends in gloomy cities were all too eager to mimic happy posts by their sunny friends at much higher rates. This “emotional synchrony,” as scientists have defined it, helps drive info through the internet like a virus, essentially “infecting” people for brief periods of time before passing on to someone else. The trends die when people stop sharing.

This, of course, means that the word viral is fairly apropos. The most highly transmissible viruses tend to follow what’s known as a logistic growth curve, that is, a growth pattern that gains early momentum, expands to exponential growth, slows at the intersection of population size and timeline, then eventually plateaus as hosts become less available (either by dying off, forming immunity, or quarantining themselves from the virus).

Example of a logistic growth curve.
Example of a logistic growth curve.

The biggest difference is that viral videos don’t make you sick and kill you. Viral videos appear to be a form of seemingly innocuous and temporary psychological disorder, like a blend of Rustin Cohle’s description of a religion virus and the Red Dwarf crew’s discovery of the luck virus.

 

Newsworthy ≠ Viral (Part 1)

While the positive effect that a video has on its viewers could be directly proportional to its virality, the opposite can be said for its newsworthiness.

You might think that if something is newsworthy enough, it will go viral no matter how depressing it is. After all, “newsworthiness” earns its name for its sheer ability to draw in spectators (think car crashes, murder scenes, public officials’ sex lives, etc), so let’s test newsworthiness against virality by using some big name events throughout history.

One of the most tragic and newsworthy days in American history occurred only a year into the 21st Century. 13 years later, the most watched video of the September 11th terrorist attacks (a homemade video by a lady working just blocks from the Towers) sits at about 30 million views. Yet, there are numerous other videos pertaining to 9/11, and what I found after compiling a most-viewed list was that 9/11 suffers from a lot of things that prevented it from becoming a viral video event.

The first and most obvious is the sheer number of videos of the events themselves. While September 11th videos in general are almost guaranteed web traffic, 9/11 as a viewer event doesn’t enjoy the same viewer consolidation as a music video. For instance, the 20 most-watched videos under the keyword “September 11 2001 attacks” had an overall viewer count of ~139 million. That’s a lot, and you might think that the remaining 9/11 videos on YouTube would bring the total 9/11-themed viewer count up to the billion range. Right?

The short answer is no, but let me take a quick detour to explain why. If math hurts your head, skip through this chapter to the graphs at the end.

 

What Does a Billion Views Look Like?

It takes 999 single millions (plus another million) to get to a billion. Sounds easy at first, but think about that next time you’re talking about becoming a billionaire. That’s either 999 (+1) videos of at least 1 million views (wow), or some other very large number of videos with less views. When I last checked, there were 319,000 results for the keyword “September 11 2001 attacks,” and that would require at least 3,000 or so views each just to hit 1 billion.

Now, I know what you’re thinking. Surely, it’s not that difficult given that the first 30 have at least a million each. In fact, the first 30 have about 151 million views overall (averaging to 5 million each). The first ones put a huge dent in that billion goal, but as the viewer counts get lower, that billion goal gets farther away.

When you get down to #100 in the most-viewed list, you’re at 59,000 views, and a little calculus gives us a decay rate of -0.063 over the first 99 videos’ viewer counts. At that rate, you’d theoretically hit 1 view per video within 273 videos. To hit a billion, those 273 videos would actually have to have 3.66 million views each, but every video below #15 has less than 3 million views. This rough calculus leaves us far short of the 319,000 videos supposedly pertaining to the topic, and the remaining 318,727 videos would have to have at least 3,137 view counts each to hit a billion.

Hopefully, this puts “a billion” into perspective for you.

Feel free to hand count them and send me the results (with all URL’s dutifully recorded), but consider for a moment that most of the videos near the bottom of the list have less than 100 views, and all this math includes videos in “parts” (Part 1, Part 2, etc.), spam videos not actually pertaining to 9/11, and recycled footage under a different name. Finally, let’s not forget that the keyword “September 11 2001 attacks” will pull up multiple events, since the attacks themselves took place at the World Trade Center, the Pentagon, and a field in Pennsylvania, which remains one of the best exhibitions of bravery by normal, everyday people.

Viewer counts for 9/11 attack videos look strong at first, but drop off considerably after just the first few videos.
Viewer counts for all 9/11 attack videos on YouTube look strong at first, but drop off considerably after just the first few videos.

Now, this isn’t a growth curve. It’s more of a simple representation model that demonstrates how 9/11 viewers are searching for eye-witness accounts of the event, and how, after the first few, they likely give up. Even the 30 million view video is pretty useless, since someone deleted a lot of important footage, like the second plane hitting the South Tower.

For some valuable time series data that does demonstrate growth, compare the graphs below, which show the growth trajectory for daily viewer counts of: 1) the most-viewed 9/11 video, 2) Baby, and 3) Gangnum Style:

Sept 11 vid trend
The most-watched 9/11 video never went viral.
Baby vid trend
Justin Bieber’s first mega-hit and its life patterns.
Gangnum Style vid trend
Psy’s world record YouTube video and its life pattern.

Baby has almost twice as many thumbs down than up, proving that even something that a majority of the viewers consider garbage can still go viral. The most important thing to note from these graphs, however, is that the 9/11 video with 30 million views never went viral, it just had spikes of activity that I won’t take the time to investigate (though, I am curious what caused those spikes).

Meanwhile, Baby started off strong, plummeted almost immediately, soared up to 2.5 million views per day, then began its gradual decline into mediocrity where it belongs. Gangnum Style didn’t start strong, but steadily climbed from zero to about 12-13 million views per day, bouncing up and down there and spiking in January 2013 before slowly drifting down to current levels.

Finally, I’d like to note that these daily viewer graphs for Baby and Gangnum Style do not resemble the logistic growth curve above because that type of curve is based on cumulative infections. If we did cumulative views of Gangnum Style, for instance, it would look like this, which is logistic in form:

The cumulative growth of Gangnum Style more closely resembles a logistic curve.
The cumulative growth of Gangnum Style more closely resembles a logistic curve, albeit a shallow one.

 

Newsworthy ≠ Viral (Part 2)

In the end, perhaps the most newsworthy event in American history falls short in viewer count to two of the worst songs in the history of mankind. Hold on, let me repeat that: Every 9/11 attack video on YouTube combined doesn’t come close in viewer count to a 4 year old Justin Bieber song.

Perhaps a majority of people just don’t want to relive it, or maybe the content disaggregation keeps any specific video from reaching its target audience. Even then, there were probably other factors at work keeping 9/11 from becoming a viral video.

For example, the date on which the event occurred would be important, as it was well before YouTube earned a monopoly on visual content and before the proliferation of camera/video phones.

You also have to account for viewer psychology, such as the possible tendency for September 11th enthusiasts to read about the event (that’s the thing you’re doing now, with your eyes and frontal cortex..) rather than watch it online, while music video enthusiasts have no choice but to click “Play,” and YouTube is free (for now, anyway).

On the flip side, 9/11 viewer data benefit from single viewers hitting multiple videos of the same genre, while multi-counts for music videos require rewatching the same video by respective fans. This would imply that, because 9/11 videos are all different aspects of the same event, it would benefit from series’ of clicks from a single viewer whereas a music video might get boring after a few sessions.

Side note: In case you were curious, multiple viewer counts on YouTube can come from any IP address, according to this study, meaning that one viewer can contribute to multiple views even from the same device. 

We’ll have to wait for something comparably terrible and newsworthy to take place in this age of video proliferation in order to test depressing, tragic, newsworthy videos against cats and K-pop singers on the viral scale. Unfortunately, Moore’s Law dictates that our expanding technological capacities change our social experience daily, a change that we experience two-fold every two years (or 18 months, depending on your techno-paradigm). This means that comparing Any Instance #1 against Any Instance #2 is quickly becoming an impossible task unless they both take place on the same day.

In short, a video will eventually triumph over Baby, and even Gangnum Style, but it likely won’t do so on the same platform, for the latter at least.

 

If Content is King, Currency is the Throne Upon Which He Sits

According to The Guardian’s “Seven Golden Rules” for making a video go viral, #6 states that the first 24 hours are absolutely crucial. This is where the viral nature of a video can mimic the calculus of something like compounding interest, where investments made up-front define the video’s growth trajectory.

Gangnum Style‘s massive popularity can largely be attributed to its tremendous growth in the months following the initial post. Wikipedia tells us that Gangnam Style increased to an average of >9 million views per day within just two months. From the Guardian article, we find that early lip service can help, too. Remember the Double Rainbow guy? That video sat idle and didn’t see growth until it was discovered early by one of Jimmy Kimmel’s writers. It was publicized on the show, and now has over 40 million hits.

*Update: several hours after starting this article, Gangnum Style was at 2,046,868,910 views*

Timing, it seems, is crucial for the success of viral videos, and while music videos like Gangnum Style are quickly replaced by newer hits (like Psy’s Gentleman, which garnered 22 million views in the first 24 hours), those newer hits continue to send traffic to the older ones. Moreover, the success that the older one fuels both hits’ popularity even more, and so on and so forth. While daily views continue to decline after a peak, the overall views for older videos keep going up.

But timing doesn’t pertain strictly to the performance of a new video in the first months. It also pertains to how current the video’s content is. In other words, historical events, no matter how important or newsworthy when they took place, will never go viral.

For instance, would the Kennedy assassination go viral today?  Let’s assume for the sake of viewer count consolidation that only one copy of the video existed. As it happens, this is partially true already: The Zapruder Film. Granted, there were other videos taken of the Kennedy motorcade in Dallas, but none as visually descriptive or valuable as Abraham Zapruder’s.

Like September 11th videos, there are lots of Kennedy assassination clips on YouTube, many of them conspiracy theory documentaries or open-source educational vids, but the most-watched Zapruder Film on YouTube has 4.2 million views, a slow-motion video of the event with a timer. If you expand the search to any video associated with the Kennedy assassination (i.e. the actual event itself), you end up with only one video with more views. It’s a high-definition digital copy of the Zapruder Film, and it shows in graphic detail the second bullet spraying blood and skull fragments all over the interior of the Lincoln Continental limousine. It only has 5.34 million views.

Can you believe that? Enhanced footage of the Kennedy assassination, widely considered the most analyzed video document in history, has less views than 37 separate fart videos on YouTube. Then again, YouTube’s viewer count isn’t what it used to be since Google realized they could monetize it, so taken with the previous sentence, we can indeed say that 5 million views aren’t worth shit.

*Speaking of shit: Gangnum Style got to 2,049,201,823 over the three days since starting this article*

Now, you may be thinking “Wait, you’re talking about a current video of a past event. If the Kennedy assassination happened today, the event in itself would be current, therefore, it could possibly go viral.”

Good point, reader. Let’s talk about that.

 

Historical Drama is a Spectator Sport with an Ephemeral Fan Base

Television doesn’t just get between you and your chores, it creates a barrier for newsworthy events going viral, too.

If you take a look at some of the most newsworthy events in the last half century, you realize that today’s internet users don’t care about them nearly as much as those who lived through them. While this may seem pretty obvious (I mean…duh, right?), the chart below helps to highlight just how much more important historical events are to contemporaries vs. internet users today.

This chart includes some of the most newsworthy events of the last half century, their original viewership (if available), and their most-watched YouTube video’s viewer count (also if available).

The data above indicate that while media attendance at the time of an event like the moon landing might have been astronomical (Ha!), the number of internet users interested in that event today doesn’t come anywhere close. That is, the moon landing video on YouTube with the most views sits at just under 11 million, while the number of people tuning in to watch the original moon landing in 1969 has been estimated at 500-600 million worldwide, and 125 million just in the United States.

The same can be said for other events. 95 million people watched the O.J. Simpson chase in 1994, but the most-watched video of the very same footage has only 1.2 million views. A year after the chase, Simpson’s verdict garnered even more publicity, conservatively bringing in 107.7 million viewers. Some initial CBS estimates went as high as 150 million. The latter would make the Simpson acquittal the most watched television event in U.S. history, but the view counter on YouTube for the exact same event sits at 790,000 (less than 1% of the original viewership).

Drama – disasters, violence, infidelity, etc. – makes for marketable news, and the historic events above are all excellent examples. It seems, however, that there is an expiration date on drama. That expiration date is debatable, since it looks like the historical events sampled that translate best to YouTube are the ones that took place most recently. Internet coverage of the 2011 Japanese tsunamis following the Tōhoku Earthquake benefited greatly from handheld devices and quick uploads. I know this because I was watching live feeds from Japan on international chat boards the day of the catastrophe.

Daily viewer counts of the most-watched 2011 Japanese tsunami video show how viewership of current events drops off immediately
This daily viewer count graph of the most-watched 2011 Japanese tsunami video show how viewership of current events drops off immediately following the event itself.

 

It’s worth noting that none of the sampled videos in the chart ever went viral. Even the 31 million view Japanese tsunami video’s daily traffic went crashing to the ground within a few weeks, most of its traffic having taken place the week of the event itself. This could be an indication that such an event is the most viral and behaves like Ebola, which is highly contagious but disappears quickly due to its high mortality rate. If this is the case, though, maybe the term “viral” is too broad to apply to all popular internet videos. Maybe we should start categorizing them by viral type (e.g. influenza-curve video, Ebola-curve video, HIV-curve video, etc).

Any event that occurred before YouTube became a collection point for visual content never had a chance to go viral, so a follow-up study would need to cover only those newsworthy events that have taken place over the last 4 years or so. I think what we’d find that other videos regarding major events would have a lifespan curve similar to the Japanese tsunami video.

The only video pertaining to these events that might be considered viral would’ve been the benefit concert for Haiti Earthquake victims, which has over 141 million views. I can’t tell you if it resembles a viral growth rate, because stats have been disabled for that video. I also don’t really care, because all the contributing artists suck, and even NPR named it one of the Worst Ideas of 2010. When NPR turns its back on a third-world humanitarian effort, you know there’s something fundamentally shitty about it.

*…almost as shitty as the fact that Gangnum Style has 2,050,656,568 on Day 5 of this article*

 

Conclusion

Is there a conclusion?…It appears that newsworthiness and virality aren’t just non-proportional, but in some ways, inversely related. The primary reason for this, I believe, is two-fold.

First, most people absorb newsworthy information at the time of the event and never rewatch it. In the age of television, millions of people were glued to their TV sets watching news filter in on the Rodney King trial, the acquittal of the four officers who beat him on tape (another one of the most analyzed pieces of film in history), and the subsequent riots that broke out across Los Angeles. It was a big deal.

22 years later, the beating tape has more YouTube views than the acquittal that set off the race bomb, but neither of them come close to capturing just how significant the event was for people in 1992. Remember, this was the kind of Helter Skelter craziness that Charles Manson had predicted only two decades earlier, after similar riots (Watts) in the very same city. Martial law had to be instated, and President Bush put his domestic military forces on standby.

Second, contemporary newsworthiness usually pertains to some form of human drama, and while the human mind latches onto drama while it’s happening, it quickly shifts gears to something else once it’s over, or whenever their media outlet decides ratings aren’t high enough to warrant continued coverage (..like that Malaysian jet they still haven’t found). Studies have shown that the brain prefers good news (or new news) over bad news, surprises over expectations, and that our collective attention span as humans is shrinking at a dangerous rate, all of which help explain why an entire nation can be herded like cattle to the next crisis by Mr. TV.

Simply put, nobody cares about the Simpson verdict anymore (ignoring the referential spike during the George Zimmerman-Trayvon Martin proceedings), and even something as awe inspiring as the moon landing, one of those rare, non-disaster, historical landmarks, has less YouTube views than the Challenger space shuttle explosion.

Are we supposed to believe that people are more interested in the death of seven astronauts than one stepping foot for the first time on lunar soil? Maybe, since the explosion grabs your attention more than a footstep and a bungled quote. Should we be that surprised? I’m sure if you went back and checked, you’d find that media outlets quickly ditched Simpson in favor of John Bobbitt, and Armstrong in favor of Aretha Franklin, Muhammad Ali, and Teddy Kennedy.

As usual, there are caveats to this inverse relationship. Sometimes virality can lead to something becoming newsworthy, the best recent example being Psy’s share of YouTube traffic which has spawned an infinite loop of articles on the topic. Conversely, sometimes things that make it on the news make it even further online, such as Antoine Dodson and the bee charming field reporter, both of which were second tier news stories that have earned a combined original video viewer count of 78 million (this doesn’t even count remixes and spin-offs). Still, that’s why these are mere caveats and not exceptions; their newsworthy-viral trend lines still cross somewhere on the X/Y axis, indicating an inverse relationship (i.e. they were newsworthy before viral, and vice versa, but not both simultaneously).

One day, however, newsworthy will become synonymous with viral. The rising popularity of internet television is highlighting a shift toward more web-centric media. Newspapers have already migrated online, and TV will soon follow. Conventional television will not exist in the next 20 years, and the fact that there is still such a thing as “cable television” amazes me given how so many people are moving to wireless internet, handheld devices, and satellite radio/television.

I admit this isn’t a very scientific approach, mainly because I don’t have access to a lot of crucial data. Regarding YouTube, I’d like a Google analyst to pull the complete viewer counts for all my keywords. I’d also be curious to know just how many views all videos on this list of events has. Google could also tell me how many times each keyword has been searched, which might be another indicator for current interest of past events. Beyond what Google can provide, I’d love for Nielsen to provide hard numbers for TV attendance, and I’d love to find a way to get the worldwide totals for viewership (if you didn’t notice, my chart stuck to nationwide viewership, meaning even the Princess Diana numbers were UK-specific).

Until I get these data, however, I’m stuck with this article. To be honest, I got half-way through this thing and realized that it was ballooning out of control, so I had to taper my methods and data collection quite a bit. Still, there’s some good stuff in here, and if it didn’t interest you, perhaps a music video will.

On that final note, Gangnum Style has 2,051,457,502 as of the 6th day of this article. That’s 3/4 of a million views in less than a week, and I’ve never even watched it.

 

Follow us:Facebooktwitterlinkedinrssyoutubeby feather
Share this:Facebooktwitterredditpinterestlinkedinmailby feather