April 7, 2007
YouTube’s Favorite Clips
By DAN MITCHELL
ON YouTube, copyrighted video clips of movies and TV shows are far less popular compared with noncopyrighted material than previously thought, according to a new study.
On their face, the results could have serious implications for YouTube’s owner, Google, and the media companies, most notably Viacom, with which it has been negotiating. But not everyone agrees.
Vidmeter, which tracks the online video business, determined that the clips that were removed for copyright violations — most of them copyrighted by big media companies — comprise just 9 percent of all videos on the site. Even more surprising, the videos that have been removed make up just 6 percent of the total views (vidmeter.com).
Google is in negotiations with big media companies in hopes of reaching agreements that would allow YouTube to feature clips of movies and television shows. Viacom, owner of MTV Networks, is suing Google over the showing of copyrighted material like clips from “The Daily Show” and “South Park” on the site.
“This finding,” writes Henry Blodget on his Internet Outsider blog, “is the opposite of consensus, which assumes that Big Media videos account for a small percentage of total videos, but a large percentage of views.” It means, he adds, that Google has “a lot more leverage in the YouTube-Big Media negotiations than was commonly thought” (internetoutsider.com).
But the consensus might not have been so far off after all, writes Adario Strange on the Epicenter blog at Wired News (blog.wired.com/business). The study is flawed because it examined only those videos that YouTube removed after receiving a complaint from a copyright holder, he writes. It “fails to take into account the vast number of copyrighted videos that slip under the radar daily, existing on YouTube sometimes for months before any removal request is made.”
How large that number is, however, remains unknown. Far from this giving Google a leg up in its negotiating power, Mr. Strange says just the opposite may occur. “None of this will slow the Viacom lawsuit,” he writes. “In fact, the report’s finding may even bolster the company’s claims. Vidmeter found that Viacom was the leader in terms of pirated content on YouTube.”
Vidmeter explains how they track videos from across the most popular sites:
How Vidmeter Works
Vidmeter gathers data from across the web to provide an accurate representation of the most popular online videos. While it is impossible to tell the exact number of views a given video has received from every website and every download, Vidmeter gathers the reported view count from a large cross-section of web-based video sites, giving a very close indicator to the relative popularity of a video. Vidmeter gathers this data like so:
First, Vidmeter's software automatically records the numbered of views and comments from the top listed videos on Addicting Clips, Atom Films, Break.com, Brightcove, Daily Motion, Google, Grouper, iFilm, Metacafe, Myspace, Revver, Veoh, vSocial, Yahoo, and Youtube once per hour.
Second, Vidmeter editors "merge" similar videos on multiple websites. This allows Vidmeter to count the views of a video on all websites as a single video, giving a more accurate ranking of a video's popularity and not just a URL's popularity.
Third, Vidmeter automatically ranks videos for the day by counting the difference between that day's view total and the previous day's view total. The most viewed videos are the most popular and ranked highest. For videos that are new (that do not have a previously recorded view count), the first view count is listed under "before" as we cannot necessarily tell on which day those views occurred. On some days videos will show a negative comment number, this is because some networks allow users to delete comments and thus lower the total comment count.
Fourth, the latest view and comment counts are set as the "all time" counts for the videos and they are ranked accordingly.
Vidmeter's resulting list of video provides a great tool for industry watchers to track trends in online videos, for marketers to see what's hot, for makers and advertisers to track their video's popularity, and for eager video watchers to get their fill of the most poplar entertainment online.
And Vidmeter's response to criticism (plus a link to the full report as a PDF:
Mar 31, 2007
We have completed a study on the number and popularity of copyrighted videos on YouTube. Download the full PDF report at:
In summary, we found that of the 6,725 most popular videos on YouTube, only 621 had been removed due to copyright requests. Views to the removed videos made up less than 6% of all recorded YouTube views.
Follow Ups (Apr 5, 2007):
With all the controversy surrounding our recent report, we would like to take a minute to respond to some of the comments.
1. Counting only Removed Videos
One of the biggest criticisms directed at the report is that it underreports the number of copyrighted videos on YouTube because it only counts videos that have been removed and does not include copyrighted videos that have not been removed. Our response to this is: yes, as we noted in the report, this is true. However, as it also states in the report, the goal of the document is to provide a “general estimate” which we feel the report does well.
We created this report in response to the question, “Is the majority of traffic on YouTube going to copyrighted videos?” And the study, even being a general estimate, clearly says no. Here’s a bit of math to demonstrate:
We found that slightly less than 6% of the top-video views went to removed videos.
Even if only half of copyrighted clips in our study have been removed, then the total views to copyrighted videos would be approximately 12%.
Even if only one-third of the most popular copyrighted videos have been removed, then the total view count is around 18%. 18% would be a fair chunk of traffic on YouTube, but it’s far from the majority.
2. Skewed to Favor Non-removed Videos
Another point that has been brought up is that non-removed videos are skewed with a higher view count because they are able to keep accumulating views while removed videos stop as soon as they are removed. Again, this is true, but as we stated in the report: this may show that there is a demand for such copyrighted material, but in its current state it does not contribute to a significant portion of YouTube’s traffic.
3. Non-random Sample
The point was made that the report is flawed because the videos sampled in the report were not random, but taken from the most-viewed list. To this we respond that: YouTube’s traffic falls into a long-tail distribution where the most popular videos receive exponentially more traffic than the less popular videos; therefore, in order to get the best idea of YouTube’s traffic, one needs to look at the most-viewed videos. To put this into perspective:
Viacom’s most-viewed removed video was I Write Sin’s No Tragedies with 6.9 million views. That video received over 500 times more traffic than Viacom’s least popular removed video about Sarah Silverman with 12 thousand views.
It’s important for us to stress, as it states in the report, that this report is designed to provide a fact-based estimation. The report should not be taken to say that 6% of all of YouTube’s traffic goes to copyrighted videos. Instead it says that 6% of traffic we’ve monitored in the most-viewed videos has gone to copyrighted videos and our conclusion from the data is that the vast majority of traffic on YouTube goes to legal videos.