Professor Ramesh Sitaraman has developed a scientific method to study viewer reaction to online videos. The computer science professor at the University of Massachusetts in Amherst, Mass., shows in a study that most online video viewers are impatient and will abandon viewing a video stream if it is not instantly available or smooth. While this may seem obvious, scientific rigor in testing this behavior has broken new ground that could prove invaluable to businesses trying to promote their products.
Collaborating with Akamai, a company that facilitates faster Web communication, Sitaraman sought to quantitatively demonstrate how video stream quality causes changes in viewer behavior. His study was based on “Big Data” he told News India Times, involving 6.7 million viewers, who played 23 million videos for a total of 216 million minutes.
“By using this volume of data, we developed an interesting scientific method – a quasi-experimental design used earlier in medical and social sciences – and applied it to viewer behavior. That makes it a big scientific contribution,” Sitaraman said on the phone from Chennai where he is visiting his mother.
By taking pairs of people who were identical in most other ways except in terms of their viewing experience, Sitaraman and his team were able to scientifically say that it was performance that affected viewership.
“People care about video performance and the goal of content providers is to give a good experience, captivate the audience, and make them want to come back,” he noted. So the question he posited was: If a video does not start right away or function optimally, how quickly do viewers abandon it? Six months ago, in a eureka moment, Sitaraman came up with the technique to analyze this kind of data and how that affects human behavior.
“People are generally patient for the first two seconds. But if you take a 100 viewers as the universe, with every additional second of delay (buffering or interruptions) makes 5.8 percent of them leave,” he said. There are also finer points to the study – for instance, people are more willing to wait on a mobile instrument than on a home or office broadband connection.
“Value defines behavior. If it is a movie you want to view, you will wait longer than if it some short clip,” he said. People with broadband are more impatient, his study showed. “On a high broadband, if the video is not en par, you will lose all viewers in 10 seconds. But on a mobile, barely one-fifth leave within 10 seconds,” he said.
The team also studied viewer habits and loyalty to websites in the event of frequent freezing.
Sitaraman said he has been interested in mathematics and science since early childhood, and that he “really, really wanted to discover new things.” Joining the Indian Institute of Technology, Madras, came “naturally” he said. But the opportunities for research were limited in India, he adds, so he left to pursue a Ph.D. at Princeton University.
Sitaraman has been involved with the mathematics behind making networks work better, even before the Internet emerged in the early 1990s. He carried over that research into the field of the Internet. When some of his colleagues decided to start a company with the goal of making the Internet more reliable, he joined them to help build one of the largest versions of the Internet – Akamai network – which makes up 25 percent to 30 percent of the World Wide Web, hosting some 100,000 servers around the world in almost every country, Sitaraman said.
According to his biography on the UMass website, Sitaraman was the principal architect, helping build Akamai Technologies, and helped pioneer the Internet-scale distributed networks that currently deliver much of the world’s Web content, streaming videos and online applications. His research focuses on foundational issues in the design of large Internet-scale distributed systems, communication networks, cloud computing, and global Internet services. Sitaraman is a recipient of National Science Foundation’s CAREER Award.
