Zürich researchers create model for the success of internet videos
A mathematical model will in future, it is claimed, explain the success of videos on the YouTube internet platform. Researchers Riley Crane and Didier Sornette at the Swiss Federal Institute of Technology (the Eidgenössische Technische Hochschule or ETH) in Zürich have described an appropriate attentiveness spiral that shows when YouTube videos will become big hits. A report from the ETH's internet service says it follows physical laws, at least in part. The two scientists want to refine their system further. In the medium term, it could yield a kind of trend-monitoring system on the internet.
Using systems he programmed himself, Crane spent two years watching the audience figures for five million videos on YouTube. He was primarily interested in those films that attracted the most attention, being watched at least 100 times per day. Only 10 per cent belonged in this category. Crane subdivided these in turn into junk videos that attract a lot of attention for a very short time, viral videos that spread like an epidemic over broad social networks, and quality videos that suddenly arouse attention, instead of showing a slow rise in popularity. Their popularity spreads rapidly and then slowly ebbs away.
Crane discovered, for example, that the ebbing away of attention shown by viral videos can be described with the mathematics used for modelling the aftershocks that follow earthquakes, which he calls "epidemics-type aftershock models". As Crane explained to the ETH internet services, this suggests that a social system may function in accordance with defined rules, in just the same way as a physical one, and can thus be described mathematically. The model, he said, even made it possible, on the basis of tendencies, to recognise whether a video could become a magnet for spectators.
The report continues that the mathematical model could be used, for example, to monitor online book purchases in real time. By constant comparison of data, marketing specialists could determine at an early stage what book was a potential best seller. Crane and Didier are in negotiations with Amazon, the internet bookseller, on integrating their system into that platform.