Video Essentials

Humans or Machines: Who Curates Online Video Better?


With the amount of online video growing so rapidly that no one can keep up with it, online video curation is growing in importance. Consider that YouTube now receives 60 hours of uploaded video every minute. Many businesses already curate videos to show on their sites, mixing them with self-created videos. By curating appropriate videos, they can create instant video libraries for their site’s visitors. Creating large libraries of only homegrown videos would be cost prohibitive for most companies.

A South by Southwest Interactive panel called “Man Vs. Algorithm: Online Video Curation Face-Off” looked at the contentious issue of who curates videos better, humans or machines? Which is better able to choose videos that will broadly appeal to a site’s visitors, or selectively appeal to individual visitors based on their surfing history?

Giving credit to machines, panelist Marc Hustvedt, head of strategic partnerships at Chill, pointed out that no human visitor could keep up with the sheer volume of new videos around. Machine curation is needed simply to keep up.

YouTube is the leader in machine video curation, and Hustvedt pointed out that it does a pretty solid job. Most YouTube visitors have an “unarticulated want,” said Hustvedt: they know they want to be entertained, they just don’t know what they want to see. YouTube’s algorithms combine video with users preferences, including how long visitors chose to view certain videos, to determine what else they’d like to see. The system works: 60 percent of the clicks on the YouTube homepage come from the recommended videos section, and they perform 200 percent better than videos from the Most Viewed, Top Favorites, and Top Rated lists.

Machines aren’t the perfect curators, however. They don’t have a sense of humor. They don’t know what will surprise or shock the viewers. Hustvedt was joined on the panel by Neetzan Zimmerman, editor of The Daily What, a site that specializes in showing viral videos before they’re viral. It was Zimmerman, for example, who jumpstarted interest in Rebecca Black’s Friday video when he featured it on his site. On March 11, 2011, he embedded the Friday video. On March 14, Rebecca Black was a top trending topic on Twitter.

“it was the worst thing I’d ever seen,” said Zimmerman. “It was just a phenomenon from the very beginning.” For the record, Black still hasn’t thanked him.

Computers can’t prank people; they can’t troll. “Sometimes you just post videos to shock your friends,” said Zimmerman. No computer is going to discover a new video that’s great just because it’s awful.

The curation solution, Hustvedt and Zimmerman agreed, was a combination of machine and human power. Computers can scan through millions of videos and look for matches to previous likes, but they can’t (yet) match human ingenuity. For something to go viral, it still takes that human element. It’s not humans versus machines in the curation battle, it’s humans and machines.




Discussion

Comments for “Humans or Machines: Who Curates Online Video Better?”

Post a comment