Can a computer look at this video of a diseased cat and know that humans will laugh at it? Probably not.
I just attended a session at SXSW called “Man Vs. Algorithm: Online Video Curation Face-Off”. It offered the following description:
Human or Machine? Who really knows what videos you’re going to like best? Online video curation is in a heated battle between algorithmic curation and human, editorial selection. Let’s settle this debate once and for all.
Settle it they did not. The session’s moderators, Marc Hustvedt, co-founder of TubeFilter and Head of Strategic Partnerships at Chill (a position and company I’ve never heard of), and Neetzan Zimmerman, editor of The Daily Wh.at, were supposed to actually do some kind of experiment during the session. It didn’t pan out because, as Hustvedt tried to explain, it would just be too hard to do in a session that short.
Instead they talked about video curation on the web, both the kind done by computers and the kind done by editors like Zimmerman, whose track record is somewh.at (I couldn’t resist) prolific. The conversation of man vs. machine really boils down to editors vs. algorithms. YouTube, for example, parses all kinds of data to determine which videos humans would like to see.
However, humans can do better because of that impulse that tells you “this is a something.” The above quote from Zimmerman is about the viral video “The OMG Cat,” which is just a short clip of a cat with lockjaw. The expression makes the cat look like it is surprised, like it is more human. Zimmerman rightly points out that a computer can’t see the inherent value in something like that.
The virality of online content is something that seems almost impossible to measure. Zimmerman mentioned in the session that he can’t really explain in finite terms what it is that makes one cat video better than another, but he certainly knows it when he sees it. It’s possible that there is something happening in our bodies when we have a “gut feeling,” something more chemical than visceral, that computers will one day be able to replicate.
Anyone who has ever launched a story or a video or had a scoop knows that it’s almost impossible to explain how it is that something goes from just existing to being an event. When you write (or, critically, read) online long enough, you can start to tell what’s going to hit and what’s going to get lost, but there’s no perfect way to predict it. It seems for a computer to fully understand viral videos, they would have to fully understand humans.
Thankfully, I think humans are going to be necessary in the editorial game for years, if not generations, to come. Now enough blogging about blogging; I’ve stories to find.