The Beast With A Billion Eyes

In just seven years, YouTube has become the most rapidly growing force in human history. Where does it go from here?

  • Share
  • Read Later
Youtube

David After Dentist

(3 of 8)

(In case you missed it, Nyan Cat is one of the more mysterious and quintessentially YouTubey video phenomena ever to have crossed the viral threshold. It's a video of a cartoon cat with the body of a Pop-Tart that runs--scampers, really--through outer space, with a rainbow trailing behind it. Nyan Cat was originally created as an animated GIF by a 25-year-old Texas cartoonist named Chris Torres, who posted it on a comics website on April 2, 2011. Three days later, a YouTube user who goes by sarajoon [her bio reads, "I am the video word made flesh"] added a sound track, a maddeningly cheery Japanese pop song called "Nyanyanyanyanyanyanya," and reposted the GIF to YouTube as a video. The mashup went viral and finished 2011 as the fifth most watched video on YouTube, with 58,949,289 views. Wait--make that 58,949,290.)

The first line of defense against YouTube's runaway chaos, and the first angle of attack for a hopeful viewer, is search. YouTube gets a billion search queries a day; if they were tallied separately from Google's, YouTube would be the second largest search engine on the Internet. But searching for videos isn't like searching for Web pages. It's harder. Computers can read Web pages because they are made out of words: if you're looking for information on echidnas, a search engine can send you to a website where the string of letters "echidna" occurs a lot. But a computer can't watch a movie. It can't look at the huge string of 1's and 0's that make up a video file and know that that movie depicts a spiny, insectivorous, egg-laying mammal native to Australia and New Guinea. Computers are blind and deaf. To a computer, a movie is a black box.

This is one of YouTube's core organizational challenges. We help guide the poor, blind, deaf computers by attaching verbal descriptions to our videos, but unfortunately, we're not very reliable. Sometimes we'll title a video something like "For your viewing pleasure, this is a short film of a monkey riding backward on an echidna," but we're just as likely to call it "LOLOLOLOL this thing is amazeballs!!!!!!!" We're unpredictable that way.

YouTube can't watch videos, and it can't trust what we say, so instead it watches what we do. If you search for "echidna," YouTube will notice which of the search results you click on and will infer that that video is more echidna-y than the others; next time, it will be ranked a little higher. It will also notice if you watch the whole video or give up in the middle, and which video you watch right after it, and whether you post that video on your blog, and if you leave the site after you watch it or hang around for a while. It uses all that information to deduce things about the contents of the videos and improve its search results accordingly.

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8