We Bunnies just got back from the Online Video Conference that was put on by The Daily Reel, and while all these viral thoughts have been spinning around my head, I had an idea about how to make the YouTube viewing experience better.
Whenever I watch a video on YouTube, the “related videos” section rarely includes any videos that are actually related. I think this is due to the fact that what YT deems “related” are just videos that have the same tags. But videos with similar tags are rarely similar. For instance, watching our quirky animation “Junk Science,” the top related video is about global warming (junk science vs. real science). That has nothing to do with the things I like to watch.
Well, I’m sure you know how Amazon.com tracks purchases, and when you’re looking at a book it tells you that “People who bought this book also bought these books.”
This would be a much better solution for YouTube. YT should track which videos people are watching, and then tell me which videos those other people are enjoying. Related videos should be less tag-related, and more about “people who liked this video also liked this other video.” I think that would be stuff I would actually want to see.
You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.
4 Responses:
November 2nd, 2007 at 7:02 am
Even a simple improvement to their tagging system would go a long way, since it seems they don’t currently allow for tags longer than one word. In your “Junk Science” screenshot, for example, “south” and “park” appear as separate tags, as do “civil” and “war.” Searching for “swim park,” “civil junk,” or “science war” would all presumably bring up your video. Giving the YouTube consumer a more accessible way to label his video would eventually allow more relevant videos to surface, at least within search results.
In addition, number of views is presented as a ranking system; the videos related to “junk science” are sorted highest to lowest. But this is deceptive, since number of views is not necessarily related to a video’s quality or relevance–”The Great Phrenologist,” for example, could be a beautiful but undersold media masterpiece, while “Global Warming…” could be posted on my sister’s MySpace.
On the other hand, watching a YouTube video is much less of an investment than purchasing and devoting time to a book. One of the main reasons YouTube has grown to be so wildly popular is accessibility—I believe I’ve heard you call this its crap-making flash machine—and that accessibility has resultantly cheapened the video delivery and video consumption processes. If you decide to click on a marginally related video, say “Sexy girl with vibrator,” and end up disappointed after a less-than-titillating minute-38, it’s no big deal. Not the same sort of cheated you feel after, say, turning over the last page of a Neal Stephenson novel.
Beyond this, I (and millions of other poor souls) have found it’s often fairly entertaining to traipse through “relevant” videos and just explore, much like pleasant stroll through Wikipedia (see http://xkcd.com/214/). YouTube really succeeds when it delivers on its potential to delight you with something you weren’t expecting to find. The reverse of that isn’t failure, necessarily, but, I think, just mildly boring.
November 2nd, 2007 at 11:30 am
I’d be interested in seeing if contextual recommendations work for short-form video. I haven’t seen it done, which must be for one of the following reasons:
1) it doesn’t work
2) there isn’t enough data to do it yet
3) no one has bothered to try
#1 is credible, while #2 is only credible if there’s actually no interest in doing so (aka, #3).
I’d suspect #3, actually, since Lauren’s point is very astute: YouTube makes their money off of ad revenue. They don’t need conversions to sales in order to be successful, they just have to keep you on the site. So the bar is that much lower – just give the user something that keeps them clicking and you’re just fine. And since they have a lot more content to wade through than Amazon (actually, that’s just a wild guess), their chances are much higher in this regard. When the whole experience is self-contained, what works is sometimes just good enough.
Of course, when the experience is the product, it wouldn’t hurt them to innovate in those areas where there’s noted room for improvement. I’d hate to lose that raw, unpredictable, exploratory nature that currently exists… but I believe there’s a way to retain that while also increasing relevance.
November 2nd, 2007 at 12:18 pm
Well, I guess my whole post was based on personal experience. I never surf YouTube. I’ve tried multiple times, but everytime I click on videos in the “related videos” section, I never get videos that are related, and I never enjoy them. Therefore I generally just follow links to YT that friends have emailed me, watch that video, and then get off. I can only imagine I’m not the only one that does this.
I am a person that does traipse Wikipedia, because all those other articles are completely relevant. For example, if you’re reading about Watergate, and click on the link for Nixon, you’re not going to get a link to an article about the movie, “Dick.” You’re going to get an article about the life and times of Nixon.
Lots of people do just surf around YT, though, there’s no denying that.
The tagging improvement is a good call.
November 2nd, 2007 at 4:42 pm
Also, yes, Neal Stephenson needs to hire someone else to write the endings to his books.