Comments for Video PageRank: Google Searches for The Holy Grail

Video PageRank: Google Searches for The Holy Grail

2007-03-14T21:23:21Z

In yesterday's interview with Matt Cutts from Google, on the topic of next-generation search, we touched on how Google is tackling the issue of indexing (and advertising around) online video. Matt implied that Google is looking for an equivalent to PageRank for online video:

"...in the Web we have this notion of reputation - which is PageRank, it's how many people link to your site and it's also the quality of that incoming set of links. So it's fun to think about things like reputation in video search - whether it be for Google Video or YouTube - because you don't have links necessarily. You might have things that are somewhat similar to links, but you look at the quality of the users, the quality of the ratings. I think in lots of ways it gives Google good practice to think about the power of people, and the power of trust - and how to apply that in a lot of different areas."
(emphasis mine)

One word which Matt didn't use, but which probably sums up what Google is trying to achieve with online video indexing, is relevancy. Google wants to find ways to rank video, so that the most relevant results display first. Also relevancy is key for contextual advertising. While all this sounds obvious, for video relevancy is a largely unsolved problem right now.

According to Matt, the key values for online video search are: reputation, power of people and trust indicators. The problem though is how to measure those values in video. With textual content, Google found the winning formula back in the mid-to-late 90's with PageRank - which uses links as the main measure of relevancy. But so far no company, not even Google, has found the equivalent of PageRank for video.

]]> Read/WriteWeb's John Milan was recently musing on this. He told me that recently he went to YouTube to watch a couple of videos, and was surprised by what he didn't see - advertising. The only thing he could find was a somewhat subtle Verizon ad supporting YouTube's mobile content. When he went to view videos, there was an ad above the video, but it was advertising by YouTube itself - promoting new features. John's conclusion was that Google has not been able to crack digital relevancy as they have with text.

The stakes for this ability could not be higher. If anything, images seem to be more compelling than words. How else to explain the explosion of YouTube, or the many pictures that make it to the top of Digg, or a single pretty rainbow (from which John got nearly 100k page views from a Digg)? Creating the algorithms for digital media relevancy represents the great race being run among Google, Microsoft, Yahoo, Digg, MySpace; and anyone else looking to monetize digital content that people love to see.

Image derived from photo by m.michael. Thanks also to John Milan for the info in the last two paragraphs.

Comment from John Milan on 2007-03-14

2007-03-14T22:33:13Z

I think there's another consequence of Google's difficulty providing relevance for digital content, specifically the YouTube content. I don't have stats offhand, but it *seems* like its becoming more common for ad shops to produce what look like amateurish videos, but are in fact professionally produced and scripted advertisements, and place them on YouTube. In my opinion, this is tacit acknowledgment that Google does not have a relevancy solution and therefore advertisers are doing the next best thing.

Comment from Bill Olen on 2007-03-14

2007-03-14T22:52:16Z

This is a relevant topic, but my bet is that Google is holding off on selling contextual advertising until the legal heat dies down. They have plenty of data to test algorithms on, and they don't need to tip their hand to competitors yet.

Comment from Ali on 2007-03-14

2007-03-14T22:53:24Z

I think that's where the use of adding tags and a description comes into play. So if the video you are viewing gets tagged with cats, thread, balls and kittens I would probably think that the Google ads that show are relevant to that, plus picking up text from the description.

As far having relevant ads within videos, probably I guess they would be auto inserted depending on the tags chosen.

Comment from incze on 2007-03-14

2007-03-14T23:03:54Z

No need to find Holy Grail here. Google is plenty with relevant data here (traffic, item length - and how long visitors watched an item, user ranks, ranks of users who had ranked, summaries, discussions, keywords...). If you're pragmatic, then you will come up with an ordering in finit time. Let's name it "beta". And beta periods can be pretty long at a decent company.

Comment from mark cuban on 2007-03-14

2007-03-14T23:13:24Z

The lack of ads has nothing to do with relevancy. It has to do with copyright laws. If Google doesnt have a license from the video owner, they cant show ads.
If you do a viewsource, you will see the ad tags which define a user id, along with category and demographic info. If the user id reflects an uploader that Youtube has a deal with, you will see ads. If not, not.

In terms of video search. There needs to be a differentiation between internet video search, and website video search. Google only does indexing of video on their owned sites.

Maybe they think they can host all the worlds videos ?

Comment from Everyday Weekender on 2007-03-14

2007-03-14T23:25:00Z

would you really be surprised if google thought that?.. why can't the host all the worlds videos?.. they have an incredibly huge server farm and it wouldn't surprise me if that was the plan :)

Comment from John Milan on 2007-03-14

2007-03-14T23:47:58Z

@mark -- First off, congrats to the Mavs! Maybe this will be the year.

I don't understand why Google can't place a series of text ads to the top, left, right or bottom of the video? It makes sense that they can't edit the video itself, but surely placing relevant content around the video is ok.

After hopping to you blog post on the subject, it seems to me that the Viacom lawsuit as absolutely a negotiation tactic. I know The Sopranos has a huge audience, but it seems to me it could be even bigger with Google's distribution. Therefore, if and when Google figures out video relevancy, Viacom and the rest would be lining up at the table to get the biggest cut of ad revenue possible.

If there's one thing Google has acquired, it's eyeballs. I also wonder if, while perhaps neglible for a hugely popular show like The Sopranos, the extent of Google's 'Long Tail' effect on a video library might also be enticing to the Viacoms of the world. As long as Google figures out relevancy.

But then again, my day job is project management :)

Comment from Rob Rustad on 2007-03-15

2007-03-15T23:16:29Z

Basing video search relevance on tags is too easily spammed and leaves you with more needles-in-the-needlestack. Metadata doesn't work well -- we all know that.

Just as web analytics has moved from tracking pageviews to user-events, I believe the holy grail will come from event-driven algorithms combined with behavioral correlations -- people who liked this, also liked this.