Let's say you are Google and you own a very popular video site called YouTube. What do you do? Here's what Google did:
* YouTube was the first video site added to Google Video when it was relaunched as a video search engine. Even if Google added other sites, YouTube dominates the search results. Of course, YouTube has a lot of videos and a strong community that provides feedback, but Google Video still can't provide the right balance between YouTube and the rest of the sites.
* as part of the Universal Search, YouTube videos that appear in Google's search results have extended snippets that include thumbnails, links to related videos and the full video can be played inline. Google doesn't do this only for YouTube, but Google Video and YouTube are the only video sites for which you can play videos directly from the search results page. This decision was probably influenced by the fact that Google can't control the performance for other video sites and those sites didn't want to lose traffic.
But what if you are Yahoo and own a photo-sharing site called Flickr? The site doesn't have a dominant position like YouTube (6.42% for the US in July 2007, according to Hitwise) and an image search engine can easily integrate images from other sites.
Yahoo also shows extended snippets for Flickr images, but it commits the cardinal sin for a search engine: forget about relevance and promote arbitrary sites. For example, the first page of search results for [Google Reader] only includes Flickr photos, while you can find many other relevant images on the web (in fact, the first 49 Yahoo results are from Flickr). Other queries also show a very-difficult-to-justify Flickr domination. While Flickr is a great place to find photos, it's not very relevant if you want to find Vista screenshots: 9 from the first 20 results are from Flickr and all of them show Vista wallpapers. A search for minimal surface includes a single result from a math-related site, while the rest of the images are from Flickr.
The conclusion is that it's difficult to have a search engine and sites for user-generated content. You can be tempted to arbitrarily increase the influence of these sites and show biased results.
