Back in December 2006 I sent in a reply to Google’s “Big Idea Challenge”;
Google responded with a phone call asking for details and they said they’d send me some information via email, I never heard from them again. With Google’s latest spate of recommendation gadgets and their latest product change, “web history” (something I have been waiting for for a while because I was fed up of never being able to search my history — an offline version would be nice and more secure), it seems relevant to post what I submitted.
I called it Google Channels, for lack of better words and to embrace Google’s brilliant naming tradition (see: Froogle’s death, one of my favourite play on words). This was my pitch:
A free, automated, user specific and editable channel to facilitate the discovery of entertainment on the internet.
Finding entertainment is very much a different process to an information search. For information, a user, for the most part, knows what he or she is looking for and creates an appropriate search query to find what they need. For current entertainment search processes, a user must know what they want to read, watch or listen to before beginning their search. This, in many respects, is contrary to the act of discovering fresh entertainment. In mature mediums there exist two paths — one for the acquirement of media and a second for its discovery. Internet search acts predominantly as the former similar to a cinema which shows a film you chose specifically to watch. Examples of the latter are television, radio and libraries; each provides a selection of entertainment services for the discovery of new quality content.
The internet today offers a unique worldwide medium for text, video and audio; it has quickly become the ultimate entertainment, communication and information hub. Google, with aims to organise the world’s information, has made great strides in providing tools for finding information fast, and via Gmail and Google Talk (amongst others) online communication is being made easier. With the emergence of flash streaming and legal movie and music downloads it is now the entertainment aspects of the internet that must be addressed by Google. The first steps have already been taken with Google Video and YouTube. I believe Google’s next big innovation should be to improve the accessibility and discovery processes for the wealth of quality content available on the internet, to match the second element of Google’s aim: to make content universally accessible.
The closest the internet currently has to an entertainment discovery service is “StumbleUpon” a site that allows users to recommend web pages so that readers may stumble through the internet finding its hidden sweet spots. Other undertakings include; The “Venice Project”, from Niklas Zennstrom and Janus Friis, creators of Skype, which sees their focused efforts in creating a recommendation driven Broadband TV service for quality video content delivery and discovery. Last.fm, a British based company, collates music listened to and generates free customised radio stations for discovery of new music. I believe now is the time for Google to begin developing a service in this field, as internet content begins to escape the confines of the PC box. Last.fm’s growing popularity and success is proof that there is a demand.
For Google Channels to solve these entertainment search issues it needs to offer two things; first it must deliver interesting content and secondly this must suit the user’s tastes. Google already has a head-start with regards to content delivery; its extensive crawl index, video and book libraries present the essence of a solution (though to my knowledge no Google owned music or audio database currently exists). To show content that is relevant, knowledge of the users’ tastes must be known and items must be grouped by similarity. Google Sets is already the foundation of these grouping functions — searching for television shows like “Family Guy” and “The Simpsons” yields “South Park” as a top suggestion. Knowledge of the user can be obtained through existing data or by user input.
Creating the Channel
Google’s personalized search data can give an accurate portrayal of a user’s tastes (e.g. top search terms), as does their rating trends, labelling actions and Google Talk’s music trends. Collating this data with similar neighbours and data from Google Sets, a selection of materials matching the user’s tastes is produced without needing search. Hence a user-specific recommendations channel is generated, creating the basics for a discovery driven internet service (an extension of the pre-existing personal homepage gadget “Interesting things for you”). With refinements through categorisation this channel can be split into genre specific bands.
The alternative approach is to ask the user for a few things they already enjoy — favourite music artists, websites, television shows and books may be key pointers. Using a grouping structure a list of recommendations can be generated, allowing relevant content from indexes (Books, Google Video, RSS feeds, etc) or established channels to be combined into a personalized channel. As Google Channels evolves the accuracy of recommendations will improve.
Implementation becomes the next question; channels may be video, audio or written, or an amalgamation. Video channels may be a generated stream of video not unlike television channels, with one recommended video continuing directly on from the previous — this stream may play live to multiple parties, play from a chosen point (e.g. select first video from a list), be skipped through or downloaded. Music channels may act like those on Last.fm, playing music similar to artists the user already knows, they could also be geared towards podcast discovery or online radio which may then be played in-browser. Reading lists would act much like existing feed aggregation services, however could contain a degree of useful automation and Google Book recommendations to inform users of literature, sites and news that interests them yet lies outside of their confined internet corner.
User editing adds significant benefits:
· Remove subjects that are uninteresting.
· Actions when viewing channels may include:
· “add similar items to my channel”*,
· “add this item to my channel”,
· “add items in this label to my channel”,
· “create new channel based on”,
· “label item/channel”,
· “ban items in this category”,
· “ban this item”,
· “blog this item/channel”,
· “share this item/channel”,
· “combine channels”,
· “recommend item/channel”
· “invite user to chat”
· “rate this item/channel”
· Recommendations can evolve using rating systems that promote good content and ban bad content.
· Complete control of channel could bring true “internet TV” to life.
* e.g. “add sites similar to Slashdot.org”, letting users find domains that match their normal reading materials without the trouble of building a specific search query to find them.
Once channels are created the next logical steps are sharing of channels, channel labels, recommended channels, RSS channel feeds, top channels lists, searchable channels database, public and private channels, embeddable channels, sponsored channels, channel collaboration, Google Talk integration for chat whilst viewing, content databases allowing creators to add their work directly to channels (much like the existing YouTube channels system), exclusive content or subscription channels.
Revenue streams for Google and content creators are also instantly apparent. Google would know the full specifics of the ‘now playing’ content. Hence incorporation of Google Ads is only a stone’s throw away. Video channel features can be interlinked by relevant video ads, and similarly for audio. For incentives to create channels owners may receive payments from a pay per click or impression initiative. The cost of displaying ads on a channel may also be proportional to the number of regular viewers or readers a channel has.
Creators of content (e.g. a television network) may wish to receive more return and have greater control of proceedings; a corporate control panel may be an option. Similarly an advertising control panel may be helpful to large advertising corporations that want to specifically control when and where their ads are shown.
Other benefits to Google are the natural categorisation and sorting made by users of channel content, which will not only improve recommendation quality but also add to the usefulness of Google search results; for instance a fuzzy search that returns results matching the users search aims but not necessarily matching their query; particularly helpful for those having trouble refining search terms.