I attended the Search SIG Audio Search event last night. Great group of speakers. Thanks to Jeff and Dave (beware of blatant ads for SimplyHired... can we trust this blog anymore? :) ) for kicking off what will surely be an amazing group going forward and a huge benefit to SD Forum members and the Silicon Valley community.
During the session, it was pointed out that the cost of every error associated with audio or video search is the time and pain of having to watch or listen to something we weren't looking or hoping for. There was a strong consensus that social networks and recommendations will play a critical role in helping people discover and filter interesting content in the future. However, there is still a big problem when users want to go beyond their social network (or at least how to help the one or two users in a group who find everything), and discussion ensued about the use of technologies and/or tagging as means for search to eliminate or lower the costs of bad results.
We know that technology has proven its worth when it comes to text search, but even when audio or video can be successfully converted to text, too much context is lost. Community tagging has been successful for photos, but one or ten words to describe a photo seems more manageable than the tags it would take to portray a 2 hour film. More importantly, in both instances, it's not that bad results don't exist, it's just that the costs to the user are so much less painful.
It seems that in audio and video, users need true and complete descriptions to help reduce pain and time wasting. No one on the panel seemed too thrilled about relying on content generators to do it. None of them wanted to call their customers lazy ignorant slobs, but we got the message. So I wonder if these audio and video communities could tear out a page from Wikipedia. In this example, a global community has delivered a commitment to providing truth and completeness to descriptions of people, places and things with stunning results.
I am sure there are some very interesting categorizations out there of the types of people who use Wikipedia (please post any you know about). If you've ever been in sales, you've been categorized as a farmer, hunter or gatherer and I do think that the analogy (not to sales, but to farming, hunting and gathering) might hold true here. There are those people who lay down the information, those who are addicted to making it deathly true and complete, and the rest of us who reap the benefit. It's not mutually exclusive, however, and Wikipedia benefits because so many people can participate in discrete areas of expertise. Sure there are bad and lazy farmers, but the hunters are always pleased to find raw content on topics they never thought to raise themselves and go to work. Gatherers of the content provide feedback or assistance from time to time in small, important ways when so moved. This balance of the various types of actors has resulted in a remarkable commitment to truth and completeness.
Not all areas of audio and video search revolve around the need for true and complete descriptions. You may not care what something's about, but you want to find something funny, or what your social network thought about it. However, when it comes to learning what something is about, and when it comes to understanding the context of audio or video, truth and completeness is critical to help lower time and pain. I think humans can play the most important role here and I think it has to go beyond tagging. Tagging will still be important both for categorization, socialization and personalization, but the wiki could be the killer app for descriptions.
The implementation scenario for these audio and video communities could be simple: make the description field a wiki. Let users self describe their content in the description field and then let the community go to work. Some rights holders will be too scared to allow it, so maybe it could be a check box for the rest of us users who aren't so uptight.
Wikipedia already provides living examples of what I am talking about (see Rambo). Sure, there will be problems initially. Jerks will mess with people's descriptions and there will probably be a lot more lazy farmers in the audio/video world than with Wikipedia (since every farmer there is by virtue of their participation a writer). However, the community will work itself out. Hunters will appear, gatherers will provide feedback and a path to true and complete descriptions for the mass of audio and video that is being unleashed might be formed.
Look Ross, I caught wikifever!
Comments