Yes, I understand that text search remains the single most powerful tool for finding information online. That blogs are discovered through search.

I know also that developers are working the problem of auto-generating transcripts of video, that can then either be included within a videoblog post or embedded within the movie as metadata with hooks to the outside that would make the transcript searchable.

At the Vloggercon session called “Brainstorming the Future“, someone suggested that even now videobloggers should voluntarily post their own transcripts so that their content is searchable. (I don’t remember who suggested this, so I suppose having a transcript of this session would be really helpful to me making this point right now).

I even posted a transcript of a recent video I’d done that wound up being ten minutes long, because it contained a lot of ideas that I feared would go unheard by those who didn’t want to sit through a ten minute video. And sure enough, the transcript was linked to by two people within hours of my posting it. (Number of links to the original video: 0.) Additional comments followed. It’s also shown up in a Technorati search as I was looking for background for this very post.

The problem is, video is only sometimes about ideas or information or “content”. Sometimes it’s about laughter, or beauty, or delight.

Consider music. I am not going to find my next favorite band through a websearch of their lyrics, whether that search is based on linking algorithms, or social recommendations, or tagging, or what have you.

Similarly, I would never have discovered one of my favorite video series of the year through search.

People discover new music in a number of ways. My own sources include: friends (recommendations, mix CDs, etc.), reading about it (magazines, blogs, etc.), or hearing it for myself (broadcast / online radio, selected or filtered at some level by a human disc jockey or programmer). In all cases I need to hear it for myself to assess whether I like it. The experience of sampling additionally helps me contextualize future recommendations from the same source (friend, writer, dj), so that for example I know my pal Russell is a good source for power pop, but he also has a weakness for jPop that I do not share.

I understand that there are key differences between music and video–music can be enjoyed while doing other activities, and also I am more likely to enjoy a song over and over again, both unlike video.

But I believe that I will come to rely on similar systems of targeted recommendations to help me sort through the expanding universe online/IP-delivered video.

More effective than search, then, are Amazon’s “People Who Viewed this Page also…” and TiVo, and Pandora, and other experiments that aggregate and track the behavior of crowds to generate socially-driven, automated recommendations. Some of these are pretty successful, because they fill in a portion of the gaping hole in search–the vast universe of ideas, creativity, products, that are unconnected to anything that I already know. Beyond what I can currently imagine I might like.

But of course the problem here is that each of these systems is closed. Amazon can’t adjust its recommendations based upon what a user has purchased elsewhere (in a store, or through iTunes), although it attempts to do so with its “I already own this” button and rating system. But in general you’re dealing with a database that tracks the behavior of users within a single technological ecosystem (or destination site).

But maybe YouTube can reach enough ascendancy in its volume of users to build such a system of recommendations. Or Google Video. Perhaps MySpace will figure it out. I’ll have more reflections in a future post on how some current technology solutions for recommendations are working, and how they might be improved.

But of course as functional as such a system could become, there’s still the thorny issue of a sustainable business model. Most of the videos people are watching online are unconnected to a financial transaction.

Developers working on video transcripting, random access, and search capabilities have no such concerns. There are millions of hours of corporate communications, conferences, university lectures, training materials, and other easily monetizable videos containing taggable and searchable “content” that is currently inaccessible and offline, even within corporations’ closed intranets. So go ahead and keep working on that.

I expect you’ll learn things along the way that can be of use to the freeloaders out here on the WWW. We’ll sure appreciate that.