I was part of a panel discussion yesterday at terrific seminar in Sydney about commercializing online and mobile video.
It was the inaugural "Video Rules" seminar, organised by AIMIA and Venture One.

On the panel were representatives from Akamai, Adobe, Google, the Australian Broadcasting Corporation (ABC) and Pixelmill Films.

On the question of how to optimize video for search engines, there were a couple of interesting perspectives.

A couple of the panellists thought it will be speech-to-text technology that drives the optimization of video for search engines, and makes the spoken word discoverable.

I took a slightly different view: that speech-to-text might be great for generating transcripts that can be published as HTML pages, to create a keyword longtail for the publisher.
(Capturing more viewers as a result).
But for as long as search engines behave as they do today – constantly changing and fine-tuning their algorithms – automated metadata extraction techniques (like speech-to-text) will lack the precision to keep up with all the changes to algorithms.
Which means there will still be the need for a high touch, ‘organic’ or manual dimension to how we optimise video at the pointy end - on top of the longtail embodied in static transcripts, for example.

Of course, I would say that: Tagmotion is all about manual tagging. 
Ray Kurzweil has an opinion about the shelf life of manual processes like pattern or face recognition........
that is, how long we've got before machines can do the job as well as we can.
That also defines the shelf-life of the Tagmotion technology, incidentally.
Gulp.
More about Ray's prognostications in my next post.