Now, I'll happily admit that I didn't watch his whole video... I've got a very heavily interrupt driven life right now, and 20 contiguous minutes aren't easy to come by. What I did see seemed off the mark to me at the time, so it went into the background.
Dennis Howlett however provided this bit of catalyst which brought it back into focus:
I spoke to Robert about this in the context of what matters to business people. He agrees these services are not ripe for prime time business use. He also agrees that humans don’t scale - the premise upon which Mahalo’s human built directory operates. But neither of these things are the point either.
Human's don't scale? That's big news to me... do you think that machines spit out ALL of the web? It may be true that spam engines make up most of it... but the VALUE is all from people.
The information about what's a good page, and what is crap does exist.. in the minds of whoever just looked at the page... the trick is to extract it at low cost, without opening the door to gaming the system too much.
I've tried (and obviously failed) to explain the idea of adding external data to existing web content... and this is a near perfect case for it. If we can somehow capture the a "vote" on a page, but lock it into identity to prevent gaming... we can continuously add value to search results.
Of course, this is the least amount of data possible to capture... one bit... but linking it to identity makes it far harder to corrupt or game. If you can see the value in this... why not go the rest of the way and add tagging and commentary to the mix as well?
I hope someone makes sense out of this... I look forward to feedback.