I have to say that I was not terribly impressed with his examples of real world implementations of this standard as it applied to devices and networks. Automatic calender sync on smart phones? Don't all smart phones do that kind of thing already? Transferring photos automatically from the camera to a networked computer or device? That can be done now with cameras that incorporate bluetooth or that use something like Eye-Fi.
I also don't like the supposed symmetry or maybe it's just the terms he uses to describe this stuff. Name to content, content to name. That's a bit too facile and I think it misses the point entirely. We don't want to know what the name (translate as location, i.e. URL, I think) of the item we are seeking is. We want to know what that content is. That's more than a name however you want to describe a name in this scheme. It's really identity.
For instance, in describing Google's search engine, he says that Google basically creates an index of web pages based on keywords on those pages. However, whatever Google uses to model the relevance of hits or our own minds are left to parse if these results are what we are looking for because the actual content on the web does not represent itself to us. It does not describe itself to us. We have to infer a description from a constellation of keywords or by inspecting it directly.
But what if every webpage out there described itself and its content with sufficient metadata, then we need not infer anything but we have those descriptions as directly searchable. Would you even need a service like Google then to crawl over the web and describing everything it it's already cataloged?
Topic: Very interesting talk: (Read 2286 times)