gibbergibber - 10:15 pm on Jul 20, 2006 (gmt 0)

"Will the Semantic Web be useful for finding the "best" widgets, or comparing prices? Probably not. But don't throw the baby out with the bathwater. There are plenty of other things it can be useful for, and for these applications Norvig's argument is a red herring. "

That's true, but then that's not a "semantic web", that's a "semantic database".

I got the impression that Berners-Lee wanted most of the internet to become machine-readable, but that isn't possible if you just want to rely on trusted sources who have no reason to lie to you.

I think in this instance Google is right and Berners-Lee is wrong: the only way to tell if something is reliable is to see how other reliable sources react to it. The reason for this is simple, contents will never be machine-readable, spammers will always find a way of disguising their websites, so only a human can actually tell what is spam just by looking at the contents.

Berners-Lee's defence, that a semantic web could tell who had written information, doesn't really make sense because he doesn't explain how you verify that authors are who they say they are, it's just as vulnerable to deception as anything else.

It's not that Berners Lee is promoting bad ideas, it's just that they're totally impractical for an open web. On an intranet, yes, but not on the internet as a whole.