It's amazing to me to still see the RDF semantic web drum beating even after its proven itself to be a failure. It's not that the idea is bad, or even that the

It's amazing to me to still see the RDF semantic web drum beating even after its proven itself to be a failure. It's not that the idea is bad, or even that the syntax lacks sugar, but it fails on three fronts: 1) People have to use it. It is completely possible to build pretty much any modern service you want, without RDF. The only thing RDF buys you for your trouble is some vague hand waving about joining the semantic web, for which no really useful services exist as exemplars of why you'd take the time and expense of building the RDF meta-data in the first place. 2) It's easier to do it the many wrong ways than it is to do it the couple of right ways. So I bothered to build a triple that says "the sky/has the color/blue" and somebody else does "the sky/is/blue" and somebody else does "the sky/is/sky-blue" and on and on. I've quickly arrived back at the same problem that the semantic web was supposed to solve, finding information. But instead of information, now I have an infinite list of qualities of the sky to deal with. 3) On single documents, the RDF data is generally crud. Other meta-data (e.g. topics, key words, summaries, entities, etc.) are far more useful than a bunch of ham-handed XML jamming up the meta-data of my document. The field itself lies at a certain kind of problem. The claim is "if we have enough semantically described data, we can do all kinds of great reasoning things with them". The problem is that there will never be enough semantically described data. At least not in the direction the field is going. Spending millions of man years building out triple-store databases just isn't really a terribly useful endeavor. Particularly when all I want to do is a find a web page about dancing dogs and dudes getting kicked in the nuts. The field itself describes all of its problems as "we just need more data". It's like the uncanny valley hypothesis. "These robots aren't human enough yet because we just haven't spent enough time making them real". When ultimately nobody cares if their robot house maid looks like a human, they just want it to take out the trash. Even if, by some miracle the world went mad for a year and RDF'd all of its web output. The result will still be only marginally more useful than the search systems we have today.