phil ringnalda

Hey, whoa there. Always keep in mind that the rocket scientists still need astronauts to pilot their damn machines.

Yeah, that about sums up my feelings on RDF too.
Is there any wonder I switched from RSS 1.0 to 2.0 on my blog?

Is there any sort of tether on this hot air balloon? They’re so far up in the stratosphere now that I don’t know how much further they can go without dooming themselves to irrelevance.

Specifications can be complex, that’s alright. Frankly, that paragraph seems pretty straight forward but I’m an ”applied mathematician”.

Here’s a paragraph from another specification
(On Preemptive Negotiation in HTTP):

”This is the name which seems to have been given to negotiation which only uses two messages. (Not some people’s idea of negotiation at all, but the only negotiation originally designed into HTTP.) In this case, the client must transfer to the server all that the server needs to perform the calculation.

”This has to date been done in the Accept*: headers, of which there is one header type for each value of i, and one actual header sent for each value of v(i) which the client can handle (if you like, for which qv(v(i)) is nonzero), nd the q attribute on that header have qv(v(i)). If the header is given but q is omitted, it is assumed 1.

”The algorithm on the server depends very much on the sort of thingste server is serving. If the serer can do automatic conversions, it can calculate when to do them on the fly. Let’s take the case of a server which has for each value of (say) j a document whose quality is Qs(j) and a length L(j).

”The benefit of document j to the client is then

Qc(j) = Qs(j) * PRODUCT{i, qv(v(i,j)) }

”as above, and the net benefit after waiting for all those bytes is the difference between that and the cost. A negative net benefit would suggest that an image, etc, should not be shown.

”The v(i,j) are just the v(i) for the jth document.

”The Qs(j) can of course be calculated or allocted beforehand for each file. The server choses the file for which the above expression is maxiumum.”

Yeah, that’s kind of hairy, and in 1992, when it was written, you’d all probably say the same thing here as you’re saying of RDF. (”C’mon, text-based protocols? whatever. these guys are nuts.”) But 10 years later it’s a non-issue. We never have to think about things like cost metrics or negotiation in HTTP. And now we all use HTTP because other people have done the work to make it easy to use.

The same needs to happen for RDF or else it won’t be successful. Whether or not it will happen, I’m not sure. At least Mozilla uses it heavily. That’s a good first step towards decent tools showing up.

Oh, I didn’t mean to imply that there shouldn’t be specs that I’m too dim to understand. The world would be a very drab place if it only included things I’m able to grasp. I just objected to Ben’s pointed link implying that if you don’t get that abstract model you shouldn’t say word one about RSS 1.0, particularly whether or not it’s the best place to keep your RDF (I’m beginning to think that it’s a really crappy place for it, the more I try to do real work with the RDF in RSS).

Oh, I see what you mean.

I’m not sure what’s up with the smarminess, I think they’re trying to ”counteract” Dave Winer’s hatred of RSS 1.0 by claiming he doesn’t understand RDF and therefore he shouldn’t have an opinion. And even if that were true, this doesn’t seem to be the way to address that. (I liked your piece on RSS being the ”Cops” of the tech community right now).

Where do you think a good place to keep your RDF is? My understanding is that RSS 1.0 isn’t really RDF, it just sort of looks like RDF. RDF seems to be about assertions and relationships and I don’t make many assertions and don’t explicitly create long-lived relationships. In fact, I try hard to create short-lived relationships.

I’m wondering if making a site XHTML compliant and embedding RDF in there might be interesting/useful. Of course, that’d probably double the size of it and I don’t really want to do that. I suppose multiple representations of the page (index.xml and index.rdf) would make sense here.

I should finish reading this book. I might have a more useful opinion after that.

Not to try and stir the stew but I thought John Wiseman had an interesting post about the promise of RDF

Oops, got distracted for a day or two there. Where do I want my RDF? I want it in a separate file, linked from each XHTML file with a <link> tag, with stuff that describes that particular *static* resource one time in the best way I can, not a temporary file that keeps describing a few of my most recent additions over and over again in a format that’s based in part on what Netscape needed for a portal last century. If the resouce I’m describing has a title, then I’ll give it a title (and it’ll be a universally recognized dc:title, not some limited rss:title), and if it doesn’t then I won’t use a faked up title element. If a Semantic Web crawler comes crawling, the last thing I want to tell it is ”here’s the last 15 little blurbs I wrote, and be sure to come back in an hour when I’ll tell you about the exact same 15 things.”

But I’m a little bitter about RDF and RSS right now, I think.