I was reading this post by Nova Spivack when I finally understood that there are two semantic webs. Nova calls them The Intelligent Web and The Data Web. (I think this is in some way related to the longstanding debate about Semantic Web vs. semantic web, which never really made any sense to me before). To put a slightly different spin on it, I like to think of these as the Orderly Semantic Web and the Chaotic Semantic Web.
The Intelligent Web is what one might imagine when reading the scenario at the beginning of the famous Scientific American article. I believe that part of the story is meant to get the imagination flowing, more than anything else. People who take it too literally have had trouble buying into that idea of the Semantic Web, not because the technology is so far-fetched, but because it seems to require a sort of Utopian information society where all information is free and everyone has agreed on a common descriptive language to communicate that information. If people see that as a prerequisite for the success of the semantic web, it’s easy to see why they’d dismiss it as “never gonna happen”.
Read a little more deeply and you’ll discover a more realistic vision: The Data Web is a semantic web that’s more like the familiar World Wide Web, in that it can be distributed and you don’t need consensus. People can create whatever content and information they want and meaning can emerge later, when there’s enough data for new processes to dig into that data in significant ways. Google is the game-changing Web 2.0 tool that arrived to make sense of the wild west that was Web 1.0, at the time. Similarly, it may turn out that the tools that will truly make sense of Web 3.0 (in other words, The Data Web) won’t emerge until some future phase – let’s call it Web 4.0, just for fun.
So why did I rename these two models? It makes it a little easier for me to state this: the Orderly Semantic Web (roughly analogous to what Spivack calls The Intelligent Web) is never going to happen. Universally agreed upon upper ontologies? Not likely. All data open and free, and never misused? No chance. The totality of human intelligence expressed using nothing more than syllogisms and first order logic? Set your mind at ease, Clay Shirky, this won’t be necessary.
The Chaotic Semantic Web is the one I’m interested in. All we need to make it happen is for people to start generating and exposing more data and metadata. We don’t need agreement, and we don’t need to understand how it will all be used. Some time in the future all this data will be useful in ways we can’t currently anticipate, and I’m OK with that.
Sure, this is a less predictive approach, and more faith-based. I’m not saying that we don’t have to seriously consider all the challenges and issues, I’m saying we shouldn’t let the fact that we don’t have answers stop us from proceeding. Personally, I love the messiness at the fringes of things, and I’m pretty sure that that’s where the value in all this is going to come from.
[Note: I made a few small changes in the second and third paragraphs to clarify my reference to the Scientific American article. 11/27/07]
Meaningful Data 
I wonder why people keep pointing to the Scientific American article, which is where we laid out the vision of the data web, as the opposing view.Not only don’t we argue for some sort of agreed upon consensus model of AI and some sort of syllogistic model, but the article actually builds an opposing view to that — the idea is that local ontologies, built by any group of users, can be used to organize and integrate parts of the data on the data Web.I wish people criticizing our article, especially Shirky, had actually read it..
I am currently blogging my reply – look for it at http://www.mindswap.org/blog under the entry “Shirkying my responsibility”
-Jim Hendler
Pingback: Mindswap Weblog » Blog Archive » Shirkyng my responsibility
This has been said elsewhere , but just to be clear; the “Chaotic Semantic Web” is what the idea has always been.
The “Orderly Semantic Web” is a straw man. Nobody thinks that everyone is just going to magically agree on a few ontologies or even a few data formats; it’s going to be a “fractal mess” (to quote Tim Berners-Lee) for a long time to come.
I don’t know where the idea that the SemWeb is about the straw man comes from, but it seems to have taken a better hold on people than the truth.
Hi,
Actually, Tim Berners-Lee predicts exactly what you call the “Chaotic Semantic Web”. If you refer to his book “Weaving the Web”, in the chapter where he’s describing the Semantic Web vision (I think it was 13) you would realize that Berners-Lee is expecting pretty much the same thing: agreement on ontologies and meanings will emerge as people will use the said technologies (RDF, OWL) more and more. There is no need for an agreement on terms as a prerequisite for the semantic web – it’s basically a process.
It’s not 2 different things – it’s one semantic web that may evolve (hopefully) into something more orderly as more and more people and organizations begin to use the same terms (=ontologies) to describe things. In the meantime, people will just point from one ontology to another.
So for example, if I have an ontology that defines the properties of a ‘Person’ entity, I may have a ‘last name’ property.
Your ontology can have a similar notion of a Person, but with a property named ‘surname’.
So now, when we want our two programs to cooperate, we would only need to point from ‘last name’ to ‘surname’ saying that these 2 properties actually mean the same thing. There’s no need for an actual agreement on terms for our programs to cooperate.
Thanks for your comments, Lior.
Based on your reaction – and James Hendler’s, the force of which sort of surprised me – I fear that I didn’t give the proper context to this post. I didn’t mean to say that Tim Berners-Lee, James Hendler, & and Ora Lassila have this rigid view of what the Semantic Web should be. Their body of work clearly shows that they don’t – for anyone who has read it.
What I meant is that the Scientific American article starts out by painting a picture that oversimplifies the concept, and many people never get beyond that. I’ve noticed this over and over in my own conversations with people, and I’m trying to find a way to get beyond those preconceptions. In my experience, most of the people who dismiss the concept of the Semantic Web do so on the grounds that they think the intention is that it be The Intelligent Web.
By nature, I break things down and define them. If I have to split the Semantic Web into two parts, two phases, or two models in order to explain it better to people, then that’s what I’m going to do. Just because I’m trying to aknowledge and respond to these misguided notions, doesn’t mean I fall into the camps that believe them.
James,
I’ve left a more comprehensive response to your post over on your blog, but here let me just say that I did not mean to mischaracterize your article. I am responding more to the preconceived notions of people that I have actually spoken to about their understanding of the Semantic Web. Now, perhaps some of those people read the article and didn’t get much further than the tale of Pete and Lucy, or perhaps they didn’t read it at all.
I’m not saying that the ideas in the article are anything short of brilliant – I’m saying there’s still a need to bring these ideas into terms that the mainstream can relate to. Unfortunately, the part of the article that attempts to do this overreaches a little (in my opinion) and as a result, has caused some of the misguided notions that you, yourself, are so rightfully concerned about.
Brandon,
I totally agree with you that “Nobody thinks that everyone is just going to magically agree on a few ontologies or even a few data formats” and I think I pretty much said so in my article. But the fact is that there are people who think that that’s the intention and they, therefore, think the Semantic Web is impossible.
I set out to present those people with another view of the semantic web, in hopes that it would clarify some things and dispell some myths. Instead, I seem to have sparked some controversy among people who, as far as I can tell, already share my perspective.
Hi Rachel, ok, hopefully I’ve corrected my mistake (changed title of my post), thanks.
Nova
Actually, I think your original title was correct. The part I question is the idea that Hendler is refutting criticisms of the semantic web. While that’s partially true (especially in the case of Shirky & O’Reilly, it seems), his post is mostly about adamantly denouncing things that I said that he actually agrees with.
Pingback: Mindswap Weblog » Blog Archive » ..but wait, Rachel’s one of the good folks
Pingback: Jim Hendler on the Chaotic Semantic Web « The “Meta” Internet: The genesis of a “virtual” Silicon Valleys leveraging the power of the Internet.