In normal circumstances, I have the highest regard for Clay Shirky. A copy of his book “Here Comes Everybody” is making its way to my great delight to the top of my reading stack. His talk at the London School of Economics is a classic lesson in presentation and delivery.
However, his article on the Semantic Web is a bit of a shocker. He begins the the article asking “What is the Semantic Web good for?”, a phrasing strongly reminiscent of Edwin Starr asking “War: what is it good for?” which of course demands the response “Absolutely nothing.”
To save us all from chanting in unison he provides us with his own answer. “The Semantic Web is a machine for creating syllogisms.” Even for 2003, when his article was written, that wasn’t true. The stated aim has always been to make information more relevant by making it more meaningful and more accessible.
At first I thought that Clay had made the error of mistaking triples for syllogisms, since at first glance they have the same form as a syllogism. But one only has to realise that what something looks like doesn’t necessarily mean that is what something is.
Syllogisms were what we would call in our post-Einstein days thought experiments written up in two books by Charles Dodgson of Oxford University in the late 19th Century. We know him better today as Lewis Carroll, the writer of Alice in Wonderland and other stories. They are an intellectual conceit. Triples, on the other hand, while coming in three parts like a syllogism, have a structure and content that enables Semantic Web developers to create and use them within a given context. Syllogisms are based on their own internal logic and usually go nowhere. Dodgson himself referred to them as sillygisms.
Clay goes on to note, quite correctly, that syllogisms aren’t very useful. But since neither triples nor the Semantic Web has very much use for syllogisms this is hardly relevant.
He then goes on to talk about the Semantic Web’s proposed uses. He mocks the idea that users may have to fill in multiple fields to retrieve information. He indicates that Jeff Bezos of Amazon won’t be losing any sleep over this. Since this refers to anything that would help customers find the items they are looking for, then I cannot see how Jeff Bezos could possibly be threatened by the arrival of a new and better technology.
Amazon’s search engine works fine (up to a point). All you are doing is matching a search query with a database of items you already have. But what if the item, subject, object you were looking for had parts or aspects of the answer needed spread over several databases. Databases that in a pre-Semantic Web world could well be oblivious to each other’s existence and with the corresponding inability of the data contained inside them to be accessed or made relevant in any kind of useful way.
The Semantic Web is not the replacement of what we have now with something shinier and newer, but is essentially a variation on a theme. It is a progressive step forward into the future of web usability and information retrieval. So to do something more complicated or which requires deeper digging than just a single entry search on Amazon or Google, then it may be necessary for people to fill in an extra couple of boxes to help the system to help them find what they want. Really, that is not such a big ask, is it, especially when you consider what you get in return?
I have to quote the next part directly:
This example sets the pattern for descriptions of the Semantic Web. First, take some well-known problem. Next, misconstrue it so that the hard part is made to seem trivial and the trivial part hard. Finally, congratulate yourself for solving the trivial part.
Apart from the kettle calling the pot black this is just mean and Clay should be ashamed for himself for descending to common room sneering. But the question starts to arise: why is he so down on the idea of the Semantic Web? Since he has already misstated the meaning of the Semantic Web, I am beginning to think that he is taking a contrary position for the sake of contrariness. (One cannot tout oneself as a thinker if one is thinking the same as everyone else.) The mean comments may have been intended to provoke a response (or a flame war) from interested parties, but it looks a lot like grandstanding to me: grandstanding in a very unpleasant manner.
Before we know it, he is back to misrepresenting syllogisms as Semantic Web thinking, and he goes on to the next paragraph which is headed “Meta-data is not a panacea”. And why is that? We’ll never know, as he is back to abusing syllogisms once again. After that, he turns his focus on to ontologies. (To be fair there is a debate still going on about their necessity or the degree of their necessity.) However, he claims that we don’t need the Semantic Web because we have RSS, and because RSS was easy to develop and the Semantic Web isn’t, therefore it must be superior. He asks us to look at how simple RSS is compared to the Semantic Web. Unfortunately, RSS is for distributing already found information. It is not a lot of use when trying to organise a trip to Finland by accessing airplane databases, ferry databases, hotel databases and so on.
Clearly since Clay has nothing much to say about ontologies, he uses RSS as a counterpoint. Which, unfortunately, for him it isn’t.
In the next section, he chooses to put the Semantic Web in the same box as Artificial Intelligence (AI), claiming that AI was found to be sub-optimal which in turn will be the fate of the Semantic Web. I don’t want to start defending AI in this article, but suffice to say thanks to developments in neuroscience, AI development is thriving quite happily. In short, the brain in a vat has been set aside and work is now focused around the idea of embodiment. As far as Semantic Web development is concerned, there is no objective place where you can say that the work has reached anything like a dead end. Obviously, there were plenty of sub-projects that didn’t pay off, but that is the nature of all experiment and development.
Finally, the conclusion isn’t really a conclusion. Since Clay has clearly misunderstood the role of the basic technologies of the Semantic Web, citing their irrelevance is irrelevant in itself.
So why have I taken so much time to take down the work of someone I usually admire. The answer is that what he says is important outside and what he says is taken seriously by business leaders and other leading thinkers. The fact that this article is seven years old does not matter because as we know nothing on the Internet ever goes away. If you type in “semantic web cricticism” to Google, Clay’s article appears as the first result.
But to contrast all this negativity, I would like to cite a few examples from the recently published Pew Research Center report entitled “The Fate of the Semantic Web“. It is an easy read and it is well worth downloading the PDF.
In this very nice quote, the writers summarise why it is so hard for people outside the environs of Semantic Web research like myself to get a grasp of what it is about.
The concept is so revolutionary that people have difficulty describing it in just so many words and its proponents self-consciously struggle to describe the ‘killer app’ for the Semantic Web that will make users understand its power – and support its creation.
Its ineffableness is what makes it so hard not only for the general public but also for database owners to grasp that it is a really, really good idea to get that data tagged and the various protocols installed.
This is why it has been slow to take off, but also why it will never be left to gather dust on a shelf somewhere. The Semantic Web is just too powerful a technology to fade away, but it needs awareness and acceptance. And as the quote points out, without the killer app it is hard to incentivise people to climb on board the train that leads to greater accessibility and relevance. It is happening but slower then it could be.