- The Semantic Web - Can it Really Work? -
Someone recently sent me a link to a fascinating article by Tim Berners-Lee, the guy who invented the World Wide Web in 1989. These days, apparently, Tim and his friends are working on something called The Semantic Web, which (they claim) will revolutionize the very thing that revolutionized communication less than a decade ago. I’d encourage everyone who is fascinated by this kind of stuff to read the entire article, but just in case, here’s a quick summary:
The current World Wide Web is a vast collection of documents that are generally meant to be consumed by humans, not computers. Automated search engines, therefore, are doomed to consistently return some percentage of false positives (e.g., an article about 4-man singing groups to someone looking for a local “barbershop”). The Semantic Web would tag data with descriptive information (called an “ontology”) so that an advanced form of search engine (called an “agent”) could perform this function more accurately. So, the barbershop quartet would be tagged as a singing group, the local barber would be tagged as a business offering haircutting services, and the search “agent” would determine (or ask) which the user is looking for and disregard results relating to the other. The authors go on to predict that given this new “understanding,” agents will be able to accomplish much more than simply searching for information. They say that agents could eventually string together common sets of tasks, and perform useful functions, such as arranging a doctor’s appointment (search for the nearest doctor, search for your insurance coverage, search for an appointment time that is free on both the doctor’s calendar and yours, book the appointment, send you confirmation on your preferred device).
Now, I realize that contradicting Mr. Berners-Lee is something one does with great peril. After all, everyone probably called him crazy when he said the World Wide Web would take off, and today, many folks couldn’t imagine their jobs (or even their lives) without it. I also realize that much of what I’m about to say is 20/20 hindsight (the article was written almost a year ago, suggesting that things are at best moving slowly). That being said, I wouldn’t dare suggest that the Semantic Web won’t happen. Instead, I’ll simply suggest what I believe to be some of the challenges that it faces that the World Wide Web didn’t face in the early 90’s.
People are Watching
The most glaring challenge, I believe, is the success of the World Wide Web itself. The World Wide Web was created when nobody was watching. The lack of a spotlight allowed Mr. Berners-Lee to set all the original standards, and then fend off (or incorporate) changes as its popularity grew. Today, successful applications of his invention have made the mainstream press hypersensitive to issues like data privacy, data ownership, and over-automation. Current writings on the subject seem to address the Semantic Web based on what’s most efficient or most technically feasible. They ignore some questions that simply weren’t asked last time around:
With the World Wide Web, these kinds of issues were only broached after the technology had reached (passed?) critical mass. Their answers have, and will continue to alter the speed and degree to which the technology is applied, but they are no longer able to prevent its creation or precipitate its demise. The Semantic Web may need to deal with these issues much earlier in its young lifecycle.
Money Makes the Web Go Round
Money is another challenge that the Semantic Web will have to deal with earlier in its lifecycle than the World Wide Web did. Mind you, I’m not talking here about the money required to build it. I firmly believe that the world’s universities, governments and corporations have and will continue to contribute. Instead, I’m talking about the willingness of service providers and consumers to participate in it without first being assured of additional profits or improved service.
It’s hard to conceive of it now, but for the first 5-7 years of the World Wide Web’s existence, everything on the web was free. Universities provided electronic access to their libraries, even if you didn’t pay tuition there. Brokers gave away stock quotes – real time or otherwise. Newspapers and magazines made the full text of their publications available for free, even though others were paying for it in paper form. Software was freely available for download – no strings attached. Small businesses and private individuals created web pages just to say “Hello World!” and didn’t concern themselves with what it cost or what the benefit was. These entities were willing to use the web for the satisfaction of proving that they could, or for the “leading edge” reputation that came with participating. In fact, when companies first began charging for web content, there was a great deal of skepticism that the public would accept it at all (the Wall Street Journal’s subscription based web site comes immediately to mind). Today, several companies generate more than $1 billion in revenue on the web each year (some can do it in just one quarter), and charging for web content is not only acceptable, it is essentially the norm. Just about every site you visit these days is either willing to sell you something, or will link you to a site that does.
The Semantic Web will not have this 5-7 year honeymoon. People will immediately begin looking at its potential for profitability. Take the doctor’s appointment example from above: for the doctors and insurance companies to be included in the search process, each of them would have to make some investment in the new technology (software, hardware, or both). Many will be hesitant to do so unless they can be convinced that it will generate more business, or at least lower the cost of their current business. The people who need doctors will only use the new “agents” if a substantial percentage of the doctors in their area are participating. Otherwise, they’d be better off using traditional search engines (or the yellow pages!) and booking the appointment themselves.
This challenge can be addressed using artificial “test beds” – closed systems that are willing to try the new process and work out the kinks for the sake of research, not profit. Universities immediately come to mind. If a university could deploy a “semantic intranet” that helped students register for courses, purchase books, or achieve other common goals (the semantic dating service, perhaps?), then a model of success could be established, and others may be willing to give it a shot.
Interface, not In-Your-Face
Back in 1999, I wrote “The Mom & Dad Test” about technology’s need to be invisible to the non-technical user. In it, I suggested that the web browser would soon be surpassed as the primary vehicle for web content delivery. Today, we have WAP-enabled cell phones, WebTV, and internet-enabled radios, cars, pagers, and personal digital assistants. None have replaced the browser in terms of dominance yet, but I think the principle still stands: if you’re going to solve a problem that I don’t know I have, you better do it in a way that is extremely easy for me to understand, extremely simple for me to use the first time I see it, and completely independent of me understanding how it works.
While the browser may not be the ultimate delivery tool, the World Wide Web has passed the Mom & Dad Test. My parents, historically technophobic and unwitting namesakes of this metric, have come to understand that if they point at the phrase “Today’s New York Weather” and press a button, the device they’re using is going to tell them what the weather is like in New York. This is certainly a problem they didn’t know they had (they’ve been very happy watching the weather report on the news all of these years). On the other hand, it’s very easy to understand (this tells you what the weather’s like right now), it’s simple to use (point here and press this button), and neither of them know, nor do they care, that the information is traveling to them from the National Weather Service’s web server farm, through Yahoo’s data feed, across a world wide network designed to withstand a nuclear attack in thousands of pieces called packets, which are then reassembled and displayed to them on their screen as cute little pictures of suns or clouds.
Passing the Mom & Dad Test will be another significant challenge for the Semantic Web. In his article, Tim Berners-Lee provides a more detailed “booking a doctor’s appointment” example than I did above. When I read through it, I couldn’t help but notice that the users, Pete and Lucy, seemed acutely aware of what their agent was doing “behind the scenes.” For example, when they weren’t satisfied with the results, they instructed their agents to change their “search parameters” and re-execute the search. I realize the example is meant to show the full power of the agent, but if you ever say the words “search parameters” to the Mom’s & Dad’s of the world, they’ll be pulling the yellow pages out of the closet faster than you can say “ontology.”
It is important to realize that the absence of a service like this in people’s lives is a problem they don’t know they have. Many of them are just as happy calling around to various doctors’ offices as they were waiting until 6pm or 11pm to find out if it’s going to rain tomorrow. To succeed, therefore, the agents built on the Semantic Web need to work just about every time, require no training, and display both their questions and their answers in a way that people are already able to comprehend (e.g., through an automated voice on your current telephone).
God’s Speed, Tim Berners-Lee
Allow me to conclude by saying that as a consumer, I hope Mr. Berners-Lee succeeds. If I could magically create a world that included these kinds of services, I would do so in a heartbeat. As a technophile, I also hope he succeeds. The possibilities created by the existence of such a network would be as exciting, if not more exciting, than the World Wide Web that preceded it. These statements suggest that either a) I’m a member of a strange minority, or that b) one day, he will succeed.
If the history of innovation has taught us anything, it is that technology rarely stands in the way of achieving our goals; it merely dictates the timeline. I don’t know what the timeline for inventing the Semantic Web will be, but I do believe that once it’s here, there will still be plenty of work required to make it a reality.