Semantic Web Adoption

As Malcolm Gladwell would put it, the question is, "Will the Semantic Web ‘tip,’ and if so, when?"^[115] "Tipping" is that transition that turns a disease into an epidemic or a fashion into a fad. For the Semantic Web, the questions are, "What needs to happen as a prerequisite?" and "What might hold it up?" In this section we'll examine some "accelerators" and "inhibitors" to the Semantic Web reaching escape velocity.

Inhibitors

Some of the inhibitors, such as the complexity of the environments and the lack of good examples, may be overcome simply by the presence of a few good books and some Web sites with some good examples as to how to put this technology to use. These items are currently missing, but I don't see any reason why they would be permanently missing.

A bigger inhibitor is the inertia of getting people to mark up their pages with this additional markup language. As of right now, this is a deterrent. Expecting the average Web site owner to correctly tag the content on his or her site, given the current support for this activity, is expecting too much. However, as we mention in the section on accelerators, motivated merchants (and others) will find a way to get this done, and an industry will sprout up to help them.

Cory Doctorow has a fun and provocative, if cynical, view of the problems. It attempts to rely on what people coding this type of information correctly will run into.^[116] One of his seven points is that people will not be objective about this coding if their incomes are at stake. The seven points read like the seven deadly sins.

Lack of reliable and useful ontologies is an inhibitor. People have to "commit" to a particular ontology or group of ontologies. There isn't much out there that most people would feel comfortable committing to. With a few exceptions in the medical and scientific fields, most of the ontologies I've seen are experiments and prototypes, and they do not have any commercial backing that would comfort those who would participate.

Another inhibitor is the fact that many end users are not competent to define semantic queries and to program software agents to do their bidding on the Internet. As things stand now, these are nonstarters. However, this could easily change with the right tools.

The final inhibitor is performance issues, or perceived performance issues. In the early 1990s the conventional wisdom was that user interfaces had to be tuned to subsecond response time to get maximum productivity for knowledge workers. Along came the Internet, and it was found that people were willing to trade the 5-to 10-second delays for access to information. The Semantic Web will go the same way: Performance will be a block until there is something worth the wait.

Accelerators

Merchants, as well as other purveyors of information and services, want to be found. They want to be found even more than the consumers want to find them. We now take for granted something that we wouldn't have believed 10 years ago: that the average cabinet maker, corner restaurant, and bed and breakfast would willingly code up a description of their offering in an arcane language called HTML, build an application behind that to allow strangers to transact with them, and pay "positioners" fees to help them climb in the search engine rankings. The tools have improved and no one codes in raw HTML anymore, but the fact remains that with their businesses at stake, businesspeople will do some strange things.

In that vein, all it would take would be a search engine that began using semantic tags for relevance, and Web sites would have semantic tags. The more popular the search engine, the more rapidly this would happen. It appears that Google has no current plans to use semantic tags.^[117] Google's owners may feel that this would level the playing field at a time when they don't want it leveled.

General Magic, the company that built an early personal digital assistant (PDA), also had an architecture for agent-based software before the Web. General Magic was ahead of its time, but time has a way of recycling good ideas. One of the issues General Magic wrestled with and seems to have solved was the problem of how John Q. Public programs his software agent if he can't program his VCR. The programming language turns out to be "I want," as in "I want a cheap vacation in Mexico." Now, where we would go with this, given that we have (or will have) ontologies and other useful tools at our disposal, would be an informed dialog with John. "When do you want to go?" "Is any area of Mexico more interesting than any other?" "What do you consider to be cheap?" And so on. Armed with this, and perhaps a few more questions the agent may uncover after its first few laps around the Internet, the agent goes out to do John's bidding.

This is far more than Priceline. This is not some prearranged deal. The agent can make any sort of offer you want, and it can be legally binding or not. But what is far more interesting than the usefulness of this to the consumer is the usefulness to the companies being visited. These companies will keep track of the agent's requests. Over time they will say, "You know, that is the twentieth request for an Aztec pottery trek we've had in the last month. Let's put one on." This is much more valuable than market research. These are real offers, or at least inquiries, that in many cases will have been dollarized.

The final accelerator will be a tool that would allow Web site developers to easily and accurately semantically tag their Web site. Tools that do most of the hard parts of this are already available from companies such as Applied Semantics, Inxight, and Semigix.

^[115]Malcolm Gladwell, The Tipping Point. Little, Brown, 2002.

^[116]Cory Doctorow, "Metacrap: Putting the Torch to Seven Straw-men of the Meta-utopia," Aug 2001. Available at http://www.well.com/~doctorow/metacrap.htm.

^[117]David M. Ewalt, "The Next Web," Information Week, Oct 14, 2002, p 40.