When I read a recent article on Facebook (Nasdaq: FB)’s plan to offer a question-and-answer feature (as do several other players in the search space), two thoughts immediately ran through my mind.
The first was to wonder about the possibility that the proposed model will actually work, and the second was how it might make money. (The skeptic in me looks for the problems before embracing the possibilities.) Here is my take on the short-term outcome of Facebook's plan.
The initial difficulty of semantic search, which would be used to help users get answers to their questions, is for the technology to correctly understand the question. Here is a case in point. I typed the following question into a standard Google search box: “Who said showing up is half the battle?”
The first of 1,180,000 hits was a link to WikiAnswers showing that this was an uncategorized question, meaning one that no one had answered. Refining the search with quotes around “showing up is half the battle,” the WikiAnswers entry still rated high, number 2 out of 4,990. Scrolling down that list, I discovered that Woody Allen said that “half of success is showing up…” This is plausible; I have no idea whether it is correct, but it was the closest that I came to attaching a name to my quote and question.
I did not ask the original question for rhetorical reasons. I wanted to use the quote and give credit to the author. The query to Google illustrates the difficulty that traditional search engines have interpreting questions. They are excellent at finding strings of characters, and when you put quotation marks around phrases, the results are generally excellent.
In the example, I was not looking for all the places where my quote was found on the Internet; I was looking for the original source so that I could give it proper attribution. Semantic technologies are getting better at interpreting questions, and it looks as if WikiAnswers understood what I was asking and knew that it did not have the answer in its content base.
The reason for using the quotation was prompted by these questions: How can social “answering engines” work if no one shows up with the answers? Will we be able to trust the authoritativeness of answers found through a Web of social sites from which the contributors will presumably come? What will motivate people to contribute answers to trivial questions or even important ones? At what point does the complexity of a question, or the way it is phrased, eclipse the ability of the semantic engine to interpret it? How will the engines correctly target the appropriate social sites where answerers would be found?
As with all semantic software technologies, there is a tremendous amount of trial and error, testing and tuning, required for users to ask questions the right way. There will be even more challenges for the software to get to people with the right answers. How many adopters of the technology will still be around when, and if, successes and reliability begin to accrue?
It did not matter to me if the correct quote is “half the battle is showing up” or “half of success is showing up,” because the sentiment is still the same. Mine was a trivial question, but for millions of important questions, will a critical mass of people show up in social spaces with the right answers?
If not, interest will soon wane, and widespread use will be limited to special applications. Some of these already exist to address very serious challenges. You might want to take a look at one good example, Innocentive, which uses social tools and search creatively to bring specialists to problem-solving.
The answer to my second question, “How will they make money?” is easy: Charge for answers. But if questions are not correctly understood and the right people aren’t found to answer, we are back to a tough spot, and this is not a short-term winning business model.
What about long-term possibilities? Niche applications might be the successful opportunity to pursue. That’s where I would place my bets on most semantic technologies.
Greg Sterling, founder of Sterling Market Intelligence, who was quoted in the July 26 LA Times article cited above, probably has it right: “This whole notion of Q&A and human-powered search is really valuable and powerful, but not so far done in a way that is effective across the board. It's really about how they can provide targeted responses to questions people have. You have to provide a great experience every time.”
— Lynda W. Moulton consults at LWM Technology Services on knowledge management strategies for enterprises.