While librarians across the country wrestle with virtual reference desk service schedules, Wondir, Inc. (http://www.wondir.com) has come out of a 2-year beta and launched a live Q&A service that “connects people with questions to people with answers.” Using a combination of search engine, instant messaging, and other technologies, it aims to allow “answer-full” volunteers to provide personal responses to queries and to generate automatic links to appropriate Web sites and news articles. In the course of development, Wondir has already tallied more than 1 million Q&A base pairs with answers coming from 100,000-plus people—more than 75,000 of whom have registered in the “Wondir community.”
Matthew Koll, who was formerly founder of Personal Library Software and Fellow at America Online, is founder and CEO of Wondir. Koll said: “There are millions of people in the world who are eager to share their knowledge, experience, advice, and opinions—[and] who want to reach out and help other people solve problems and answer questions. The key is getting the right question in front of the right people at the right time.” Koll emphasized the community aspect of Wondir. “Think of Wondir as a virtual town square. And the more crowded it gets, the easier it is to get the answers you need.” Laura Horn, co-founder and director of Wondir, calls it the “helpfulness gene.”
The technologies integrated by Wondir include advanced search, instant messaging, e-mail, database management, categorization, browsing, reputation systems, and push and pull—not to mention extensive anti-spam filters. When responding to questions, Wondir includes not only answers from people, but also documents retrieved from a search of rotating Web search engines. The search engines include some from German-based answerbus.com, AlltheWeb, some vertical search engines, and even an occasional Google search. Although Koll considers the search engine results “the least interesting part of what we do,” the completeness seems to give a certain gravitas to Wondir’s answer pages.
Wondir integrates answers within a search engine model to guarantee speed, ease-of-use, and to support a business model based on targeted advertising and sponsored searches. Unlike Google Answers, no monetary bidding for answers occurs in Wondir, although Wondir indicates that some answerers may receive “tips” from satisfied users. According to Horn, however, the PayPal-based tips are rarely used. Psychic rewards—a thank you, a good rating, feeling smart and generous—seem to fuel the online community, according to Horn.
These days, the Web appears to abound with question-answering services. Even Google has recently integrated a Q&A feature for short reference-type answers. However, Koll said he really doesn’t consider other services as competitors. The other services use different approaches. For example, Google Answers has a coterie of some 500 people who negotiate their answer services; Yahoo! Advice uses a similar approach. Although Koll appreciates the use of experts, he wants to offer “much broader experience, both reference knowledge and life experience. We’re more open. There’ll be a lot of value if we can make it work.”
Some specialized services offer an array of tools similar to Wondir’s, but in very specific niches. Other competitors, such as Advicenators (http://www.advicenators.com), Answerbag (http://www.answerbag.com), or LookSmart Live (http://live.looksmart.com), may have some of Wondir’s features, but not all of them. Businesses can create Wondir profiles that direct possible customers to their sites for further information. Wondir also supplies a syndicated service to other Web sites—blogs, search engines, social networks, organizations, businesses, etc. Examples of such syndicated partners include http://www.ichef.com, http://www.theautochannel.com, and http://www.ratemyteachers.com. Other Web sites can add a Wondir question box and/or a customized “stream-of-questions ticker.” Wondir partners may also create private categories and link to parties in the Wondir community. Koll believes that these partners may form more effective subcommunities of answer sources in time, especially as customized tickers present only focused subsets of questions.
According to Koll, answerers have four ways to see questions: by checking the ticker, by browsing a QuestionBoard by category and paging through 20 questions at a time (currently the most successful route), by forwarding e-mails of questions in a category (Koll’s preferred route for the best answers), and by users pushing questions to someone recommended on a results page. In the near future, Wondir may officially launch an RSS feed; unofficially, Koll said that RSS is already available somewhere on site.
Privacy protection is set by the answerer. If registered members choose to divulge information about themselves in a WonderID profile, they may. In fact, they may create more than one profile identity and supply different information in each profile. Wondir lets people promote themselves, their businesses, their causes, etc., using profiles. Koll hopes that questioners will get in the habit of checking out profiles and use the information to evaluate potential answers and answerers.
Once a Q&A session concludes, the transaction records become immediately searchable for anyone with a similar question. A question control panel shows questioners all items they have asked in the past. Even unregistered users can click on MyWondir and find answers to questions they have asked that took some time to receive answers, according to Koll. Wondir uses a cookie to identify people.
Free registration allows users to become members of Wondir; they’re then eligible for special features. For example, registered members can create profiles linking to their home pages or blogs or some promotional effort. They can set up New Question Alerts to ensure notification when someone asks a question that matches their interests. They can have answers to their questions automatically e-mailed to them.
Members can also rate the quality of answers. Highly rated members will probably draw more questions. The rating system tracks around two factors—the volume of answers given and the helpfulness (quality and value) of answers. Green stars appear next to answerers’ listings when ratings are available. Wondir also has a few dozen moderators who help build the taxonomies and monitor performance in key categories.
Scanning questions listed in a category nearly always finds some eyebrow-raising—if not eye-popping—discrepancies. (“Should I make a career in healthcare?” appears in the Movies category. Looking for the job page at the Yakima Canutt Memorial Clinic for Injured Stuntpersons?) Koll says that some of the problems result from the inevitable mechanical confusion caused when an auto-categorization program encounters human language. But one source of confusion, he admits, stems from a procedure they have often discussed but have decided to leave unchanged. If a questioner selects a category and then enters a question, Wondir leaves the question in that category, regardless of whether it fits. So questioners scanning questions posted in a particular category who suddenly think of something they’d like to know in another area may just go to the question box and enter it without changing categories. Koll says that the system tries to prompt people to double-check their category choice, but he admits people usually ignore the prompt. Failure to properly categorize and keyword one’s questions means that they may never reach the most expert subcommunity of answerers, warns Koll. If customers disagree with Wondir’s automated categorization of questions, they can correct the category assigned to their questions. The Wondir staff also categorize Q&A \ pairs into the “Wild View” or the “Mild View.” This can help users avoid material that they might find offensive. Wondir also has a reporting system set up for abusive responses, although they warn users not to report poor answers as a form of abuse.
As the press release announces, however: “Anyone can ask. Anyone can answer.” I pointed out to both Horn and Koll that some people—information professionals come to mind—might not consider that policy something to advertise. Both Horn and Koll recognized the quality issue and spoke about how they hoped to deal with it. Horn hopes that organizations—businesses, government agencies, etc.—will come to them directly and help them serve potential customers and constituents. In fact, she hopes that librarians who work virtual reference desks may become partners. Koll said that initially Wondir had hoped to build on and around the virtual reference community, but that nothing they did seemed to “find traction” with virtual reference librarians. So Wondir decided to go its own way, but, according to Koll, “it is important to us to get our facts and tone right.” He still hopes connections to the virtual reference desk community may grow in the future. For example, he would love to share the Q&A base pair database built by the Library of Congress and OCLC’s QuestionPoint project. He approves of anything that builds their community and provides validated sources to produce better answers. The balancing act, according to Koll, is to acquire the “most authoritative answers, while still keeping it easy to use.” Koll considers most “Ask-a” services to be “cumbersome and slow, while Joe Consumer wants instant gratification.”
Commenting on Wondir, Steve Coffman, vice president of product development at LSSI and a controversial, vocal observer of virtual reference services, said: “I admire Matt Koll for trying. It surprises me that he has gotten as much traffic as he has. On the other hand, if I have serious questions, I want expertise. Who are these people answering Wondir questions? Who has time to sit down and even put in a detailed profile? Do these people have a life? I don’t have the time to sit down and answer all the questions I get, and I’m not sure I would trust anyone who did.” On the other hand, Coffman salutes the thousands and thousands of questions Wondir already gets each day. He pointed to a study by Joseph Janes in which 162 virtual reference desk operations reported a combined total of some 2,600 questions per day.