information is not directly disclosed to other users. However, these solutions work by blocking unwanted exchanges, where unwanted is usually defined by how costly the exchange is to the answerers. Even if the models are 100 percent accurate, mediation may not be ideal.

Consider a social Q&A service designed to be used by askers and answerers with some sort of higher-level shared goals, such as employees within a company. There may be scenarios where answerers should be interrupted to incur an immediate cost in order to help another who is working on more important and urgent tasks.

port help-seeking and help-giving. I found that allowing people to pay for help can indeed increase overall welfare for parties involved, although there may be overhead costs in deciding the value of getting help and the cost of giving it [ 11]. If users are expected to make these types of decisions many times a day, a complicated market may actually incur too much overhead cost and reduce potential gains. Related studies on pay-for-answer Q&A services also show that paying can improve overall answer quality [ 7, 8], and the length of the answers [ 9], but not necessarily the quality of the single best or chosen answer [ 9].

One area of research that needs further exploration is how financial

rewards impact who uses these systems and the social relationships between users. Are these users more financially motivated than users on free social Q&A services? Related research in behavioral economics suggests that financial rewards can change the framing of the interaction between people, from social to transactional [ 11, 12]. Using financial incentives for Q&A may reduce the amount of social interactions that are vital in sustaining an online community. Further research is needed to determine if this also applies to social Q&A sites.

❝How can we better design
social Q&A sites so that they
are more sensitive to users’
needs and constraints?❞

Market Pricing

Market pricing offers a third approach to improve existing communication technologies. The basic idea is that the askers need to pay or commit to pay the answerers for the answers. Frivolous requests may be reduced because the incremental cost of sending a question forces initiators to be selective, sending requests only if they believe the value of the answer is more than or equal to the price [ 6].

Also, the use of pricing can allow initiators to signal the value of communication abstractly, without disclosing personal information. Additionally, financial rewards may attract more answerers. This idea has been explored by Google in Google Answers and is currently incorporated into various pay-for-answer services, such as Mahalo Answers, Just Answers, UClue, and AskBright.

I’ve also demonstrated a pay-for-answer service with a real-time communication channel: mimir [ 7]. Question askers can broadcast their questions to answerers directly, while offering mims, the virtual currency. On the other side, answerers can filter incoming requests based on the mims. When they do choose to answer, and if their answers are chosen as the best answers, they are rewarded with mims. Figure 1 shows the mimir interface.

Does paying help? In general, research indicates that financial rewards do affect the quality of the answers [ 8, 9, 10]. In a laboratory setting, I explored the use of markets to sup-

Applying the Solutions

Of the three types of solutions to improve social Q&A— context-sharing, intelligent mediation and market pricing—context-sharing may be most appropriate for Q&A services to be used with close family and friends who would use the contextual information in a positive way. Intelligent mediation can help Q&A between strangers as it reduces unwanted communication requests.

However, users may prefer to handle certain requests themselves when communicating with others with shared goals, as there may be cases where partners’ gains outweighs one’s own costs. Finally, market pricing ensures that both askers and answerers benefit from the exchange. It’s important to note that the market pricing system does not need to use real money. In a corporate setting, company store credits can be as the currency of exchange, and tokens can be used when interacting with family and friends.

While these three solutions are fairly distinct, they are not incompatible. In fact, perhaps the most optimal solution is to leverage a hybrid system. For example, one could utilize both intelligent mediation and market pricing for social Q&A between strangers or other weak social-ties. As the relationship between users strengthens, the system could be gradually transitioned to use context-sharing and simple pricing rules with tokens.

This design will protect users from spam when interacting with strangers but offers flexibility when interacting with closer and more frequent communication partners. While more research is needed to explore hybrid strategies, it is cer-