4 ms·
A game-based approach to question answering services.
To play this game, you would submit rules consisting of query patterns and answers.
Sample rule:
Query pattern: when dawkins OR evolution book
Answer: September 2009.
When a person enters a query (e.g., When will Dawkins' new book be released?), you get back all answers of matching rules ranked in a way that takes into account not only the pattern match relevances but also the number of rules yielding each answer and the reputations of the rule makers.
You would then identify the best answer, which would reward all submitters of rules having that answer and penalize all submitters of rules having other answers.
To help users decide which answer is correct, one could show them the rule patterns so it would be more obvious whether an answer is intended for the query given.
What is the incentive for picking the best answer?
One possibility is to keep track of your question and answer for future reference. Another possibility is to improve future results by rewarding those rule makers who tend to give you answers that you like.
How would players know what users might ask? You could have a ranking of popular query topics for the last hour, day, week, etc. In this way, players could focus on the sorts of questions that are likely to give them more points.
Players would get feedback on how well their rules are doing so that they can try to improve them.