Q: How do you know when your question-asking site is broken?
A: When you get 104,003 questions for the President.
I applaud the Obama administration for soliciting online questions for the President’s online town hall. And they let us all see the questions that our fellow citizens (of the US and the world) were submitting. Excellent!
But if you get that many different questions, it’s pretty much guaranteed that you really got far fewer unique questions. If people can’t easily find the question they had, they asked it again. This dissipates the votes on the questions as well.
I don’t know how to fix it other than by manual intervention, or possibly automagic natural language processing, or some such. Or maybe you could show people questions like the one they just posed (through just a little bit of automagic NLP) and offer to let them vote for those questions rather than pose their own. This might cause some clustering around questions: Why ask “You, dude, when are you going to make pot legal? PS: You can come by our place in White Plains any time if you do.” when you’re shown that the question, “Do you support the legalization and taxation of marijuana?” already has 983,455 votes?
If you head over to http://stackoverflow.com you will see that they start with the questions and a search box.
Recently on one of their podcasts Jeff Atwood(lead of SO) talked about why they did that and basically said that it was to encourage people to look to see if their question had been asked and possibly answered. They are encouraged to vote on questions and answers to push the “best” ones to the top.
Another (simple) approach is to conduct voting by presenting a random selection 5 (or so) previously asked questions and ask which one people like the best. That helps to prevent “gaming” of the system.
hmm thinking out loud:
openess principles of opensource doesn’t always scale if the community is huge.
Denial Of Service attacks work well on Single Point of Failure.
People mix Transparency with Traceability.
That is definitely a concern for many portal admins. One of our clients, Boy Scouts (http://ideas.scouting.org), uses all of our features to handle it:
1) Search box to allow the person to first search for their idea
2) The ability for the user to flag an idea as a duplicate (for the moderator to later merge).
3) Functionality whereby the moderator can easily search for and merge duplicate ideas.
Rob
This item connects well with the 3/26 posting –
http://www.computer.org/portal...../x2exp.pdf
in terms of searching natural linguistic data…
Suppose the Q&A site could, as David suggests, prompt the questioner to aggregate their questions with those submitted by others, then get bigger consensus on important issues.
What interests me in this is the possibility that such processing may sharpen points and ideas rather than give the kind of Gallup-poll mushy choices we’re used to from mass media.
[...] The Reference Section: Gene’s blog a post on this topic from our regular host David Weinberger [...]
nice, really nice!