During the presidential debate tonight, I'll be doing an experiment with some of my NPR colleagues. As we did during the VP debate, we're inviting Twitter users to help us fact-check the candidates' statements. So if you hear them say something that seems inaccurate, prove it. Try to track down a primary source that sheds some light on the claim one way or another - a speech transcript, YouTube video, etc - and tweet the URL with the tag #factcheck. We'll then monitor the results and use them as we do our own factchecking on the NPR Vox Politics blog.
Meanwhile, another experiment occurred to me: can you use Twitter as a form of distributed dial-testing during the debate? If you've watched CNN during the previous debates, you may have noticed the dial-test data they display on the screen. A group of people are sitting in a room with a device that has a dial on it. As they hear stuff from the candidates that they like or dislike, they turn the dial to reflect how they feel about it. CNN then averages the dial test results and maps them on the screen.
Imagine if we used Twitter to do the same thing on a mass scale. The simplest way to do it would be to ask users to post tweets with a 1-10 numerical score whenever they have a reaction to a candidate's statement, then tag it with the keyword #dialtest. You could then follow the search results using Twitter's search engine and get a feel for how Twitter users are reacting to the candidates.
Of course, it'd be great if you could do it in a more sophisticated way. For example, I'd love to see some sort of application that could observe the search results for #dialtest and average the numbers included all of the tweets in rapid succession - like every 10-15 seconds - and then retweet the average through another Twitter account. Ideally, you'd want the app to be smart enough to ignore numbers submitted outside of the 1-10 range, and maybe limit the number of tweets from an individual user to a few times a minute so a user can't skew the average. Similarly, I'd love to see the app let users register themselves as supporting a particular candidate or as undecided, so you could follow dial test averages for each category of user type, since the Twitter community probably skews towards Obama supporters.
For tonight, maybe we could just have users tweet something simple, like #dialtest Obama 7.5 if they wanted to give Obama a 7.5 out of 10 for a particular remark, then monitor the search results. I think that's the easiest way to get started. Meanwhile, some of you could also try the new Twitter plotting tool Plodt, which tracks tweets that reference a keyword and assign it a numerical score, placed between asterisks. For example, if you wanted to give McCain an 8.0 for a comment of his, you'd tweet *McCain 8*, with the asterisks included. If you want to try this method, be sure to follow @plodt on Twitter first so they know to track your tweets.
Update: Okay, here's how it's gonna work.
Step 1: Follow @plodt on Twitter.
Step 2: each time you want to rate a candidate's statement, format your tweet like these examples:
#dialtest *McCain 7.5* Good answer on Iran
or
#dialtest *Obama 7.0* Like what he said re: bailout
By including #dialtest in your tweets, everyone will able to follow along using this Twitter search page. And for those of you who are more visual, the tweets will be plotted on a graph using Plodt.com. The graph will only accept your tweet if you follow @plodt on Twitter and surround your ratings with asterisks, like the examples above.
You should now be able to access the Plodt Web site. And like I said, you'll need to follow @plodt on Twitter for your tweets to be processed, though.
Anyway, this is all just a nutty little experiment, so please take the results with a grain of salt. In the meantime, I'm reserving @dialtest on Twitter, just in case.
-- Andy Carvin, aka acarvin on Twitter
