The way we tweeted

What does a tweeter like me add to a meeting like ScienceWriters2010?

As promised, you can compare the twitter vs. blog coverage. To make this easier, here are the tweets from one session (Get the Numbers Right), including mine and others'. And for comparison, Sandeep Ravindran's blog post of the same talk. Obviously Sandeep's is easier to read post-meeting, with its luxuries like full sentences and organizational structure.

My biggest question, after a full day of tweeting, is this: Who are we tweeting for?

• For those who couldn't make it to the meeting? Twitter users from afar have been replying to meeting tweets, RTing choice lines, and thanking attendees for tweeting.
• For those who are in other sessions and need to time-shift? I had high hopes to use the tweets this way myself, but it relies on good transcript tools. I'm not convinced that the right tools exist...yet.
  
• For those who are in the same session and want to discuss it? In other words, a backchannel. A few of us were using the tweets this way.

I was a programmer before I was a science writer, so it's interesting to compare the twitter coverage here to IRC backchannels from technology conferences. In those cases, people's comments were rarely just repeats of what the speaker said; usually they were critical, questioning, or providing more information. Groups of people would get into a discussion of a related subject, and yes, they might be IRC-ing with each other instead of listening to the talk.

Today's tweets often just propagated pithy quotes. Hey, what is twitter for if not pith? One typical selection was Tom Siegfried's comment that the red flags for wrong science (first report on something, hot topic, contradicts previous knowledge) are the same as the recipe for newsworthiness. Another was the question put to the panelists of "Building the Big Book": If all books on earth disappeared and you could save just one, which would it be? (don't worry, @vanessadamico tweeted their answers too.) Whenever a speaker said something surprising or funny or beautiful, a spray of tweets would appear all at once repeating it.

Live tweeting was an interesting experience for me: I had to decide, before I forgot what the speaker said, whether it was worth repeating. I tended to err on the side of "yes" - stats show me as the day's most prolific tweeter.

It's not easy to know what to tweet. Several times during a talk, the presenter would introduce a topic - like "reporters often get odds ratios wrong" or "statistical significance doesn't mean what you think it means" - that it took a while to explain. I wanted to tweet the idea, but how to do that when you haven't digested the explanation, and aren't sure if you'll understand it by the time the speaker finishes explaining? One audience member missing the point isn't a big deal, unless they're trying to report it live. You can see this happen in the transcript: tweets flounder around until an "aha" moment when several people tweet the conclusion.

Related problems include: what if you missed tweeting the beginning of the story, but the middle is so good you can't resist? (I tweeted anyway, about Gelman's sex ratio study.) What if you forget to tweet something? (Like the full name of the speaker who I referred to by initials for the next half hour - oh well.) What if you get a fail whale mid-talk? (Try, try again.) And the classic tweeter's dilemma: what to do when there's something that just can't be squeezed into 140 characters? (Wait for the blog post!)

One of my favorite things about tweeting today was providing extra information referred to in a talk. When a speaker showed an article or mentioned a topic that is worth reading up on, I quickly googled up the relevant link and tweeted it as part of my coverage. This is similar to the idea of google jockeying, which could be a fun experiment for a future meeting.

As for time-shifting, it's not easy on twitter. The #sciwri10 tag separates meeting coverage from off-topic tweets, but with four tracks of talks, that hashtag covers four overlapping topics in the same 90 minutes. To make the transcript above, I had to hand-pick the tweets on my chosen talk, which was tough because a concurrent but unrelated session was way more popular. What might help is an agreed-upon tag for the talk or the track. I tried using both #sciwri10 and #stats for the stats talk. (That at least worked better than the online commenting talk, where I attempted to use #onlinecommenting before dropping it because, geez, 17 characters?!)

How do you make a twitter transcript anyway? If you've got your hash tag, you can do a twitter search or use a tool like What the hashtag. Unfortunately I couldn't find any transcript tools with fine-grained control, like the ability to include or exclude certain users or secondary tags. (Are they out there and I missed them? Please comment!) Twitter seems to have removed the RSS link on users' profiles, or else I might have done the job with a perl script. This wouldn't be the first time that people's use of twitter outpaces twitter's abilities. Remember ScienceWriters08 where we had to use third-party tool twemes.com just to search for our hash tag?

What do you think about twitter coverage? How have you been using tweets during the meeting? Who are YOU tweeting for?