In the span of four minutes and 36 seconds, 500 messages containing the word “Obama” were posted to Twitter on the afternoon of Election Day. The flurry started 27 seconds after 3:55 p.m. PST, when a livejournal blogger from Kentucky posted this simple tweet: “Obama 08!”

During that time, an average of 108.7 Obama tweets were published every minute.

What about Senator McCain?  Starting at 3:50 p.m. that day, an average of 50 messages with the word “McCain” appeared on Twitter every minute. There were 500 McCain tweets in the span of ten minutes.

Last week’s Twitter traffic for both candidates peaked on Tuesday, but I suspect the high water mark was set sometime during or after the president-elect’s acceptance speech. Unfortunately, I am unable to identify the peak hour and/or minute of traffic.

I’m working now on a post that will tie the Twitter activity to the main topic of this blog: the almighty (hyper)link. 

In the meantime, I just want to share these numbers and my methodology, with the hope that a bored programmer wil figure out what time the Obama Twitter traffic peaked.

Trust me, if you aren’t a statistician or programmer, you should stop reading now!

My methodology

  1. Use the Twitter Search API to retrieve the last tweets that contain a keyword on a given date.
  2. Use Yahoo Pipes to aggregate five pages of results. The Twitter Search API returns a maximum of 100 tweets per page. Aggregating will create a feed with 500 tweets.
  3. Use Pipes to sort the feed chronologically.
  4. Still in Pipes, create a new feed that contains only two items: the most recent post from the original feed, and the oldest post.
  5. Rename the title of each post to reflect its timestamp (measured in seconds). 
  6. Calculate the “elapsed time” by taking the difference of the two timestamps.
  7. Divide 500 (the number of tweets) by the elapsed time. This is the number of tweets per second.
  8. Multiply the number of tweets per second by 60 (the number of seconds in a minute) to arrive at the number of tweets per minute.

Linked to from this post: