Twitter at Supersonic – some early data

The obligatory Wordle. Click for legible.

The Twitter experiment at the Supersonic Festival went pretty well, I think. I’ll be playing around with the data and writing a full report soon but in the meanwhile here’s some very rough and ready stats for you. (All these are subject to me getting the numbers horribly wrong – I’m not a data-cruncher by any definition so caveat emptor, etc.)

Excluding those posted by me to the @supersonicfest account and a few irrelevant posts I spotted and removed (more cleanup will be needed) there were 789 messages sent to Twitter between 8pm on Friday and 2am on Monday containing one of the following:

• @supersonicfest
• #supersonic
• supersonic festival

Be aware we’re not counting mentions of just “supersonic” as many of these were not related to the festival (jet planes, Oasis fans, Queen lyrics, etc). Sifting of the “supersonic” search for relevant tweets would pick up a fair few (I was watching it over the weekend) but I don’t have time to do that.

Those 789 tweets were sent by 146 individuals. We could take this as a percentage of those attending but be aware a number of Twitter users were listening to the live audio stream so it can’t be assumed they were all on site. They were all part of the audience in some form though.

The distribution of tweets per users looks like this:

Click for bigger

The mean average tweets per users was 5.5
60 people (41%) sent one tweet over the whole weekend.
24 people (16%) sent more than 10 tweets.
8 people (5%) send more that 20 tweets and accounted for 275 or 34% of all messages sent.

Looking at these it seems like a very few dominated the Twitter feed, but this chart showing the number of tweets per user made me think twice:

click for legend

More than half the feed was made up of people not sending much at all and Flowdeep’s torrent of tweets don’t look quite so dominating.

Anyway, much more on this to come. If you’d like to play with the data here’s the base transcript I’ve been playing with or you can grab them yourself from this search. I’d really welcome some proper analysis by people who know what they’re doing with this.