A Twitter Follow-up

Following on from my earlier post and from James' analysis of the possible use of Twitter, I was interested to read an item on the BBC's web site entitled "Twitter tweets are 40% babble".

I always enjoy seeing how statistics are used in the media - write the story first and then find or spin statistics to fit, would seem to be the prevailing methodology. Just looking at the headline. It would be possible to say "Only 40% of tweets are babble" or even "Tweet are 60% non-babble" - both of which give the story a completely different spin while imparting the same data. Looking further into the text of the piece, it is revealed that 37.5% of tweets analysed were "conversational" and that 8.7% had "value" in that they passed on "items of interest".

Given that Twitter was designed as a way of saying what you were "doing now", in other words "meaningless babble", I think that finding that 8.7% of 2,000 Tweets were valuable is pretty good. Having just returned from holiday, I'm tempted to say that less that 8.7% of my inbox was valuable (not your emails, of course - they're always valuable) and I would not be keen for anyone to measure the value in my spoken communications. I should also like to mention that the 2,000 tweets analysed were recovered over a 14 day period and, at this moment, Twitter is showing 14,564 messages are being transmitted per minute. I'm not sure how good a sample is, therefore, of only 2,000. Their methodology (available here) seems flawed to me - 200 messages per day for 10 days (working days only apparently). They offer some different analysis from their dataset - best times to Tweet etc - but I'm surprised that they seem so definite in their conclusions given how small their population was. The analysis was carried out only on US Tweets (although I'm not at all sure how they managed to identify those) and between 11am and 5pm (US time - probably CST but they don't say for sure). Not only that, but they appear to be confused about some features in Twitter - repeating the false contention that one cannot secure one's feed. I'm disappointed in the analysis (surely they had time to analyse more that 200 Tweets per day) and in the BBC's rather lazy use of it. The BBC's item ends with a note that the study will be repeated quarterly. Unless the size of the study it much enlarged, I don't think this will help much.

Would the results be different if a more full study was undertaken? I don't know - but at least we would be able to be more confident in any analysis which was undertaken.

Twitter is a social medium - one designed for ordinary people (well, tech-friendly ordinary people) to communicate. It should not be a surprise that most people do not need to say things of value. As James' post mentioned, the business use of Twitter and other social media is in its infancy. Patience, BBC, patience.