credit: Brothers_Cider Flickr

Glasto is fast approaching, with all of its musical, social and artistic surprises. We wish we could go. But thanks to the rise of the social web we can still participate through social media. We can ride the build up, hear the latest reports from the festival itself and then bathe in the after-glow with the festival mashups. That got us thinking at Face. With all this data floating around, there was certainly something fun we could do.

After much deliberation and debate, discussing topics from analyzing brand sponsors to what goes into a “Festival Survival Kit”, we finally landed on fashion. We were struck by the number of people who spend each year trying to predict the key fashion trends for festivals. Could we do a better job using social media monitoring? By looking at what people are saying before the mud bath at the end of June, making predictions, and then seeing what actually happens during the weekend, we can, at least, know what to wear for the next festival, since we can’t go to this one.

Prediction is Difficult but Valuable

But this is about more than finding out what to wear to Reading. According to Seth Godin, explanations of the future are much more useful than explanations of the past. It’s relatively easy to look at past behavior and explain how that led to the current state of affairs. Lining up cause and effect is simple after the fact, but is not quite as easy to figure out beforehand. There is a science of observation behind trends analysis and prediction, but it is a difficult problem to tackle.

Looking at past behavior has long been held as the best guide to the future. One of the techniques that stock traders have in their tool boxes is a company’s past performance. Looking at how a company’s stocks have behaved in the past is often a good predictor of its future performance. But not always. The problem is that trends analysis is not always linear. If something is increasing now, it may not continue to increase at the same rate in the future.

Stock traders have had centuries to practice their trade – and yet predicting the stock market is still far from an accurate science. Though stock analysts employ many different kinds of data in creating their predictions, much of Finance is based on quantitative number analysis. Predictions can also come from more qualitative approaches, such as relying on expert judgment. However, increasingly the dependability of experts has been called into question. In The Wisdom of Crowds, James Surowiecki suggests that under the right conditions, the masses often make better decisions than single experts.

Social media analysis is a mixture of these approaches. It has the quantitative rigour of stock analytics, the voice of experts from their online publishing, and the aggregation of mass opinion. Rather than asking people questions, we look at how they behave by applying statistical analysis. Our main goal in analyzing the Glastonbury festival buzz is to see if this combination works. And find out what to wear at Bestival.

The Next Steps

We have already created the lexicon and started the search in our Pulsar social media buzz monitoring system.

Pulsar is busy gathering up mentions for every type of garment we could think up. We are looking for the leading brands and the festival fashion influencers.We’re even tracking different ways of referring to the festival, from Glastonbury to #glastofest. It was not an easy syntax to figure out – what do you think “buy + Glasto” would come up with? Bet you it is not what turned up. But now that the refining process is done, we get to sit back and see what the program returns. The program has been at it since Monday, 9 June, so we’ll give it another week or so before getting back to you.

What type of data do you think we will we get? That’s what we’ll be discussing next as part of this Glastonbury Prediction series. Sign up via RSS to stay on-top of this developing exploration. What do you think the issues with using social media for predictions are?