And we all thought that it was so prescient 😉 Apparently a study has just come up with the astounding claim that Twitter cannot predict the outcome of elections. Unpacking that a bit and trying to remove the headline from the actual study content, I am unsure who was making the rather silly claim that the tool itself was capable of providing specific new insights. There is a lovely collection of social expression there thought that undoubtedly can be mined effective to determine some sentiment. However, in fairness should we see Twitter in a light much different from the ‘scientific’ polls that are conducted by agencies claiming to be able to give predictions of the same calibre? MIT Technology review reports on a study by Dr Daniel Gayo-Avello that dismisses predictive claims (valid) and suggests the absolutely unrepresentative nature of the constituency renders any Twitter datamining moot as a means of predictive insight. After a quick look at this study I am unconvinced. Armed with a knowledge of the demographic bias social media and twitter in specific has it clearly can provide qualified insights on a particular sector and how it may cast a vote. I think these are appropriately corrective statements in raising a level of caution to not overhype the clairvoyant aspects, but I still suspect that the studies themselves tend to adopt the same rather strident swing in the other direction and do themselves a disservice. Caution all around, but nonetheless worth considering.