If you ever doubted the quality of "expert opinions" and the predictive value of the odds of bookmakers, this year's Eurovision Song Contest gave you every reason to stick to this scepticism. Around 120 million viewers in 43 countries watched the Eurovision Song Contest on Saturday May 14, 2011. For most viewers, and certainly for those who had been calling in or sending text messages to support their favourite candidates, it must have been a heart-stopping show. And the winner was: Azerbaijan. Yes, exactly the country that not one expert had included on his list of favourites.
Looking back, song contests, just like future events in general, have an impressive history of wrong forecasts, mostly from people who consider themselves experts on the basis of some nebulous claim to credibility. For example, in the realms of pop music, there are plenty of has-beens who played the washboard in a Skiffle band forty-something years ago and then switched to journalism to pay their loans and bills, which is an absolutely honourable reason, but… It was probably just this sort of expert who spread the word about the most likely winners of this year's Eurovision Song Contest. The Irish super-quiffed twins were hot, as was the Finnish troubadour; maybe even Lovely Lena stood a chance again – and, of course, there were also the Swedish and Danish mainstream pop-rockers – but no one seemed to have Azerbaijan on their radar, or even Italy as a possible runner-up. Teradata's CTO Stephen Brobst assessment of marketing in the social media age is also true in this case: If you are relying on gut feeling instead of analytics, you're bound to be lost!
As data journalism gains momentum, I dare to predict that this year's Eurovision Song Contest will be remembered as a landmark shift from the kind of gut feeling expertise described above to a much more sober approach to predicting results.
Because data analysts – even if they were totally tone deaf – had good reason to consider the modest Caucasian country as a potential favourite. In the week before the contest, the German weekly newspaper "Die Zeit" published the results of an analysis of voting behaviour based on the last 10 years of Eurovision Song Contest finals. Marcus Weber, a mathematician from the Zuse Institut in Berlin, had analysed every country's votes. The software he had used identified six clusters of countries, which were very similar with regard to their favourite music. As the original illustration is not available online, you will find a visual representation of the results on a map (which was created by using a free tool from our partner Tableau Software.)
Further evidence that the actual result didn't come as a total surprise had been provided by the Guardian data blog. Based on Eurovision voting data, the blog predicted as early as Friday that "Serbia, Azerbaijan, Greece or the Ukraine seem most likely to win".
An analysis of Eurovision data adds some interesting insights. Although Azerbaijan has enjoyed considerable support from the countries of its cluster, excepting Armenia, it also received high points scores from countries like Malta, San Marino, Croatia, Iceland and Austria. But just see for yourself which country awarded how many points to Azerbaijan.
This means that the "expert" role will no longer be an easy source of income for failed musicians. Sorry guys: data analysts or data journalists will take your place!