Eurovision: bloc voting or cultural clusters

Duncan Watts wrote an op-ed in NY Times on The Politics of Eurovision. There he writes:

I had heard about this practice, of course, whereby geographical and cultural neighbors tend to vote for each other, and nobody votes for Britain (well, except for Malta). But it was startling to see just how flagrant it was. The Scandinavians all voted for one another; Lithuania gave 10 points to Latvia (whose entry, bizarrely, sang in Italian); former Warsaw Pact countries voted for Russia; and almost nobody voted for Britain (surprisingly, Ireland did — and, of course, Malta).

Indeed, there has been some debate on bloc-voting in 2007. The map on that page does show that East got better scores than West:

ranking.png

Eurovision has been studied by academics a couple of times by now: Derek Gatherer titled his paper “Comparison of Eurovision Song Contest Simulation with Actual Results Reveals Shifting Patterns of Collusive Voting Alliances” and Anthony Dekker has a paper The Eurovision Song Contest as a ‘Friendship’ Network. The titles are not very forgiving, and here is an example of a chart from one of these two papers (neither of which has been authored by Duncan, of course):

blocs.png
(there isn’t a single Baltic country in the area denoted as “Baltic”, but they are all in the Balkan peninsula).

Serbs and Albanians are pretty much in a state of war, yet they seem to be aligned together in a “friendship” network? The same goes for Greece and Turkey in 2005, and many other similar pairs. I guess the young form a voting alliance for the preservation of hip-hop and metal, and the old too form into an opera friendship. It’s not preferences – it’s politics! It’s not musical taste, it’s alliances! It’s not quality, it’s who is friends with whom! It’s not the fact that musicians in Britain, Ireland, France, who top the charts, already have an excellent means of commercially deploying their music beyond the confines of their country. It’s rare for a first-class musician from the West to try for Eurovision: why expose themselves to public ranking and ridicule in case of failure if they can already sell lots of music by other means. But not so with the East; for the musicians there, Eurovision is the best way of going beyond their borders.

It’s true that Balkan music is overrepresented in the voting scores: it has many small countries with similar musical taste and little population. But as this Excel spreadsheet shows, Serbia would win even if all other countries were prevented from voting. In case only the 1994 members were allowed to vote, Turkey would win instead, followed by Serbia in the second place. The results are quite robust with respect to countries that are allowed to vote. If, however, we weighed the votes by population, allowed countries to vote for themselves, and excluded non-1994EU, the winning order would be Turkey, Greece, Serbia. So, even by population-weighting the votes, the results do not change much.

There are some interesting nonlinearities. It’s known that novelty value plays a big role at Eurovision: and there will be little novelty value to British, Irish, Spanish and German music that are so successful and ubiquitous in the marketplace: they won’t get novelty scores like some more exotic types of music will: the past winners include goofy entries such as Finnish monsters, Ukrainian warriors and an Israeli transvestite. On the other hand, anyone who has tried listening to classical music knows that it takes some exposure before you can enjoy music, just as it takes a certain amount of exposure to literature to enjoy poetry.

The reaction from Britain was quite harsh, dismissing scoring as pure politics. But the UK song should be examined in the context of the 1990 winner on a similar theme. Google can translate the lyrics and you can compare them to the UK entry. Among other events, that 1990 song was what inspired the chain of events that led to the break-up of Yugoslavia (where some of the states were pro-EU and others against), and possibly other nations too. For musical quality, compare it to the UK 1997 winner.

So, in summary, there are many models that explain correlations in data. A cluster in votes can be interpreted both as an alliance to win the majority, or it can be equivalently interpreted as a group of countries that shares cultural preferences. One interpretation is cynical, the other is respectful. If you are an author that doesn’t distinguish between the Balkans and Baltic, you might find it hard to decide on the right interpretation. A respectful one is a safer bet.