Saturday, 10 May 2014

Demonstration Game

This week saw the release of the Budget. Last year I took the time to analyse the main areas of spending and the demographics they played to. As it turned out it was quite difficult to break down the ABS data to a manageable level that informed our predictions. It was an interesting exercise, but no points could be awarded for correct predictions – much like a demonstration match at the Olympics.

This year, however, there is something else to look at. This weekend we will find the results of the International voting in an election run across the length and breadth of the European Union. International politics will come to the fore. Diplomatic ties will be strained and strengthened. Smoke, wind and fire pour forth from the floor.

I speak, of course, of the Eurovision Song Contest.

It is no secret, of course, that the voting in Eurovision is at times more political than a reflection on musical talent. The Nordic/Scandinavian countries will back each other to the hilt.* Turkey and Armenia will strive to put the past behind them once more.** Ukraine and Russia will go head-to-head on an illuminated screen floor concealed beneath dry ice and loaded with enough pyrotechnics to blow up the British Houses of Parliament.***

This year, however, Australia has sent Jessica Mauboy to perform. Technically Australia’s not in Europe (although similar geographic inconveniences has never stopped Israel participating). Instead, we were just be appearing as an act in the second semi during voting for the real contestants. It was an interesting exercise, but no points were awarded for the performance – much like a demonstration match at the Olympics.

The semi-finals are over, however a brief analysis of past semi-finals results may have informed our expectations. Certain countries are well known to favour each other – this may be due to political reasons, geographical closeness fostering a familiarity and friendliness to other nations or songs being sung in common languages (e.g. the Most Serene Republic of San Marino might sing in Italian, which might isolate non-Italian dominated countries a little but appeal to the Italian vote). This was a bigger rort back when the winner was picked by a small jury of representatives. In 1998 the phone voting system was introduced, but because unpredictability and lack of bias are boring we are now on a hybrid system.

Most importantly, it is not possible to vote against a country. This means that nations can afford to narrow in on the support of a few select nations without a significant backlash from others. This also means that there are situations, for example in the current Crimean Crisis, where the results are counter intuitive. Pro-EU (aka anti-Russian) forces outnumber pro-Russian (aka anti-European Union) protestors in Ukraine. As a result, one would expect the Russians to do poorly off of the Ukrainian vote. However since the anti-Russian vote will be split across so many other options, the consolidated Russian support might mean top marks for the former super-power. A pox upon first-past-the-post voting systems.

Semi Finals

First off, here are the total points awarded to the countries on the y-axis by those on the x-axis since 2008 when the semi-finals were first split into two pools.

And here is the average score since 2008, adjusted for the times each country has not participated.

Given that only half of the x-y combinations will match up each year (countries in pool 1 cannot vote for those in pool 2), and the average pool size is around 18 participants, an average score would be roughly 1.6 (the actual average of all scores is roughly 1.484). The standard deviation is higher than this (1.643) and therefore does not offer a convenient way of grading favouritism. Instead, let’s graph each nation’s support for each contestant to determine if there is a particularly strong peak of support anywhere.
 Okay. Let’s not, then. Instead, 1.0 and less is arbitrarily considered poor, up to 2.0 is fair, up to 4.0 is a lean towards a nation and anything above 4.0 is a very strong lean. And, go:

Slightly better, I suppose. While the obvious approach is to simply see which 10 teams do best from each voter in the same pool as a prediction, the green votes above are the key ones that we can rely on. Whether you see this as the historical strength of each nation’s musical ability or other factors, certain patterns emerge. Greece is historically very strong with green ratings from 13 different countries and favours the smaller nations in the Balkan/Eastern European sphere. Russia does well off of the Baltics and other peaceful neighbours (Poland, Azerbaijan, Lithuania and Latvia blue, Estonia green) but poorly from the major powers (red from France and Portugal, grey from UK, Spain, Germany, Italy, Sweden & Norway) and neighbouring nations fearing a military invasion (Georgia, Ukraine and Belarus all red).

So, lets take a look at, on past voting, who can be expected to do best in each pool:
Where countries have drawn, the availible points are averaged. These are rounded off to the nearest whole number above.

Comparing the rankings (total score) with the actual results (red countries in the grids were eliminated) It is clear that there is no real predictive power in this methodology. In fact, the bottom three and top four in each pool went straight through, leaving the battle to the middle of the field.

Instead, I’m going to have to pick the songs manually. Countries can get support for being genuinely good (see Norway) or, more commonly, for ridiculous costumes (see Iceland), special effects (see Montenegro) or bizarre songs (see Estonia). You can also get into the finals for being Scandinavian or, in a tiebreak situation Baltic/Balkan.
For Pool 1, I predicted the ten to continue to the finals as Armenia (the only pyrotechnic-dependant performance), Latvia (splitting the Hipster vote with the Netherlands and scooping up the often overlooked cake-baking vote), Sweden (Scandinavian vote), Iceland (slightly less lucrative Nordic vote, reinforced by technicolour costumes), Russia (for using singing twins and a guy whose only roll is to stand behind a collapsible sunrise), Ukraine (giant freaking hamster wheel), San Marino (impressive use of video screens with dry ice), Portugal (all the drums, and a half-naked black guy), the Netherlands (reaping more than half of the votes of hipsters dominating the west of Europe) and Montenegro (impressive figure-skater/effects coordination).

With 16 contenders I was guaranteed at least 4 of my 10 would get through, so the possible scores for this prediction range from 0 to 6 guesses inaccurate. Latvia and Portugal dropped out for Azerbaijan and Hungary, so that’s 8 out of 10 correct, but should honestly be a score of 4/6 or 66%.
For Pool 2, my pick was Malta (hipsters with the obligatory acoustic guitars, double bass and drums (plus a piano and a… a… I don’t even know what)), Norway (perhaps the best act on actual musical ability), Poland (for traditional dress and sensual – even erotic – clothes washing), Austria (for supporting a bearded transvestite), Lithuania (for the Tron-inspired outfits), Belarus (for Mo-town junior), Switzerland (whistling hipsters with obligatory acoustic guitar, double bass and drums (and banjo and tambourine)), Greece (a rap/R&B/dance number appealing to a completely separate demographic, backed up by a trampolinist), Romania (Mohawk and fake circular piano) and one other. The choices for the last place are varied, with different attributes to pick from: there’s Ireland with a Bollywood/Riverdance combo; Slovenia with an interesting flute opening; and there’s the Former Yugoslav Republic of Macedonia with Voldemort in a white hoodie.
With 31 contestants between the two semis we actually earn one tossup, which I would probably use here if the results were not already known. Instead, I’m going with Georgia due to their electric guitar wielding mobster and a guy with a parachute.
15 contestants means 5 of my 10 predictions are guaranteed to be correct. Georgia was a gamble that didn’t pay off. Lithuania also failed to capitalise, and Slovenia and Finland (Damn you Nordic bloc vote) replaced them. So, 80% again on total results, but 3 out of 5 (60%) once adjusted for mathematical certainties.

With that less-than-impressive background, we need to look at the automatic finalists:
UK: Britain seems to think Eurovision is a song contest of some sort. Nothing too zany to report on here.
Spain: Based on the preview movie clip, there may actually be rain falling on the stage for this one.
Germany: Accordion, Double Bass and Mime artists – I’m a little concerned France and Germany may have switched acts.
France: Big hair and a burning desire to grow a moustache – I’m very concerned France and Germany may have switched acts.
Italy: Futuristic Punk. ‘nuf said.
Denmark: Seems to be competing as Belarus’ older brother.

With the exception of 2010, an auto-finalist has not won since 1997. In that time, the Nordics have won six times. If you want to pick a shortlist of winners, the Big Five and the Host is not the set to pick. That’s not to say they cannot win, but I suspect given the cost of hosting the contest, none of them want to become regular giants of the contest (compare Ireland in the 90’s).

And that brings us to:

The Finals

A lot of gimmicks that got countries through the semis – circular pianos and Wiggles-inspired suits – will lose their appeal the second time around. I think the strongest contestants from pool 1 are Russia, Ukraine and the Netherlands. For pool 2 I’d back Norway, Switzerland and, at an outside chance, Poland or Austria. The first-time performers are not hamstrung by this seen-it-all-before fatigue (except perhaps Denmark), so France may get a boost there too.

To pick one, I’d have to tentatively circle Ukraine, with Norway a close second and the Dutch in third. The Netherlands will overshadow the Swiss for the hipster vote, and I think Russia and Poland will appeal to similar voters, splitting the votes. Austria may come from nowhere, but I doubt it.

Since I am operating with completely untried (or, realistically, no) predictive methods, these predictions will not be counted towards the total points of this blog. I should, perhaps, have done the same thing with last year’s conclave and the United Nations Security Council vote the year before that (although the UNSC results have not been factored into the current score at Infographinomicon). It is an interesting exercise, but no points will be awarded for correct predictions – much like a demonstration match at the Olympics.
* Although exactly where these boundaries are drawn is open to debate. With Denmark into the finals by default, this does not necessarily result in predictable preference flows.

** Although Turkey has chickened out this year.
*** Calculation not performed. However, the gunpowder to be used by Guy Fawkes et al amounted toa cubic ton. Measuring gunpowder by volume is complicated by grain size, but will generally resolve to the equivalent weight within 10% accuracy. So allowing 1.1 ton (roughly 998,000 g) of gunpowder to blow up parliament seems like a pretty good benchmark. If you know where to buy blackpowder substitutes (that link may land you on certain government watchlists) you can get the stuff for less than $20/pound ($18.29 plus shipping, limit 48 lb), or go full-on V for Vendetta for just over $40,000. Granted the cost of fireworks will cost more for labour, colours etc. and can vary hugely with the choice of pyrotechnic, and because these are payed for by the performing countries, exact expenditure is difficult to track down. But even if the fireworks are 10% gunpowder by cost, that’s $400,000 spread between 37 countries. Given that most contestants give their performance at least twice, that’s only $5000 per act on fireworks. Add in the opening ceremony etc. etc. etc. and Big Ben doesn’t really stand a chance.

No comments:

Post a Comment