Voting Insight: The Impact Of 2016’s New Split Voting Written by Ben Robertson on February 21, 2016 | 2 Comments

The revelation of the biggest change in voting since 1975 was what the EBU annouced as news broke on Thursday morning.

In some sense this statement of huge upheaval is true, the presentation of votes with the big finale is a change and will lead to a guaranteed moment of tension and spectacle. Presentation aside, when it comes to the scoring this change does not significantly alter the context of the Contest to anywhere near the same level as the introduction of televoting, and is nowhere close to the monumental change that SVT’s douze points system created in 1975.

The new results are broadly similar to the system used in recent years. The five jury members will still have the equivalent of ten percent of the national vote, and they continue to rank songs from first to last. Viewers’ votes cast from ‘in front of the telly; still equal half the available points from their country. The one somewhat subtle difference is that these votes are never combined together, and that has some repercussions in the final results.

ESC Insight has examined the previous ways that the 50/50 split was calculated. The previous method of combining before points are allocated allowed jury rankings to drag down and kill a televote. The growing consensus was that the successful Eurovision songs were those that took no risks and did not attract any negativity leading to last year’s Grand Final being besieged by boring ballads. The televote is now liberated from the jury’s conservative restraint, so how does that change the system?

The Methodology In Detail

The EBU has published the full splits from each individual juror for 2014 and 2015. That means we can look back at the last two years of results and look at the impact of the new voting system over six events (four Semi Finals and two Grand Finals). Results from earlier years use a different jury system that does not directly rank all the songs, so cannot be used for a direct comparison.

What the new system means we have to do is count the final jury ranking as a straight first place to tenth place, with first place receiving the country’s douze points from the jury. The televote ranking from that country is then also converted into a top ten, with 12, 10, 8, 7, 6, 5, 4, 3, 2, and finally 1 point being awarded to those songs.

The big headache here is what to do with any disqualified juries and televotes. In the new rules any jury or televote scores not counted will be ‘calculated by the result of a pre-selected group of countries’. This process has already been confirmed by the Reference Group, but we are awaiting further details on how it the groups and the scores will be calculated.

San Marino is unable to record a separate televote from Italy, so has always used their jury results 100% since their debut in 2008 (Image: EBU)

What we are going to do in this analysis is to model this process using the groups given for the drawing of Semi Final lots as the pre-determined countries. This means the jury and televoting results of countries in the same pot will be averaged out and then put into the score from the missing country as required. This is a likely approximation of what would happen and probably a better representative result than just to replace the score with the jury or televote, which is what happened under the old system.

For example, this was how I calculated the missing Georgian televote in 2015’s first Semi Final. I have calculated this sample below with Georgia able to score points until when replacement points are allocated. Countries in the same pot competing in same Semi Final were Russia, Armenia, Belarus.

Sadly this technique is not possible to use in 2014’s first Semi Final, as both Albania and Montenegro missed the requirements for a valid televote, and they were the only countries from the same pot in that Semi Final. Clearly this is an issue to be addressed so the group of including countries would need to have a bigger sample size, but the method provides a suitable approximation for what would happen.

In terms of results, this ‘arranged vote‘ does show notable changes. You can see above here that this approximation would give a televote for Georgia awarding 10 and 12 points to neighbours Russia and Armenia. Yet the Georgian jury result, which replaced the missing televote in 2015’s system, awards the top scores to Moldova and Belarus. In the calculated results, using the averages from the other countries is more likely to represent televoting trends, but that is not a guarantee.

San Marino in 2015 found itself in a pot containing Belgium, the Netherlands, Ireland, Greece and Cyprus. The microstate would find itself awarding five points to Poland. This is from what we can assume is diaspora voting from the Netherlands, Belgium and Ireland. However the top four scores for San Marino’s ‘televotes‘ now are cast to the same four countries which won the televote across the entire Contest (and yes, Italy did get the 12). This is a better guess for what people’s opinions would be in the landlocked nation than giving their jury a voting strength double that of other nations (Latvia’s extra 12 points from the San Marino ‘televote’ would sink to zero).

Aminata would have dropped from 2nd to 3rd in qualification as the jury vote was replaced by combined televotes from other countries (Photo: Derek Sillerud)

This is a mathematical quick fix rather than a direct replacement for a televote. This modelling has the risk of being very awkward. If Azerbaijan did not record either a jury or televote, then their pot countries could be used to make a ‘best guess’ for what it would be. This though would very likely result in Azeri points going to neighbouring Armenia incredibly high, which would not go down well with many politicians in Baku. In the post-Final aftermath of data could it ‘appear’ that Armenia received the Azeri top score?

At The Top Of The Leaderboard

The results of winners is of course most important. Austria would have still won in 2014, and Sweden would have still won in 2015. Both winning margins would be over 100 new points and would still be clear cut, nothing changes that (and yes, 100 new points is a lot, but remember with both jury and televote giving separate twelves, that’s the equivalent of 50 old points).

Indeed, the top of all the leader boards remain relatively unchanged. The top six that ran away from the pack in Vienna still do so, and the top six are equally unchanged in 2015. The only difference in positions within them is that Italy’s Il Volo would leapfrog Russia’s Polina Gagarina to claim the silver medal in 2015.

This lack of notable change should not be a surprise. These songs have done well because they were able to attract top scores from both juries and televoters. Effectively splitting the results into two separate competitions helps a few songs which scored well from one or the other to increase their point average (such as Italy in 2015 with televoters, and Latvia from the same year with the juries). However, with many thousands of points extra to play for, the difference here is only big enough to come into play in a very close competition.

Il Volo’s televoting appeal would now not be dragged down by juries, moving them to second place last year.

It is worth saying here that if you compare the leaders results with their score they actually received in Eurovision (doubled, to account for the double counting now in place), then the leaders’ new results are a slight percentage drop from their 2015 ratings. This is because the spread of points is now far greater with consensus not needed, so countries with only a jury or televote high score from a country will pick up a score, taking some points that would otherwise be dominated by the leaders. With the old 50/50 system, we’ve been looking towards at many winning songs coming in around the 300 old points mark. 600 new points in 2016 would only be reserved for landslides, so expect a winner to average around the 500-550 new points mark. It seems certain Rybak’s 387 (old) points record will topple in May.

No More Nul Points?

The extra spread of points filters its way down to the very bottom and causing some notable changes. Austria and Germany, last year’s nul pointers, would now receive 40 and 29 new points respectively. The UK would have been last with 16 new points. France, marooned on 2 old points in the 2014 Grand Final, would escape up to the breathing room provided by a cushion of 6 new points. Lots of these countries proportionally get more points than you would otherwise expect, their poor form limiting the need for  jury and televote points from the same country to combine and award them a score in the old system.

Electro Velvet would now finish in last place (Photo: Derek Sillerud)

To say it is impossible to get nul points in this new system is absolutely not true. What it does mean though is that you have to get nul points in not just one Contest, but the two parallel Contests of the jury vote and the televote. Brutally, the probability is therefore squared. If we said for example, that each year there was a 1 in 2 chance of getting nil points, then that could be modelled as being 1/2 x 1/2 = 1/4 – in reality a nil points result is far less likely than that.

This minute probability does assume the jury vote and televote are mutually exclusive, which is not true (last year’s Grand Final jury and televote splits correlate positively at +0.71, which means a strong albeit not perfect link between the two), but it does demonstrate how the chances of nil points is much reduced.

The Midfield Scraps Change Both Little And Lots

The midfield at the Eurovision Song Contest is often dominated by which countries qualify in the lower positions from the Semi Finals. In the three Semi Finals we can measure using the methodology above we find no change in the thirty qualifiers. Ewan’s article on release of the news included Malta as qualifying above Azerbaijan. This is only correct if you assume the jury votes can be used to replace the televotes. Instead averaging the televotes of those countries in the same pots as those with missing data drops Malta down to two points below Azerbaijan’s ‘Hour Of The Wolf’. This was still closer than in the real voting last year, but not enough.

It does seem though that Portugal would have taken San Marino’s qualification spot in 2014, with the extra televotes helping steer ‘Quero Ser Tua’ over the line. Four countries did not record a valid televote in this show, and you would expect more of those votes to be cast for Portugal than with the jury replacement scores. Sadly this calculation can not be completed due to both Albania’s and Montenegro’s lack of televote.

Amber, 5th in the jury vote of her Semi Final, was only 12th with televoters (Photo: Derek Sillerud)

These beneficiaries of Malta and Portugal highlight the downward kick divisive songs had in the Song Contest over the last few years. Portugal in 2014 was a televote favourite, but jury scores regularly pulled it down to 11th and 12th in the combined voting. The opposite was true for Malta in 2015, where a high jury score was cancelled out by finishing at the bottom of some televotes. By splitting the two contests apart for jury votes and televotes those songs that don’t have agreement do get a small but important boost.

Some of these boosts are big enough to catapult a song up the Grand Final leaderboard. Cyprus’s strong jury supported song  ‘One Thing I Should Have Done’ scored 11 old points on Grand Final night last May. That’s equivalent to 22 new points, but our model sees the island nation score a lofty 71 new points. Poland, with a loyal diaspora vote ensure televote shines above the weak jury score, jumps from the equivalent of 20 new points to 54 new points. There’s an equal kick for televote favourite ‘I’m Alive’ taking it from an effective 68 new points to 100 new points. You would be unsurprised to hear that the big midfield winners in 2014 were Poland (televote) and Malta (jury), now both entering the left hand side of the leaderboard.

The biggest losers on the other hand are not dropping anywhere near the same amount of points as some of these examples above are gaining. The biggest downward kick appears to be 25% of the final points score. In both its Semi Final and in the Grand Final, the song with the lowest drop in score was Slovenia’s ‘Here For You’ from Maaraya. A good, competent song with a fan base of televoters and jury, neither of which lifted the song to top either ranking. Separating the voting into two parts weakened Slovenia’s scoring here, dropping to 6th in the Semi Final and 20th in the Grand Final (compared to 5th and 14th respectively). Other fallers in the field include ‘Hour Of The Wolf’, ‘Adio’, ‘Wars For Nothing’, ‘Silent Storm’ and ‘Dancing In The Rain’.

In a tight year you may notice one or two fewer songs in the category of gentle, unremarkable performances qualify to the Saturday night show, replaced by those that divide opinion a little more. These changes are not fundamentally different. The biggest percentage increase of points for a qualifying song in this methodology is sixteen percent, equivalent to around 20 new points in 2016. This is the same kind of level we would expect from having a friendly voting bloc drawn out in the same show. It is not enough to radically alter the make up of qualifying countries.

Nudged In The Right Direction

Over sixty years there have been times when voting has changed dramatically to alter the nature of the Song Contest. The introduction of televoting and the Semi Finals are obvious modern examples, but even the break in 1975 to a system of douze points radically spread out the number of points awarded in Eurovision. The 2016 system does the same, it is likely that each country will award a wider geographic spread of points to anything between 10 to 20 different nations.

This headline of voting spread loses impact because most of the difference is away from the dramatic end of the scoreboard, making it very hard to score a big zero and to make everybody a Saturday night success. With televoting being revealed afterwards in one huge big lump, all the acts get a chance to shine when their votes are revealed. With this system you could reasonably expect each act to get double figures (only one song would miss out in the last two Grand Finals), giving at least something to celebrate when the camera shines on them. Yes, this idea you will all know from Melodifestivalen, and even when the reaction is shock to a low score, it makes ‘good TV’.

Out of forty qualifiers to the last two Grand Finals, this voting system (probably) just changes one of them. We see it being a little easier for countries with diaspora, or with strong jury-focused or televote-focused entries, but the level of boost is only as significant as having another friendly neighbour or two in the same show. This may mean in tight years the middle-of-the-road entry gets squeezed out a few times, but it’s likely to only be a few times.

This is a change in the right direction to evolve the Eurovision Song Contest. The voting system is now steered back towards trying to score from either the jury or the televote, rather than trying to present an average song that reaches the mid-table of both jury and televote rankings. This encourages more bravery, rewarding success in each of the two separate voting shows, and treating them as equal competitions, rather than demanding that performers try to maximise the mathematical return involved in the country-by-country combination. All of this clearly visible to the public, as all the juries vote via spokespeople on satellite, and the televote comes after as one exciting block presentation.

This will generate excitement, which is what the Eurovision Song Contest has been lacking through the recent voting sequences. The change is not going to radically alter the competition side of Eurovision, but this one small step might be a giant leap for bringing back some wonder to the Song Contest part of our beloved show.