The robots are coming!
‘In the dystopian future there is only one kind of dance: the robot.’ No, you aren’t reading an extract from the Peter Crouch autobiography, but the lyrics from a hit song by the Flight of the Concords, the New Zealand duo who tool New York by storm before sliding back into musical obscurity (at least for the time being) via shows at the Glasgow Hydro and Edinburgh Fringe. Although their timeline is questionable (they predicted robots would take over in the year 2000, presumably in light of the millennium bug) you can’t argue with the general sentiment.
The rise of artificial intelligence is certain to impact the beautiful game in the years to come. The 2018 Football World Cup saw an unprecedented level of data analytics in terms of individual players and teams, which heralds a change in how we watch and interpret the performance of players and coaches. FIFA, through mix of data experts and camera systems integrated within the stadiums, were able to provide an unparalleled insight into what was happening on the pitch. The report was far removed from yesteryear when we relied on ‘expert’ opinion to tell us about playing styles and star players; it was an objective account of the formation teams adopted and their success in exploiting space as a result. This, however, is just the start.
The detailed reports that were part of the World Cup could be defined as data analytics, evidence of our improving capability to track the action on the field and report this back in a meaningful and coherent way. There have been a number of contributions to the popular media on this topic and people can agree or disagree about the importance these metrics have to their own team.
At present our understanding of limited to single games or perhaps seasons. But the increasing sophistication of machine learning means this data could result in our ability to ‘model’ the way that teams and players perform in critical areas of the field. Machine learning is about the ability of technology to digest and analyse huge quantities of data. It is the same technology that directs personalised adverts to our Internet browser based on commonly visited sites and search history. In football (and sport more generally) this means systems that can analyse every game being played at the professional level in every country around the world and then begin to build up detailed models of the way players move when in different formations and against differing oppositions. Indeed, the promise is that with enough data we might be able to predict, or at least reliably represent, the styles of play apparent in different leagues and competitions. Such data, when analysed and interpreted in the right way would lead to the sort of artificial intelligence that would herald a new ear in how we watch, prepared and select players in the future.
One of the perceived downsides of artificial intelligence is that it will, in time, erode jobs that, until now, have relied on human endeavour and labour. For example, driverless cars, robots on production lines and drones delivering our weekly shop. In football and sport it may be the relatively new roles occupied by performance analysts that are most under threat. As camera systems and software become more sophisticated it is easy to imagine a time when the manual operation of a camera is no longer required, heralding the end of the performance analyst or at least those who cannot develop skills in other relevant areas such as coding and the analysis of big data. Around 15 years ago I met with a start up company who were promoting a camera that could stitch together different views to provide vision on the whole field of play, removing the need for a camera person following the ball or two seperate cameras, one watching the ball and another focusing on the shape of a team off the ball. Cameras that offer this capability are now commonplace on the touchline of amateur teams all over the world. The march of technology. Intelligent cameras are able to track the the players and ball independently of an operative and feedback information on position and key performance metrics. The resultant data is added to a super computer tasked with understanding how teams play over the longer term. We will be able to understand the actual impact that a different coach has on the way a team plays, or, how new players change systems. Where this data is held by private companies, clubs will find themselves paying for the privilege of understanding how their team and its opponents are likely to play. A genuine arms race.
Some will argue that whilst different countries hold different broadcasting deals this is newer impossible. Maybe so, but it is hard to imagine that in the future club football will be played within the geographical boundaries of individual countries. Global leagues with streaming rights are more of a ‘when’ rather than ‘if’ and there are already companies such as WyScout with access to video clips from around the world. With this in mind could scouts also be at risk? Of course, artificial intelligence and machine learning can’t tell you anything about a players personality or attitude, but artificial intelligence may not be limited to on-the-pitch data. Our biometric data is already being recorded, stored and analysed. Applications within our smartphones know how much exercise we take, how much sleep we have and where we have been. In this sense, the availability of this data to clubs looking to purchase a player, when combined with an in-depth understanding of their performance may well be better than the traditional methods of recruitment.
I accept that for some all of this will be regarded as anathema. Technology replacing experience that can only be gleaned from standing on touchlines, notebook hand and discussing the players on show with other ‘nowhere men’, to coin a phrase used by Michael Calvin. There are, however, significant advantages to the adoption and use of artificial intelligence in football and sport.
The more we understand about the movements that players make in certain areas of the pitch, the more we are able to create models and algorithms that can predict how individuals will react when their teammates, opponents and the ball in different locations. Although there is an inherent chaotic nature within most sporting events, when millions of movements are analysed there are likely to be common features reflected in the different playing styles and formations. When incorporated into virtual reality software, such knowledge may allow young players to play matches against opponents from Mexico, Croatia or Japan, experiencing different styles of play and technical skill sets, all from the comfort of their own home. Not only is this a benefit for the players, but also for the clubs It is hugely expensive to arrange tours or attend tournaments that allow youth players to pit their wits against players from around the world - far more costly than multiple sets of immersive headsets or Google glasses.
It is not just exposure to international styles and formations that clubs may value. As technology advances, clubs may be able to develop online and virtual training sessions that players can interact with and coaches can comment on. In this sense teams will be able to recruit and train players from all over the country with no need to travel to the training ground, except for matches and competition. As far fetched as this may sound it would be an answer to a very real problem; there are still players and athletes who spend more time traveling to training than they do, actually training.
Anyone who has witnessed training sessions held the day before a game will be able to immediately spot the players not starting; they are the ones resplendent in pink bibs, moving lethargically at the coach’s direction and trying to provide some degree of opposition to the players start to start the match. The formation of the ‘opposition’ may have been gleaned from scouts’ reports on two, three, maybe four or five performances if you are lucky in the hope that this will give the players an indication of the challenges they are likely to face.
Artificial intelligence, machine learning and the advancement of robotics have the potential to change all this. Imagine a time when football clubs have 11 robots that can be programmed to move in the way that the players they represent have been coaches to do, not over the course of five games, but since the start of their career or, perhaps since the team started playing together under their current coach. Anyone sceptical about the capacity of robots and technology to deliver such advancements should visit the Robotarium at Heriot-Watt University. As technology progresses, such ‘shaping sessions’ will likely be conducted using virtual reality where players see themselves and their teammates against the next opponent in 360-degree high definition and are able, with their coaches, to interact and manipulate that environment to best prepare themselves for the game. This approach will have the added benefit of enabling athletes to not undertake physical activity the day before a game, further enhancing their preparedness.
So does all this spoil the beautiful game? Can something that has captivated our collective imagination for so long be reduced to numbers, predicted movements and algorithms? No, of course it can’t. A computer can’t replace the flair with which Maradona glided through half the England team or when Leigh Griffiths guided his free kicks past Joe Hart, not once, but twice. These events are exceptional, and as such, rare. They are also, however, the exception that proves the rule. Much more van be predicted: the mundane sideways pass, the third man run, the strikers movement, the patterns of play that coaches install in their players, and the habitual movements individual players make and tells before they are about to run, pass or check. It is this that computers and the cameras they are linked to are watching and analysing, waiting until the moment they feel they can predict what happens next with a degree of certainty that makes the owners take them public. At this point clubs will have a choice to make; do they exploit the competitive advantage or hold on to nostalgia.
This article was first published in Nutmeg, the Scottish Football Periodical in March 2019.