How Close Are We to Having Robots as Managers in the Premier League?

Dean Jones@DeanJonesBRFootball Insider at Bleacher ReportNovember 23, 2018

CORRECTS NAME TO JAN HENDRIK BERLIN - Jan Hendrik  Berlin of the Nao Devils team repair robots prior a soccer match of the Standard Platform Liga competition at the RoboCup German Open 2018 in Magdeburg, Germany, Saturday, April 28, 2018. Around 200 teams with more than 1,000 participants from 16 countries demonstrate the state-of-the-art in robotics with competitions such as in soccer, rescue and service robots until Sunday, April 29, 2018. (AP Photo/Jens Meyer)

Arsene Wenger has a vision that one day robots will be standing on the touchline as managers.

The former Arsenal boss made the claim in an interview last weekend, also stating that he believes social media polls will be used to decide substitutions. The comments sparked a mix of intrigue and despair. The Mirror labelled it a "frightening prediction," while the Daily Mail referred to his claims as "startling."

Yet in an age when data and statistics are ever more embraced by top clubs around the world, perhaps the idea of technology taking an even tighter grip on the Premier League over the next 20 years should not come as much of a surprise.

It would not be the first job in history to be replaced by machinery. Check-out cashiers, rail-ticket sellers and factory workers already have stories of how their lives have been impacted by improvements in technology.

Replacing a human football manager with a robot would no doubt make for sensational headlines, but how seriously should we take Wenger's prophecy?


Wenger predicts robots could replace managers in the future! Watch 'Arsène Wenger in Conversation' here 👇 https://t.co/u6wqLygQlE … Available online for the next 7 days! #Arsenal #AFC https://t.co/odjXCsBIuF

The Frenchman's logic stems from the advancements he saw over the 22 years of his Arsenal reign, and social media undoubtedly became one of his biggest frustrations—as fans had a louder voice than ever before.

Now that he is out of the game, it seems the impact of Twitter is still lingering. 

"I've said many times, you could imagine the next chairman who says that the social networks can make a change in the second half," Wenger said. "That will become more and more entrenched. It will happen. I personally would not accept it. I'm from the old school in that respect. But we're going in that direction."

Formula E embraced the social platform four years ago, giving fans a chance to tweet drivers during each race. The number of tweets throughout the event were totted up so that, during the last lap, the driver with the most mentions received a power surge to his electric car.

Football has seen an example of in-play fan interaction, too. An annual pre-season fixture in South Africa allows fans to choose the starting lineups and then vote for subs.

"It is in the Carling Black Label Cup," Brandon Portnoy, a social media specialist at SuperSport, explains to Bleacher Report. "Kaizer Chiefs and Orlando Pirates play and, essentially, fans vote for their starting XIs through SMS and then have the power to make in-game substitutions through SMS, too."

Portnoy believes there are advanced ways Twitter polls could be used to trigger substitutions. 

"Imagine player shirt numbers were built with technology that changed colour based on a social media poll," he said. "If you vote for a specific colour, it could mean good, bad or fantastic. Then we'd see which players fans are enjoying watching, and which players they hate. That could probably be implemented in years to come—although it would be at a large cost."

Dean Jones @DeanJonesBR

Twitter polls to decide your team's substitutions... Would you be into it?

The rise of data has completely changed almost every aspect of the game. Statistics now evaluate transfer targets, wearables judge player condition and fitness in training and matches, and tactical reports are gathered using algorithms.

"Today we live in the world of artificial intelligence," Wenger told Andy Gray and Richard Keys on beIN Sports. "You are flooded with information at half-time. Sometimes you would be flooded with emotions but today, you have to meet the scientists who sit in the stand collecting the data. You sort out what is important—the modern manager is basically a guy who has to trust his feelings but also select three or four that can help him be a bit more efficient."  

It is almost impossible to be successful without embracing technology. 

"In the finance world it is impossible to work without Bloomberg," says Matteo Campodonico from WyScout. "Similarly, without all of the data we collect it would now be impossible to fully understand the world of football without being able to check how other teams play or scout through computers."

WyScout is a tool used by clubs, players, agents—anyone involved in football—to provide a blend of data and video footage of talent across the world. 

Traditionally, clubs would rely on scouts to unearth new players, and then follow up on those with potential. These days, scouts increasingly take their lead from computer-generated data.

"Data can predict and understand the game", Campodonico tells B/R. "You can try to find the new Neymar in quite an easy way. You can see how Neymar plays today: how his stats and data measure up. And then you can compare this to where he was at other points of his career.

"When looking for a player that can become like Neymar, you can suddenly do this at a click of a button—simply by comparing player stats to exactly how Neymar was playing at a certain stage of his career."

With computers able to generate a list of potential signings with such ease, the game has almost taken on a feel of Football Manager. Anyone could lean upon such information with a sense of confidence—and that's not where the tech help ends.

The game has become so far advanced that artificial intelligence can be used to analyse and reconstruct team performance. Croatia had great success at the 2018 FIFA World Cup, by embracing STATS Edge—a tool that compares playing styles and conducts game-changing set-play analysis.

You can even start to predict how an opponent might react to a specific scenario based on their past behaviour. 

With all that in mind, why do we still rely on a human as manager or head coach?

"AI has come a long way, but one thing we can not currently compare to is the way a human reads a game," Dr. Patrick Lucey, vice-president of AI at STATS says to B/R. "We don't have the senses—the visuals, the language between the players, an understanding of the personalities in the team.

"We have been tracking data since 1999, but to get closer to reading a game like Arsene Wenger can, we would need to go beyond tracking dots. We would need skeletons, faces, and we would need to know which way people are looking. There are so many nuances that we would need a humanoid robot, but we are a long way from something like Westworld."

A robot takes part in an adult-size category football match in the humanoid league tournament at the RoboCup 2017 in Nagoya, Aichi prefecture on July 30, 2017.  
In the four-day RoboCup 2017, about 3,000 researchers and engineering students showed off the
AFP Contributor/Getty Images

Robots do already exist in football. 

RoboCup is a tournament that was founded in 1996, and the organisers have a dream: By 2050, a team of fully autonomous humanoid robot footballers shall comply with FIFA rules to beat the World Cup holders. It certainly seems ambitious.

"The latest competition was held in Montreal earlier this year, with over 5,000 robots from 35 countries competing for glory," explains Robbie Dunion, an avid follower of artificial intelligence and robotics.

"Robots aren't very talented footballers and it's not much of a spectator sport, but the capabilities that are being developed and tested could eventually be transferred to other applications and technologies.

"I think that the future for robotics is much more around machine learning and the role that artificial intelligence can play in the stadiums, changing rooms, dugouts, TV studios and press boxes rather than directly influencing proceedings on the pitch."

Dr Lucey adds: "There is too much of a mismatch in the way robots move around. Robots' mechanical parts won't move like humans—learning from observation is better."

A robot takes part in an adult-size category football match in the humanoid league tournament at the RoboCup 2017 in Nagoya, Aichi prefecture on July 30, 2017.  
In the four-day RoboCup 2017, about 3,000 researchers and engineering students showed off the
AFP Contributor/Getty Images

Yet robots do have a track record of outdoing humans. IBM's Deep Blue supercomputer defeated world chess champion Garry Kasparov in 1997 and Google's AlphaGo computer program beat a master Go player two years ago.

Transferring such success to football is difficult, but producing a robot coach is more likely than ever fielding a superteam of 11 robots. 

Professor Peter Stone is a computer scientist at the University of Texas at Austin and has first-hand experience of putting robots through their paces in a football sense.

"I view all this as a wonderful technological opportunity, whether coaching or playing with robots," he explains to B/R. "It doesn't mean people lose their jobs because of it. It is just a challenge to see how far we can develop the technology.

"I think of the two, it would be easier to produce a coach. With players, there is the physical aspect of running, of being agile, of playing the ball in the right way and to be realistic—not just shoot a ball at 90mph at a certain angle that will make it unstoppable.

"Decision-making is a challenge, but from a technological point of view we are further from building a robot to do what Lionel Messi can do than we are to building a robot that can make decisions on what substitutions to make or which tactics to use.

"With the role a coach plays, it would also be easier to do this experiment at a football club using AI.  Whether it will ever happen could be down to public acceptance and what the players and organisations think of the concept."

One of the other key issues with robots is teaching them the rules of the game as instructions—something a human already understands instinctively. And unlike chess, the outside influences of football, such as psychology of players and reacting to match situations, are factors that really matter.

Daniel Polani is a professor of artificial intelligence at the University of Hertfordshire and one of the leading men at RoboCup. Asked whether Wenger is right about robots becoming managers, he also has reservations.

"I have no doubts they can serve as decision-helper—but the problem for me is that humans are imperfect executioners. Look at Pep Guardiola, he likes to work with the very best teams. Some say with this team anyone can win—but that's not true, I think it's naive. Even good teams can crash badly with the wrong manager.

"What happens, in my opinion, is that Guardiola is super-meticulous so only the best teams and best players can implement what he wants carried out. 

"The reason he needs good teams is because they are the only instrument that can generate the music he is striving for.

"A computer would have the same problem. They could have a good strategy but people may not carry it out."

CARDIFF, WALES - SEPTEMBER 22: Pep Guardiola Manager of Manchester City shouts instructions to his team from the dug-out during the Premier League match between Cardiff City and Manchester City at Cardiff City Stadium on September 22, 2018 in Cardiff, Wal
Athena Pictures/Getty Images

Robots are never going to take the job of someone like Guardiola, but with so much to offer the game there is still a high chance robots will eventually find a way on to the touchline.

Wenger's vision maybe not be exactly right, but robot fan Dunion reckons a day will come when we see robots as assistants to a human boss.

"You can also see from the recent Amazon documentary on Manchester City just how much importance Pep Guardiola places on the use of data to analyse performance," he says. "AI has the ability to take this analysis to a whole new level, providing real-time analytics which can factor in learned situations and scenarios.

"Rather than simply using statistics such as distance covered and pass completion, which are blind to the context of the game, AI can help to explore more in depth metrics based on a players specific role or the overall tactics and game plan.

"So a robot analytics coach or assistant manager could 100 per cent be a possibility in the future. In fact, I think that it will be a necessity...and I really hope the first one is called Robot-o Mancini."