All products featured on WIRED are independently selected by our editors. However, we may receive compensation from retailers and/or from purchases of products through these links.
The future looks fun for gamers. With Google's streaming service Stadia set for release in November and Microsoft and Sony PlayStation announcing a rare partnership to compete with it, competition is fierce to decide who takes pole position in the race to become the next major online platform. Gamers will be spoilt for choice, and they are right to be excited.
Most of the questions about these platforms have come from gamers, desperate to know more about the technical possibilities: how much is the service going to cost? Will it work on average internet connections? Will VR become a future feature? And what impact will it have on the gaming industry as we know it?
These are all important questions. Most of them, however, focus on what is being given to the gamer, rather than what the companies building these new services are getting in return. Obviously, profits are important. But are these platforms gunning for other, longer term assets? What about data, for instance?
New cloud-gaming platforms have the potential to generate far more behavioural data and insight than social media platforms do today. Just think about how much insight organisations like Microsoft, Amazon or Google can glean from analysing how billions of people play computer games every day.
To begin with, there is the surface data that can be collected from simply signing-up to the new service. This could include credit card details and financial information, geo-spatial data showing your location, who you play with online, which games you choose and the conversations you have with fellow gamers.
In the future, however, cloud-gaming platforms could be even more audacious in the analysis of how people play computer games. The platforms might be able to gain insights into decision-making processes, how alliances are formed and broken, how individuals respond to stress, what motivates and demotivates them, how they learn and how they strategise. You can learn interesting things about human behaviour and intelligence from analysing video game play.
A growing body of research is indeed trying to learn about human behaviour from video game behaviour. It includes recent studies, such as “Human Learning in Atari” from Harvard’s Department of Psychology. Here, researchers observed people playing Atari games to understand how humans learn in comparison to AI programmes. The researchers found games to be “an excellent testbed for studying intelligent behaviour”.
Samuel Gershman was a member of the “Human Learning in Atari” research team. He argues that “most psychology experiments could be construed as video games.” Psychology experiments need to take place in a controlled environment in which participants complete some contrived task, so that the researchers can draw inferences about general behaviour. Video games are not much different. They provide an artificial environment, enabling researchers to examine behaviour with the necessary experimental control, but in the same breath they possess some of the complexity found in real-world situations. “Video games sit somewhere in between control and chaos, and can provide a microcosm of behaviour,” says Gershman.
Another person that has experimented with analysing human behaviour through computer games is John Laird, a professor of engineering and computer science at the University of Michigan. Laird pointed out that “there are many different levels at which you can learn about human intelligence and behaviour through computer games.” For example, researchers have analysed how humans react to events, by measuring how quickly game players click to respond to an in-game threat or situation. Researchers have also tried to make inferences about how humans create tactical and strategic plans, as part of teams and groups.
These insights into intelligent behavior can also be used to refine and improve artificial intelligence. The analysis of game playing has been influential in the field of imitation learning. “The idea behind imitation learning is that I observe the behavior of an agent, (it could be a human or a machine), that is really good at a task,” Gershman explains. “And then you try and find a policy that is similar to the behaviour exhibited by the expert agent.”
The artificial intelligence-focused lab at DeepMind (a subsidiary of Google) has long recognised the potential of imitation learning. The firm honed its AI by training it on Atari games and, more recently, real-time strategy game StarCraft II. Learning through games has underpinned DeepMind’s approach to understanding intelligence from the get-go.
Improving AI-training through the analysis of gameplay could open up exciting new avenues of research. "If you can set up situations in games that match real-world situations, you can help inform machine capabilities," Laird says. Imagine the creation of social games, to better understand and model complex social environments, or driving games to help autonomous vehicles navigate the roads.
Of course, there are limitations to making inferences about human behaviour through how we play computer games. People may act differently in virtual environments and in real-world environments. And virtual environments are never as accurate as real life. Both of these factors present limitations as to how much can be extrapolated from computer game behaviour.
Limitations aside, the world’s main technology giants have clearly spotted the potential of gaming platforms. The best games draw huge numbers and inspire hours of gameplay. Nothing proved this more than the explosive growth and popularity of Fortnite, which in November 2018 registered 200 million active accounts. This could mean a lot of behavioural data and see gaming platforms dwarf traditional social media in size, scale and behavioural insight.
As well as the opportunities, there could also be risks regarding surveillance. In her book The Age of Surveillance Capitalism, the Harvard professor Shoshana Zuboff, argues that we are currently witnessing the emergence of a new form of economy, which Zuboff has branded "surveillance capitalism". According to Zuboff, in this new economy, major technology companies are locked in a race to collect as much behavioural data as possible. What you think, what you feel, how you write, when you walk to work: these companies are interested in collecting every conceivable behavioural trait about their users. They use this in order to predict and manipulate future consumer behaviour for commercial advantage.
Could cloud-gaming platforms be a new front in the surveillance capitalist economy? It’s hard to know at this stage, as most companies have been fairly secretive as to how their gaming platforms work (Google did not return a request for comment). But we might expect different companies to take different approaches to privacy. Apple’s new Arcade platform claims to “respect their users privacy”, a stance that falls in line with their pro-privacy track record and public stance. In a recent interview, the team at Google Stadia mentioned that privacy will be “at the user’s control”. They did not specify, however, what the default privacy settings would be, what sorts of data the systems would be collecting and how this data would be used.
Answers to these questions may arise sooner than we think. Cloud gaming platforms are on the cusp of being launched, with many of the design decisions influencing the ethical and social values of these platforms already resolved. The rise of cloud-based gaming platforms is exciting – but it will be fascinating to see who is playing who.
This article was originally published by WIRED UK
