Secret Level
THE UNSUNG HEROES OF DEVELOPMENT
Game designer JOHN DENNIS explains how to get difficulty balancing right
Before entering game development, John Dennis was “an art teacher who really didn’t enjoy teaching!” as he puts it. A friend who was working at Team 17 encouraged him to apply for a game designer job there, which, after some hesitation, he did. “I was the first game designer they’d had.”
Though at the beginning he wasn’t sure precisely what a game designer did, he was clearly a quick learner. He stayed at Team 17 for 15 years, and had advanced to the level of head of design, leading a team of 14, by the time he left. After a few years at Activision, he moved on to working with New Star Games and Five Aces, where he’s been for the past ten years.
The game he worked on for Activision was the mobile CoD game Call of Duty: Strike Team. Balancing difficulty for that involved a combination of things that may seem obvious, such as enemy health and accuracy, enemy effectiveness at taking cover and so on. As a racing game, New Star GP involves different variables, but the principle is the same. “We knew we wanted different tyre types,” says Dennis. “We knew we wanted fuel load, and for the fuel to affect the handling of the car. We knew we wanted dynamic weather in races. We knew we wanted component damage. The original concept for the career was that you’ll start racing in the 1980s and race all the way up to the 2020s. Each decade of racing will be more difficult than the last one. The tracks will get longer and more technical, the opponents will get harder. And also we’ll make the economy slightly tighter as well. The first decade will be very generous with the money, and it gets slightly tighter as we go.”
TYRE YELLOW RIBBON
This is the highly effective method of introducing elements to learn at an early stage (eg tyre wear) which become more important as the game progresses.
On top of that, New Star GP also uses adaptive difficulty. “If you beat [AI opponents] by a lot, for example, they jump up their performance curve more than they would if the race is very, very tight. Certainly in the earlier decades of racing, they might step back a bit and go, ‘We’re probably a little too far ahead of the player at this point.’”
Dennis shows me graphs of AI performance across the game’s decades, used to produce a literal difficulty curve. As each decade becomes more difficult, the adaptive element becomes less forgiving. During the Early Access period however, with only the first few decades available, some players wanted more of a challenge. APro difficulty was added, which makes things more difficult by dialling the adaptive element all the way down to produce AI opponents that want to win regardless o
