What can game developers learn from road safety?

Ben Lewis-Evans used to work in road safety. He's now a UX researcher with Epic Games, but finds his previous career informing his current one in a variety of ways. Speaking at the Game UX Summit in Toronto this week, Lewis-Evans described the road safety concept of "The Three Es" for the game developers in attendance.

Each of the three Es is a different tactic to help increase road safety--Education, Enforcement, and Engineering--each of which Lewis-Evans said could be applied to game development to help address unwanted and anti-social behavior from players.

Each tactic has its own benefits and drawbacks, Lewis-Evans said. In road safety, education might consist of traffic signs warning people not to drink and drive or ad campaigns against distracted driving. For games, it may be terms of service players agree to when they install a game, or a code of conduct posted on its official website. Education is relatively easy and inexpensive, he said, but it's also the least effective on its own.

"The community will generally react well to education efforts initially, but if they're not backed up by something else, that can turn," Lewis-Evans said.

"Even though we like to believe that it does, awareness of knowledge does not usually result in behavior change just by itself"

There are some ways to maximize education efforts. For example, keep it close to the activity you're trying to improve. A code of conduct policy on a website might not be remembered by players once they boot up a game, but reminding them of proper etiquette and behavior on loading screens just before gameplay can make a bigger difference.

"Even though we like to believe that it does, awareness of knowledge does not usually result in behavior change just by itself," Lewis-Evans said.

Enforcement is one common solution suggested as a correction for bad behavior, but the goal is not to punish people; it's to deter them from behaving poorly in the first place. Whether dealing with traffic safety or trolls, there are three major components to effectively deterring people with enforcement, Lewis-Evans said.

The most important component is certainty. People have to know that if they violate the rules, they will be caught every time. After that is swiftness of punishment. People need quick feedback when they break the rules, or they're more likely to ignore them. The third, and least important component for Lewis-Evans, is severity. Having draconian punishments that may or may not ever be invoked isn't necessarily a better deterrent than a quick, minor punishment administered instantly whenever people break the rules.

These sort of deterrents can take several forms in games. Lewis-Evans said it's important that offenders get detailed feedback about why they've been punished. Instead of just telling them they've been dinged for poor behavior, it might be better to say they've had six complaints levied against them during three different matches, and from four different parties. It might also be helpful to explain to them how that's not normal, telling them perhaps what small percentage of other players have received this many reports in that short a span of time.

It's also important that the people who report bad behavior in-game are rewarded for their efforts, with Lewis-Evans suggesting an automated thank you to notify them that their report had been reviewed and acted upon. If developers want to continue having people participate in such reporting systems, it's important they know that their efforts to report other players are working.

"Even if there's a small amount of toxicity in a big, general channel, that can become the norm"

Like education, enforcement also has its downsides. If developers are relying on player reports, that hurts the certainty and swiftness of their response. There's also concern about false positives, or as Lewis-Evans cautioned, the perception of false positives. People will of course claim that accusations against them are false, and that can become a big issue in the community. It's also harder to draw lines around certain bad behaviors. For example, if a player keeps dying to the enemy in a MOBA, are they just not very good, having a bad game, or intentionally "feeding" the enemy to sabotage their own team's chances?

Punishment can only teach what not to do, Lewis-Evans said, and it also tends to focus on repeat offenders, even though in his experience, most offenders are just one-off problems where someone has had a particularly bad day. Lewis-Evans touched upon a less-used fourth "E"--encouragement--as another way to promote good behavior, pointing to League of Legends' honor system or DOTA's highlighting of players with very small number of negative reports as examples.

Finally, Lewis-Evans discussed what he sees as the most effective solution to these problems: engineering. On the road, you can try signs advising people to drive safely or high fines for speeding tickets, but neither can prevent head-on collisions the way a physical divider between lanes of traffic can. Fortunately, there are plenty of ways to handle this in games.

Lewis-Evans started with communication systems, noting that Epic doesn't have built-in voice chat in any of its games. After surveying players, they found that people value voice chat to speak with friends and family very strongly. There was some value seen in communicating with strangers and teammates, and little value in talking to opponents. But when it comes to using voice chat, most people talked to friends and family, and even though few saw value in chatting with opponents, even fewer actually used it for that.

Even trivial decisions can have an impact, he said. For example, League of Legends used to have global chat on by default. When they changed that to a feature players had to opt-into, they found it reduced negative behavior, even though the overall volume of chat was more or less unchanged.

"In a lot of team games such as MOBAs or first-person shooters with defend-the-flag modes or whatever, we glorify getting kills, and then we hardly reward at all defending, or playing a defensive role, or supporting others"

"That's really important because even if there's a small amount of toxicity in a big, general channel, that can become the norm," Lewis-Evans said. "That can become, 'This is OK, we don't see people being punished for it.' And this kind of behavior goes around."

Online games that are winner-take-all can also be a problem. If there's no reward for losing a match, people will abandon the game as soon as the tide turns against them. Developers can discourage that by giving people small rewards for losing and having them build up over time, giving them incentive to gut out a tough loss.

Matchmaking is another place where developers can engineer against toxicity. In team games where it's important that each team have a well-balanced roster of players, developers can either give players prompts to tell them what their team needs, or better still, let them pick what role they want before the matchmaking process, and then arrange players onto teams that are naturally well-balanced.

More core gameplay elements can also be tweaked to promote cooperation. For example, when a player in Destiny uses their super ability, they spawn a bunch of little orbs that can be picked up by other players to help boost their own supers.

"Even if you're the most selfish player in the world who never wants to help anybody else, as long as you're using your super, you're helping other people. And they know you're helping them. They're aware of why you're helping them. These systems are built into the game to encourage positive team play."

Just as engineering can foster positive behavior, so too can it unwittingly encourage the sort of toxic behavior developers are trying to stamp out.

"In a lot of team games such as MOBAs or first-person shooters with defend-the-flag modes or whatever, we glorify getting kills, and then we hardly reward at all defending, or playing a defensive role, or supporting others," Lewis-Evans said. "It's all about the kill-death ratio."

Ultimately, Lewis-Evans said his goal is simply having everybody be nice to each other online.

"Maybe this never happens, but I do think it's a goal that's worth working towards."