Cars and the trust problem

Raph Koster took part in a huge research about levels of trust and grouping in video games. Please understand that “trust” in game theory isn’t about moral betrayal but simply failure to cooperate for any reason, like “you can’t trust Anne being on time, she is often late”. They point out that multiplayer games must be designed focusing on the trust level of the players: “a game might fail if it calls for very high trust, but people tend to play it with strangers”. They also set the goal of using the game to improve trust among people.

For example killing a random mob together needs very low trust. Even if the other guy does zero DPS, you are no worse off than playing alone. Acquaintance games need some level of trust, mainly trust in reciprocity (I do my part, you do yours) like the one you’d expect from a WoW pickup group: don’t pull mobs, don’t ninja, don’t spam hate, do some DPS. Games that call for friends include incomplete information, blind moves, synchronous coordination, for example Mythic Raiding with a fixed team. True high trust games start featuring permadeath mechanics, prisoner’s dilemma, self-sacrifice for the good of the group, and so on and can only be played with people you really trust. The only in-game equivalent I can think of is account sharing (which is banned exactly because GMs don’t want the drama when this trust is abused).

They point out that games with non-overlapping roles require trust. If you can’t heal yourself, you must trust that the healer can do his job, while you can carry a sucking DPS. The problem (which they implicitly try to solve but fail to) is that players like high-trust mechanics (like classes, complicated bosses, epic adventures with lots of people), but without the needed trust, the experience turns out horrible. They also point out that making a high trust game is easy as the mechanics are self-evident. Anything “realistic” is necessarily high-trust, as the humankind is built on high-trust cells like families. Then Raph arrives to the elephant: “What’s hard is designing for high and low trust at the same time.” and while provides some tips how the game should look like, none of them is new or non-obvious. The advises are “do X but don’t overdo it” kind of “sweet spot” euphemisms of “well, we have no clue how to solve it”.

Now comes the part where genius Gevlon solves what so many researchers couldn’t. Except I don’t have to, because it’s already solved.

Driving cars on the road need the highest level of trust, just like Raph’s favorite example: flying on the trapeze. I must believe that none of the other motorists will crash into me, causing serious damage in my car or even killing me. The other motorists are total strangers, I don’t even have time to look at their faces. Yet traffic functions properly. It also has great scalability, works at night when barely anyone drives and in a rush hour too. If we answer why traffic works, we can easily design games that have high-trust gameplay shared with strangers.

The solution is external authority. There is a third party in the equation: the state that does the following:

• Formalize the patterns of driving, informing all drivers how they should drive in order to do it safely. As long as everyone keeps these directions, no accidents will happen. These rules set clear responsibilities: it’s always clear which car shall yield to the other.
• Issue driving permits to those who are trained and tested to be able to follow the above directives.
• Punish those who break the trust by performing actions that could lead to accident (speeding, drunk driving, ignoring road signs and red lights). Seriously punish those who cause accidents.
• Enforce mandatory insurance system to guarantee that any monetary damage caused by a “trust-breaker” is paid.

How would it look like in a video game:

• The tasks of the players are clear and responsibilities set. For example if a mob must be kited, one player must pick up a kiting item which defines him as the kiter of that mob. From there no one else can help kiting that mob, that’s his responsibility.
• The game system issues statistics to show how the player performed previously in that role. For example when a raid leader seeks a kiter, he can look up the applicant’s “kiting score” which he earned by doing content as kiter.
• The game should explicitly call out who failed, which act both as punishment for the failer and removes the burden from the raid leader. For example if the kiter moves the mob right on the raid, killing them, the game itself should issue a raid warning “Adam failed to keep Evil elemental away from Betty who is now dead.” Of course this assumes that the raid locations are defined by the game, so if Betty wanders off and hugs the elemental, she’ll be called out.
• Problems caused by fails must be player death or non-damaging penalty (like an undispellable debuff). Avoidable damage punishes the healers, not the failers.
• Raid repairs after wipe is on the failer, enforced by the game system.

This way a Mythic progression raid can pick up a complete stranger as the game guarantees that he has past experience in doing the task (even if not on this boss), calls him out if he fails and punishes him instead of forcing another player to confront him. Not only such game allows trusting the stranger, but helps building trust by providing strangers who likely not abuse it. Groups are more likely give a chance to someone who likely succeed than a total random. By observing him succeeding and helping in a task they cannot do without him, they will have more trust in him, likely even adding him as a friend for further raids.