Designing Multiplayer Team-based Mechanics without adding Frustration

Introduction

I was recently playing a game of Heroes of Newerth (an online multiplayer 5v5 team game) with a few friends, and all of a sudden the following chat appeared:

Hon rage

What might surprise you (if you haven't played this game), is that this abuse came from one of our teammates, rather than the opponent, and was directed at one of my friends.

My friend had not spoken a word, he was not behaving badly - he was simply playing the game, and he made a simple mistake which led to his character dying in game, giving the opponent the advantage.

What I have noticed is that certain games bring out the worst in people. It's not that angry people are attracted to this type of game, it's that normal people are turned into angry people by playing this game.

So why is this? How can we design games to prevent this?

After examining the games with the biggest issues, certain patterns in the game design have emerged - there are some running themes with badly behaved teammates. These games share negative design patterns, which provoke and encourage bad behaviour, and can be solved with corresponding positive design patterns.

Firstly, lets look at the easiest, most obvious, and yet in some ways most useless pattern - removing / reducing teamwork. I consider this a non-solution. If you are changing your game to remove teamwork, you will get less teammate frustration, but are also removing a core part of what makes team-based games fun. Also known as "Throwing the baby out with the bathwater".

Teamwork vs Individuality

Some games are more at risk of teammate frustration than others. The biggest factor in whether teammates will get annoyed at each other is how reliant on their teammates players are, and therefore how much a player can accomplish on his own.

Every team-based multiplayer game sits somewhere different on the scale:

Teamwork wins <---------------------------------> Individual skill wins


Team-focused games include Team Fortress 2 and World of Warcraft arena. TF2 teams are closely tied together, with a medic healing their teammates, and their teammates protecting the medic. Once the medic dies, their teammates quickly fall. Similarly in World of Warcraft arena, healers must heal correctly, teammates must control the opponents at the right time and protect their healer. Teamwork is absolutely essential to success and without it individual players have no chance.

In Team-focused games, an organised team of unskilled players can beat a disorganised team of skilled players

Individual-focused games include Counter-Strike and Call of Duty. One person can kill the entire enemy team, without any support from his teammates. One good bullet to the head can kill each player. If you're good at these games, you can join a server and get a great personal score, no matter who you're playing with. Having a good team obviously helps (and at the highly competitive end, teamwork can become essential to win), but compared to other team-based games, individuals can really shine on their own.

In Individual-focused games, a disorganised team of skilled players will beat an organised team of unskilled players

There are advantages to both design directions:

Individual-focused games - Advantages:

Team-focused games - Advantages:

Generally, if you are making an individual-focused game, you'll have less problems with teammates shouting at each other. However, the design challenge is creating and keeping team-focused mechanics without introducing teammate frustration, so you keep all the benefits of teamwork without bad teammates ruining your gameplay experience.

Rule #1: Avoid Negative-spike mechanics, and design in Positive-spike

The first concept is spiking success vs spiking failure. If a healer is healing perfectly 99% of the time, then 1% of the time he fails and a player dies, that player will typically ignore his healer 99% of the time, and get angry 1% of the time. This is what I call a "negative-spike". Success is normal, failure stands out and is heavily punished.

For example, negative spike occurs often in World of Warcraft raiding (25 players defeating a large boss together), where the team succeeds simply if everyone does their job reliably. If someone casts the wrong spell, or stands in the wrong place, or fails to press a button in time, everyone dies, and that failing player gets shouted at.

Positive-spike is when small failures occur all the time and are expected, so success stands out. For example a player is surrounded by enemies in Counter-strike, he's scared and he calls for a teammate to help. Everyone expects him to die but his teammate throws a perfectly placed frag grenade - it kills 3 enemies and the player survives. He was expecting to die, and yet his teammate (surprisingly) saved him with an excellent grenade throw. This is a spiking sucess.

If the grenade didn't kill anyone and the player died as expected, he wouldn't have been upset, but yet the success was a spike and that feels good. Success spikes are more easily remembered, talked about, and laughed about - those are the memories which stay with players ("Remember that awesome thing we did?") and bring them back to your game again and again.

Rule #2: Remove any 100% reliances on teammates

If you cannot progress until your teammate does something, this leads to frustration. For example, a teammate in Counter-Strike has the bomb but he doesn't go to the bombsite, or he stays at spawn or goes the wrong way, this will cause anger in his teamates.

Another example is Team Fortress Classic, which had certain walls in some levels which required a certain class (demoman) to blast a hole through. In combination with server-based class limits (e.g. only 1 demoman allowed per team), if you have a demoman on your team and he does not blast through, your team cannot progress and will get angry.

There were even some community-made maps which placed heavy reliance on other classes performing certain actions, e.g. an Engineer to fix an elevator, a Spy to open a door, etc. This is a really bad way to encourage class diversity and "teamwork", and leaves the door wide open for frustration. Fortunately the feature was not carried over into Team Fortress 2.

A way to overcome this problem is to provide alternatives. Team Fortress 2 has health packs dotted around the levels so you aren't completely reliant on a medic. If you get hurt and your medic dies or refuses to heal you, you can simply pick up a health pack or go to a resupply station. You aren't 100% reliant on the medic for healing.

Rule #3: Make it hard to correctly judge teammates tactical choices (back-seat playing)

If a teammate is doing something wrong, and it's obvious, it's easy to get annoyed at him. For example, if you are on low health, and the medic instead heals someone that is already full health, which causes you to die ("Why are you healing him? You should be healing me!") it is obvious he is doing it wrong, so it is easy to be annoyed.

On the other hand, if you have no idea what his correct action is, it is much harder to point and say "you're doing it wrong".

Fixing this is actually quite complex, and requires 3 components:

Depth means multiple viable tactical options. If teammates have lots of choices, and it's difficult to calculate the correct choice, then it's much harder to point to the exact strategy and say "You should be doing that". Without depth there is only one sensible thing to be doing, and doing anything else gets shouted at.

Imperfect information is where you deliberately obscure some game state information from all players so they cannot make perfect decisions (they must guess / estimate). For example, if you know exactly how much health a teammate has, and you know exactly how much you can heal, it is possible to calculate (and later argue) precisely what the correct course of action is (this is bad).

Alternatively, if the correct choice of action depends on e.g.:

then everyone will come to slightly different conclusions. Any tactical choice based on those conclusions is much harder for teammates to say definitively "this was the wrong choice".

Make it impossible to analyse and calculate what the correct course of action is, by making it impossible to perfectly identify the situation.

Hidden personal information is where each player has personal information that their team is not aware of. For example, if you don't know how much health your teammate has, you can't blame him for hiding on full health. If you don't know how many bullets he has left, you can't blame him for reloading. If you don't know whether his spells are available or not, you can't blame him for not using them.  

By making knowledge private you prevent players from disecting their teammates actions.

Rule #4: Bad Teammates should ALWAYS be Better Than Nothing

This rule sounds obvious, and yet so many games fail to adhere to it. Teams should never be better off without a player, no matter how bad he is. If a bad player is more damaging to the team than no player at all, then obviously players will get annoyed at him.

For example, are there shared team resources which teammates can squander? Can teammates get in your way? Can they cause you harm directly or indirectly?

Some examples where bad teammates can be worse than no teammates at all:

There are plenty of guilty mechanics. If you are considering including the design for a mechanic like this, think "Is it possible to achieve the same gameplay result without punishing the whole team for 1 player's mistake?"

Summary

If you are designing a multiplayer team-based game, here's your checklist to avoid teammates getting frustrated: