Toward a Grammar of Multiplayer Gametypes

Two things strike me as odd about the multiplayer gametypes in video games which model human-like players in a 3D arena. First, they are generally based on familiar real-world game rules, like Capture the Flag or King of the Hill, which is not odd, but then they rarely stray outside those rules. The other thing is, as a designer, there doesn’t appear to be any grammar for how those games are built, or how they can be compared or analyzed. What might such a grammar of multiplayer games look like?

First, some context. When I was a kid, often on summer nights a bunch of neighborhood kids would gather in a local field to play games. 2-Flag Capture the Flag and Jailbreak were the go-tos, Tag or Hide 'n' Seek if we weren't feeling organized, and we'd play for hours until our parents called us home. There might be minor variations in the rules (flags hidden or in the open; a jailbreak frees everyone or only one player) but that was grist for debates. It was common for players to propose new rules, playtest them and usually discard them.
 

Playground rules in video games

20 years later I was at Bungie Software during the development of the sci-fi FPS Marathon. It featured multiplayer between ethernet-networked Macs, a rare thing at the time. The primary multiplayer game was what Doom players called “deathmatch” and we called “Every man for himself”. But the sequel, Marathon 2: Durandal introduced new multiplayer modes: Capture the Flag, Kill the Man with the Ball, King of the Hill, Tag – all analogs of playground games familiar to us from real life. Though the great majority of FPS multiplayer games actually played was always “just shoot everybody” those new gametypes added different kinds of strategic routes toward winning games. The LAN parties that Marathon, Myth and Halo CE all spawned proved to be exciting social activities as well.

Bungie's later games, the Myth and Halo series, carried on this tradition of adapting playground multiplayer game rules into RTS and FPS multiplayer. These served Bungie's games well and went on to be elaborated in the many interesting gametypes in the later Halo games. Halo Reach’s Invasion gametype made big strides toward a dynamic set of game goals.

(Here is where I should do an analysis of the many interesting gametypes in LoL, TF2, Overwatch, Garden Warfare, many others - but I’ll leave that for a future post.)

A grammar engine

10 years after that, now a designer at Wideload, I worked on an entirely different kind of game called Disney Guilty Party. It was a turn-based, cooperative, family-friendly mystery game on the Wii, where a group of players worked together as detectives to solve mysteries – about as unlike a fast-paced competitive shooter as you can get.

We made an engine which generated mysteries, naturally called the Mystery Engine. We populated it with over 3000 clue items, each tagged with nouns and verbs that identified the core meaning of the clue. The Mystery Engine used this data to relate clues and generate randomized mystery scenarios which were non-obvious and solvable. There were billions of unique possibilities with this relatively small set of data.

It used a grammar of nouns and verbs to create these mysteries. While the mysteries it generated were admittedly pretty simple, it's the only real random-mystery-generator in a game that I'm aware of. Disney acquired Wideload for that game, and Patrick Curry, Fletcher Dunn and Nick Schrag were awarded a patent for that engine.

Sadly, the Guilty Party franchise did not survive at Disney and Wideload went on to work on a prototype for a real-time, competitive multiplayer game set in a superhero universe. Some of the tech developed for the Mystery Engine was applied to a new engine, the Enigma engine, which was now designed to randomly create superhero adventures.

Grammar to gametypes

One key to the drama in a superhero adventure is the Dilemma. The hero/player is confronted with a hard choice - either save a single target which means a lot to them personally (i.e. their beloved), or save dozens of civilians from the villain's evil schemes (i.e. a train car full of people). The tension of that choice is what makes the story engaging. A villain would just let the civilians die, but you are a hero - what will you do, how will you resolve this conflict?

For our multiplayer superhero game, we wanted to make a dynamic goal-generating system, where the player would begin the match with one game goal in mind, but then the engine would introduce Dilemmas - either adding new goals, transforming the base goal or even superseding it with an unexpected, greater goal.

We wanted a goal randomization system such that few matches would play the same, players would often be surprised and rise to the new challenges, and they would come away from the match with interesting stories about the twists and turns of their match. Finally, we hoped that this system could take the place of writing a lot of authored narrative - the game would tell the player their changed goals and circumstances, and their imaginations would make sense of it without a ton of exposition from us.

In the course of designing this Enigma engine with the team, I worked up a list of nouns and verbs which might be randomly shuffled to create a huge variety of game goals and rules. In fact, it seemed to me that a scheme like this might be able to systematically describe every playground game, every sports game, and every videogame using similar rules - any game played with multiple human-like avatars on a field. It also offered the possibility of creating entirely new gametypes which no one had thought of yet.

A possible grammar of multiplayer gametypes

Here's what we arrived at – this was basically the first-pass of a written design, and was only lightly implemented in-engine. But the prototype did work with a subset of these nouns and verbs. I consider this a starting point for the development of a better scheme.  

The 7 nouns in Enigma:

The 10 verbs in Enigma:

There are huge caveats to point out with those lists – they were made with our intended gameplay in mind, and many of those verbs above could be broken down to finer atoms. For example, "delivering" could be broken down into “acquiring the flag” "moving the flag", and then "entering a site with the flag". We decided at the time that the granularity of “delivering” was appropriate for the design of the prototype we were making. We had a subset of these games working (and pretty fun!) in the engine when the project died.

The big idea

So, could a collection of nouns and verbs like this be useful for systematically describing a wide range of multiplayer gametypes, in both real and virtual worlds - sports games, playground games, videogame multiplayer? Here are some examples: