09 January 2012

Guest Column: Games and Simulations, with Dr James Sterrett

Following up on last week's GameTalk - "Games or Simulations" - we have a guest article written by Dr James Sterrett, an instructor at the US Army's Command and General Staff College. Please note that these are his ideas and are not reflective of official US Army policy, doctrine, canon, religion, or other official imprimatur.

Brant and I have cheerfully sparred over the distinction between games and simulations over the years.  What follows is my take, focused on training & education, in two different variants.  The first is useful as a snappy comment, while the second works better analytically.  In the end, both point to the objective of an activity as more important that the software (or paper rule set) being used, and neither variant is perfect.

(read the rest after the jump)

Variant 1: The only difference between a game and a simulation is the presence of a training objective.
You and I can sit down with JCATS, an official Army training simulation, and play it for the sheer fun of blowing up each others’ tanks – which sounds rather like a game.  Or, we could use the exact same software to conduct a staff training exercise.  We could also take Battlefield 2, a first-person shooter marketed as a game, and use it to conduct crawl-level training on basic techniques for an infantry squad.  In each case, the tool is the same, while the objective is different.
This definition makes a decent amount of intuitive sense – games are fun, simulations are for training.  It’s also a means of turning pointless complaints about an exercise driver - “we should use a simulation instead of a game” - turned into a fruitful discussion of what the exercise requires, regardless of the label attached to the driver.  However, this variant is not a useful analytic tool.

Sid Meier said that a game is “a series of interesting choices”. That’s great design guidance, but not so useful as a definition.  Greg Costikyan defined a game as “a form of art in which participants, termed players, make decisions in order to manage resources through game tokens in the pursuit of a goal.”  Costikyan’s is a good definition of a game (as opposed to a sport); see http://www.costik.com/nowords.html for the full article explaining the statement.  However, Costikyan’s definition doesn’t help us make the distinction from simulations – but Costikyan and Meier, both brilliant game designers, have a close focus on decision-making.
Both of their definitions imply that Candyland and the rest of the Snakes & Ladders genre are not games, because there are no decisions in them – though they can serve as good training on how to play a game: obey the rules, take your turn, good gamesmanship, and so on.
So Candyland isn’t a game, but it is a simulation.  Of what? Candyland is a simplified variant of Snakes and Ladders, which originates from India, where it was a morality lesson on life, with the ladders being virtues, the snakes being vices, and the endpoint being Enlightenment.  This brings us to…

Variant 2: A game is a simulation with decisions and goals.
Per the Defense Modeling and Simulations Office, a model is a mathematical description, and a simulation is a model iterated over time.  Thus, I can have math that describes how a missile launches from an aircraft.  As a single series of equations, it is a model. Translated into a series of iterated steps, it is a simulation.
Equally, Candyland simulates a particular model of life.  Neither the missile simulation nor the Candyland simulation involve any decisions. Thus, per Costikyan, they are not games.

When we add decisions and goals to a simulation, it becomes a game.

That means that, not only is Candyland a simulation... and an NTC rotation is a game.

Oddly enough, we called such exercises “war games”, once upon a time.

By: James Sterrett


So, what do you guys think? Where's the line between a "game" and a "simulation"? What's the difference and how should we distinguish them?


Brian Train said...

Thanks for posting this. I suppose Variant 2 is closer to my thinking about what we call "wargames" certainly, in that the number of individual decisions and turning points in the play of the game can be quite large, whether you are handling this with dice or letting a computer do it - all in pursuit of a goal that can have several layers to it.

But variant 1 reminds me that I'm not sure we should be drawing a line between "game" and "simulation" at all, because of the lessons even a fairly simple game can impart on both fun and training levels.

Perhaps we only feel driven to draw a line between the two to make the person above us, the one with the money and decision power, happy since he has such a dichotomy in mind.

Anonymous said...

I think the DMSO definition is targeted towards scientific models and simulations, and is therefore too strict to use within a gaming and/or training context.

For example...the computer program developed for the TV series "Deadliest Warrior" appears to be a simulator in so far as you a) create a set of parameters and variables, b) populate said parameters and variables with values, c) create an algorithm that calculates outcomes using the values, and finally, d) run the algorithm numerous times. The output of this simulation will be stastical display showing likely outcomes. If you want to modify any value to see its impact on outcomes ("what if" scenarios), you would need to run then entire simulation, again. In fact, this is exactly what is done with the kind of simulators that are used for modelling climate / weather patterns, oil field production, etc.

So why can't we use this model to our field? Because our games and/or simulators are by defacto objective-based exercises. The objective may be quite simple (cross the finish-line, first), but it still exists. No one starts a game of snakes-and-ladders and says "let's calculate the entire range of possible outcomes - starting with the greatest number of dice rolls and ending with the smallest number of dice rolls - that can be used to move our pawn across the board. And once we are done that, let's determine how far we can move along the board if we only roll the dice 3, 5 or 7 times, each. And then let's extrapolate our answers to determine what the outcome would look like if we played on a board with 1000 squares."

I would instead define a game as "a form of art that is based on a set of abstract concepts and rules where players try to achieve a goal by making decisions and utilizing resources"...and a simulator as "a form of art that uses emperical data, experience and knowledge to reproduce, in whole or in part, a real world experience or environment."

So Candyland, etc are all games (as are many "sim" games, like sim-city, etc)...but Stealbeasts, A-10 DCS and x-plane are simulators.

Jack Nastyface

Brant said...

Jack - I gotta disagree about CandyLand and Snakes & Ladders. There's zero participant decision-making. There's no game when there are no decisions to be made. It's an exercise in card-flipping/die-rolling and following directions, but it's not a game.

Anonymous said...

So if Candyland is not a game...it's also not a simulator, as it doesn't reflect any type of real-world experience, except in a completely metaphorical sense.

Do we need to introduce the element of competition instead of choicein the definition of a game?

Most humbly and Ever at your service,

Jack Nastyface

EastwoodDC said...

Jack writes: "No one starts a game of snakes-and-ladders and says -let's calculate the entire range of possible outcomes ..."

No one except a mathematician. Snakes-and-Ladders is a fine example of a Markov chain, and with a little Google you will find precise mathematical descriptions of the game. :-)

Perhaps Snake-and-Ladders is a bad example. Randomness can make a game more interesting by adding uncertainly, but games do not need to be random at all (Chess, Prisoner's Dilemma). Randomness (if any) seems to be part of the game setting.

Now by Variant 2, Prisoner's Dilemma (PD) is a game if you play it once, and a simulation if you play it repeatedly. Yet there is always a choice and an objective, so it's a simulation even if you play it just once.

[Interestingly, sometimes the best way to repeatedly play PD is to decide randomly.]

I'm having a hard time thinking of two players games that are NOT simulations. I think games that involve physical skills (marbles, Billiards, Tennis, Golf) might be distinguished as NOT simulations, though they could be training. Paintball, dodgeball, and even tag are simulations of a sort. I can't point to a single aspect that makes the distinction clear.

What about one players games? Can we find a line there?

Anonymous said...

A few points...
1) according to others, Snakes and Ladders is not a game...it's a sim.
2) Just because you can study outcomes using math doesn't make anything either a sim or a game.
3) Using Variant 2, Prisoner's dilemna is ALWAYS a game because it always involves choice and it has goal.
3) Back to my point...if a 2 player game involves choices, it is, using these terms, a game...not a sim.
4) ...which is why I suggest that the word "sim" really has two definitions - first (sim-1), a largely scientific and mathematical construct that is used to model a condition, environment or outcome, and second (sim-2), an interactive representation of a real-world environment that is based on known and observed conditions.

So it would be entirely possible to create a sim-1 simulation of "how many fairies can dance on the head of pin"...because an algorithim (with variables) could be devised to create an answer. You could NOT, however, design a sim-2 simulation of this...as there are no such things as pins..err...I mean, fairies.

FWIW...game designer Will Wright once described his well known design Sim-City as "...neither a game nor a sim, but a toy."

yours once again most respectfully,

Jack Nastyface

James Sterrett said...

Will Wright's designation of SimCity as a toy is something I really should have worked in - and he says that because it has decisions, but not goals - just like a ball. Once you add goals to the decisions you make with the ball, you have soccer, baseball, etc - a game.

It's true there are two things meant by "simulation" - I avoid Jack's meaning #2 because I'm working with the Army, and that meaning #2 is essentially the gamer meaning of the word simulation (which would be called virtual simulation in Army-speak - see below). Even in the gamer definition, though, the simulation need not be real-world or involve observable data. I-War was a great simulation (definition 2) of a spaceship that has never existed, for example; you can make a similar case around Bethesda's Elder Scrolls games.

Brian is correct, that we wind up having a distinction largely because of people with money and power, many of whom find the term game distasteful, yet then get hung up on the desire to bring in the games "the kids" are playing, and run aground trying to figure out the difference.

Dan, I don't think you are going to find a line. All games have rules; those rules are the simulation, and they encode the model. Just what is being simulated is sometimes completely unclear to me - Tetris doesn't have any clear analogue in the real world - but it is not required that the model describe the real world. (Not required according to me - DMSO claims the model is of something in the real world, but then they turn around and model stuff that doesn't exist yet. :) )

The side note on Virtual - the military recognizes live (real people using real systems), virtual (real people operating simulated systems) and constructive (simulated people operating simulated systems) simulations. The definitions more or less work but you can poke big holes in them. If I play Steel Beasts, I've got both virtual and constructive elements (and behind every virtual sim is a constructive sim in any event).