Jump to content
  • Advertisement

Archived

This topic is now archived and is closed to further replies.

Diodor

Abstraction vs. simulation

This topic is 5204 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

It appears to me that the general game design thinking is that deeper, "more accurate" simulations are the way to create better games. I find this school of thought in the obsession with advanced photo-realistic visuals, in the hunt for the ultimate Matrix-like RPG, in the dream of a strategy game with such an advanced AI for each of the units that you can give general orders that are propagated through a hierarchy. Sim games do it. Black and White did it. The recipe is to create an extremely complex system, then give the player some control over the system and let him have fun with it. Since the system is so complex, learning how to control it is an interesting experience. Abstraction is viewed as a necessary evil, as a compromise to the realities of technical problems. One reluctantly abstracts away everything too difficult and expensive to simulate. Abstraction is recognised as important because it can make the difference between a finished project and vaporware, but is otherwise shun. So what am I complaining about? It is obvious that every game requires both simulation and abstraction. Firstly, complexity is not desirable. The best games have few rules and deep gameplay. Games (not limited to computer games) are enjoyable _because_ they are superficial, not in spite of it. Adding more rules should be regarded as a design failure, not something that is universally desirable. Secondly, it is the abstraction that defines the game - it limits the range of activities of the player to a small subset ("gameplay") that is considered "fun". Different abstractions create different types of gameplay. Several abstractions are widely used (eg. controlling movement is reduced to pressing a few keys) and indeed their effect on gameplay is so profound it is often taken for granted. Those abstractions that make the game world seem as concrete and "real" as possible are prefered (the visual criterion is slowly giving more room to physical and AI realism) - but this choice may be very limiting. Game design should always start with choosing the right abstraction. Emotionally engaging gameplay, deep political intrigues cannot simply be added on top of a RPG/RTS/TBS/shooter/whatever, but may well be a lot easier to achieve using a good metaphor.

Share this post


Link to post
Share on other sites
Advertisement
quote:
Original post by Diodor
It appears to me that the general game design thinking is that deeper, "more accurate" simulations are the way to create better games.


Depends upon the genre, doesn''t it?
quote:

Abstraction is viewed as a necessary evil, as a compromise to the realities of technical problems. One reluctantly abstracts away everything too difficult and expensive to simulate. Abstraction is recognised as important because it can make the difference between a finished project and vaporware, but is otherwise shun.


I rather think you''re overstating the point here. Abstraction also makes a game playable. If there was no abstraction, only highly trained professionals could play the simulation. Generally, such a simulation isn''t called a ''game'', but a ''training program''.
quote:

So what am I complaining about? It is obvious that every game requires both simulation and abstraction.

Firstly, complexity is not desirable. The best games have few rules and deep gameplay. Games (not limited to computer games) are enjoyable _because_ they are superficial, not in spite of it. Adding more rules should be regarded as a design failure, not something that is universally desirable.


Depends upon the game. ''Easy to learn, hard to master'' games have very few rules that interact to produce complex behaviour. But games like AD&D have many rules. Part of the challenge is figuring out how to work within the framework the rules provide. Also, a lack of complexity is not caused by having few rules, a lack of complexity is caused by having a stupid game. And stupid games tend to get boring pretty fast.
quote:

Secondly, it is the abstraction that defines the game - it limits the range of activities of the player to a small subset ("gameplay") that is considered "fun". Different abstractions create different types of gameplay.


So isn''t it good if, say, an RTS sports advanced AI that lets players interact with their units as groups, but which can be bypassed for control of individual units? Wouldn''t such a game achieve "fun" for more players than one with a rigidly fixed set of gameplay responsibilties?

Isn''t conditional complexity a good thing?
quote:

Game design should always start with choosing the right abstraction. Emotionally engaging gameplay, deep political intrigues cannot simply be added on top of a RPG/RTS/TBS/shooter/whatever, but may well be a lot easier to achieve using a good metaphor.


Isn''t emotionally engaging gameplay the point of games? A game that doesn''t inspire some kind of emotion (be it power, wonder, merth or whatever) isn''t a game worth playing.

But, I prefer a game that''s flexible enough that players can experience a variety of emotions depending upon how they play it. Go around blowing stuff up, and feel Power. Go around looking at stunning scenery and visual effects, and feel Wonder. Go around looking at the jokes, and feel Merth.

Share this post


Link to post
Share on other sites
quote:
Depends upon the genre, doesn''t it?


More advanced graphics, better physics, better AI, more units, more items, larger maps - I''d say there''s a trend here and it''s pretty genre independent.

quote:
I rather think you''re overstating the point here. Abstraction also makes a game playable. If there was no abstraction, only highly trained professionals could play the simulation. Generally, such a simulation isn''t called a ''game'', but a ''training program''.


Yes!

quote:
Depends upon the game. ''Easy to learn, hard to master'' games have very few rules that interact to produce complex behaviour. But games like AD&D have many rules. Part of the challenge is figuring out how to work within the framework the rules provide. Also, a lack of complexity is not caused by having few rules, a lack of complexity is caused by having a stupid game. And stupid games tend to get boring pretty fast.


I consider games like AD&D a step in the wrong direction - they _are_ fun (for some people at least), but they are at a point where even more rules would not make the games more fun.

It can also be argued that AD&D only has a few meta rules that define the game world: players are described using statistics, all the actions of the players depend on the statistics, equiped items can modify the statistics, etc. Understanding the meta-rules allows one to play the game without understanding all the vast volume of rules.

quote:
So isn''t it good if, say, an RTS sports advanced AI that lets players interact with their units as groups, but which can be bypassed for control of individual units? Wouldn''t such a game achieve "fun" for more players than one with a rigidly fixed set of gameplay responsibilties?


The reason the current RTS abstraction (select units give "move here" orders) creates fun gameplay is very much the same as with 2d screen scrolling shooters. It''s all about dealing quickly with an overwhelming amount of events. High AI would not necessarily change that (the way the units in Age of Empires form pretty looking battle-groups) and if it would the game would be seen as "less" fun by RTS pundits (and hopefully "more" fun by other people).

My point is if you want to create a different kind of strategic gameplay, there''s no reason to keep any of the usual stuff you find in a RTS (units of different kinds, a 2D map etc.)

quote:
Isn''t emotionally engaging gameplay the point of games? A game that doesn''t inspire some kind of emotion (be it power, wonder, merth or whatever) isn''t a game worth playing.


Emotion is the result of (some) games, not the subject of those games (my last post was ambiguous here). This is not because it can''t be done in games (look at the smilies above - yes, it''s that easy) - it''s because current abstractions don''t support that kind of gameplay.

quote:
But, I prefer a game that''s flexible enough that players can experience a variety of emotions depending upon how they play it. Go around blowing stuff up, and feel Power. Go around looking at stunning scenery and visual effects, and feel Wonder. Go around looking at the jokes, and feel Merth.


Wonder and Merth are not inherent in most games I play - they are borrowed from the other arts.

Share this post


Link to post
Share on other sites
Abstraction isn''t always a necessary evil. It depends on what the designer wishes to model. I believe that analyzing what you wish to model, and how this model affects the emotional and mental state of the player is the key to successful design.

The trick is in figuring out what you want to model. But in order to figure out what you want to model, you must first decide what kind of experiences you want to give to the player. I personally tend to prefer systems that are more realistic because generally, realistic systems are more consistent. They require less denial of plausibility and as a consequence can feel more real and immersive for me as a player. This is why I prefer games like Ghost Recon and Rainbow Shield over games like Unreal Tournament or even Medal of Honor (i mean come on, there''s a bunch of first aid packs all over the place that heal you?).

But some people prefer more drama in their games. They are not so much concerned with the causal relationships between an effect, and how that effect was achieved, as they are in just knowing what actually happens. To many players, the key to enjoyment is what they do, not how they do it. For this style of play, it''s not necessary to achieve any great level of realism or accuracy.

If anything, I''d say that there is a greater trend to abstract many details rather than simulate them. Look at the dearth of Simulation games (either flight sims, tank sims, or realistic wargames). Look at how few realistic shooters there are versus the "run around and blow stuff up" shooters.

If there is an appeal to simulation, it is in understanding the details. As the old saying goes, "God is in the details". I believe that only looking at the effect, and not understanding how such an effect was achieved is somewhat dangerous. It could lead to a false sense of achievement, by equating one''s success to skill, when really it may have been nothing more than dumb luck. Because simulation style games tend to be more consistent by having more details to worry about, it allows astute players to figure out which variables they have to control to maximize the chances of their success.

Once you have figured out what kind of experience you want to offer your players, you then have to extract the essence of the gameplay in order to model it. This is when you should decide whether certain things can be abstracted or not. Sometimes abstraction is also necessary simply because in the real world, we don''t understand the variables involved in the process. In which case sometimes we can only use stochastic modelling to try to determine the probability of certain outcomes.

In my opinion, abstraction should be used when causality is not important to the desired experience you want to get across to the player. For example, in my games, I like knowing that there''s a difference between an M16A2 and an AK74. If both weapons are treated the same (by abstracting the details), I feel as if I''m no longer in a real environment. On the other hand, I don''t mind how every player moves at exactly the speed, even though in real life, many people are faster than others. To me, it is not so much a non-issue, but one that isn''t as important in evoking the flavor that you want out of your game.

Share this post


Link to post
Share on other sites
quote:
Original post by Dauntless
The trick is in figuring out what you want to model. But in order to figure out what you want to model, you must first decide what kind of experiences you want to give to the player.


Yes!

quote:
I personally tend to prefer systems that are more realistic because generally, realistic systems are more consistent. They require less denial of plausibility and as a consequence can feel more real and immersive for me as a player. This is why I prefer games like Ghost Recon and Rainbow Shield over games like Unreal Tournament or even Medal of Honor (i mean come on, there''s a bunch of first aid packs all over the place that heal you?).


Realism is definitely a good thing to strive for - but what facets of reality will be brought into the game? I believe a game that chooses the first-person 3d rendered world as an abstraction of reality is very limited by that very decision.

Let''s assume I want to make a role-playing game with an emphasis on intrigue, politics, with very smart and aware NPCs (all of these being realistic traits). If I start the design by creating a 2d map, items, classes and the rest of the usual RPG stuff, the very things I wanted in the first place will be pushed aside and downplayed. A visual/physical realistic abstraction does not support politics - but would it be possible to throw away all that stuff and find a higher abstraction where only the politics exist in the game world?

Share this post


Link to post
Share on other sites
I consider abstraction and simulation as two points on the same spectrum, the choice being where on the spectrum each game system should be on the spectrum rather than being one or the other.

For example, look at how games with projectile weapons calculate where they hit etc. Games tend to use what I call an "attack roll" (abstract, like D&D dice rolls, could also be called a model) or "entity simulation" based system for this.

The attack roll system will put values (range, target size, random numbers, projectile data, etc) into a formula and the hit result is the answer to the formula.

The entity simulation creates a projectile entity with position, velocity and so on and lets the simulation step deal with it, in this case updating its position and checking for collisions. If/when a collision occurs interaction code is called to determine the result based on the entities involved.

The point is that the attack roll system is not completely abstract and the entity system is not completely simulated. The attack roll system handles individual projectiles, but could be abstracted to calculate the entire combat outcome instead. The entity system doesn''t simulate the rifling effect on the bullet; instead it abstracts that into an accuracy statistic.

Personally, I prefer games to be nearer the simulation end of the spectrum because of the emergent gameplay aspect and as Dauntless said, they tend to be more consistent and have believable scenarios.

One quick example is cover giving protection from projectiles. In an attack roll system cover rules have to be explicitly defined, including any cover size and placement modifiers. In an entity simulation it is implicit - the unit gets the benefit of cover because the projectile hits the cover instead of the unit, and the size and placement of the cover is automatically taken into account.

quote:
Original post by Diodor
Firstly, complexity is not desirable. The best games have few rules and deep gameplay.


How is complexity not desirable? Chess would be less complex if it only had 2 pieces per side but would that increase or decrease its desirability as a game? Too much complexity is undesirable as well. I would define "too much" as the being if the player cannot cope with it (chess with 100 pieces per side?). It’s not a question of "complexity yes/no" but "complexity, how much for the target audience?"

Civilisation is generally considered one of the best games of all time but I doubt it could be classed as having "few rules".

quote:
Original post by Diodor
Realism is definitely a good thing to strive for - but what facets of reality will be brought into the game? I believe a game that chooses the first-person 3d rendered world as an abstraction of reality is very limited by that very decision.

Let''s assume I want to make a role-playing game with an emphasis on intrigue, politics, with very smart and aware NPCs (all of these being realistic traits). If I start the design by creating a 2d map, items, classes and the rest of the usual RPG stuff, the very things I wanted in the first place will be pushed aside and downplayed. A visual/physical realistic abstraction does not support politics - but would it be possible to throw away all that stuff and find a higher abstraction where only the politics exist in the game world?

I don''t think it matters if you define and display the physical game world in 2D or 3D, the physical game world doesn''t include the game''s politics system regardless of its abstraction/simulation level. Now if your game world is abstracted to the level of countries and federations than city politics isn''t going to be represented, but that''s not related to how the game is rendered to the player.

In your role-playing example I''d say the dialog system will have a far greater effect on the scope of the politics system then the map, items and classes, but if intrigue and politics is where you want the game''s focus to be, surely those systems should be designed first and the others made to fit around them?

Fulby

Share this post


Link to post
Share on other sites
quote:
Original post by Fulby
For example, look at how games with projectile weapons calculate where they hit etc. Games tend to use what I call an "attack roll" (abstract, like D&D dice rolls, could also be called a model) or "entity simulation" based system for this.


Games with projectile weapons rely on the "kill''em''all" abstraction - the game world is all about fighting. This happens to be an entertaining abstraction, but is it the only one?

quote:

How is complexity not desirable? Chess would be less complex if it only had 2 pieces per side but would that increase or decrease its desirability as a game? Too much complexity is undesirable as well. I would define "too much" as the being if the player cannot cope with it (chess with 100 pieces per side?). It’s not a question of "complexity yes/no" but "complexity, how much for the target audience?"

Civilisation is generally considered one of the best games of all time but I doubt it could be classed as having "few rules".


Well, I should have said not all kinds of complexity are good. Chess has fewer units and simpler rules than Total Annihilation yet it is the deeper game.

You''re right about Civ, and it''s a similar example with the popular Dungeon & Dragons games. That being said, both D&D and Civ require vast amounts of sheer work to be enjoyed, which is OK for a rather small segment of the market (incidentally, the same segment all the big publishers are fighting for). It can also be argued that the D&D and Civ worlds are so successful because of their use of some clever abstractions. These games reduce to comparatively simple and understandable rules things as hugely complicated as the entire human history. Civ is that good because of how much it simplifies.

quote:

I don''t think it matters if you define and display the physical game world in 2D or 3D, the physical game world doesn''t include the game''s politics system regardless of its abstraction/simulation level. Now if your game world is abstracted to the level of countries and federations than city politics isn''t going to be represented, but that''s not related to how the game is rendered to the player.

In your role-playing example I''d say the dialog system will have a far greater effect on the scope of the politics system then the map, items and classes, but if intrigue and politics is where you want the game''s focus to be, surely those systems should be designed first and the others made to fit around them?


I''m thinking if I want a political RPG, really there is no need to even go to the physical level of simulation. What''s the logic in starting a political game with coding say path-finding algorithms.

Abstract away the physical simulation and focus on what the game is about.

Share this post


Link to post
Share on other sites
quote:
Original post by Diodor
Games with projectile weapons rely on the "kill'em'all" abstraction - the game world is all about fighting. This happens to be an entertaining abstraction, but is it the only one?

Fallout, Ultima Underworld 2, Thief - games with projectile weapons where the game world isn't all about fighting. The projectile system itself doesn't even have to be about fighting; Thief has projectiles which are used in ways other than fighting and other games often do this with puzzle solving scenarios.

I wouldn't call "kill'em'all" gameplay an abstraction in the same sense of abstraction vs. simulation. I'd call it a genre or a style of game, whereas abstraction as mentioned in this thread (at least in my posts) means removing or simplifying elements of a system. This fits in with abstraction as the opposite of simulation and the thread title. Sorry for going on about it but I don't see how a game world which is all about fighting is an abstraction in the sense the thread title is using the word.

quote:
Well, I should have said not all kinds of complexity are good. Chess has fewer units and simpler rules than Total Annihilation yet it is the deeper game.

That depends entirely on how you define "deeper", and I don't know how you are defining it. I would think TA has a larger state space than chess, though both are so far beyond human comprehension it doesn't matter and I'm not sure that's a good measurement of depth anyway. The unit dynamics - how different unit types interact with each other and the game world - may be a good indicator but exactly what to measure is not a simple question.

quote:
You're right about Civ, and it's a similar example with the popular Dungeon & Dragons games. That being said, both D&D and Civ require vast amounts of sheer work to be enjoyed, which is OK for a rather small segment of the market (incidentally, the same segment all the big publishers are fighting for). It can also be argued that the D&D and Civ worlds are so successful because of their use of some clever abstractions. These games reduce to comparatively simple and understandable rules things as hugely complicated as the entire human history. Civ is that good because of how much it simplifies.

I completely agree and would class most or all of Civ's game systems as much closer to the abstract end of the spectrum than the simulation end.

quote:
I'm thinking if I want a political RPG, really there is no need to even go to the physical level of simulation. What's the logic in starting a political game with coding say path-finding algorithms.

In a country level political game, inter-government politics would depend on the position of countries on the globe. Hostile and defensive actions, trade routes, commodities, etc. depend on the location and connectivity of countries. If a country embargos another but goods can travel through a third, then the physical level (and path-finding ) has a direct influence on the political situation. The game may not simulate trucks moving along roads, instead it would abstract that to an average travel time between neighbouring countries. The same applies for politics between the human and orc kingdoms, or a farmer letting people use a stream on his farm for water because it gives him leverage on other village issues.

Fulby


[edited by - Fulby on May 14, 2004 3:28:04 PM]

Share this post


Link to post
Share on other sites
I raise an eyebrow anytime says that one thing or another is the way to design computer games.. whether they''re complaining that games are too deep or not deep enough, or whatever. The fact of the matter is, there''s room for just about any kind of game imaginable. There is a place for simple puzzle games and a place for advanced shooters. There is a place for deep and involving storylines, and a place for no storyline at all. There is a place for complex simulations and a place for one-click classics. There is a place for games that teach, a place for games that inspire, a place for games that cause controversy, that confuse, delight, calm, anger, or terrify, and a place for games that simply pass the time. Why must people continually demand that such-and-such is the right or wrong way to build games?

Share this post


Link to post
Share on other sites
quote:
Original post by Fulby
Fallout, Ultima Underworld 2, Thief - games with projectile weapons where the game world isn''t all about fighting. The projectile system itself doesn''t even have to be about fighting; Thief has projectiles which are used in ways other than fighting and other games often do this with puzzle solving scenarios.

I wouldn''t call "kill''em''all" gameplay an abstraction in the same sense of abstraction vs. simulation. I''d call it a genre or a style of game, whereas abstraction as mentioned in this thread (at least in my posts) means removing or simplifying elements of a system. This fits in with abstraction as the opposite of simulation and the thread title. Sorry for going on about it but I don''t see how a game world which is all about fighting is an abstraction in the sense the thread title is using the word.


Simulating a physical world reduces what the player can express in that world to physical actions - just because the player must move, fight, take, drop, climb, jump, aim, look, search, dodge, avoid in order to do anything, the game must be about those physical things. Anything else is condemned to be merely a distraction.

If Fallout and Ultima are anything like ADOM, they''re about fighting - pick 10 random moments in the duration of a game - what is the player doing in 9 of those moments? Even in Thief, I bet 9 players out of 10 black-jack their way through the game.

quote:

That depends entirely on how you define "deeper", and I don''t know how you are defining it. I would think TA has a larger state space than chess, though both are so far beyond human comprehension it doesn''t matter and I''m not sure that''s a good measurement of depth anyway. The unit dynamics - how different unit types interact with each other and the game world - may be a good indicator but exactly what to measure is not a simple question.


At least there are less books written on the strategy of Total Annihilation than on that of chess.

There is deep gameplay - each decisions requires (or even better - allows) a lot of thinking and there are complex rules - simply learning the game is a lot of work. IMHO, the former is desirable and the latter is not (though the two sometimes go hand in hand and sometimes, as in the D&D example, learning the many rules is a fun endeavor in itself).


quote:
In a country level political game, inter-government politics would depend on the position of countries on the globe. Hostile and defensive actions, trade routes, commodities, etc. depend on the location and connectivity of countries. If a country embargos another but goods can travel through a third, then the physical level (and path-finding ) has a direct influence on the political situation. The game may not simulate trucks moving along roads, instead it would abstract that to an average travel time between neighbouring countries. The same applies for politics between the human and orc kingdoms, or a farmer letting people use a stream on his farm for water because it gives him leverage on other village issues.


How about a game that is not as much about what resources each of the players control (be those countries or RPG stats and items) but instead about what each of the NPCs think of each other and the player, what each tells each other, about the relations between them and so on and so forth.

Share this post


Link to post
Share on other sites

  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

We are the game development community.

Whether you are an indie, hobbyist, AAA developer, or just trying to learn, GameDev.net is the place for you to learn, share, and connect with the games industry. Learn more About Us or sign up!

Sign me up!