Definitions: What’s a Game State?

After reading some of your articles on depth, I have to ask. What exactly counts as a “state” in a game? Depth is characterized by the number of non-redundant states, but I’m not clear what counts as a state.

Have you ever used an emulator? You know how there are save states in emulators? Save states capture the current value of every single variable in the game. This is basically what I mean when I say a state. A state is the current value of every single object, element, or variable in a game. Any time something changes, even slightly, in a game, that is a new state, strictly speaking.

For an easier to imagine example, in a board game, a state would be every possible combination of pieces on every space on the board.

In most video games, it’s harder to think about every possible state because many games have things like velocity, or extremely minute and precise position tracking. Things appear to be more analog thanks to the use of floats and sub-pixels.

This is why I specify that depth does not count redundant states, because this level of minute detail goes far beyond what is significant in the majority of cases. Many of these tiny states can be said to be negligibly different from other states around them. Determining what states are non-redundant is unfortunately still a matter of interpretation. I count analog spectrums of information tracking as being more deep than a digital spectrum, but only slightly. I believe there are diminishing returns for additional subdivision of states. You only gain so much for going from an integer to a float in my book (look up data types if you don’t know what a float is).

This will probably help.

What’s Deep as a Fighting Game?

I’ve seen you claim that fighting games are probably the pinnacle of depth. How much of that would you say is related to them being multiplayer–depth being squished out by players themselves? what about strategy games or good competitive shooters? Any SP game you consider to be as deep as them?

Dante from DMC4 is more deep than any individual fighting game character ever. Maybe not more deep than an entire fighting game, but more than any single character in one. The number of options and ways he can combine them are absolutely tremendous. DMC3 dante with the style swapper might be even more deep than that (considering he has more styles, more moves in individual styles and more weapons, he’s almost inarguably deeper), though, shit, something about the way he’s animated manages to make his style swap combos look less impressive. Maybe the lack of color flashes and hud notifications.

Some speedgames are ludicrously deep, like Mirror’s Edge, Half Life 1 and 2 (amazingly in completely different ways), Mario 64/sunshine, Dark Souls 1, ocarina of time (I say reluctantly), Ori and the Blind Forest, Ratchet and Clank, Super Monkey Ball, Metroid Prime, at least one castlevania game, at least one sonic game, F-Zero GX. Though I’d be hesitant to put them on the level of a fighting game.

RTS games absolutely go toe to toe with fighting games, they might be even more deep than fighting games, I don’t really want to make a call there.

Go is up there with fighting games, no doubt, probably chess too (though you could argue that the development of the meta has shifted the relevant field of depth into a smaller range).

Quake 3 and UT 2004 are close, but I think fall a bit below fighting games, less options, less complex neutral.

But I mean, there’s a reason I play fighting games, and it’s because they’re the best games around. They’re games that I can put more research into than any other and have more to verifiably show for it. They’re games that I can confidently say I don’t understand completely and have a lot left to learn, where Dark Souls is one I think I do understand completely.

And is being multiplayer part of why fighting games are so deep?

Being Multiplayer is a part of that, a lot more fine interactions are tested in a multiplayer environment that can’t be tested in a singleplayer environment. Player psychology is a very real thing that can be played with and experimented with to produce favorable results. In a singleplayer game I don’t have to consider my opponents adapting to me, I don’t have to consider common habits that some ice climbers players have and others don’t. I got 2nd place at a tournament last monday, facing an ice climbers player twice, probably the third best ICs in the state. The first time I beat him 3-2, the second time I beat him 3-0. I realized between those games that if I stayed on platforms, he’d eventually get the climbers out of sync, as his offense was too weak to overwhelm me and that was my chance to attack, separating them, and that he loved to shield to bait me into a shieldgrab, but I could run up to him and grab him, usually getting a free followup and there wasn’t a lot he could do to stop me, and he definitely couldn’t wobble me. So the game changed from me poking him a lot and getting wobbled, to preying on split up climbers and fthrow fsmashes. In a netplay match I lost in marth dittos the first round, then won the next two by counterpicking to snake, pure intuition that the player was unfamiliar with the character despite it being an amazing matchup for marth. In street fighter I recently realized that I need to just walk into their range and hit them with mediums sometimes, I need to start comboing into spiral arrow off my light jabs, because I usually get a few of those for free off my pressure setups, but the grab afterwards is less free, and they don’t do much damage by themselves. In multiplayer games, it’s a constant cycle of improvement and thinking about common player tendencies.

This is why icyclam’s claim that bots are better because they’re harder is such bullshit, because human tendencies create depth in the form of bringing game states into relevance that normally wouldn’t be, and computers simply cannot recreate that.

Local top PM players Kysce and Flipp went to Shots Fired 2 recently, teaming in doubles, and Flipp’s Snake plants C4 all the time, leading Kysce to say, “Got ’em!” whenever he does. However Kysce knows Snake well enough to know that sometimes he doesn’t have the stick, but it looks like he does, so he says “Got ’em!” even when the stick isn’t on them, and this occasionally confused the other guy, even enough to make them kill themselves or irrationally shield or airdodge when they weren’t really in danger, opening them up to attacks. I played a friend in third strike, and jumped when I hear him do the fireball motion. AI doesn’t have that sort of internal model of self or other, and Desk’s video on SFV survival shows it. Survival mode itself shows it.

Depth and Meaningful Complexity

What is the difference, if any, between gameplay depth and meaningful complexity?

I use them as synonyms personally.

Possibility space is essentially raw complexity, it’s all the possible states and outcomes, even the pointless and redundant ones.

Depth is what you get when you cull the possibility space for only the relevant and non-redundant states. It’s the complexity that’s left over.

In layman’s terms, we know that games are better as they get more complex, but we also know that some games are complex on a surface level, but end up being simple in execution. Like, you have to sort through a ton of variables, but the actual sorting algorithm is simple, despite being time consuming.

Games need to become more complex overall to have more meaningful complexity, however the way in which they’re constructed can lead to a more complex game that isn’t more meaningfully complex.

The enemy of depth is optimization and redundancy.

How a Simplified Input Game can be Interesting Too

I don’t get the hype behind Rising Thunder. Simplifying inputs? And you like Divekick, yet you’ve said that complex inputs are more rewarding (when I asked about PM’s input leniency compared to Melee, or when discussing wavedashing in Melee).

Divekick explores a unique strategic space. It’s fast, and there are things to learn about the game. I played one friend in it and seriously beat him every game for like 20-30 games in a row. I perfected him multiple times during that. One of my friends actually figured out a new way to advance safely, by jumping, then kicking near the end of the jump, which I would normally do at the beginning of a very shallow jump.

There are actually some advanced techniques in divekick, like performing special moves requires pressing both dive and kick at the same time, but there’s a short leniency period, so you can kara-cancel into a special move. For example, Mr N can kara-cancel his kick into his hover, allowing him to effectively kick for a frame before hovering, kicking with less commitment. And of course he can keep doing this as long as he has enough meter.

Rising thunder has unique character designs, like Crow, who is so unique he couldn’t exist in another fighting game because of his invisibility power, or Vlad who has a weird air dash with a meter you can expend as you like, or Dauntless, who has some really amazing combos and unique special moves, along with one of the rare normal anti-airs. I like that with Chel I can cancel sweeps into fireballs (which I almost never get to do in SF, it’s only possible in SF2), and the combos are reasonably interesting. It’s cool to be able to see someone else do a combo I’ve never seen before, then start doing it on them. The kinetic advance system is also cool, it’s like FADC, except you can also jump out of it. Not to mention that combos do get rather execution-heavy at a high level. There are even link combos, usually from M into L.

In its own way, it’s interesting that all the special moves are on single buttons because it very much changes the amount of time you can execute moves in. I remarked on picking up Chel that it was like every character with an anti-air special was a charge character. I mean this in that you can instantly react to jump-ins with just a button tap. I was so trigger happy at first that I even reacted too soon in some cases to jump-ins, whiffing completely, because I expected my fingers to be slower. And the cool-down periods, much as I dislike the use of cool-downs as a balancing measure, do actually add a strategic element to the game, so if your opponent whiffs an anti-air, you know jumping in is safe for the next few seconds. Chel’s projectile has the cooldown negated if it hits the opponent too, meaning that you can keep up fireball pressure as long as the fireballs are hit or blocked, clearly pointing to neutral jump as an answer to Chel’s fireballs.

The other thing is, and I admit this isn’t speaking to the game’s favor, but it’s a proof of concept that even if you simplify the inputs down to the minimum possible level, scrubs won’t magically get good at the game. This is a moral victory for me.
Having inputs that are hard isn’t something that’s strictly speaking a good thing. I don’t think any fighting game needs a pretzel input ever again. I think the move away from FRCs for GG Xrd was a good thing and made the system more interesting, even if there were some OSes that worked in 1.0.

I think that the difficulty of an input is something that should correspond to how helpful the result is. It’s not something that can easily be judged. The difficulty of a given input should be relative to how rarely it needs to happen, so you get easy inputs most of the time, hard sometimes, impossible rarely.

The bigger compromise here is the depth of Rising Thunder in part because of the input system they chose. There are less options, less ways to modulate options, and thus the game is more strategically flat. Having movement commands act as a modifier on top of normal button presses allowed for a larger range of moves to be accessible at once.

Also seriously, wavedashing isn’t hard. You can learn to do it in 30 minutes or less.

Is Depth not Enough if it’s not Stressed?

Would you say that a deep system in a game is not enough, the game must appropriately push it? thinking about this while playing Bulletstorm, which has great weapon variety and ways to defeat enemies, but encounters don’t really demand it, and you get points for doing combinations to compensate and in DMC series, often you can just get used to the attack patterns of enemies to avoid them, and spam a basic attack, and you can get through even Dante Must Die with that, and the game basically becomes God of War, despite having far greater possibilities. Whereas Quake for example ammo depletion forces you to change weapons, and enemy positioning forces you to use movement options and level geometry more fully, so you necessarily get a more complete exploration of the system’s possibilities through the game itself, and those possibilities are actively being tested.

To an extent yes. I have two measures of depth, absolute versus relative. Absolute is the amount of differentiated game states in the game as a whole. Relative is the number relevant to the playerbase. There might be a huge depth inherent in a game, but the playerbase will only access a certain portion of it. To that extent, the game developers must be cognizant of how the game will actually be played. Creating depth relative to the playerbase is a matter of balance. If you have one weapon for which ammo is abundant which does the highest DPS to enemies in all possible situations, then the amount of depth relative to the playerbase will shrink in relation to the absolute depth of the game. If you have enemies which can be dispatched easily regardless of weapon choice, then the game loses the depth of strategic weapon selection as the situation demands it. If you have a ton of weapons that are each only good at killing one enemy, options that only do one thing and don’t even slightly overlap, then you lose all the scenarios when you use a normally suboptimal weapon for an optimal situational purpose (and a lot of alternate solutions, the game becomes more puzzle-like). If you have a ton of options that all do the same thing, then they lose differentiation, which also is a lack of depth.

Designing for depth is about making it so as many different components of the game are relevant in as many different situations as possible. That’s why I have my 3 criteria shorthand, a given option should have it’s own niche (different role from other options), it should have multiple uses, and create different outcomes based on manipulation of the option or circumstances (like allowing jumps of different heights, getting more distance from different timings, more damage for better timing/positioning, inheriting properties from other variables). It occurs to me that I should add a 4th criteria, for whether an option has an interaction or synergy with another mechanic.

From a design perspective, I feel it’s stronger when a ruleset is enforced by winning or losing, by barring progression or recognition of victory until you actually succeed. It’s one thing to have scoring or time as a supplementary objective, but for most people, that feels really weak. Most people don’t care, I usually don’t care. Difficulty can act as a selective force, pushing players to try different things out to optimize their play, which is what brings out the depth of the game.

Yes, I’ve considered how you can just get by spamming a basic attack before and discussed it with a few different friends. I didn’t like how multiplayer games push us to master everything about the game, but single player games seemingly lack this capacity to always push us higher. How do you even grade high level play in DMC? It’s a matter of style and creativity? How could you make a pass/fail system for that type of thing? The truth is, you can always win literally any beat em up game by spamming the weakest attack and not getting hit. This isn’t true of another human opponent, because they adapt. You can’t spam the same thing forever because they’ll just start doing the thing that beats your spam every time. Though theoretically, you could win a match against a human opponent by spamming if you just always did it at exactly the right time.

If you push for scoring systems, then more often than not, solutions become incredibly rigid and repetitive, rather than vaguely allowing anything that manages to get through.

Yeah, you could probably beat DMC3/4 by spamming nothing but stinger, but I don’t imagine it would be particularly easy to beat it that way. Ninja Gaiden can be beaten with flying swallow, but it’s slow and harder than using other moves.

NG is arguably a good case study here, you shouldn’t always use flying swallow because it doesn’t have the best damage output (though it can insta-kill many smaller enemies), and it can be blocked and punished. You shouldn’t always run directly at opponents because many have ranged attacks that will punish a straightforward approach. You shouldn’t always block and counterattack because enemies have guard breakers or throws, which require you to either reverse wind away or dodge outright. You can’t always spam powerful combos because other enemies are around to hit you, and some will block and punish you (though this is only really true of enemies with super armor in NG1, I noticed the chapter 2 boss of NG2 will do this, requiring hitconfirming). Even without an adaptive AI, if you throw in moves like these which incentivize using some moves for some situations, it influences the decision-making process of the player. Oh, and you should always use the ultimate technique, it just may be tricky to find an opportunity.

Another good case study is DMC4’s bloody palace, which has not only waves of enemies, but the timer which you need to defeat enemies quickly and efficiently in order to earn more time for (with only the loose bonuses for enemy deaths, and no damage). Where in story mode you can win with anything, bloody palace is far more selective, pressuring players to not only deal damage without taking it, but to continually do it the most efficiently they can. Though if you play well enough, you can earn a ton of time and not care how much you waste. Might have paid to set a cap on the maximum amount of time you can earn, to like 15-20 minutes or something maybe.

Regrettably even Quake doesn’t require everything from you. I’m pretty sure it can be beaten with nothing but an axe. (Don’t quote me on that, a vore + shambler like in the final map might be trouble). Still yeah, Quake does a lot to make sure all the components in the system are used through ammo limitations (otherwise lightning gun or rocket launcher would wipe everything out).

Still it’s a pretty good example. Custom Gamer goes over rather frequently how even though one enemy, like a Shambler, might be easily beaten with just an axe (which he does) by itself, if you have two enemies, they can fill in for each other and prevent you from just repeating the thing that kills one enemy dead easily. If you throw in a fiend, then you can’t move into melee range for the shambler, then out again as he attacks, the fiend will hit you. You need to dodge both the fiend and the shambler at once, and the optimal pattern for one doesn’t overlap with the other.

But yeah, difficulty, in particular challenges that stress the different options the player has, requiring them to prioritize some over others, making it difficult to pick the one necessary to survive on a moment-by-moment basis, and giving each a chance to shine, is a big deal for depth.

What is(n’t) Emergence? In the comments, the author asks: what *isn’t* emergence? Some people have contested his definition (or rather, his understanding of how some people define emergence).

Seth Hearthstone addressed this matter as it applies to games fairly well before me:

I’m going to limit my explanation here to games, I believe emergence is a phenomena that exists. I believe it’s a phenomena that is critical to understanding depth. The quintessential engineering goal of games is to produce a large number of distinct outcomes, while keeping the apparent structural complexity of the game low.

For example, I’d cite visual novels as having a low factor of emergence, because the number of outcomes relative to the number of choices is usually 1:1 (though some of course carry over variables, and modify parts of the script based on circumstance). Point and Click adventure games similarly have a low factor, because the majority of interactions between objects are invalid, relying on you to find the valid ones.

To give a simple example, you have rock paper scissors, which has 3 essential elements, that can produce 6 possible outcomes. Interaction of elements to produce a large number of differentiated outcomes is a thing that can reliably be planned for and evaluated. A term used in a similar context is Possibility Space, referring to the range of possibilities created by a system.

The product of a sufficiently complex system is that among the outcomes that emerge from it will be outcomes you could not necessarily foresee. If a system is simple enough that the designer can foresee each individual outcome (at least those significantly differentiated from one another), then as a rule of thumb it is likely to not be a system that will hold player interest for long.

I would personally avoid conflating emergence with unpredictability. I think trying to judge a facet of an object for intention or foresight on the part of the designer is pointless, and emergence is a thing that can be engineered for by creating mechanics that affect variables inherited by or shared with other mechanics.

I actually discussed how this article from Less Wrong made me feel uneasy with a friend fairly recently. It’s primary purpose is to prevent people from using a buzzword as a substitute for a specific explanation, which is sensible. However at least in the field of video games, there is a clear delineation between things that are apparent results of the combination of mechanics (like wavedashing) and things that are arbitrarily implemented (like the properties of a Forward Smash attack). There isn’t an underlying mechanism that determines those according to a more fundamental set of laws (like the laws of muscle composition, organic chemistry, chemistry, newtonian physics, quantum physics as these all apply for a person throwing a right hook), they’re determined arbitrarily by the designer/animator and have none of those underlying rules operating behind them.

A Basic Introduction to Depth

I end up writing a lot about depth. It’s the only universal constant for games far as I can tell. Everything always comes back to depth, so here’s one such writing on what depth is.

Depth is in some ways the raison d’etre of games. Many people have varying definitions of it, but my personal one is the number of non-redundant game states possible within a game. Non-redundant being used as my catch-all to eliminate game states that are essentially the same for all practical purposes.

A game in which jumps are only possible at specific contextual points and always cross the gap perfectly without allowing player input in the process has less depth than a game which allows a player to jump in a perfectly fixed arc when they press the button, which has less depth than a game that allows the player to control the character while that character is in the air, which has less depth than a game which allows you to also vary the jump height by the length of time the button is held, which has less depth than also inheriting your ground velocity in the transition to the air, which has less depth than allowing another jump in midair.

To resimplify the example, imagine you have fixed jump arcs and you can press jump again to jump in midair at a specific point in the jump, this creates 3 possibilities, jump in the normal arc, jump in the normal arc with another jump at the point specified, or don’t jump at all. Lets allow you to press jump at any arbitrary point in the first jump, this creates additional possibilities based on the number of points that actually exist in that jump arc. Allowing the player to modulate the height of each jump by the amount of time they hold the button down creates a number of possibilities relative to the amount of points at which holding or not holding the button down creates divergent game states, and this is then multiplied by the number of points at which they are allowed to double jump. In an exponential fashion, each of these factors creates between them an even larger number of potential game states.


The tricky part is with obsoletion or dominant game states/strategies. Some techniques, such as correcting for bullet spray in counter strike, obsolete some game states. If you know how to correct for bullet spray, you always will. Is this depth? Are those obsoleted states redundant? I would personally answer no, they are not redundant, and yes, this technique is something that contributes to depth, but in relation to it is dominant play styles. In some games, particular options or play styles obsolete others, such as Yun and Chun being the best characters in SF3, and completely destroying the rest of the cast so badly that in the current metagame many characters simply aren’t played. A similar situation exists in smash bros, where lower tier characters get snubbed by balance issues. In some games there are simply strategies so good that everyone does them and it prevents a much wider range of strategies from playing out, such as planking in Brawl, which eliminated a ton of the on-stage game. Do these things result in a loss in depth? Compare to a game where the characters were balanced and all of them could be seen playing at a high tournament level? I’d say yes, the dominance in these play styles results in a loss in depth for the game, though it is hard to say where exactly those extra game states that are lost to this dominance go. I have no term for precisely describing it and my definition of depth is thereby insufficient.

Super Monkey Ball is a simple game to understand and to play. It has relatively few complexities in its operational methods, one of which being that special diagonal movement can result in faster acceleration, yet the particular physics system and composition of the levels creates a large number of ways the goal can be attained, representing a type of depth.

On the other hand, some games have elements that I completely feel are redundant, not just accomplishing the same goals, but doing them by methods that are close to identical. Deus Ex featured lock picking and multi-tool use, but both of these functioned identically, allowing the player to wave a lockpick or multi-tool at a given node, having a bar move and after waiting a bit, the object would activate, opening a lock or otherwise. Hacking was similar, pressing a button then waiting for a bar to finish. The sequel by comparison had all locks be little minigames with a variety of scenarios possible within that mini-game. They lost some elements of level progression related to resource management, but also eliminated looking at bars move across the screen while doing nothing or holding a button down, which represents a net gain in depth, as an additive to the depth in the rest of the system. More depth could be gained if these elements were somehow integrated with the rest of the system, becoming a multiplicative depth gain instead of an additive depth gain. I have no proposal for how this is possible in this particular example, it’s just how depth works.

Many turn based RPG games feature a number of abilities that overlap each other with only slight differences, differences in elemental type, effect, attack power, and generally function is rather common, but does little to make the game a more complex thought process as many of the results of each action are just damage and modifications to damage rates. In order to have depth, a game needs more variables in flux than just HP, such as positional values on characters, acceleration, velocity, states, and occupation of those states. A common and easy means to achieve that is through physical information, information related to the spatial positioning and dimensions of the object. Thus turn based RPGs may be complex or have many game states, but few truly have depth (and I’d like to call out SMT Nocturne as being an exception and competitive pokemon, even if the latter is a silly clusterfuck).

Dialogue is also usually a poor representation of depth or challenge, as it lacks physical information, such as spatial positioning, time sensitivity, velocity acceleration, temperature or other possible analog varaibles. By contrast, most dialogue is just a set of choices leading to other sets of choices with variables occasionally remembering prior choices, altering the outcome of future ones. In comparison to something like mario or civilization, this usually represents a low level of depth, and probably always will until better conversational AI is feasible on a commercial level (or at all).

Difficulty is important in relation to depth, as a system may have a lot of depth, but without difficulty it won’t be brought out. There will be no reason to explore it and learn what’s possible. Games are the art of challenge, and depth is a means to a more varied as well as intellectual challenge. However without difficulty, suboptimal strategies may be good enough for many players and hard counters or ideal options for specific situations may become soft or nonexistent ones. Normal pokemon may have a lot of depth, but only competitive pokemon really brings it out by forcing players to develop and learn strategies in order to overcome others who may use those strategies against them. A similar example would be many speedrun games like mirror’s edge, where many of the advanced techniques have incredible depth in of themselves, but without the pressure of going fast, are never necessary or even practical in the normal game. One example from Mirror’s Edge is the kick glitch, which has variance based on the angle relative to the wall, the time running on the wall, the angle facing away from the wall, and is completely impractical for anything but getting through levels more quickly, despite being one of the deepest and most interesting things in the game. Speedruns give many of these techniques a reason to exist, and a pressure to use them competently in order to perform well. Difficulty is related to the prior issue of dominant strategies, and again I have an issue with whether it is redundancy going on here or not, and I think there is a more precise term. It is the dominant strategy problem in reverse, where instead of having a strategy that obsoletes or represses others, many strategies are possible, but pointless.

Here’s my big 4 criteria for whether a game, or a specific mechanic in a game has depth:

1. Does the game feature a variety of options all with their own niche? Alternately does this specific option occupy its own niche?
2. Does this/these option(s) have a variety of applications in the game?
3. Does this/these option(s) allow the player to leverage it to get more or less out of it, or different results that may have value relative to the situation?
4. Can this/these option(s) combine with other options to either create new states entirely, or to generally form a type of synergy?

4 criteria depth.png

Not all options in a game need to have all 4 of these traits. Not all games have this type of structure where this criteria is even applicable. You will have to judge a bit for yourself, but this is a good starting point for analysis or design, meaning they’re important for game reviewers, game critics, and game designers. I write more about these traits here.

Mirror’s Edge’s kick glitch fits the bill perfectly. It can only be used when wallrunning on a wall that goes off a ledge. Some traversals are only possible with the kick glitch, and others are just more efficient with it, thus it has a niche. It can be used to cross gaps, move more quickly, fall further distances, and chain together other wallruns, thus it has a multitude of applications. Players can use it more well or poorly, getting more or less speed out of it, controlling the angle differently, and coming out of the wallrun at different points, thus it has variance in its application relative to the situation and to more effective use. Double jumping or variable height jumping are other perfect examples that I will leave you to deconstruct.

I wrote way more here than I intended to, I really only meant to kick off the topic. Does anyone else have anything to add on this matter? I think having an understanding of what depth is and what elements in a game add to it or harm it is an important thing for absolutely anyone talking about games critically.

Different Types of Depth

depth venn.png

I think a good semantic here is, “Potential Depth” or “Absolute Depth”, referring to the possibilities inherent in the system regardless of who is looking at the system or not, and “Relevant Depth” or “Accessible Depth” to refer to depth as it applies to the playerbase. Both can be referred to more simply as depth, and both are consistent with how most people use the word depth in a broad sense.

Another dichotomy that can be drawn is between depth and breadth. This is another component of how a system can be complex, but not deep. Dynasty Warriors has a lot of different characters, but all of these characters only have a few attacks, so there is a limited range of expression with each individual character. There is a little depth added by having so many characters, the characters are like options unto themselves, but if each one is shallow or occupies an overlapping design space, then the depth gain is small.

Design space is also an important thing to have relative to this. There may be two options, but if both are the same option, they’re redundant, or overlapping as both game states and in design space. Design space is an abstract conception of a game as a collection of elements or niches. Different characters in fighting games typically occupy different design spaces, between rushdown, keepaway, grappler, shoto, or other archetypes. Different weapons in FPS games typically occupy different design spaces, between machine guns, rocket launchers, grenade launchers, snipers, shotguns, or other weapons. In games like Call of Duty or other realistic shooters, there is typically a big overlap in design space. Like in Vanquish, the Assault Rifle, Heavy Machine Gun, and Boost Machine Gun all have an overlapping design space, they’re all rapidfire machine gun type of weapons, and generally are good at the same things. This is a good general weapon type for many encounters, so having an overlap like that can work out rather well, because it means you’re able to carry multiple of a good general use type of weapon, making it just a bit harder to run out of ammo. The downside of overlapping design spaces is that as options are more similar in their function, the differences between the options tends towards creating a clear best between them. This is the Boost Machine Gun in Vanquish, and the SCAR in Crysis for example, or Toad/Peach/Yoshi in Mario Kart 64. Mario Kart 64 – Analyzing and Tiering Overlapping design space can serve a role in a game, but it doesn’t exactly lead to depth necessarily, unless the elements themselves have depth. It is possible to balance overlapping elements, and it is a lot easier usually than balancing elements that exist in disparate design spaces. A shortcut to balancing out a disparate design space is typically to homogenize the elements so they can be easily compared to one another, while also creating overlap and redundancy, reducing the potential depth of the system.

When it comes to design space, there is also the matter of something much like a nested tree, or a russian Matryoshka doll. Design spaces can be nested inside of design spaces. An example could be characters, each character represents an element in a total design space, having their own niche they play to (like peach, jigglypuff, marth, and fox in smash bros), but then within each character there is another design space, where the options they have all have their own niche (like Marth’s Fair, Fsmash, Dair, Dtilt, counter, and dancing blade), then there are submodalities of options which have their own design space, like how the dancing blade can vary the speed of its swings and also select between up, down, and forward attacks. And all of these design spaces and elements thereof nest inside each other, while also defining the niche of the design space above them. So collectively, all the small elements add up to all the larger elements, which each cement the position in design space of those elements.

To cover overlap a bit further, in order to prevent overlap, there must be room in the overall design space for more than one way of defining an elements characteristics. If the only property that a given element has is its damage, and absolutely nothing else, then every turn, you just deal damage and whoever exhausts the other entity’s damage pool first wins. If all the different options just deal different damage, then no option can occupy a non-overlapping design space with any other option. Most games create a varied design space through the inclusion of physical or temporal differences in the options, such as different periods of time over which an option is active, different movement through space, occupation of different physical dimensions. More abstract properties are also frequently assigned to avoid overlap, such as elemental typing, super armor, invincibility, buffs/debuffs, or other factors.

And returning back to depth versus breadth, the essential formula is based on interactions between the elements. If the elements are segregated then the depth gain is additive, if the elements are integrated, then the depth gain is multiplicative. A game may have many elements side by side, such as a large number of characters, or a combat mode next to an exploration mode, next to a race mode, next to a puzzle mode, but unless these each have depth unto themselves, or integrate with each other in some way to create a greater depth than any individually. Ideally you want a few nested design spaces that each contain a number of elements with unique characteristics. The other big deal is how much options are not just separated from one another but irrelevant to one another. In Quake, you have 8 weapons, in Call of Duty Ghosts, you have 52 (counting explosives and side arms). Many of these weapon options overlap each other, there are only 11 weapon categories in Call of Duty, and to further compress the range of expression available, only 3~ weapons can be equipped at once. This decision to only have a certain number of weapons equipped at one time creates two levels of design space, a broad high-level one with all 51 weapons (because these are selected typically before the match and are long term decisions), and a very narrow low-level one with only 3 or so (the weapons you actively can switch between, each with very short term consequences on individual shootouts). Specifics such as what weapon types can be simultaneously equipped can further compress what is possible in the low level design space, as does the huge overlap within weapon types. Meanwhile in Quake, all the weapons are available simultaneously, and there is a very short delay in weapon switching, allowing multi-weapon combos to be very viable. The higher level the choices, the less that the choices can hard counter other ones and remain fair (you don’t want to lose just because you picked the wrong character or weapon loadout), leading to differences between them becoming more arbitrary and more stylistic. Thereby, Quake has a huge Depth, and Call of Duty has a huge Breadth. Dark Souls has a large Depth, where Skyrim has a large Breadth. Breadth does contribute in part to depth, but doing a lot of things in a shallow way cannot compare to doing a few things in a deep way. It is like stacking a linear curve next to an exponential one.

Comparative Depth.png

Easiness vs Depth

What are some games you like that you would consider to be easy/lack a lot of depth in their mechanics?

What’s funny about this question is you assume there is a link between those two things. Easy ≠ lack of depth. In an objective sense, there’s absolutely no relationship between those things. The range of scenarios that can be produced has no relevance to how hard it is to produce a successful outcome.

As relevant to players though: a game that has a large objective depth will only have a large relative depth if it is at the right level of difficulty. Right meaning hard enough that players have to utilize everything available to them and engage with all the depth that is there, but not so hard that only a small selection of solutions are viable.

For example, a game that is about pressing a button as many times as possible within a minute. This game doesn’t have much depth compared to nearly any other game (like chess lets say, or checkers, or getting a good time in a racing game). However it’s inhumanly hard to press that button 1000 times in a minute.

Call of Duty is hard on veteran, that doesn’t mean it’s a particularly interesting game. I had this difficulty in writing the review for Thi4f, the game is legitimately really really hard with all the custom difficulty settings turned on. Some of my level solutions were downright arcane. This isn’t interesting though, because levels consist of repeating the same dumb thing until it works basically. There’s no room for improvisation or significant problem solving, the levels aren’t even nonlinear enough to give you an alternate means to pass through them. I had a similar issue playing through Super Meat Boy, that I did not have playing mario, contra, megaman, or ninja gaiden. I felt in super meat boy that I was just repeating the same small selection of inputs again and again until I got every part right rather than acting in the moment. It felt like there was only really one or two ways to beat the levels, and that was lame.