Thursday, July 19, 2007

The Chemistry of Game Design

It has been a bit quiet in the garden this summer as I've been busy working on a set of longer essays. The first, The Chemistry of Game Design, is up on Gamasutra this morning. You can read it here.

A blurb from the article:
'“…it was clear to the alchemists that "something" was generally being conserved in chemical processes, even in the most dramatic changes of physical state and appearance; that is, that substances contained some "principles" that could be hidden under many outer forms, and revealed by proper manipulation.”

I recently happened across a description of alchemy, that delightful pseudo-science of the last millennium that evolved into modern chemistry. For a moment I thought that the authors were instead describing the current state of the art in game design.

Every time I sit down with a finely crafted title such as Tetris or Super Mario Brothers, I catch hints of a concise and clearly defined structure behind the gameplay. It is my belief that a highly mechanical and predictable heart, built on the foundation of basic human psychology, beats at the core of every single successful game.

What would happen if we codified those systems and turned them into a practical technique for designing games?'

The article describes a psychological player model and a system for visually mapping out how skills are mastered throughout a game. There is even a diagram of what Tetris might look like. :-) This essay introduces the basic concepts. In the future, I'd love to explore how these ideas might be used as part of an iterative design process.

I find systems skill atoms and skill chains incredibly exciting since they have the potential to ween us off our over reliance on reuse of existing genres. By understanding the rules behind why games work, we can synthesize new, highly effective game play from the base elements.

Feel free to post any reactions (allergic or otherwise) in the this blog post. :-)

Happy day,


  1. Congratulations Danc! Great article. I really think this is the closest thing to Game Design 101 for everyone that I've seen yet.

    I didn't see categorizing of burnouts as related to player patience, though. My wife and I are complete opposites: she will play a secret level in Super Mario Sunshine until she has beaten it, whether it is once or, literally 100 times. After about five or six attempts I get bored and frustrated. Perhaps this'd be part of a greater discussion of adaptive difficulty?

  2. Amazing read. How long did that wonderful graph take to draw? :-)

    I find it particularly interesting to look at this in the light of machine learning. In many ways, machine learning has limitations you mentioned for the human brain. It could be interesting to find those missing links in the game design by using AI.

    (A topic that fascinates me :-)


  3. "I am so awesome compared to other people"

    Yup. (:

  4. Danc,

    Good article. Formalizes a lot of concepts that people discuss in an ethereal way.

    I'd love to see someone apply that to World of Warcraft. I imagine the simplicity of WoW would be an eye opener.

  5. Very exciting article. I would like to discuss it further in detail. I have collected a number of remarks/questions and I would be honored to hear your opinion.

  6. Its like evolutionary psychology informed design; a deeper sort of propaganda than the merely sociobiological currents of the 20th century.

    For a more clear sense of what I mean, read this.

  7. Thanks for the article Danc, it was an excellent read. I'm going to try mapping out the skill chain diagram for the game I'm currently working on. I think it will be very interesting and possibly very helpful!

  8. There is a harmless typo near the end of page 7, "It is only by gaining a deeper understanding of the fundamental building blocks of design that game designers with gain the power to break free from the accidental successes of the past." - 'with' should be 'will'.

    I really enjoyed the essay. I guess you'd expect that though, since I mastered a new skill atom. Now I can chain that with my other game analysis skill atoms, etc. How do you keep self-awareness from being depressing? You are essentially trading away the long term mystical beauty of the world for the short term high of atomic skill mastery and leading your readers blindly into it too. Sorry I'm not trying to sound critical.

    Max Dama

  9. Hi. Great article: I very much enjoyed reading it. It liked it that you summarised the best bits of A Theory of Fun and Rules of Play, which in my opinion are the best game design books by a long stretch, and built some sort of framework on top of them.

    @Max, regarding self-awareness: It doesn't have to be depressing if you don't want it to. Master authors will know the structure of literature, great musicians will know music theory, brilliant artists know the rules of art theory and still produce brilliant works. In science too - we know that a glowing red sunset is caused by the refraction of sunlight through the atmosphere, but that doesn't make it any less beautiful. If anything, it makes you change your appreciation of it, making it richer, fascinating, more beautiful.


  10. Random thought -

    Eventually all skill chains have to end, right? In the case of Mario they all end with collecting coins , advancing levels, and earning lives (I would posit some gamers would score themselves on how many lives they had left).

    I wonder if some general description of "game depth" could be scored based on how long the chains were or how many branches each skill atom created or some combination of that. i.e. the jumping skill atom leads to three new atoms that I can come up with: horizontal jumping to clear gaps, platform jumping to access other areas and avoid monsters, black banging to smash blocks or earn coins. That's a deeper game than if it was just jumping pits, wouldn't you say?

  11. Fascinating article. A simple model of game interactions that actually seems to have some merit. I think just for fun, I'm gonna try to produce a partial skill chain graph for Starcraft.

    @bfblog: Yeah, I bet you could measure a game's depth by computing like, the longest chain in the graph or something. It'd be great to construct graphs for good games and bad games, and see how they compare1

  12. Mario mastery? Look at speed runs at or something. Lives remaining aren't particularly representative of top-tier skill in Mario.

    As for the longest chain in the graph being the chain with the most 'depth', that's not necessarily true. Different skills take a different amount of practice and time to master, so the number of them in a chain doesn't necessarily represent much.

  13. Great article. Thank you!

    "Our player model is simple: The player is entity that is driven, consciously or subconsciously, to learn new skills high in perceived value. They gain pleasure from successfully acquiring skills."

    It seems that I can learn more from a textbook in an hour than I can from the average game. It's probably also easier to perceive why the skills from the textbook have value. Why is the game more fun?

  14. Great article once again, Dan. Your perspective on game design is always fascinating. It seems with this article, you've finally concluded that you see games as science and not art.

    I'm not sure if I agree yet, but it's great to see your stance on it presently so clearly. Great job.

    I'm mildly disappointed that you stopped so short in your analogy to chemistry. Why aren't a group of linked atoms called a molecule? It only seems common sense to extend the analogy at least that far.

    For the Mario example, you could say that the "movement" atom, "gravity" atom, "collision with platforms" atom, and "jumping" atom all combine to form a "jumping onto a platform" molecule.

    That seems to match up with chemistry moreso than having bigger atoms that encompass smaller atoms. I know you can only take an analogy so far, but surely it can go at least as far as molecules, dont you think?


  15. Dylan:

    In short, I would say that you get more endomorphins from playing the game than you do from reading the textbook, which results in greater pleasure.

    Working within the context of the system presented here, there are immediate and highly repetitive feedback loops present in a game that allow you to control a character that moves around and interacts with things in a virtual world. In the case of reading a textbook, any small amounts of meaningful feedback you get are all in your head (ex: "Oh yeah, that makes sense. This is related to that.). It's not until you get tested on your knowledge that you can see any external feedback on what you've learned. That's a much, much slower feedback loop than Mario jumping around at 60 fps.

    Not only that, but there are certain types of feedback that we as human beings really love to get. Animation - the simulation of life - is a good example. To many, that's a hell of a lot more compelling than a couple sentences explaining table manners. IMO this is something missing from Dan's article that is really important - namely that certain atoms give us more endomorphins than others.


  16. Dylan:

    Yes according to this model, school would be the most fun place in the world. In fact, hanging around with friends in the schoolyards doing nothing is much more fun then learning all those wonderful skills in class.

  17. Maybe the answer is that you do learn at a higher rate out in the schoolyard (sports, social skills, pop culture, gossip).

    I'm not totally convinced that Halo teaches more in an hour than a Chemistry text, but maybe it does. The skills may only be perceived as useful in the context of the game, but that still fits the model. In the first hour at least there's lots of new things: how various aliens act, how vehicle control works, when to use different weapons, how to protect teammates.

    A problem with the model is the amount of fun (or at least invested time) people have with games that they've seemingly mastered already. Why the hell am I still interested in Diablo?

  18. Well, I guess I was just saying that you get more pleasure from the game, not necessarily that you learn more.

    If you could tie endomorphins to amount of learning, then you could say that you actually learn more from games. I guess I don't know if that's the case or not. Maybe you still learn more from reading that textbook.

    My point was just that you get more pleasure from playing the game, due to how the feedback loop works and the nature of the feedback itself. I'm a visual guy, so I definitely get more pleasure from visual things than from reading.

    Good question, why do you still like Diablo? :P I think you can s till learn stuff from it, or relearn something that you forgot. Plus, depending on your tastes and preferences, I would guess that some skill molecules take a really long time to burn out.

  19. The reason a chemistry text isn't more fun is because it doesn't teach you skills: it teaches you concepts. Concepts are further divorced from the pragmatism of reality than the skills that are forcibly acquired in the pursuit of goals. The appropriate analogy would instead be that you learn more from an hour in a chemistry lab than you do from a game; in such a case, I think you would agree that this is questionable: both have relatively equal educational potential, given appropriate conditions.

    Just like the Civilization games include encyclopedias that give you reference cards to help you play, the chemistry lab gives you a textbook to enlighten you as to the concepts which you can expand into skills.

    The law of constant composition, for example, is a concept. It is not a skill. You do not DO anything with the law of composition (other than fractions, which is a skill): you merely know it. Successfully mixing a reagent, on the other hand, is a skill.

    Chemistry texts are boring because you are utilizing active skills (how to read, how to memorize) in order to acquire concepts that you may feel have a low perceived value, in large part because the skills they form the foundation for also have low perceived value. Someone excited about chemistry, on the other hand, will devour the text with gusto, because the perceived value of the concepts is high, as the skills that build upon them are have high perceived value.

  20. Off of torncanvas' comment, I instantly began thinking of the atom/molecule representation in an archival sense. Diagrammatically representing multiple games in a 'genre' such as platformer, it seems to me that they should have many atoms in common, but have different molecules using those atoms with game-specific atoms. Now picture one of the common atoms (like jumping), with multiple lines radiating from it to each of the various game-specific atoms that are used with it in those games (I'm not going to even attempt a list just now :) ). Is there any doubt that designers would be inspired to combine the game-specific atoms in new ways once they could see how they all relate to an atom in common?

    I must say, I like this idea. I'm about to go read the Notes and comments mentioned by Krystian Majewski, but for now I think it's an impressive way of viewing things.

  21. Excellent article, I thouroughly enjoyed the read and find the presented system to be an excellent step forwards, I'm looking forward to seeing your future expansion of the topic and in particular how such a system might be applied to a process of design rather than simply an analysis of existing gameplay.

  22. Re: Book learning versus Game Learning
    @Michael: That is a wonderful description of the difference between skills and concepts. I've been tempted to interject with a chemistry lab example, but you beat me to it with a very nice description.

    Though I am by no means an expert in classroom learning, you could say that games differ from classic lectures and studying techniques in the following ways:
    - Games focus on experiential learning through self motivated exploration. Traditional study often prescribes rote activities such as copying, memorization, etc.
    - Games encourage the players to put their newly discovered skills to practical use. Mastery is not recorded until the player performs. Traditional study rarely takes this step, except for the occasional artificial test. In the language of skill chains, they rely far to heavily on the very limited and high burnout activities associated with partial mastery.
    - Games use feedback cycles that operate on much tighter time scales with many more opportunities for failure and iteration. Instead of a test every semester, they are giving the player feedback every second or so for many base atoms.

    What is interesting is that many of the dismissive descriptions about traditional techniques are identical to the descriptions of burnout in skill chains. Perhaps what schools need is simply need some revamped, better balanced game mechanics. :-) It would be a fun exercise to use skill chains to critique common teaching techniques in schools in more depth.

    Skill chains vs. molecules
    *grin* It was certainly tempting to continue the metaphor.
    - Classes of actions, rules and feedback techniques become our sub atomic elements.
    - Large scale patterns of player use become our astrophysics.

    Games remain their own beast, at a unique intersection between psychology, economics, literature, cinematography, interaction design and probably one or two other fields that come complete with their own models for mapping out the world. I decided that it was best to provide just the hint of a familiar metaphor to get the conversation moving. From there, the models should adapt to fit the reality of the subject, not the match with the entended metaphor. :-)

    Wonderful comments all around.

    take care,

  23. Danc,
    great article. It makes me think in a way of implementation (framework) in which you can build a videogame specifying the actions represented by an atom and having an underlying subsystem that gathers information about how is the player experimenting the videogame while he/she is playing. That framework can be used in QA phases in order to take snapshots of the player knowledge in defined sequences of time to early identify game design issues/concerns.
    For example, we can end with a conclusion like: Most of players acquire all skills in a 30 minutes session, the burned out abilities are A, B and C, etc.

    It's one step forward to Game Design Quality Assurance...

    Diego (From Argentina)

  24. Yeah, I figured you deliberately held off from using molecules for some reason or another.

    I guess for me, it hurts more than it helps to stop at atoms. I like to think of molecules, too. It gets me into the mindset of combining things more.

    I'd say one of the one or two "other" fields would certainly be art. :)

  25. "I'm not totally convinced that Halo teaches more in an hour than a Chemistry text, but maybe it does."

    I think a more relevant comparison would be spending an hour playing Halo versus spending an hour reading a book about how to play Halo.

    Thank you for the excellent article, by the way.

  26. Danc,

    I couldn't find contact information for you anywhere. Is there a way I can email you?


  27. dclayton,

    You can write to me at danc [at] lost garden [dot] com

    That is exactly where I hope a system like this could be used! :-) Think of it as test driven game design. I wonder who will build such a system that uses skill chains first?

    take care

  28. Dan, let me know if you're up for board gaming sometime. I just moved from Denver to Seattle and am ready to follow up Carcassone with something else.

    Lost your email address, too, so drop me a line at (remove the underscore).

    Peter S.

  29. I can't really see the picture clearly :(

  30. Great article! I really like the analogy you're drawing. I love thigns that inspire me to work on and improve my own designs :)

  31. Great article Danc. A few things I noted, since I just got a chance to reading your article in its entirety:

    Psychological Profiling / Player Model:

    The player described in the player model is limited, in that it describes a specific subset of the set of players: those that are "driven," usually to learn or dominate (and to dominate, they must learn first). Due to the fact that a psychological model is called for, the player model described contains the following player subsets: Timmy Power gamers, Timmy Diversity, Johnny Combo, Johnny Offbeat, Johnny Uber, and all the Spikes, who, evidently, this player model is perfect for. [See Game Design in Wikipedia, specifically Psychological Profiling:]

    So what about the Timmy Social, Timmy Adrenaline, and Johnny Deck profiles? One could argue that they aren't the mainstream of gamers, because they aren't "hardcore" gamers - however, Timmy Adrenaline would disagree vehemently, as witnessed by DOOM3's and F.E.A.R.'s success, and The Sims is proof that Timmy Social is a very large demographic that can't simply be ignored. Johnny Deck's profile, in my honest opinion, represents those players that build or generate some sort of expression of themselves or their experiences in the game, such as a guild or clan, a house or empire, etc. And they're usually rooted into the other player profiles on some level as well.

    The reason I mention this is to ask if maybe there isn't a player model that abstracts out a bit further to include a more involved psychological model. The purpose of your original abstraction was to focus on the core that is the player - but if the player is too loosely or too strictly defined, then it degenerates a critical component of the chemistry of game design. In this situation, the player is too strictly defined.

    On that same note, I believe the Wiki should be updated to include your article on game design and your blog in general, as there is a wealth of game design information available here.

    Pre-Existing Skills:

    Shouldn't the general assumption be, of every game, only that a player cues in to the screen for some sort of visual feedback, and that they know that pressing buttons or moving analog sticks is supposed to do SOMETHING, as opposed to nothing? With this same thought process comes the need for every game to have at least a very simple tutorial, if not more involved for more complex games, which defines and describes the game controller and its general actions, and the capability to learn advanced skills by utilization of mastered basic skills, thus teaching even a non-gamer that mastery and utilization of certain skills leads to game advancement and skill advancement.

    Most games today do provide a tutorial, but not all of them, and often times the "New Features" or "Tutorials" sections are nested into sub-menus that aren't necessarily clear in the first place (Example: Madden '08 by EA Sports). And in some cases, the tutorials are demonstration-based (see earlier: Madden '08, which provides nice video and audio tutorials showing exactly what to do) but don't allow the player to interact there directly to try a skill-set out and see if they can't get the basis for it down outside of actual gameplay instead of possibly trying it out during gameplay, messing up, and costing yourself time to re-load from a game death, etc. As such, these games are less effective in delivering the message to non-gamers or gamers who like directly to "jump into the action" and there's no built-in game-based tutorial. Obviously, advanced gamers may want to skip a built-in tutorial, and so they should be given that capability, just as newer gamers may want to listen/interact with the built-in tutorial and given that capability. (This goes with what you state in the Pre-Mastery of Skill Taught In The Game section regarding feelings of boredom and frustration for advanced players.)

    Skill Atoms & Skill Chains: (Objectives, Future Topics)

    Proposal: an artificial intelligence analysis framework, in comparison with the MDA game analysis framework for game analysis that you mentioned, that utilizes Skill Atoms and Skill Chains as they are progressed throughout the Player's Gameplay Life Cycle (as I like to call the fundamental game play, skill mastery and burnout flow described by your Chemistry of Game Design). If some sort of CGD architecture is built into the game itself and actively monitors a players partial mastery, full mastery and burnout of Skill Atoms, then it can eventually generate data by which an adaptable artificial intelligence can play against the player, allowing for variable levels of interaction and difficulty for the player, constantly providing a challenging stimulus without leading to frustration or boredom. Such information would be tremendously useful in any number of applications within the gameworld itself, and as a preferences experience for future game titles (and has numerable applications in the marketing world as well). In addition, this analysis could clue players into how using certain previously burned-out atoms may advance their skill-set later on during the game, by knowledge of their usage history with the skill atom in question and the necessity of learning the skill atom further down the chain for completion of their current objectives.

    Just some food for thought. I do apologize that it took me as long as it did to read such an excellent article, and I'd love to hear your reply.


  32. "Every time I sit down with a finely crafted title such as Tetris or Super Mario Brothers, I catch hints of a concise and clearly defined structure behind the gameplay. It is my belief that a highly mechanical and predictable heart, built on the foundation of basic human psychology, beats at the core of every single successful game."

    This is the essence of intelligence: the discerning of patterns within and between agents and entities. Einstein found applications for Riemann's work, work which was not intended for Einstein's purpose. Watching highway traffic isn't dissimilar from watching a hydrodynamic system.

    Intelligence, it would seem, has at its heart two things: abstraction; and application. The designers of enduring games have abstracted worldly elements from their agents and contexts, and applied them into novel sets of rules and interactions. Like art, literature, and parenting, game design is at its best when it aspires to recast patterns with new pieces.

    I haven't read it yet (used to be something of an avid Gamasutra reader), but you seem to be pursuing another avenue of abstraction in your article. Reducing what is generally considered, at least by the masses, as vacuous entertainment, to a system, with its own elements and operations. To algebra. I think I'll go read it now (or soon).


  33. Danc,

    I enjoyed the article and I can see the utility of skill chains in an iterative design process. However I disagree with you on perhaps the key points of your article and theory.

    First of all, why should Game Design be a science? Why should science be the goal of game design techniques? I don't think it should at all.

    Second, and this is something that I pointed out in one of your previous articles (, your understanding of psychology and science in general is Conductist and oversimplified. You assume too much and prove too little to have scientific aspirations.

    Players pursue skills with high perceived value over skills with low perceived value

    Humans are not creatures of pure logic. We know people exhibit consistent biases in how they weight their actions.

    Contradiction: you state a universal true on human's logical behaviour (they always seek A over B) and then state that humans are not purely logical and are biased.

    If they feel that their action has been in vain, they feel boredom or frustration.

    Maybe then feel even more excited because the game’s not over and the challenge was tougher than they thought initially. Maybe they don't think that their action has been in vain, even though the graphic says so. How would you note this down?

    The just-so story here is that playful folks that instinctually engaged in long term learning with no immediate benefit were the ones that mastered agriculture, hunting and language. These folks thrived. Those that did not died off.

    Do you personally know these folks? This knowledge may have taken centuries to spread, we can assume that it’s derived from some sort of play…but you are transforming vague assumptions into proven theories, this is not the way if you pursue a scientific model of gaming.

    The existence of a set of skill atoms that are tuned just to entertain us and that never actually lead up to a real world skill is something new to the world

    Virtual games don’t lead to real world skills? I learned English by playing a MMORPG. We can also develop mental structures by playing games. Why is puzzle solving educational? Because it trains brain functions, it gives you the possibility of deeper analysis, of observation and prediction, of trial and error. I seriously disagree with you on this point.

    -Identify where the player experiences pleasure in your game

    I don’t understand how you can identify this.



  34. This is a wonderful article!

    My first stop at the site (directed here from a comment you left at, and I look forward to reading more of your articles. I myself have toyed around with some flash game design, but had never finished a game.
    I think these essays might give me back some motivation.



  35. Great article once again. Your perspective on game design is always fascinating. It seems with this article, you've finally concluded that you see games as science and not art. By the way if you wanna know more gaming design and their working details, browse this site .