Irreparable Complexity, Game and World

(cross-posted to Easily Distracted)

I’m interested in the kind of complexity that arises through emergent processes, in which relatively simple rules governing the action of autonomous agents within a given environment can give rise to permanent structures or changes within the environment which then change the way that the agents express their rules. Unplanned systems, but often highly functional in their own way.

However, there is also complexity by design, in which a system which is consciously intended to have certain restricted purposes or functions becomes more and more elaborate over time, and more and more of its mechanisms become obscure and hidden in their inputs and outputs. I think maybe there are some natural examples of this kind of movement towards baroque complexity. But baroque complexity dies out when it becomes actively dysfunctional within some kind of fitness landscape.

Human systems can achieve this kind of opacity by accident and by intent. Accidental drift towards a system where no one really understands how cause and effect work within the system happens in institutional life all the time. Stakeholders in individual parts or aspects of a system are inclined to expand the influence or size of their mechanism. New forces or powers outside an institution are often accommodated by being incorporated within it. Procedures or heuristics used by an institution in its everyday business sometimes take on a life of their own, especially when they are incorporated into technological infrastructure and automated in some respect. Histories of past practices accumulate and become binding traditions.

Baroque complexity happens by intent when human agents with some degree of authority over an institutional system want to block off direct access or control to some of its inner workings as a safeguard against easy tampering. It also happens when someone with an interest in a particular system believes that secrecy and confusion will instrumentally advance that interest. I think there are quite a few examples of authorities who set out to make it hard for an outsider to understand how a system or process works only to find that in making it hard for outsiders to understand, they’ve made it hard for everyone, that even people in control who thought that secrecy would conceal selectively have found that it conceals indiscriminately.

I’ve found that virtual worlds, massively-multiplayer online games (MMOGs) have provided some great examples of this kind of Rube-Goldberg complexity-by-design, and have also demonstrated why this phenomenon can be a source of so much trouble, that you can end up with systems which are painfully indispensible and permanently dysfunctional, beyond the ability of any agent or interest to repair.

The underlying code of any contemporary large software application is approaching a threshold of complexity where no human agent could ever hope to understand all the possible interactions between the code, the hardware and the user. Even if a programmer can understand why a particular failure or negative event happened, they often cannot hope to understand how to reliably stop it from happening in all possible intersections of code, hardware and user without perturbing some other part of the codebase with unexpected consequences. Pull on one thread, and another may unravel.

This is especially true with virtual worlds, where the size and intricacy of the software is enormous and the practices of users are remarkably diverse and often rivalrous. Developers of a virtual world now start with established code libraries of some kind for managing the visual and interactive components of their product, but they also have to deal with and accomodate histories of user expectation and practice in previous virtual worlds.

Virtual world designers end up with baroque complexity both because their design imperatives drift naturally in that direction and in some cases because they’re trying to veil or protect some of the underlying mechanisms and code of a game from the users. Arguably in some cases, I think they may even be trying to protect themselves from knowing too much about how the world works precisely because they’re trying to keep the processes and procedures that players must follow somewhat opaque, because a lot of virtual world player behavior is about seeking opportunities to arbitrage.

This kind of complexity gets designers into trouble when there is some major aspect of their world whose dysfunctionality is driving players away, where there is some desire to fix or change the game’s systems. Baroque complexity taken too far is irreparable: you can literally get to a point where there is no adjustment of one subsystem that will not cause another subsystem to fail or produce unexpected negative consequences.

A lot of my previous analysis of the early history of the game Star Wars: Galaxies centered on this kind of problem. So much of the underlying design had a kind of Rube Goldberg feel to it, with systems and properties tethered to one another at varying levels of code and design, from how information was stored in the game’s databases to how crafting, the environment and the economy were functionally intermingled in ways that were not always how they were intended to be intermingled. I came to feel that there were many cases where the designers literally had no way out of certain problems, that fixing one aspect of the design would produce problems elsewhere, sometimes problems that could not be anticipated in advance of implementing the change. Characters advanced through developing skills within loosely structured classes, but the game design had almost no way to differentiate between the role or value of some of those classes. At launch, most classes had skills that had little value or that were simply not implemented. Fixing one skill generally broke another, or failed because other skills in other professions that were needed to properly support the fixed skill were not working correctly. The developers of Star Wars: Galaxies eventually came to the conclusion that they would just have to gut out most of the game’s design and start again. They did so in a disastrous manner, but I’m not sure they were wrong about the basic insight.

To some extent, I think the developers of the current virtual world Warhammer Online are in the same kind of pickle. In this case, one of the serious issues in the game’s design is that it is almost impossible for players to understand how to achieve victory for their faction. There are two major factions in the game which fight to control certain parts of the game environment at varying stages of the progression of the player-characters. In the endgame, both factions try to accomplish a series of difficult challenges that will allow them to attack and control the major city of their rival faction. At the moment, it is very hard to tell exactly how these systems work, and I think that is not because the players have yet to figure the system out, but because the interaction of many diverse elements in the game design is so messy that it is impossible to figure it out, possibly even for the designers.

The designers have a vested interest in keeping the system opaque. If players understand very clearly what they need to do, they may discover that the system is easy to exploit, or that one side has a structural advantage. But at some point, making a system appear opaque and making a system actually so difficult to understand that it is genuinely opaque even to its creators are actions which shade into one another.

Far more importantly, the system may simply come to seem mechanical and lacking in adaptability. Once players understand exactly what it is that they must do, how they must do it, and when they must do it, they are likely to find competition to be boring and repetitive. I think this is a major reason that baroque complexity is added by design to many human systems, games and otherwise: because they are systems which need to simulate adaptability, portability, flexibility, which need to mimic the organicism and mutability of life itself. In a way, that’s what successful art in all its forms actually accomplishes: the deliberate creation of mystery, of a work which supercedes the narrow intent of its maker. But a system which requires ongoing use, even the mechanics of an online game, needs a functionality that art does not.

————

In a limited way, I think the dilemma that some game developers have encountered echoes the vastly more consequential problems of the current global financial system. For both instrumental and accidental reasons, I think the financial system has acquired this same kind of baroque complexity, this same kind of disconnect between the top level that believes it has control over the system’s workings and numerous veiled or incomprehensible mechanisms that have been churning away busily well beyond that control. Like a virtual world whose design has functionally become impossible to easily control, the financial system may now be too complex to repair. Changing one feature may lead to undesirable and unpredictable consequences elsewhere in the system. Pulling on one thread may cause another part of the tapestry to unravel.

And like virtual worlds, there are stakeholders who have a continuing interest in the parts of the Rube Goldberg machine to which they have adapted themselves. In a virtual world that has gone badly wrong, where many players are fleeing its failure, there will always be a few players who have become adroit at using one or more of its broken subsystems. They will be the ones who complain most strenuously at any changes. The emptier the world, the louder their complaints will sound.

Players can leave all their virtual worlds for good: their ludic desires can find other expression, other opportunities. A developer who guts out everything inside of a broken virtual world to replace it with some simpler, cleaner design can hope to bring back all the lost customers, but we know very well that players who quit a virtual world almost never come back. So sometimes you stick with whatever remnant you’ve got left, no matter how dysfunctional the complexities of the design, and ride with them right out to the thinnest margins of profit before closing for good.

The difference between a game and the real world is that the capital which can move away from the broken complexities of the financial system can’t just stop circulating altogether. It needs to go somewhere, wants to go somewhere. The choice may be similar, however. Listen to the actors who’ve adapted to the dysfunctionality of the system, who’ve adapted to live on some cog of the broken machinery, and they won’t want a change. Neither will people who work within some fragment of the system that works pretty well, because they know that a fix to what’s broken has a decent chance to break what works. Gut out the whole system to try and start anew? That’s rarely possible in real life. (So far it’s never really worked with games, either.) Sometimes the best answer is to build a simple, elegant alternative to run alongside the old clanking complexity, to have the System 2.0, and hope that over time, there’s a migration from the old to the new.

Comments

[dmyers]

1.

This is a long post, Tim.

Two short points.

1. Much of this strikes me as reasoning by analogy, and if our analogies of how things work, now matter how seemingly logical or appropriate or insightful, simply, for reasons we do not fully understand, DO NOT APPLY, then we are in a pickle. I have begun to think that much of our human-brain-compatible analogies about how the very distant and the very strange parts of the world (micro-quantum-level/macro-social-level) work place us in some uncomfortable pickles.

2. There's a big mmo design problem in here, concerning how to retain customers over time (the mundane slant) and, simultaneously, concerning how to maintain and sustain novelty and "fun" in general (the more esoteric slant). In addition to your reference to SWG and WAR, I recommend taking a look at the current pvp 2.0 rules impositions happening over at City of Heroes, which are placing current CoH/V players in medias res.

3. And a third point, I guess. If we look to nature for our answer: extinction.

[dmyers]

2.

O, and great review of Spore btw.

Dont think I said that earlier.

[Cunzy1 1]

3.

Oh. Yeah not really a fan of RPGs anyway.

[N. Karlovich]

4.

I like this train of thought. Where it pertains to MMOs, I personally feel that this "adding complexity to conceal the inner workings of the system" is the equivalent of a new coat of paint on a stress fracture. Not to say that this doesn't happen all to often in MMOs. Essentially the MMORPG experience is a simulation of sorts, my character's stats versus another character or NPC with a little bit of interaction. All of the intricacies of character development, items, and abilities, along with the way these things affect each-other; this is the designer painting over the fact there simply is not enough interaction, no truly satisfying core mechanic. In games, knowing how the system works is integral to the player, and the fewer rules the better. Those rules simply have to allow for the maximum amount of diversity in order to provide an enjoyable experience. We all know that a few simple rules can create wonderful complexity, but the unspoken reverse is that too many rules reduce player choice and create the enigma problem.

[Makaze]

5.

So is the solution to go with simpler and possibly less interesting systems or head more in the direction of simulation? I don't believe the market will accept simple systems as they cannot hold interest long enough to make subscription based revenue viable.

Going more the simulationist route comes with it's own pitfalls though. The codebase would be maintainable but our control over its behavior would be significantly weakened. We would have to forgo much of the central planning and carefully managed player experience approach we take today and embrace the emergent behavior that would inevitably appear.

[Hades]

6.

Warhammer's zone control scheme is certainly overly complex. Its also vulnerable to the other faction boycotting PVP once they start to lose. So what happens is players spend all this time getting close to a zone lock, and then they watch Victory Point decay reverse all their progress once the other side stops coming out to fight. Mythic needs to simply their zone control mechanic, and make owning keeps/bfo's the central part of owning a zone.

Their crafting system is also dependent on other players. So as the population declines people have to roll ALT's who have complimentary gathering or crafting skills so they can make something.

Warhammer is a good game, but certainly needs tweaking.

[greglas]

7.

For whatever reason, it took me awhile to get around to reading / commenting on this one, but it is really interesting. Thanks for cross-posting it here.

I think you and Dave have some important things in common, Tim, in that you both have a tendency to focus on the logic of the system -- either the game system as designed or the human agent as she interprets and acts within that design. There's a flavor of formalism here that makes me a little concerned about the limits of the discussion -- it seems to me that you're both often fond of reading the virtual world a puzzle or a text rather than as a tool used for various purposes by various players.

But putting that aside, I'll stick with the "designer as author" perspective, since that's where this problem of baroque complexity seems to arise.

As an initial matter, it seems to me that it is a common problem for any good puzzle, right? The jigsaw puzzle can have ten pieces or a thousand pieces, leading to a certain trajectory of play -- the simple puzzle is completed too quickly. The puzzle with 100,000 pieces never sells because no one would want to put it together. (10,000 seems okay, though.) So people express a preference for things with *limited* opacity.

But part of the problem with the MMORPG, at least with a PvP/RvR world like CoH or WAR, is that you have multiple players trying to assemble the jigsaw puzzle in tandem. If the fastest actors get a lock on the solution, the game becomes less interesting for both those who solved it and those who must compete against them. (See Dave's problem with droning in CoH.)

The answer, I think, would be to let the puzzle's solution change. If the ground conditions vary with each day, the dominant strategies may change as well. Move the flag around the field, pick some pieces off the board, let it snow, etc. In fact, I think, ideally, this is what you'd find in a PvP game that works well -- I hear (from Nate) that Eve has some of this strategic depth.

But I suspect that you might not like that because it would be incompatible with your focus on initial design that leads to emergent complexity. If the players are the solution, the logic of the game is no longer the central matter. (See again, Dave's problem in CoH, where the dominant solution is "legal" but not interesting to the players.) If the work becomes a serial, on the other hand, we're not dealing with a fixed text any longer.

But really, I don't see any other clean solution to the problem as you present it -- if it is a puzzle intentionally designed to pleasurably impede the player, the player must be able to solve the puzzle and overcome the impediment. And then, if that's all you're concerned about, the game is over.

[Thomas Malaby]

8.

That's funny, Greg, I don't read what Tim was saying as hewing toward a game as puzzle or other set form stance at all. That is, I don't see how what he's saying disallows your suggestion that in part the actions of the players transform the game. My read is that instead Tim's granting that players can themselves transform what the game is about over time, and that in fact that designers are granting that, too. The difficulty for them is that in trying not only to live with that but also leverage it, they get themselves into some pretty challenging spots. Their contrived (baroque) complexity is an attempt to govern in an open-ended fashion, prompting but not determining how the game develops. For those questions, attention to the "logic" of the system is apt, but it does not lock us into thinking of it as a puzzle.

[Balls MacHammer]

9.

You have diarrhea of the blog, big time.

[greglas]

10.

OK, so I re-read it and here goes.

I agree that Tim is referring to two kinds of complexity -- 1) "Accidental drift towards a system where no one really understands how cause and effect work within the system happens in institutional life all the time." and 2) "Rube-Goldberg complexity-by-design". In my reading he seems to focus primarily on the latter and he associates it with virtual worlds.

With regard to the latter -- while what he suggests designers are creating is not exactly a puzzle (since he suggest that designers don't fullly know what the design at times), he suggests it does result in a largely static game structure. This seems to follow from his suggestion that complexity creates problems because those who master the complexity have investments in it and seek to entrench it.

But he suggests that part of the driver for more elaborate complexity is that: "Once players understand exactly what it is that they must do, how they must do it, and when they must do it, they are likely to find competition to be boring and repetitive." While he suggests that "baroque complexity" is an attempt to deal with this problem, I see this as very similar to his criticism of where baroque complexity leads -- to something static and unsatisfying, a broken system.

The system is only open-ended insofar as it allows for some unanticipated results. However, given that the unanticipated results lead to what I guess might be called a "bureaucratic" style of "broken" game play, which can be attributed to origin design principles and "fixed" only by a new platform, I do see this as being, at least in part, a criticism of the shortcomings of the developer as author, which might be cured (or re-framed) by making the role of the players more central.

Maybe I need to read it a third time? :-)

[Thomas Malaby]

11.

Lol, no that squares pretty well with my reading now. :-)

I guess I would add that it seems from your re-framing that we have two forces that to a certain extent come to work against the open-endedness of virtual worlds. One is the designers' desire for control and predictability. Not wanting their creation to spin out of their own control, they have an incentive to try to contrive a system that is complex enough to provide contingent and engaging experience (as opposed to boring routine), but "shallow" enough in its open-endedness that it doesn't spin out of control. The second force is the accumulation of interests on the part of *some* players and groups of players (institutions, let say) within the virtual world at hand (using "within" loosely). They come also to be invested in the status quo (this leads directly to Weberian/Foucauldian ideas).

So while routinized, "solved" gameplay runs the risk of being unengaging, it may suit the interests of both the designers and in-game vested interests. Or, at least, that would be the way I would read some of the really interesting implications here.

[dmyers]

12.

I am moved to say that "open-endedness" and "emergence" and such have begun to take on this fairy-tale quality connected to pots of gold and rainbows and lights at the end of the tunnel.

Im thinking maybe an essay to be written is "The Death of the Game" in which game rules and game goals recede into blogs about economics and 3D catgurls.

In the olden days, you either proved it, proved it couldn't be proved, or said we gotta work on that some more. And you preferred more of the former than more of the latter.

Nowadays, we seem to prefer more of the latter. Maybe it's because it's easier to not understand something than to understand it. And, of course, not understanding something has never prevented you from making money on it. Sometimes it seems to help.

[Jan van der Crabben]

13.

As a game designer who is working on a game with systems that gain a complexity that is impossible to understand even to the designer, I wish to add some thoughts here.

In most cases, the designer doesn't need to understand the exact workings of the system. He doesn't need to be able to predict the exact effect of a change. What he needs to know is what things move which effects into what direction. The rest is tampering, polishing, and re-tampering until you get it right. This is a normal process and it's perfectly fine to do it like this.

Secondly, from a game design perspective, those systems that the player needs to interact with directly (such as the attacking a city example in WAR) should in my opinion always be transparent. Complex systems should evolve through the interaction of several transparent systems. If a key game loop that the player has to interact with is too complex to understand, then this in bad game design in my opinion.

In the end, it's not complexity that matters, but whether it is transparent to the player what effects his actions will have on the immediate system that he is dealing with.