The God Concept

During lunch two weeks ago with a couple of local developers from GraphSim, we started talking about the greatness of older games -- inevitable with 30-something gamers! -- and the mid-1980's game Rogue came up as an example as one of the greatest PC games from a pure gameplay standpoint. Anyone who's played this game will have a tough time arguing against the pure addictiveness of this colorful little ASCII-graphics game, originally released by long gone Epyx (best know for another of my all-time favorites, Jumpman, plus several Olympic sports games). I have a special fondness for this game as it inspired my first real success, Kingdom of Kroz, the game that officially launched Apogee from a profitable hobby into a full-time business in 1990.

While Kingdom of Kroz and its six follow-up episodes owe a lot to Rogue's design, there's one thing important thing I didn't copy: You can finish Kroz based entirely on skill.

With Rogue, you couldn't do this. You might be the best Rogue player on the planet, and yet getting through all 26 of the game's dungeon levels was still a matter of luck, with a fail-to-success ratio something around 20-to-1. Rogue created a random set of levels each time you played that simply had too many booby traps and out-of-whack difficult situations. The designer must have thought that failure was part of the fun, and retrying over and over again made that rare success so much more sweet.

So, while I loved the Rogue's gameplay, I hated the luck-based nature of it. Finishing a game was a fluke, even for exceptional skilled players. Thus, it was this game in the mid-80's that taught me a valuable principle in game design, one that I call The God Concept.

In short, the God Concept means that a god-like player should be able to play through and finish any game, on the first try, without dying.

With Rogue, a god-like player, with unmatched ability in all the skills required to play the game, could not guarantee victory when he or she played. The game simply had too many random and unfair elements outside the player's control. All games should give the player a fair chance to win, based on the player's ability.

In my Kroz games I used the God Concept as a design principle, and made sure that a skilled player always had the means and knowledge to finish the game even though much of Kroz was randomized. The key is that Kroz was never so random as to be unfair and unfinishable.

Another important aspect of the God Concept is that, in theory, a careful player, well skilled within the genre, should be able to finish a game on the first time without dying. This may be hard, but it shouldn't be impossible. The idea is that the player should never be forced to die in order to learn something important about the game, such a how to solve a puzzle or avoid the trap that just killed him. A cautious, careful, skilled player should always be able to avoid death. Death should never be a requirement in learning how to get past a hazard, or learning how to solve a puzzle.

Though this seems like a rather obvious design principle, not all designers believe in the God Concept. In Noah Falstein's "Better by Design" column in the June 2003 issue of Game Developer magazine, a case is made by designer long-time designer Mark Cerny (Marble Madness, Spyro the Dragon, many more) that the player should be killed randomly every once in a while, let's say by breaking open a rare, unmarked crate full of explosives. The idea is that the player should be kept on their toes and not automatically thinks that breaking open crates is always a safe, robotic, and kinda monotonous thing to do. While I agree with the goal of this idea, I cannot agree that randomly killing the player is a fair solution. Ever.

I don't see the God Concept being violated too much nowadays, but the story used to be different. Especially with adventure games, where surprise deaths were common, and only after dying would the player know that a particular path or action was instant death.

I'm sure there must be current games that violate this concept, perhaps others can point them out...

Comments

[PaG]

About making opening crates a thrilling activity, I think Diablo did a nice job of that. Some barrels in the game either explode or reveal a skeleton when opened. The key here is that the explosion isn't (usually) strong enough to kill the player character and the skeleton is a pretty weak enemy. So you're kept on the edge, but don't die gratuitiously.

Or designers could remove crates-smashing entirely from games and replace that with something more meaningful...

[Mike]

I remember when I first played a LucasArts adventure and they made it a point that the main character could not die. Having played many Sierra adventures, I thought that that was a refreshing concept. I don't mind being killed in an action game, but the goal of a puzzle/adventure game shouldn't be to avoid death.

[Dan MacDonald]

Uh oh, a blog about design :) Expect a lot of comments.

I completely agree with this principle. It fits nicely with my view on what games are. Games are a form of entertainment, and thus they should be fun. A large part of making something fun, is just removing frustration. Games should not take control from the player by killing them with some random incident or some action that is entirely unavoidable. Players should always have a way out, they may not exercise it and die but at that point the player blames their own skill and ability, not the game. When the game leaves the player no other option but to die, the game is in essence killing the player; it leaves the player feeling powerless and helpless regardless of their ability. This is not a “fun” experience and arguably quite a frustrating one.

It’s a difficult situation to articulate, as are most things with game design. It becomes very evident when you sit a player down in front of a game you have created to watch them play it. The walk though a door and boom lightning strikes them and their dead (hypothetical situation), you see them get frustrated and quit. At that point you realize you’ve made a big mistake in your game design.

As a side note, I think designers may still use frustration and reward in a Pavlovian sense to make a game addictive (Everquest?), but that doesn't necessarily equate to being fun.

[David Thomas]

To me, it boils down to a question of rewards and consequences on one hand, and an issue of narrative coherency on the other.

Randomly dishing out consequences is no fun. In fact, I think that is how you drive people insane in real life--issuing arbitrary punishments for things. Random challenges are fine, as long as you have the means to overcome them.

It does seem that design can get lazy and let players figure out how to handle new challenges through trial and error--which often means dying while trying. But I completely agree that a player should not have to resort to mindless trial and error and rewind to figure stuff out (unless that is the point of the game, ala Blinx).

The narrative thing is a bit sticky. It sort of breaks the membrane of the story to have your character reincarnate instantly after doing something wrong. Poole did a great job in "Trigger Happy" of discussing the need for irreversibility in narrative, of how stories stick with us because the flow of events cannot be changed. Of course, many games would last about 15 minutes if the player didn't get these cosmic "do-overs."

So, while try-and-die is probably necessary in games, it shouldn't be used by the designer, but rather gifted to the player. Which is an around the block way of agreeing with the original post!

[Factory]

Interestingly enuff of the two major roguelike games that are around (Angband and Nethack), the one that gets more*1 mainstream attention is Nethack. Which is by far the more randomly dangerous of the two games.
But Nethack also generally has smaller and more varied dungeons, focusing on a shorter but more colourful games.
Also one could consider that compared to most games most roguelike games in their default behvior are played in an ironman fashion.*2

*1 note: I said 'more', not 'alot' :)
*2 .. but this ignored by most players.

[Brad Renfro]

"While I agree with the goal of this idea, I cannot agree that randomly killing the player is a fair solution. Ever."

"Ever" is such a strong word :) I think there will always be the type of hardcore gamer who enjoys such masochism. You could argue that deathmatch can often be very close to instant random death. An example would be popping your head into a window and getting sniped instantly without ever knowing your enemy was there. In an FPS, you don't always have a complete view of the current situation, making some of your deaths seem essentially random.

Another example is survival horror. The imminent prospect of random death is a tool you can use to evoke visceral fear. That randomness is an integral part of the unknownness that fear depends on.

I think this is all a symptom of the fact that these aren't truly games in the traditional sense. They're experiences. As long as the experience remains enjoyable for the targeted audience, I don't think they have to be perfectly tight in terms of pure design mechanics.

[Jor]

All in all some very good points, and I'd have to agree with most.

Yes 'random death' is by far the most annoying thing which can happen in a game, as it punishes the gamer for nothing. It just isn't fair. But I don't think the God Concept should always be present: see for example 'Star Trek: Borg' by Simon&Schuster. In this game you can't actually win if you don't 'die' (make a wrong decision) at least once. This is entirely fitting in the game's storyline, and makes sense there. There is no real way you could know otherwise the right path in the game without guessing.
See also the Penultima Rerolled series of mods for Neverwinter Nights: a large part of the storyline depends on the player dying at least once.

Of course this is quite different from what your post is about: death in a game where it serves a purpose is not the same kind of death which is so frustrating in others. Rogue's unbeatable dungeons (and the modern counterparts) are just a way to prolong the game by forcing the player to take the same actions over and over, rather than adding to the game somehow.

[Scott Miller]

-- "An example would be popping your head into a window and getting sniped instantly without ever knowing your enemy was there. In an FPS, you don't always have a complete view of the current situation, making some of your deaths seem essentially random."

Players accept that in deathmatch contests they may die without knowing what hit them. But, in a game designed world players expect fairness, and the ability to outwit any danger regardless of how hard -- as long as they're given fair warning.

For example, being sniped from 150 yards away in Call of Duty is okay because one shot alone cannot kill you. It chips away health, but health is plentiful in that game and after you're hit you have time to run and duck for cover before being whacked again. At which point you can cautiously fight your way out of the situation.

However, I'd be greatly pissed if a single distant, unseen sniper shot killed me each time, as that crosses the line into blatant unfairness.

-- "I think this is all a symptom of the fact that these aren't truly games in the traditional sense. They're experiences. As long as the experience remains enjoyable for the targeted audience, I don't think they have to be perfectly tight in terms of pure design mechanics."

On this point I agree. But, I still never want to be killed without fair warning. Being killed unfairly doesn't make me think, "Gee, I need to try harder next time." Nope. Instead I'm boiling with the thought of slapping the designer upside the head as I scream, "You dumbass, learn to design or go back to frying burgers!" No kidding. I get mad. ;-)

[Chris Oltyan]

As you brought them up, I was thinking back to the great adventure games by sierra I played in the 80's and 90's. Many things you did would kill you randomly, but there were "rewards" for each of these deaths. First, was the simple reward of knowing not to do such a thing in the future (don't drink from that pool Roger!) the other reward was the often comical scene of your death. We would occasionally go out of our way to kill ourselves to see how we would reach the great beyond.

Given, a careful player wouldn't do many of the things we attempted, but in an effort to solve a puzzle, you might realistically try several of these options.

I guess the question in my mind is "How Godlike of a player do you design for?" and how can you figure out that scale?

[Scott Miller]

-- "I guess the question in my mind is "How Godlike of a player do you design for?" and how can you figure out that scale?"

Chris, the easy solution is to always have some sort of fair warning clue or indicator that doing something will hurt or kill your character. In the Infocom games, for example, venturing into a dark room was always proceeding with a warning about lurking grues. You should never be punished for doing something normal or innocent, unless given a warning of danger beforehand. Perhaps this principle should be renamed, The Fair Warning Concept. That's a better description of its purpose.

[J.]

I always assumed that the All Men Must Die (borrowing from George R.R. Martin) aspect of "some" games was born of the same mental disconnect that comes from game designers either being so self-impressed, self-important or just desperate not to go back to flipping burgers that they forget to think about their game in the most basic, simple terms. Which means approaching the game as an archetypical player would -- and then /play/ it. (Scott's notion of what players might accept is one way to do this.)

The genesis of this debate, from my perspective, is more linked to designers not knowing or just not caring about aspects of their games that make them fun, and instead focusing on "wouldn't it be cool" crap that players will never experience, because they'll be throwing their controllers or uninstalling the game.

[Wyatt Jackson]

I agree with the original post, the grues are an excellent example, although not as entertaining or imaginative as the Sierra death scenes. Sierra games didn't kill you randomly though so I'm not sure they apply at all anyway.

Besides all that, I believe 'God Concept' can be expanded to inlcude more that just the random death idea. For instance, games that create a system of rules and then allow the computer controlled AI to freely break those rules indicates poor design decisions as well. Games that come to mind like this are obviously the Street Fighter series (I've broken my fair share of controllers over this one) as well as the the idea of giving a speed boost to the lagging competitors in a racing game. In a shooter, why does the enemy have auto-aim, why do they not have to reload, why do they have unlimited ammo?

You could argue I suppose that these things increase the challenge and thereby the 'fun' of the game but then why do we still choose to play against human opponents rather than AI and why does playing against the AI not necessarily prepare you in any way for dealing with human competition. Maybe this is just my personal gripe concerning the state of industry AI but it sure does get under my skin.

[Alan De Smet]

Wyatt Jackson said, "Sierra games didn't kill you randomly though so I'm not sure they apply at all anyway."

I don't think Wyatt remembers the old school Sierra adventure games. Death was certainly unexpected and fickle. The early games were rife with them. Regrettably it's been a _long_ time since I played them, so I'm having problems remembering good examples. I do remember that none of the Sierra heroes understood that falling was dangerous, if you clicked a bit too far off a ledge, you died.

In one of the early Police Quest games, you were warned to practice shooting your pistol. You were never given a "you have practiced enough" indicator, so I practiced a little each game day and played on. However, apparently I didn't practice enough, because when I needed to shoot a hijacker I simply missed. Repeatedly. Death after death after death. I needed to go all the way back to the beginning of the game and practice even more.

In one of the later King's Quest games is appears you need to map out a desert. The Oasis are few and far between, you can only find them by trial and error. Error naturally means death. (It's possible that there was an easier solution, but I found the game so frustrating I gave up.)

[Mark H.]

I liked how Prince of Persia: Sands of Time handled the death situation. Just like Max had bullet time which served a purpose for gameplay, so did this. Everytime you made a mistake and died you could reverse and rewind back to a point and retry another option of getting from point A to point B. In this case, I never have the feeling of being haunted or frustrated by dying and it allows for trial and error(without the irritation).

[Barret de Nazaris]

We have so much to learn from boardgames ("korsar", "la citta", "clans" for example)...

[Ben]

This reminds me of a game 'The adventures of willy beamish' where when you 'died' there was a specific animation that played. You got used to that animation meaning game over, so you would reload during that time fairly quickly.

However, when you reached a certain point in the game, you actually HAD to play though that animation, you wouldnt get 'game over' at the end. It was fairly sneaky/mean of them to do so. Took a little while to figure out that was the only course of action.

[Adam Vandenberg]

I was playing "Shadowgate" on the NES a few months ago, which I had never played. Jumping Jehosephat, that game is the epitome of instant unwarned death. Every other thing you click on kills you with no warning that I can tell. Was this ever considered fun?

[steve]

The whole "kill the player off a lot" mentality has to come from the arcades, right? Otherwise, what's the point? In the arcade, you need to kill the player off in order to get them to put in another quarter. At home, this isn't really an issue, so why follow the same paradigm?

I've always felt that the key to any game where death is a potential outcome is to make the player feel like they're on the edge of death, but never have them actually die. Death is frustrating and brings the game to a hault.

I'd like to see a self-balancing game where, let's say, you reach a certain health level, you turn the corner and there's a health pack, or the AI starts missing with greater frequency or acts a bit dumber.

Max Payne supposedly does some of this type of balancing, though I've never actually noticed it (and it didn't appear to be any harder or easier than most games; maybe that means it works perfectly).

[Scott Miller]

Steve, I've got a full blog coming very soon about game difficulty, which ties into this subject a bit. With Max, you're not suppose to notice it by design, but it is there the entire time trying to make sure the game is not too easy or too hard.

As for dying in games, I've given this substantial attention for over a year. I've yet to come up with a solution that avoids the fact that players must die in action games. This will probably make for a a good future topic -- there are literally a dozen angles to player death that are hard to avoid. So, whether from the arcades or not, player death carries on because there's been no better alternative.

I'd love for someone to show I'm wrong, but most times when I hear someone's solution it overlooks a key problem.

[Dave Wood]

Adam, I played ShadowGate when it was new and I found it to be a great game. I think when we look back at the old games we used to love our perspective is greatly skewed. Back then if you we're so much as touched by a single bullet, enemy or anything you we're dead. One hit, no death animation, just dead. Games have evolved much over the years and we still seem to be missing something they had in the beginning.

Dieing and respawning is basically the same thing as reseting the scenario. Why do we have to die? It's a punishment. With first person shooters it seems necessary, although creativity could lead in other directions where dying would be an opportunity instead of a punishment. Dieing is a risk, makes you feel like you have something to lose, and keeps you on your toes. Maybey that's why it's necessary.

This is a very interesting topic. Perhaps by discussing the issue and arriving at alternatives, we could create a whole new kind of gameplay. I think that is the spark the industry needs right now, innovative ways to fundementally change gameplay without creating the same scenario and risk/reward system.

[Brad Renfro]

I think death in games is important because it's a realistic metaphor that any gamer can immediately understand and relate to. There are many other abstract ways to challenge the player in an action game. But as long as mainstream games choose to use realistic themes and intuitive metaphors, the health/damage/death concept will be the usual standard, just like the concept of gravity, control of a single avatar, and traditional three-dimensional movement. I don't think gamers actually want nearly as much innovation as they claim to want.

[Jeffool]

Yo, Renfro, (just had to say that).

Don't think Scott was saying the player should never die, just that they shouldn't be randomly killed over a factor they have no control over. The above mentioned crate/barrel-smashing randomly equalling either a bonus or instant death, for example, that I believe PaG mentioned. And that the "1337"-est of the "1337" players could concievably play the game without ever dying.

And Scott, I'd love to prove you wrong... I just keep agreeing with you. One day, I'm going to have to change my mind on something just so I can get into one of those 'flame wars' I hear so much about.

Jeffool.

[Brad Renfro]

"Don't think Scott was saying the player should never die, just that they shouldn't be randomly killed over a factor they have no control over."

Whoops, didn't mean to imply that the player should never die in any game ever. Just responding to the idea of trying to create a single game that avoids player death, namely a mainstream game.

"Players accept that in deathmatch contests they may die without knowing what hit them."

Yeah it's funny how people complain about Medal of Honor's snipertown level in singleplayer but would probably go on to enjoy the same type of sniping in multiplayer :).

"But, in a game designed world players expect fairness, and the ability to outwit any danger regardless of how hard -- as long as they're given fair warning."

I think it can become a matter of shifting gameplay. In the case of Diablo's random exploding barrels, instead of having the gameplay of "avoiding" the bad barrels, you have the gameplay of potion/health management. If gameplay is about making interesting choices, the choice the player must make is not "How do I figure out which barrels explode", but rather, "When do I use and buy potions to cope with these barrels that randomly hurt me".

It's kind of like poker. You cannot know or control the cards you are dealt, since they are random. The gameplay is not in trying to psychically know which random cards you will be dealt. It's about managing your betting risk by playing odds and reading opponents' reactions.

[Scott Miller]

-- " In the case of Diablo's random exploding barrels, instead of having the gameplay of 'avoiding' the bad barrels, you have the gameplay of potion/health management. If gameplay is about making interesting choices, the choice the player must make is not 'How do I figure out which barrels explode", but rather, "When do I use and buy potions to cope with these barrels that randomly hurt me'."

Brad, I agree with you. Like I said before, it's okay to randomly knock down the player's health because the player always has the choice of not opening a chest (that might explode) unless they have sufficient health to withstand a booby trap. This then becomes a gameplay decision.

But, it's entirely unfair for a chest to kill the player regardless of how healthy/strong the player is.

-- "And Scott, I'd love to prove you wrong... I just keep agreeing with you. One day, I'm going to have to change my mind on something just so I can get into one of those 'flame wars' I hear so much about."

Heh. Jeff, this blog is still young. And my most volatile topics have yet to be unleashed. ;-)

In fact, since I don't have time to wrap up my little essay about game difficulty in time for my usual Monday post, with the holidays, family time and stuff, I think I'll post a quickie topic that I'm sure will be like tossing bloody meat into the warm oceanic waters of Great Barrier Reef.

[Noah Falstein]

I should clarify about the reference to my BBD article in Game Developer. When I heard Mark Cerny make the case for occasional random death, my first thought was that he wasn't serious, just enjoying the role of being an inconoclast, but the more I thought about it, the more it felt like an application of "art" to the science of game design. Yes, it's in principle a bad thing to randomly kill someone - but having just a tiny bit of unexpected death struck me as a master including an imperfection precisely to get people to question all their other assumptions. I don't want to make too much of this, I wrote about it in part because I thought it a controversial topic and therefore more interesting than going for a generally accepted principle. But I agree that the "God Concept" is a good overall rule to follow, and in fact as Mike posts above, it was a consciously articulated rule when we were doing early LucasArts adventure games, and in direct response to our frustration with Sierra.