A Model for a CRPG?

[M. J. Young]

Recently I was corresponding with someone whose understanding of The Big Model is rather incomplete. In the course of our discussion he portrayed his dream of designing a CRPG with drama mechanics and a narrativist agendum.

Well, I didn't want to tell him that was impossible without at least getting him to understand that probably he didn't really mean what he said. In the process, I suggested some ideas that I thought might approach what he wanted. This is an effort to develop a one-player computer-engine role playing game. I was curious what others thought would be the likely outcome of this, or whether it's entirely useless.

I wrote

One way you might approach a drama mechanic would be through a resource mechanic.  This is really karma based, but it's an unusual approach to karma.  Let me flesh this out a bit.  The player has a pool of points that he can spend.  Some other mechanic, probably a fortune/karma blend (most mechanics are of this sort--probability plays a part but so does character scores), is used by the game engine to determine outcomes in play.  However, the outcomes are somehow ranked.  Let's suppose that you've created five possible outcomes for each situation, and ordered them in a way that is stepped between two extremes, and the random factor determines which of these five happens.  The player can now accept what the game engine gave him as the outcome, or he can spend points, one point per place shifted, to move the outcome to a different one.  It might make sense in this connection to include a mechanic that every time the player accepts the outcome given by the game engine he earns a point for the pool, and it also might make sense that in situations which are more significant there is a greater range of possibilities.  You might test how it works to have options that are off the scale, that is, which cannot possibly be rolled but can be bought by shifting the outcome with point expenditures.

This way the player chooses what happens within the limits that he must sometimes accept the generated outcomes to earn the points to spend when it matters to him.

That's as close as I can get to a drama mechanic for a one player game with a computerized referee.

The problem with the narrativist aspect is that narrativism is not about the game playing out a story.  It's about the player making decisions concerning a moral, ethical, or personal issue.

Argon and Cymelda met each other as part of a crew of an interstellar trading vessel.  Cymelda is a quiet girl from a very religious world, daughter of the reigning prince, sheltered all her life, with high moral standards.  Argon comes from a warrior tribe whose attitudes have not changed much despite the rise of technology, for whom leadership is still determined by physical combat challenges, and among whom men and women have sex without any long-term commitments, children being the responsibility of the community.  These two have embarked on a rocky relationship, each finding the other very appealing.  Meanwhile, another girl on the crew, Amber, finds Argon very attractive.  We'd say she's a slut, but that's the way her culture thinks.  She seduces Argon, but she doesn't like Cymelda's morality so she makes sure that everyone knows about it as it is happening--she turns on the monitors for the room.  Captain Dack has a soft fatherly place in his heart for Cymelda, and si
nce everyone knows, he's pretty upset.

The narrativist question here is what does everyone do?  Does Argon realize that this is going to upset Cymelda?  Does Cymelda accept that Argon is never really going to understand her thoughts and her ways?  Does Captain Dack kill Argon to protect Cymelda's feelings, and would that make Cymelda more upset or less?  Will Amber apologize for her actions, and to whom?  Would Cymelda accept the apology?  Will she leave the crew and return home, abandoning the relationship she had begun?

The reason you will have trouble coding this into a game is because narrativism is all about the intangibles of what people believe, what they feel, and how that impacts their interactions.  The player has to have the freedom to create a character with values he wants to explore, and the scenario has to allow him to explore them.  In playing or refereeing this game, each person has to be able to consider what his character, or his NPC, thinks, feels, believes, and on that basis what he will do.  Canned responses based on probabilities won't work.

Now, if we use our drama mechanic from a moment ago, we could have certain fixed responses (that is, the game won't roll for them) that can be changed by spending points.  That is, does Amber successfully seduce Argon?  The fixed answer is yes; if you want it to be no, you have to spend a lot of points to prevent it.  If you do prevent it, the next step in the game is that Amber, furious at the rebuff, tries to kill Argon.  Will you have enough points to prevent it?  Do you want to prevent it?  What happens to play after that if you don't?

Maybe this will get you thinking.  In this case, though, you've got a single player trying to interact with a machine in a way that addresses moral questions.

That's the heart of the idea. What thoughts? I find it intriguing myself, but I have no idea whether it would be workable at all.

--M. J. Young

[Callan S.]

I can't see the initial problem/impossibility to begin with (so I can't evaluate a solution).

Is the problem not being able to render the players action exactly? If the game world doesn't respond at the characters actions that are a result of the address, then the player has not made an address? Eg, if events make me want to capture the captain and carve my dead lovers name across his chest, if I can't actually do that in the game, then I can't make my address?

If it is, that'd explain why I don't see the problem. You have made an address - you'd like to carve your dead lovers name on his chest. You know that, and since your the only one playing, everyone who's important to play knows it. Is there a parralel being drawn with the force a GM applies to an expression of address that doesn't suit his story needs, and a computer which is dumbly unable to support the way you'd like to express your address in game?

[Warren]

Coming from a (video-)games development myself, I would love to see something like this, but I think that is won't be very practial, mainly because of the resources required.

Games developers (and publishers) generally like the player to experience 99% of the work that has been put into the project (which is often why unlockables are fairly things like concept art and whatnot - very little effort to create). In the example framework here, even if every 'decision point' has four potential outcomes, in a 10 'decision point' game, there could be up to 4^10 (a little over a million) outcomes, of which ten are seen on any given playthrough! Nobody id going to finance (or even ever get time to complete) somthing like that.

I believe that better AI and reasoning systems would allow us to do a more Narrativist-styled CRPG; in a DiTV kind of way: A contained location is created with a bunch of AI agent NPCs, each with a set of objectives, desires and limits. The player then interacts with this setup to achieve some thematic resolution.

Warren

[Adam Dray]

With tools like Chris Crawford's Erasmatron, I think a Narrative game is possible.

Really, you need to build enough system to allow a player to address premise via the computer game. That is, you must allow the player to build a character via choices with address of premise in mind then program some rudimentary understanding of those choices into the gameplay. The game system must offer the player situations that test a player's resolve and force them to make tough choices that address a premise.

I'm not sure what you mean by "drama mechanics," so I won't comment on that. Can you explain a bit more?

[M. J. Young]

I'm not sure what you mean by "drama mechanics," so I won't comment on that. Can you explain a bit more?

As I understand drama mechanics, they essentially mean that a situation is resolved based entirely on what someone at the table wants to have happen. The classic drama mechanic is referee fiat, in which you succeed because the referee thinks you should succeed, or you fail because the referee demands that you fail. As Erick Wujcik points out in his article here (although without the term), drama mechanics are used constantly in all role playing games, as someone just decides that a character can do something without rolling for it.

However, drama mechanics go beyond referee fiat, at least in definition. They include such concepts as group consensus, metagame overrides (i.e., "get out of jail free cards"), outcome narration, and any other situation in which what happens is determined by one of the players being granted the credibility to declare what happens according to what he wants to have happen, and not according to a karma mechanic (who is stronger, or faster, or smarter, or spends the most points) or a fortune mechanic (what do the dice say happens).

The problem I see with a karma mechanic in a CRPG is that you have only the computer and the player. If the player simply decides what happens, there isn't any challenge to the game. If the computer decides what's happening, a "drama" mechanic would essentially be railroading--the programmer put it into the system that this would happen next, without reference to skill, character scores, or random outcomes.

What I think is a potential solution to this is for the mechanic to involve a way in which the player and the computer have to "agree" concerning what happens. Given the computer's limitations, this would most easily be achieved by having the computer say in essence, "I can see any one of these things happening, which would you choose?" (I imagine something similar to the Ask Jeeves engine might work, in which the player types in what he wants to have happen and the system returns one or more statements of what it thinks the player means for the player to confirm, but that might take longer in play.) Do that directly, though, and what you have is a computerized Choose Your Own Adventure book (which really is not so difficult to code, is it, Warren?). That is much less like a game, although it is certainly a drama resolution system. I'm trying to inject something else into it that constrains the player so that he cannot always make his favorite selection, and so I've thought of the points system and the random default outcome, which inject, respectively, a bit of karma and a bit of fortune into the essentially drama mechanic.

So that's what I'm thinking in terms of drama mechanics.

Does that help?

--M. J. Young

[Warren]

What I think is a potential solution to this is for the mechanic to involve a way in which the player and the computer have to "agree" concerning what happens. Given the computer's limitations, this would most easily be achieved by having the computer say in essence, "I can see any one of these things happening, which would you choose?" (I imagine something similar to the Ask Jeeves engine might work, in which the player types in what he wants to have happen and the system returns one or more statements of what it thinks the player means for the player to confirm, but that might take longer in play.) Do that directly, though, and what you have is a computerized Choose Your Own Adventure book (which really is not so difficult to code, is it, Warren?)

Essentially Myst and all its offspring (including a lot of the early-era of CD-ROM 'Interactive Movies') were nothing more than Choose Your Own Adventure books with pretty graphics/video clips. The problem with them being, as I pointed out above, there is normally only one correct path through the game, with occasional minor offshoots. That kind of railroading is fine if that's all the player wants, but if they want to express and contribute to the game/plot in more significant ways, the developers would struggle to build in all those choices, as I explain above. (aside: This is not difficult to program, being little more than a large set of "if the player does this then do x, otherwise do y" statements, but the content creation -- unless it's just plain text -- is a massive demand.)

An AI-based natural language processor (the Ask Jeeves approach) might be possible, but certainly wouldn't be trivial. Ask Jeeves might be able to deduce your intent from a query string, but providing the content of your selections (i.e. the web pages you select) would be the limiting factor again. The closest you could get would probably be a more advanced parser for good old text-adventures (aka Interactive Fiction, nowadays) "PUT BOTTLE IN WELL" kinda thing.

Sorry, but I think the limitation with CRPGs is the limited nature of interacting with the gameworld. That could be OK, if -- returning to my 'Dogs in the Computer'  example, the AIs all ran around doing what they wanted, and the PC could interact by "talking", "action", "fighting", "shooting", but I can't see any useful way for the equivalent of Raises and See narrations to be entered by the player, understood but the computer, and then (crucially) generating an appropriate See/Raise narration made in return.

Warren

[contracycle]

Well there might be one shortcut that obviates plot problems entirely.

I agree with Callan that if the player makes a moral decision, then everyone who needs to know knows.  The question then is how to reflect that in the game itself, the problems all being plot related.

But what if these decisions relate purely to NPC's?  You could perhaps railroad the character in an entirely orthodox manner if the player has opportunities to determine whose company they keep.  To do this you need a kind of reward/penalty system for controlling interaction with the NPC's such that they themselves score the character for likeability and so on, and interact accordingly.  They then contribute assets, information and advice depending on how they feel about the character.

The advantage I see here is that it is not plot branches that differ, but character interactions, and I would expect that such interactions will be a hell of a lot less work to create than, say, whole levels.  Lastly, I recommend again Deus Ex Invisible War for anyone examining the topic, because it is probably the only game I know of that was willing to deliver a 10-minute socio-political argument to the player and then ask them what they thought.  That alone raised it above the threshold of "mere adventure".

[Callan S.]

I'm seeing a rather large focus on trying to accommodate any actions an address of premise might involve, followed by someone/something taking those actions and adding to them so they provoke another address of premise from the player. And so on.

But that's just a bunch of techniques (used in gamism and sim as well)! Yes, it makes narrativism more fun. But it's not a requirement!

I have the sneaking feeling that "someone/something taking those actions and adding to them so they provoke another address of premise from the player" isn't just desired for the greater fun it causes. In addition to that and more importantly, if someone can manage to do that, they must be really getting your address of premise. They wouldn't be able to twist the knife in the wound so expertly if they didn't understand the pain you've expressed. The more they are able to 'hurt' your PC, the more they must understand your PC's hurt.

Again, it's desirable but not needed. It's possible to torture your own PC by yourself (just like with gamism it's possible to challenge your own PC (and thus yourself) in solo computer play...happens all the time). The more tools the game provides the player to lay against their own PC, the better. It's certainly harder to twist the knife expertly in ones own PC (the pain puts you off). But then again no one will screw over a PC like it's own player. The only problem is once the players surrounded their character with knives (all aimed at the most intimate regions), it's hard for the player to plunge them in. At that point I'd wager various types of program can do quite well at stabbing. And it was the player who did the expert aiming.

[JamesSterrett]

Seems to me there's really two separate topics here:  narrativist play in CRPGs and dramatist mechanics in CRPGs.

Taking the definition that narrativist play means "the player mak[es] decisions concerning a moral, ethical, or personal issue." --  then CRPGs can do this.  Not all of them do, and all could do it better.  However, Deus Ex and its sequel, Deus Ex: Invisible War are extended meditations on the morality of the use of power; and Planescape: Torment centers powerfully on the question "What can change the nature of a man?"

Given player buy-in to the situation - accepting that the situation is personally important - any of these three will create scenes that players will remember because of *their* decisions within given scenes.  [I'd argue that player buy-in is also necessary in table-top RPGs.  After all, who wants to play with somebody who doesn't give a hoot about the game?  :)  ]

An example, from Deus Ex:

You're sent on a mission to kill a man.  Your bosses have given you the usual rap sheet on why he's a very bad person.  On getting to this guy, the first thing he does is surrender.  Do you shoot him?  No?  Now your boss arrives and orders you to kill the prisoner.  Do you?  Now your boss is going to kill the prisoner personally if you don't.  Do you?  And...  If you don't shoot your boss to prevent the execution, you're complicit in the execution of the prisoner.

Yes, it's limited in its interaction options - but it's still forcing the player to address premise, and giving you 10-30 seconds at each stage in the tree - long enough to ensure you're not being driven into split-second decisions.

What's mechanically much more difficult is the dramatist mechnics - anything the players wants to have happen, happens, within the limits of the potentially competing desires of other players..

First off, there's the simple problem of the limits of the computer's programming - it isn't possible to put every possible outcome of every possible interaction into the code.

On the other hand, any CRPG with cheat codes or a save/load feature hands some pretty powerful tools to players to ensure they always get exactly the outcome they want.  It's a kludge, but if you save/load until every crossbow shot is a critical hit, or every attempt to talk somebody around on some point works out, you've effectively insisted on "your way" to the computer [and paid in the coin of your own mounting boredom over the repeated attempt - a self-limiting currency?  :) ]

On my third hand, several CRPGs come to mind [Knights of the Old Republic, Lionheart (not especially recommended, sadly - nifty setting, though), and, if memory serves, the superb Fallout] that use a "fate point" mechanic - the player gets a limited number of points to use that *ensure* a selected action *will* come out in the player's favor.  You typically get these for performing notable tasks; but where you subsequently spend them is up to you.  Not up the standards of Universalis, I admit!

[Eric J.]

CRPGs CAN adress premise.  I spend a lot of time thinking about this actually.  James, I enjoyed your examples but I'd also like to add to the list.

Baulder's Gate: Limited in your options but there is quite a range of what choices you make regarding what you do given the fact that you're a god-spawn and your half-brother is after you.  It doesn't end up in a meaningful resolution (you have to kill your brother in any case whether you become good or evil) but it does encourage you to make meaningful decisions (such as whether to try and save him)

Alter Ego: Male/Female -This is a really old game and almost unfinable now but it IS an RPG.  You go throughout a web of 'choose your own adventure' style choices and it impacts the development of your alternate persona.  These choices impact what kind of person you become, when you die, etc.  The premise is very 'You are someone else.  Who are you?'

Wing Commander Games: In many of these games you take on the role of a person but you are still given choices (in the field or out of the field) to make that have impact on the game.  It has a very specific premise: You're guy x in situation x.  What do you do?

Any MORG (my term for MMORPG): You create premise.  You address premise.  'nough said.

Any other computer game where you make decisions with meaningful impacts: In the entire game of Jedi Knight II you make one decision that changes the long running impact of the game.  You either kill your old friend who betrayed you or don't.  It's the only time you make a decision.  It's still a narrativist element.

CRPGs can have narrativist elements.  It's just hard to do and usually must be forced decision or close.  Drama mechanics are not wholely neccecary (I'd like to see a split on one of these.  I agree that they're two seperate topics.)

I'd be more interested in a discussion on how CRPGs best address premise than if it's possible.

May the wind be always at your back,
-Pyron

[Callan S.]

I agree rather fully with your post, James!

Another account is GTA: San Andreas. I found a recurring conflict of protecting my turf Vs keeping my girlfriend happy kept coming up. Sometimes with good gamist tactics I could get both. Other times I had to make a difficult choice.

II think the spiritual attributes from TROS actually show a very computer portable mechanic for encouraging address of premise.

As to actually naming each SA, I think choosing from a list with customisable elements is fine. You can still have rigid content in the game. I mean, if a players SA is 'Woo the eight sisters of the mind's eye' or 'Kill pirates' you can still have content that revolves around saving a princess...it's just that the reason for saving her will now correlate with the SA listing (She will teach you to woo, or she will show you where the pirates are).

[JamesSterrett]

It's true you wind up with forced-decision mechanics; the limit on content puts us in that bind; but the form of narrativism does that too; what's narrativism without meaningful choices?  And to be meaningful, they must be crux points, greater or lesser, in a plot arc: places where a decision of some sort *must* be made.

However, it shouldn't feel arbitrary to the player - in either tabletop or CRPGs.

I don't have this fully fleshed out, but I believe the key lies in creating context.

Narrativism requires player buy-in on the premise.  We create this in a tabletop setting through social dynamics - agreeing on what we want to play next.  A computer game has a tougher sell, since it has to draw you in and sell you on the premise from a cold start. 

However, once the "sale" is made - the player buys off on the setting and the premise - then setting, premise, and plot need to come together to make the forced choices make sense.  If you can offer options A through D, but a bought-in, premise-addressing player can easily see a logical option E, you have a problem.  Equally, if choices C and D are silly or stupid, you may have a problem.

[An example: playing through Knights of the Old Republic as an evil character, my wife and I found ourselves repeatedly irritated by the stupidity of many of the evil options.  We wanted more options to "seem fair and be foul" instead of having to choose to be a thug.  The light-side choices were better crafted to create dilemmas between doing the good thing and doing the safe/most profitable/easiest thing.]

How to make all this work together isn't exactly a science; if you swan around on www.gamasutra.com you'll find plenty of articles on "how to make good plot arcs", "how to make lifelike NPCs" and the like, alongside articles on using the latest 3D rendering tech.  (On a more general note than RPGs, I highly recommend Ernst Adams' articles on game design.)

[Callan S.]

Getting someone to buy in is a dead end road. You'll end up choking on theory if you try and figure out how to force them to buy in by sheer neatness of content. ( http://www.indie-rpgs.com/forum/index.php?topic=16181.0 )

Customisable content is about as close as you can get. Even then it's almost an illusionist technique, when they can and do configure the game toward something very suitable to them. Then they think it was the game that made them buy in.

[JamesSterrett]

I agree you can't get buy-in from somebody who doesn't want what you're offering.  (I'd tend to say the group in your example was dysfunctional in this respect - there was no agreement between them and you on what kind of play there should be.)

Let me try a different tack of explaining what I'm driving at.  :)

When there's a reasonable degree of agreement on what ought to be happening, or, in the case of a computer game, the player has a general idea of what the game is about, and has decided to be interested enough to investigate by playing it,  --  you have people who are *willing to* buy in; they are willing to invest meaning in the game.

Yet even such situations can lead to failure.  Why?

Somebody, somewhere in the process, has screwed up.  In a computer game, the player hasn't been given enough hooks, or hooks of sufficient quality, to draw them in - even when they're willing to be drawn.  The hooks, in turn, cause the place to place emotional value on things in the game - and this gives meaning to their actions and decisions.

For example:

In the shoot/don't shoot dilemma above, the scene could, in theory, be the entire game.  It also would pack much less of a dilemma for the player.

Part of the weight of the scene comes from knowing - because the game has told you, one way or another - a significant amount about the various characters in the drama.  You've gotten to know your boss already through direct interaction and from other character's comments.  Various people (including your boss) have offered opinions on your mission before - save lives or take them? - and you've gotten feedback on your actions from characters.  It sets you up to expect that the game will react appropriately to your actions; and it sets you up to care.  (Plus, as a added dimension, the guy you're to kill is offering information about a mystery the game has presented already - and if you or your boss kill him, he won't be able to tell you....)

Without that weight of context, there's much less emotional content *available* for the scene, and it loses power.  Without the player buying in to that context, the dilemma fails.  If the context is presented badly, the player has to work to gain it, and the probability they will care drops.

Thus the trick is to deliver the context - the reasons to care - deftly enough that the player absorbs the context on an emotional level, not just an intellectual level.  Grab their guts, make them care, and when they get to one of those crux moments of addressing premise, they'll be fully engaged.  Ensure they your set of options matches those that they are going to think of (and again, context helps by limiting what they think is logical!) - and the player's decisions are, indeed, their own. Fail to allow for the player's desired decisions, and the process breaks, leaving a frustrated player (see our complaint about KoToR, above.)

[Which means you've got a good point about the process being illusionism; though I'm not sure where the boundary between illusionism and not-illusionism lies: if the set of possible reactions is covered, but it's laid down in advance, is it really an illusion?  :)  ]

[Callan S.]

I think your still trying to figure out how to get them to buy in. Now that you've expanded on the problem, I think it's an example of being promised one thing, but given less. Take your KOTOR example...did the light side choices make you feel a promise was being made about the intensity of the dark side choices?

It might be less a nar thing and more a social contract issue with the computer game designers.