Holy cow Oluseyi! I turn away for like less than a day and I have like a billion lines of text I gotta read now! Geeze! [smile] i'll have to get back to this topic after I catch up.

Interesting. I could use a different representation of tension for myself.

It could be fun implementing this with a simple interfase. Im considering giving it a whirl myself with a text interfase of multiple choice, to see how it works.

edit: as in stdin/stdout... no parser, just multiple choice. Don't get excited ;)

I'd like to take things the other way a bit, if I may, and return to the initial post...

Quote:

Original post by Oluseyi Prologue The majority of gaming genres are predicated on doing - swashbucking combat, stealthy pilfering, white-knuckled strategizing, athletic competition - in large part, I suppose, because of the fact that such active interactions with other people or with objects are very well regulated. It is comparatively trivial to delineate between valid and invalid interactions, and thus provide structure for the challenge or set of choices that comprise the core of the game.

I disagree. There is nothing about those 'involved' activities that makes them any easier to deal with vis-a-vis validating interactions; however, what they do make easier is the development of a model that people will buy into. An experienced skiier or martial artist will tell you that skiing games or Tekken are nothing like the real thing - and they're right. What the developers of those games created were approximations of those activities, models that contain the 'main' elements of the activity - lean left, lean right, punch, kick - easily seperable activities, events within a system, that they then set into a dramatic context. Remember, we could just as easily call them Action #001, Action #002, Action #003, and so on. The particular dramatic contexts you've picked out just happen to be good skins for those systems - the game systems imitate the real-life ones well enough that to the uninformed the difference isn't visible. It doesn't mean that the system will fit the dramatic context perfectly - I don't know of any Tekken-style games that let you pull out a gun and blast your opponent in the head, for example.

This is something that's been around in games for ages - people know that they can't hit the button on the flat-texture control panel because "that's not the way the game works." Interestingly, one of the comments I've heard people making about Half-Life 2 is "If you can think of something, chances are the game will let you do it." They've developed a model for their world that matches the rules of a real-life confict closer than most other games, and this is surprising people.

Quote:

Trivial, that is, when compared to the complex interactions, rules, heuristics, intuitions, emotions and other factors that influence dialog and drama.

Right, the systems governening those dramatic contexts in real life are orders of magnitude more complex.

Quote:

We want more drama! I'll be honest, twitch reflex games and obsessive micromanagement simulations simply hold little appeal for me, and even my trusty old sports games have gotten rather stale. They're fun for playing with other people, but sometimes I want relatively short, engrossing games that I can play alone in arbitrarily-sized increments, like owning a DVD of a good movie. I want great stories and characters, but not just as the backdrop for a bunch of running and gunning or hacking and slashing, nor as "mind candy" for puzzle gameplay.

I just want to pick up on something there - you said owning the DVD of a good movie. While I think...

Quote:

Original post by Rosecroix When you describe it I see for my minds eye a very advanced version of Carmen Sandiego, very much a thriller; a researchheavy detective story. I would belive constructing an engine to create one of those would be quite different from the soapopera emotionemphasized game I first envisioned you wanted, something much closer to Friends or Frazier than say CSI.

Quote:

Original post by Oluseyi The other reason for actor shifts from primary to secondary is narrative arcs. In long-running narratives such as television shows with large casts, not every character can play a central role in every episode. Consequently, pertinent to the issues of a particular episode, a certain subset of the cast (sometimes as few as a single character) becomes primary while others become secondary. Since our objectives here including both dramatic gameplay and episodic gameplay, a soap opera or primetime drama type game is not out of the question, where each play session corrals a subset of the characters, establishes a premise and sets the protagonist/Actor (user) off on a quest to resolve all tensions.

In case you hadn't guessed it already: TV. The fourty-hour epics that you go on to talk about, Oluseyi, are the game industry's answer to film; but what of TV? I can sit down in the evening and watch an episode of The West Wing, or Stargate SG1, or Buffy the Vampire Slayer; each has drama and inter-personal interaction packed neatly into 1-hour segments for easy consumption. Once you know the premise - the west wing of the white house; a team of people who use an alien artifact to visit other planets; teenage girl fights demons and saves the world - each episode is pretty much self-contained. (Sure, there are some long-running story arcs, but you can usually pick things up pretty quickly without watching earlier episodes). Show me a game that packs a self-contained 'adventure' into a one or two hour space, complete with narrative resolution, and we'll have something interesting...

Quote:

I've been watching a lot of old movies recently, and one thing I consistently marvel at is how taut stories and tight narrative direction create just as exciting and visceral experiences as an explosion- or effects-laden "blockbuster." Violence and sex are powerful elements, but they motivate, create and alter the situations in these films, not resolve them as in modern films and games.

Agreed - I tend to find the stories much more exciting than the explosions and the effects. The Spiderman films both affected me pretty deeply, but when I think back to them, it's not the fights and explosions that I remember - it's scenes like the wresting ring in the first one, or where he stops the train in the second.

Quote:

As I mentioned earlier, I'd prefer a compact narrative. I find most contemporary games too long. "40 Hours of Gameplay!" is a heck of a lot of time, especially when you factor in all the dying and loading, and it is frequently achieved through repetition of an action.

I think a claim that '40 hours of gameplay' is still a target for anything but some RPGs is a claim on shaky ground. I usually hear 10 hours being throw around as more of a target. But anyway...

Quote:

I like games where I can play through the entire narrative in a single sitting without raising the ire or suspicion (that I'm a hopeless geek) of a significant other: we're talking three hours, tops.

See above re: TV [smile]

Quote:

Obviously, I'd also like a design that awards penalties without forcing repetition or precluding resolution/success. Clearly this requires a fair amount of dynamism in both design and implementation, such that each play through is sufficiently unique.

So, let's just be clear where we stand: We want the system to be less clearly defined (or at least, more complex, such that mapping it out takes longer), but we still want to penalize the player when they fail to abide by the rules of that system. OK, just so I know how low we stand [wink]

Quote:

The target audience for this would be casual gamers or other gamers taking a break. The objective would be pick-up-and-play simplicity, which precludes gamepads, complex keyboard maps or cluttered interface. We have a two-button mouse on a personal computer, with a keyboard for auxilliary input.

By which I think we mean 85% of the market. Let's not understate the importance of this, if we get it right - we're talking about real games for the wage slave, here. Successful games in this category stand to make a lot of money.

Quote:

In addition, it'd be nice to have greater depth that simple dialog trees (match-the-response-to-the-prompt). Somehow, using that mouse and a simple interface design philosophy, we want to be able to move the protagonist/avatar, speak to others, observe surroundings and specific objects and so forth. We also want to stay away from Lucas Arts-esque use-object-with-other-object-or-action-in-thoroughly-nonsensical-and-non-intuitive-fashion gameplay.

Well, we're talking about stepping back and defining a looser model that allows us to abstract a greater number of things. OK. And I guess that if we then wanted to transport our 'regular' brand of gameplay across, we'd simply do something like put a gun into our game world and allow the player to pick it up and use it. (The idea of doing something small, like a handgun, and intending that the player only ever use it to fire a single shot, is one that could easily arise in a dramatic situation, after all).

Let's not be too hasty in discarding our special effects-laden blast-em-ups. Just because they don't have drama, doesn't mean that to achieve drama we should head in the other direction completely.

Imagine a game where you command a squadron of fighter craft, in a space setting. The game features space combat as a fairly major component, but you also have to 'manage' your squadron - that means both issuing orders in the midst of battle, but also getting to know the people you command. If you need to send someone out on a 'Do not engage' mission, do you trust Maverick enough to do it? How will you account for the fact that two of your pilots have fallen in love with each other? What will you do about the insubordinate sonombitch who keeps refusing to follow orders? The interplay between characters - and the cast of characters need not be very large - creates a whole new system for the player to keep track of, on top of the familiar 'point at the bad guy and press the trigger' that games like Freespace have used for ages. It could also be easily made into a TV-type game, with the squadron being sent on a different mission each episode - the 'main' story for each episode is resolved in course as the squadron complete the mission, while the interpersonal relationships can be carried forwards into the next episode.

It's the term that Ken Perlin used in his talk at EDF - psychology hacking. People love to be amateur psychologists, to make predictions and inferences about the way people think and behave. That's why soap operas are so popular - because people can get together the next day over the water cooler and compare notes, tell each other their predictions, and generally compare psych-hack penis sizes. We can do the same thing in games, but we don't need to throw away everything we've built up to reach that point.

Quote:

Original post by superpig Quote: Original post by Oluseyi Prologue The majority of gaming genres are predicated on doing - swashbucking combat, stealthy pilfering, white-knuckled strategizing, athletic competition - in large part, I suppose, because of the fact that such active interactions with other people or with objects are very well regulated. It is comparatively trivial to delineate between valid and invalid interactions, and thus provide structure for the challenge or set of choices that comprise the core of the game. I disagree. There is nothing about those 'involved' activities that makes them any easier to deal with vis-a-vis validating interactions; however, what they do make easier is the development of a model that people will buy into. An experienced skiier or martial artist will tell you that skiing games or Tekken are nothing like the real thing - and they're right...

and there lies the difficulty. Not everyone is a skiier or a martial artist... but everyone has to deal with people every day. It is not that the model is so difficult to make, it is that it has to be much, much better to be accepted.

Quote:

Imagine a game where you command a squadron of fighter craft, in a space setting. The game features space combat as a fairly major component, but you also have to 'manage' your squadron

and this space combat requires reflexes and fairly complex controls, thus being a turnoff for the audience stated in the prologue. Unless, there was the option to make combat happen automagically (and conversely, an option to keep the combat and just skip the double-guessing and trickery). Then you'd have BOTH groups of gamers interested, otherwise you'd have half of each.

Also, in a military setting, interactions are much simpler, and could probably do without all this system. Consider morale in the Total War series. That pretty much models if your unit (or wingman) is going to follow an order or not (with some tweaking of course). Who cares if you hurt Maverick's feelings?


back on the technical side, i wrote some things and more questions arised.

Mainly, there's knowledge, and how to represent it. For example, let's say that we have the pre-designed knowledge of "mary had a little boy". And the feelings of each NPC towards that. Okay so that was simple.
Now Mary kisses Jay, but Elaine (running outta names already) sees them. Elaine rushes to tell James.

Now, Elaine would have to add this new knowledge to her list of known stuff, and then pass it on to James.
What would she store? The concept of kissing and a link to Jay and Mary? if done that way, how does one decide if that was good or bad?
The other option is creating a new concept that goes "Mary kissed Jay". But again, how to decide how the NPCs feel about that new, unlisted knowledge?

Secondly, i tried putting trust in there, for the effects of allegiances on the spreading of knowledge. I considered each NPC having a trust rating for every other NPC. But then i realised that trust is relative.

Say Mary was James' girl (of course!), and Jay was his long time ex who she still adores. James will believe everything Mary says, except when it is about Jay... he'll be suspicious about such kind of knowledge.
How can this be modeled? what if Jay is suspicious about what Mary may let Elaine see? (ok so we could limit such calculations to one-on-one only, for complexity's sake).
Also, what if Jay is suspicious about Mary regarding, say, drugs? say she's a recovered crack addict, he trusts her always unless its about crack.
How to model that?

Third: gossip. This could be simplified by not carrying a trail of NPCs who passed it on, and just passing the fact and who just gave it to you. This way it would work and be simple, i believe.

Quote:

Original post by superpig I disagree. There is nothing about those 'involved' activities that makes them any easier to deal with vis-a-vis validating interactions; however, what they do make easier is the development of a model that people will buy into.

The point was not that the real life activities were easy to validate; the point was that their electronic entertainment analogues mapped cleanly to valid and invalid interactions, which is the basis of their popularity and status as staples of gaming.

Quote:

In case you hadn't guessed it already: TV.

The reason I didn't turn to TV for an analogy is the time constrained nature of TV so far. You must be watching a specific channel at a specific time on a specific day to partake of the episodic entertainment. A DVD, on the other hand, is something you can enjoy whenever and however you like. The perfect analogue would be television on-demand (can't sleep at 4:07 AM? call up a specific or random episode of The West Wing, or whatever your preferred show is), but that isn't a reality anywhere in the world just yet.

Quote:

So, let's just be clear where we stand: We want the system to be less clearly defined (or at least, more complex, such that mapping it out takes longer), but we still want to penalize the player when they fail to abide by the rules of that system.

No.

When we speak of penalties, for instance, we're talking about how the system responds to a user's choices in terms of how to solve a certain problem. In most games there is only one way to solve a "puzzle." In this system, we wouldn't so much specify puzzles with solutions as we would layer objects with intrinsic properties and have the system recognize when an object no longer constitutes an impediment to a specific aspect of user progress. Say our user arrives at the gates to the mansion of an eccentric businessman whom he believes may have valuable information, but who does not receive any visitors. How does he get in? The gate and fence - as well as any guards, dogs, spotlights, camera and other security systems - will have intrinsic properties and update cycles, and the objective is to somehow play properties against each other - have the dogs set off the alarms - or disable properties such that the user can advance to the next challenge or reward.

Quote:

Let's not be too hasty in discarding our special effects-laden blast-em-ups.

Nobody's suggesting we do. All that's happening is that this discussion is placing that aside, as it isn't inherently pertinent, and focusing on what is unique here, which is the dramatic gameplay and dynamic construction of narrative.

Quote:

Original post by Madster back on the technical side, i wrote some things and more questions arised.

I honestly think it's a little early to talk implementation. I like to really turn an idea around in my mind, consider all the ramifications, let it fester and soak and grow and shrink and mutate, so that when I do start modeling systems, I have a very definite abstraction to work from. Talking implementation too early leads to half-assed, inelegant systems.

That said, I will point out that it is more efficient to have a single repository of knowledge that all NPCs dip into, and simply store a sequence of record IDs in the NPC itself. The available fundamental interactions would have to be specified by the designer/writer (Mary kisses Jay relies on the fundamental action "to kiss" [romantic]), and each one would have an associated intrinsic morality or response value. If two people who are not otherwise attached kiss_romantic, then the evaluation shifts to an investigation of who the involved parties are and the "morality" of the observer/respondent. This way, some characters could be offended at homosexual kisses while others would be happy that "Steve finally found someone." Similarly, Elaine might be scandalized that Mary kissed Jay, but Carol might be all for it.

Quote:

Secondly, i tried putting trust in there, for the effects of allegiances on the spreading of knowledge. I considered each NPC having a trust rating for every other NPC. But then i realised that trust is relative.

Correct, but I had this conversation two years ago, so I'll try to scrounge up an accurate summary of our (then) conclusion. The relationships between characters form a directed graph, such that A might trust B 73%, but B trusts A 94%. Now if A does not know C but B knows C, A's evaluation of - and thus trust value for - C will be a product of trust(A, B) and trust(B, C), with a little randomness thrown in for good measure. As A and C have direct interactions, however, trust(A, C) (and trust(C, A), of course) will change.

Obviously, at this point, our graph only assigns a single trust value to the connection between characters, which doesn't allow us to model more complex situations like different trust values depending on domain. We could expand the trust function, of course, to yield trust(A, B, domain), or we could employ a data-driven design that allows for per-character pair, per-domain trust modeling.

Again, this is why I think it's a little early to talk implementation. We haven't discussed enough aspects of the idea to grasp it completely enough to try our hands at decent code.

Quote:

Third: gossip.

Two things: First, we want information/knowledge to percolate through a society over time, with the most connected nodes clearly being the first to know (and with trust or familiarity ratings accelerating/decelerating the spread between characters); Second, we want rumors to change form as it is carried from mouth to mouth, in a sort of "Chinese Whispers" system. I'd discussed this, too, two years ago, but I can't remember what we arrived at. I'll have to call up my buddy to see if he can jog my memory.

Quote:

Original post by seanw I think a speech interface would be amazing though.

I agree. I think that speech, facial expression and gesture are our most natural forms of communication, but there are challenges with them beyond even the technical hurdle of high fidelity data acquisition. Some people struggle with grammar and/or lexicon, and we don't want to penalize them for aspects of who they are outside the game. Gestures don't translate well across cultural boundaries; one of my favorite examples is how the American "swinging arms in front of torso with fist meeting palm at bottom" is interpreted as an idle loop, but it means "do you wanna fight?" in other parts of the world.

Some facial expressions are "universal," but the precise interpretation given on each face exhibits high variability.

Quote:

In comparison, for a graphical game you'd need a horrible GUI button, a pull-down menu, a keyboard button etc. for each type of action you'd want to do...

Not necessarily. One of the challenges will be finding the appropriate interface, which will undoubtedly draw from a wide variety of preceding genres, but one of the things I said early on (first in Madster's thread, IIRC) was about the prospect of "playing emotions" - using a symbolic language of sorts where a statement is annotated with a sequence of emotions. It's still fairly complex-sounding, but, as I said above, I think it's a little early to be talking implementation.

Quote:

Hmm. There's quite a divide with that issue. I never really liked illustrations much because they tend to go against the mental image you build up, which I find more powerful given good writing.

Remember that these games are supposed to be fairly quick, though. Reading a few paragraphs that describe the immediately visible environment is a significant time investment in comparison to looking at and seeing the mountain, the precipice, the club.

Quote:

Interactive music would be a nice feature...

Absolutely.

Quote:

I'm pretty much saying I reckon you could achieve your goals with a modernised text-adventure which had beefed up natural language recognition (so you could phrase sentences in quite complex ways), having the vast majority of user actions accounted for by the game and using speech for input (anyone who can talk and know not to put more than one action per sentence could use it). Maybe that's not what you want, but I'd love to see what it would be like.

Noble and fascinating as that sounds, we have to ask the question: is this appealing to users? Honestly, it's not appealing to me. Pages and pages of descriptive or narrative text on a computer screen make my eyes glaze over. That's what books are for.

Quote:

Original post by Madster and there lies the difficulty. Not everyone is a skiier or a martial artist... but everyone has to deal with people every day. It is not that the model is so difficult to make, it is that it has to be much, much better to be accepted.

True enough. That's why seeking ways to keep the complexity down - such as restricting the number of people involved - is something I think we have to think about.

Quote:

and this space combat requires reflexes and fairly complex controls,

Reflexes yes, fairly complex controls hell no. I could design you a space sim which used only the mouse and two mouse buttons.

Quote:

thus being a turnoff for the audience stated in the prologue.

Those audiences aren't mutually exclusive...

Quote:

Unless, there was the option to make combat happen automagically (and conversely, an option to keep the combat and just skip the double-guessing and trickery). Then you'd have BOTH groups of gamers interested, otherwise you'd have half of each.

... but you knew that. I don't think it's a question of 'half of each,' though, I think the proportion of each group who would be prepared to put up with both activities is much greater than half. And besides, if we're sticking with my space combat example, there's nothing stopping the player from ordering the other members of his squadron into combat first while he hangs back, away from the action. (That might get some squadron members thinking that you're a coward, but then you get to deal with that).

Quote:

Also, in a military setting, interactions are much simpler, and could probably do without all this system.

The purpose of this isn't to model morale, it's to model the complex social interactions within a tightly held group such as a fighter squadron. Those interactions will have an effect on morale - just as they'll have an effect on whether a pilot is likely to be reckless or cautious, to protect a fellow pilot or neglect them, and so on.

Quote:

Consider morale in the Total War series. That pretty much models if your unit (or wingman) is going to follow an order or not (with some tweaking of course). Who cares if you hurt Maverick's feelings?

Not played Total War, I'm afraid, but the point is this: you should care if you hurt Maverick's feelings. It will affect his behaviour, his priorities. He may turn insubordinate; he may put in for a transfer (in which case you lose one of your best pilots). And let's not forget the other pilots who are friends with Maverick - you mess with him, you mess with them too. Due to the player's actions, the entire squadron could undergo schism. What will you do then?

A major point is this: the units are not interchangable. Each one would be distinctive, unique, not just by callsign but by character. The player should feel differently about sending Maverick to his death than they should about sending Gooseman.

Quote:

Original post by Oluseyi The point was not that the real life activities were easy to validate; the point was that their electronic entertainment analogues mapped cleanly to valid and invalid interactions, which is the basis of their popularity and status as staples of gaming.

That's quite a leap of faith there - that the clarity of the system is what makes it popular. Guess the number is a system with very clearly defined rules and processes, but I don't think you'll catch many people playing it for fun.

Quote:

The reason I didn't turn to TV for an analogy is the time constrained nature of TV so far. You must be watching a specific channel at a specific time on a specific day to partake of the episodic entertainment. A DVD, on the other hand, is something you can enjoy whenever and however you like. The perfect analogue would be television on-demand (can't sleep at 4:07 AM? call up a specific or random episode of The West Wing, or whatever your preferred show is), but that isn't a reality anywhere in the world just yet.

Delivery mechanisms are a different discussion, really. Television-on-demand is the best analogy for what I'm getting at, yes. But "that isn't a reality anywhere in the world just yet?" Tell that to my five season's worth of Stargate SG-1 DVD boxsets. [grin]

An episodic model could be adapted for one release per week, or a 'series' of 24 episodic games around the same premise being released simultaneously, or whatever. It's not really important.

Quote:

When we speak of penalties, for instance, we're talking about how the system responds to a user's choices in terms of how to solve a certain problem.

Ah, then you're using the wrong word. I think you want 'consequences.'

Quote:

In most games there is only one way to solve a "puzzle."

Bollocks! Here's a puzzle for you: there's two guards on the door I need to go through, patrolling back and forth. I can try and sneak past them, I can shoot them, I can create a noise elsewhere for them to distract them. That's three main categories of solution that any game supporting field-of-view and sound-event systems will allow (and that's most recent FPSes and third person shooters).

Quote:

In this system, we wouldn't so much specify puzzles with solutions as we would layer objects with intrinsic properties and have the system recognize when an object no longer constitutes an impediment to a specific aspect of user progress. Say our user arrives at the gates to the mansion of an eccentric businessman whom he believes may have valuable information, but who does not receive any visitors. How does he get in? The gate and fence - as well as any guards, dogs, spotlights, camera and other security systems - will have intrinsic properties and update cycles, and the objective is to somehow play properties against each other - have the dogs set off the alarms - or disable properties such that the user can advance to the next challenge or reward.

That's nothing new, though. As I said before, most modern shooters will give you three solutions to the guard-on-the-door problem; but there's nothing explicit about those solutions, they arise through the properties of the guard (has a field of view, reacts to sounds in the environment). All the game cares about is whether you get through the door without dying. In your example, all the game would care about is whether the player gets into the mansion (without dying, if that's a possibility); it'd be up to the player to spot properties of elements of the system (dogs: will chase meat; alarms: can be triggered by dogs; guards: will investigate alarms when they go off) that they can use to develop solutions.

Quote:

Nobody's suggesting we do. All that's happening is that this discussion is placing that aside, as it isn't inherently pertinent, and focusing on what is unique here, which is the dramatic gameplay and dynamic construction of narrative.

OK. I was just a little concerned that the discussion seems to be headed towards developing a game which is purely based around drama and dynamic narrative, which I think would compromise the real goal; we don't want to develop a new genre of 'dramatic' games, we want to make all existing genres more dramatic.

Quote:

Original post by superpig That's quite a leap of faith there - that the clarity of the system is what makes it popular.

I made no inference about popularity. All I said was that they provided an easier - and thus more easily implemented - mechanic.

Quote:

Ah, then you're using the wrong word. I think you want 'consequences.'

That's debatable. Some consequences are penalties, but not all penalties are direct consequences of user action. Shoot (or offend, for a more benign variant) a character and you may lose the ability to solicit information from him. Miss a train, however, and you may lose the ability to interact with a certain character who was going to be in town at a certain time - but that's not a direct consequence, since you didn't know of him at all. The latter is clearly a penalty, not a consequence.

Quote:

OK. I was just a little concerned that the discussion seems to be headed towards developing a game which is purely based around drama and dynamic narrative, which I think would compromise the real goal; we don't want to develop a new genre of 'dramatic' games, we want to make all existing genres more dramatic.

Yes, but we want the latter as a side-effect, almost. Once we can demonstrate that a purely dramatic game is viable both technically and economically, then it's a given that the techniques developed will be incorporated into other genres of gaming, given how competitive the industry is. Interestingly, making existing genres more "dramatic" just might cure them of their overwhelming sameness, which might draw persons like myself back to FPSes and so forth.

Quote:

Original post by Oluseyi Quote: Ah, then you're using the wrong word. I think you want 'consequences.' That's debatable. Some consequences are penalties, but not all penalties are direct consequences of user action. Shoot (or offend, for a more benign variant) a character and you may lose the ability to solicit information from him. Miss a train, however, and you may lose the ability to interact with a certain character who was going to be in town at a certain time - but that's not a direct consequence, since you didn't know of him at all. The latter is clearly a penalty, not a consequence.

I never said direct consequences. The problem I have with the word 'penalty' is that it's exclusively negative. You talked about penalties as being "how the system responds to a user's choices in terms of how to solve a certain problem" - that response could equally be positive (e.g. the player is rewarded for successfully resolving the problem without spending any money, or without killing any people). But we're getting into semantics and away from the point.

I'll oblige and get back to what the rest of you want to talk about [wink]

An idea I've thrown around before is a very simple graph of concepts and opinions. Take Joe NPC, living in his house on Bitstream Avenue. Joe, his house, and Bitstream Avenue are all concepts in this graph, as well as other things - the sky, his neighbours, Jane NPC - and more abstract things, like God, or violence, or betrayal. Joe has an opinion on all of these things, a set of feelings (which may or may not be mutual) - God is good, betrayal is bad, Jane NPC is sexy. These opinions will not only govern the actions he decides to take himself, but also how he reacts to other events - events that may not involve him directly. If I walk into the room and loudly proclaim that God is bad, Joe begins to build a picture of me, mapping out the opinions I hold on concepts - and, depending on how similar my opinions are to his, will balance and adjust his opinion of me accordingly.

The real flexibility of the system lies in what constitutes an 'opinion', and that's not something I've figured out yet. The early demos of the system that I've built simply used a float value from one concept to another, where negative values indicate 'bad' opinion and positive values indicate 'good' opinion; I had two NPCs who liked each other, I slapped one, and the other one didn't like me for it. However, it's a far too simplistic a model - say we're talking about Joe and Jane NPC again, and I waltz in and give Jane NPC a passionate kiss. Joe observes this, and infers that my opinion of Jane NPC is highly positive; so is his, so we'd get on great (Hilarity, polygamy, and threesomes ensue). Poor slighted Joe has no way to differentiate between respect for someone and sexual relations with someone. There are other similar problems that result from only having a single number representing an opinion. So I think a more useful model for an opinion will probably have a set of emotions ('love,' 'envy,' 'lust,' etc) and weighting values for each.

Also central is the idea of the 'incident.' When I'm french-kissing Jane NPC, a new, temporary concept is created in the graph, representing that action. Joe can thus form an opinion on that action, can remember it, can change his feelings on it later (perhaps after a little heart-to-heart with Jane NPC). Incidental concepts also provide for 'rumour' type systems - other NPCs, who didn't even witness the event (perhaps do not even know of one or more of the participants) can form opinions on it. "Hey, did you hear about what Jane NPC's been up to? She's a right slut..." And of course, the way that NPCs tell each other about an event will be influenced by their opinions... "The guy kissed her, but she was all fighting and pushing him away."