Procedural content and emergent gameplay, a brainstorming

[TheGreatGodPan]

I think the procedural world sounds a hell of a lot more fun than the pre-scripted shit we've had to deal with.

 

[Mangler]

http://www.jabberwacky.com

Might be good for a crazy person you have to get some information out of...
:wink:

Otherwise most important characters have full written dialog, with the occasional AI written support lines to 'fill in' the blanks.

Also to try and handle situations not covered by canned dialog.

Somebody would otherwise need a horde of good writers, trying to cover every situation. Which they can't because there are far too many variables in these systems.

Procedural systems just arn't good enough for the 'Main' plot, however they are perfect for lots of small 'fed-ex' type quests and stuff.

AI technology just isn't quite there yet.[/u]

 

[galsiah]

In your example, you created a very nice chain of logical straightforward events, which anyone could figure out. No food on the markets->the supply chain is cut->Bandits. From players perspective it's fairly easy to figure out what's the first step. However, procedural content might create situations where the player will be at loss. For example, the same farmer might complain his tools got stolen. It just might be the case of a perfect crime...nobody will be able to give clues because the thief successfully got away with it.

Sure - if it's the perfect crime, then perhaps the player won't make progress, and will need to quit. The idea that the player character should always be able to win at everything only seems reasonable because we're so used to that being true. If players stopped thinking of role playing games as games, but instead as worlds, then they should expect not to be able to solve every problem.
If success is always possible and straight-forward, it's less rewarding.

Even in this "perfect crime" situation, the player is not really lost: as soon as he gets the information that there has been a theft, he has leads - even if it just means asking neighbours, guards etc. Just because there is no solid clue does not mean the situation has no structure - the fact that something has been stolen provides structure on its own. Presuming the player knows what was stolen, he might also guess at a motive - e.g. jewels might have been stolen for sale, so ask at jewelers. If the item has little resale value, perhaps the reason is personal, so ask NPCs the victim knows first...

Perhaps the player's enquiries will turn up nothing, but he can still go about investigating with a sense of reason and purpose. I think it would be more rewarding to be making focused, perceptive enquiries, but come up with nothing, than to randomly talk to every NPC until you bump into a quest trigger - which is all too often how things proceed at present.

What is rewarding / frustrating for the player is the process - not the goal. A player will become frustrated when he has no way of knowing if he's doing the right thing, or if he is getting anywhere. A player who investigates a theft by asking reasonable questions of relevant NPCs knows that he's doing the right thing (presuming he is aware that there is no arbitrary "trigger" to find). He is also making progress, since after he has exhausted all leads, he can deduce that there is no way to find the criminal, and simply move on. There is no need to become frustrated that he can't find the trigger, since there isn't one. He hasn't failed as a player / character, since there might well be no solution - perhaps the thief was just really good.

I'm fairly sure things could be arranged so that problems were rarely impossible to solve - at least when it seemed reasonable to the player that he might be able to solve them. Having things occasionally too hard / impossible / impractical to solve doesn't have to be a bad thing though - so long as the player is aware of this possibility, and he has a clear idea of ways to attempt to solve the problem.

One of the main reasons players become frustrated when they fail is because all the things they tried were, to their certain knowledge, pre-determined to fail. There is only one trigger, and until you find it, you are wasting your time. With procedurally generated, need-satisfaction goals, this is not the case - and it will not seem the same to the player. He has not been wasting his time trying possible solutions, since any of them might have worked - and indeed may work next time.

As a matter of fact, all your examples seem highly theoretical to me. Frankly, I was talking about the practical side of procedural content, as something that is possible to implement today.

So am I. What I've said isn't easy to do, but it certainly isn't impossible at present. (voice acting aside)

Unless I'm wrong, I see absolutely no way to implement the system you are talking about...making npcs act upon a set of needs, is scripting.

Not really, but that's not your point, so I'll move on .

Making them come up with a dialogue isn't. In your example, I really have no clue how to make the city merchant tell the player about the cut supply lines, or how the military will share about the bandits’ attacks from the west.

Of course this sort of thing needs to be prepared, but it's not really that hard. I was careful to set out everything in very general terms.

For a merchant, he can fairly simply be set up to answer any of the following:
Do you sell X?
Where do you get your X?
Do you ever receive deliveries from X?
When did you last receive a delivery (of X)?
When will your next delivery (of X) be?
How many deliveries (of X) do you get per week (/month)?
Does X ever come here to shop?
Do you open on day X?
What is your daily turnover?
What are your opening hours?
Have you had any unusual deliveries recently?
Are you due to receive any unusual deliveries soon?
etc.

As well as general questions, such as:
Have you seen X?
Do you know X?
When was X last here?
Has anything odd happened recently?
etc.

All that's necessary for the above is for merchants to store information regarding deliveries, visitors, purchaces, trading partners... along with the items in question. Answering the above just involves filling in the gaps in various stock responses. Do you get wonderfully interesting, poetic dialogue? No - but you can certainly get the plain facts.

Organising the player's range of possible questions might be a bit tricky, but it's not rocket science either. You'd probably just need a few subcategories, and the ability to fill in "variables" in each question - either from a drop-down menu type interface, or through typing.

For example, asking a merchant "When was your last delivery of grain?" could involve:
Talk to merchant
Select "Enquiry"
Select "Delivery"
Get options:
(1) When was your last delivery (of...)?
(2) When will your next delivery (of ...) be?
(3) How many deliveries (of X) do you get per week (/month)?
(4) Do you ever receive deliveries from ...?
(5) Are you due to receive any unusual deliveries soon?
(6) Have you had any unusual deliveries recently?

Select (1)
Fill in gap with "grain"

And you've asked.
The merchant can then check his delivered items list in reverse order, querying for "grain". The first time he finds it, he can return the time, and perhaps the amount received. If he doesn't find it, he could check his list of scheduled deliveries, and based on whether he should be getting deliveries, either say something like:
"We never get any grain deliveries."
Or
"We haven't had any grain deliveries for at least <length of delivery records>, which is odd, since usually we get ... per ..."

Of course, if the merchant didn't like the player character, he might merely tell him to mind his own business, or something similar.


The above is certainly possible. I'm sure it's not the best / smoothest implementation, but it would probably work to some degree. You would get simple answers most of the time, but that's usually what you'd want. Given that there would be no "trigger" dialogue, the player wouldn't be tempted to take a Morrowind click-on-everything approach to dialogue. He'd know that the only results of his enquiries would usually be a direct answer to his question, so he'd only ask something if he wanted the answer.

The answers wouldn't be unique in style, but they would be specific to the individual the player was talking to. In that sense it'd be better than Morrowind, since it would really matter that you were talking to a certain NPC.

Quantity over quality is really a bad idea when gameplay is concerned. As excited I am about the possibilities of procedural content, it won't make fun games. I was thinking about combining the good old human designed quests with stuff generated by PC, but it seems impossible.

Flowery dialogue != quality gameplay.
I think that involved dialogue is an important part of an RPG, but its absense from one quest does not mean that quest has bad gameplay. Procedural quests would just be one style of content, in the same way that combat is another style of content. Procedural quests are not better or worse than hand written quests - they are different.

I don't think a good RPG could rely entirely on procedural quests, but neither could it rely entirely on combat. That doesn't mean to say that including procedural quests would be a bad idea, any more than including combat would be a bad idea.

The idea that it's impossible to do good procedural quests seems a bit defeatist to me. I agree 100% that it's hard to do even reasonably, and very hard to do well. That doesn't make it impossible.

Many current techniques in real time computer graphics or physics would have seemed impossible ten years ago. Why are they now possible? Because a lot of people have spent a lot of time working hard to come up with new methods to attack the problems. If the same amount of effort had been put towards generating procedural content in RPGs, without results, then I'd agree that it seems impossible. It hasn't, and I don't.

Doing procedural quests on a small scale with the fulfillment of NPC desires as goals seems quite a reasonable target. Given that it has rarely been tried, I don't see the lack of current results as any indication of its impossibility.

Why do you think the players want this? It just doesn't make sense, to wander around in a world full of people with problems (ranging from mundane to epic), some of them might be irresolvable...I mean, where is the fun?

First, please allow me a brief (well it was supposed to be brief) semi-rant:

Why must people persist in the idea that the purpose of games is "Fun". The idea that games like Tetris, Mario, Zelda, Fallout, Tekken, Pro Evo Soccer, Total War, Civ, Space Empires... all have the same goal is utterly insane.

Games are entertainment - that is not the same as "fun".
The "fun only" era of games (if it ever existed) went out with pacman, pong and space invaders. The games industry has matured since then, and spawned a variety of genres, each with a different style.

Allow me to compare things to the film industry. How many of the following films would you describe as "fun"?
One Flew Over the Cuckoo's Nest, Barton Fink, The Remains of the Day, Boys in the hood, Memento, Casablanca, The Deer Hunter, Do the Right Thing, Twelve Angry Men, Le Mepris, The English Patient, American History X, Training Day, The Usual Suspects, The Pianist, Leon, Arlington Road, Seven.

I'd say that all of the above films entertained me. I'd probably call some enjoyable. I doubt I'd use the word "fun" to describe any of them - and they're not all similar, by any means.
If you said "You can't put that in a film, since films are supposed to be fun.", people would laugh at you. I don't see that saying the same thing about games these days is much more sensible.

Most of the time I play Civilization, I am staring at a static screen, filled with outdated graphics, thinking and planning. This is challenging, and entertaining, but not "fun" (not most of the time anyway). When I play a Mario game, Doom, or a sports game, I'm usually looking for a bit of light entertainment.

An RPG is somewhere in between. It is certainly not something I play for a bit of a laugh. The "It's not fun, so it doesn't belong in the game" argument does not work for an RPG, since RPGs are about more than light-hearted fun. An RPG should put me in the role of a character, and allow me to feel part of a cohesive world.

A world where I can solve every problem is not convincing. A world where an NPC only raises a complaint if it is possible for me to solve it quickly, is not convincing.

Why am I playing then?

You are playing because you want to feel you have a role to play. You want to have to make important and influencial decisions. You want to display your character by making tough choices which don't make everyone happy. You want the NPCs in the world to seem as though they have lives, and to see you as a character in the world, not as a Super-All-Purpose-Problem-Solver.
You don't want to be able to do everything, and you certainly don't want everything to be easy. You want to display your character's unique strengths by succeeding where non-experts in that area would surely fail.

Perhaps most importantly, you want to do all this in a world that convinces you - that draws you in to another reality. Not a world that seems to be constructed around you to give you a series of carefully paced challenges. Rather a real living world where you are one of many individuals. A diverse world with many complex problems.

Maybe it is your destiny to save the land from some dark peril. However, it should not be your destiny to turn the world into a wonderous fairy-land where every NPC loves you and wants for nothing, since you have satisfied every one of their needs.

It is not a question of "realism", but of coherence. Wonderous, perfect fairy-lands, where the player is the saviour of everything and everyone, and can always succeed at everything, do not seem convincing. They are not coherent worlds, so the player is less likely to be drawn in to their "reality". If you stop treating the player like superman, you at least provide the chance for him to feel part of a reasonable world.

 

[crufty]

I've often thought that would be the ultimate nethack...make the QUEST itself procedural. Rather than always going down to get the amulet of yendor, what if it was???? hint hint for anyone looking to take a crack at it

 

[AlanC9]

Perhaps the player's enquiries will turn up nothing, but he can still go about investigating with a sense of reason and purpose. I think it would be more rewarding to be making focused, perceptive enquiries, but come up with nothing, than to randomly talk to every NPC until you bump into a quest trigger - which is all too often how things proceed at present.

Wait a second. Even in a scripted game, players typically don't randomly talk to all NPCs in hopes of finding the quest trigger. You look for the trigger where the trigger is likely to be found, which happens to be with the exact same people you'd be talking to if the quest was procedurally generated. It's reason and purpose either way; one way you're trying to guess the designer's intent, the other you're trying to guess what the procedural algorithms did this time. And you could brute-force the quest the same way too, regardless of how it was generated.

One of the main reasons players become frustrated when they fail is because all the things they tried were, to their certain knowledge, pre-determined to fail. There is only one trigger, and until you find it, you are wasting your time. With procedurally generated, need-satisfaction goals, this is not the case - and it will not seem the same to the player. He has not been wasting his time trying possible solutions, since any of them might have worked - and indeed may work next time.

This doesn't seem convincing to me at all; perhaps we have a different psychology. The failed actions are pre-determined to fail whether a designer's scripting or procedural algorithms generated the quest. I just don't see one being more frustrating than another -- unless the procedural quest really is insoluble, in which case I am wasting my time right up until I gave up. Gave up in frustration, that is; I can't imagine not being frustrated in that situation. Failure is never enjoyable even when success isn't possible.

I don't think a good RPG could rely entirely on procedural quests, but neither could it rely entirely on combat. That doesn't mean to say that including procedural quests would be a bad idea, any more than including combat would be a bad idea.

The thing is, you're really playing a different game when doing procedural quests from when you're doing scripted quests. I'm not sure it's such a good idea to mix both in the same RPG. Though this means that having generic dialog in the procedural quests is a good thing, since the player always knows what kind of quest he's working on.

It is not a question of "realism", but of coherence. Wonderous, perfect fairy-lands, where the player is the saviour of everything and everyone, and can always succeed at everything, do not seem convincing. They are not coherent worlds, so the player is less likely to be drawn in to their "reality". If you stop treating the player like superman, you at least provide the chance for him to feel part of a reasonable world.

I think this obscures something fundamental about the distinction between fiction and reality. Fictional worlds have coherence precisely because they are designed. It's just that most of the worlds in RPGs are designed badly. No objection here, but then we're back on the old question of how we keep the procedural stuff from breaking the designed stuff. Of course, you can get around this by going sandbox.

Or how about procedurally generating story elements? I'm surprised nobody's brought this up (or did I miss it?). We can conceptualize fiction about as well as we can conceptualize real human actions, maybe better. Fiction seems to be more popular than reality, so instead of trying to simulate a real world, why not try to simulate fiction?

 

[Nog Robbin]

I guess the other question would be how efficient could the system be, how many characters could it maintain, and at what sort of level.

The system used in OB is supposed to keep people moving according to their schedules even if the player isn't there - but at a far lower level of detail. The actual action only takes place if the player is near enough to potentially see it.

So - suppose an NPC has a job, and to do that job they have a single required item. If the item breaks or is missing they must find the item before they can do their job. How long would that NPC use trying to find/replace his missing/broken tool before he gives up? What steps should he take to find it (if missing)? At what cost level is it even worth his time not just replacing it (assuming he can buy one)?
Can NPC's change job? Is "guard" a job?

Would any of the above be possible if the player wasn't near enough to witness it - or at that level does it just boil down to essential needs? Are other details simply assumed?

 

[Zomg]

Wow, galsiah, those posts are making me feel like I'm listening to the St. Crispin's Day speech.

Or how about procedurally generating story elements?

Ever heard of the concept of emergent narrative? Think of playing a game of Civilization - for any given game, I'd bet you could write a compelling two or three page mini-history of that world, a virtual Guns, Germs, and Steel. Stories are surprisingly self-organizing in procedural and reactive worlds - just try to play X-Com, a game with zero elements of literature, without a story popping out at you.

 

[Stark]

the discussion thus far has been about generating basic NPC needs and letting it act out on its own, and hoping in the process that it will generate some coherent quests out of it.

the flaw with this approach seems that it
1) does not guarantee compelling gameplay--most probably end up generating very simple one step quests (fetch me food/retrieve my sword)

2) implementation issue--when to stop the simulation?

i am surprised no one has mentioned a far more effective way of generating random quests: come up with a set of basic quest templates, allow them to mix and match, and assign random actors to it. Also generate random paths to solution.

the trick is making sure the end result of the mix and match of these quest templates make coherent sense. the quest templates are allowed to mix and match for two reasons:
1) the number of possible quest types would be alot more than the sum of the quest templates
2) allow element of surprise to player (one template may switch to another quickly)

3) generic but hand crafted dialog with NPC may still be possible

example:
You went to fetch an item from an NPC (fetch quest template) only to discover he is murdered (murder quest template with one witness) which turns into a guild rivally case (guild rivally template). You can then inflitrate/storm/bribe into the rivalling guild to steal/retrive the item and kill/persuade the guild rival.

I believe implementing this way generates far better result than giving emotions and needs to NPCs...

start with a 20 or so basic set of templates, and allow random numbers to generate them out. obviously some templates do not generate well with others. as i said, the challenge is in making sure the generated quest is coherent and makes sense, without illogical twist and turn.

 

[Zomg]

The point of AI needs and similar types of systems is not generating quests, but rather establishing reactivity.

 

[Stark]

The point of AI needs and similar types of systems is not generating quests, but rather establishing reactivity.

but most of what's discussed before is about generating quest out of "AI".

anyway, reactivity without a quest tied to it would be pretty pointless too. there's only so much one can admire the reactivity before getting bored with it (think Fable). for it to be useful, it should generate a quest for player to interact with.

 

[Imbecile]

I believe implementing this way generates far better result than giving emotions and needs to NPCs...

.

I agree. Often fudging the gameworld can provide a much more convincing set up, that is far less like to go wrong (and for a a lot less work) than going the whole hog on NPC AI based on http://en.wikipedia.org/wiki/Maslows_Hierarchy_of_needs, or whatever.

As a basic example, trying to get a decent working economy based on supply and demand would be pretty damned difficult. Presumably you would need some items that are consumables (food) that people can continue to produce, sell, buy and consume. How would this work with swords and armour though? If weaponry cant be produced they cant make money, If they can be produced, they need to eventually disintegrate otherwise you will end up with loads of weapons selling for 0 gold. Not to mention the issue of shops with no items, or economy crashes.

It all seems like a lot of work that could effectively be replaced by an algorithm saying something like:

Each town has a fixed set of prices for goods that rises and falls between 80 and 120% of the base value. Every 24 hours these values shift up or down by 1-3%. At 100% there is a 50% chance of the price rising or falling. At 90% of base price the movement is 75% likely to be upwards and 25% likely to be downwards. The base prices would be logical so mining towns would have cheaper gems and weaponry, and ports would sell cheaper (generally speaking) fish, spices etc..

Random quests are definitely a good idea – but why not in conjunction with the tailor made quests?

Why not have various people who will continue to offer you random quests. A merchant house (deliveries, finding general items), An alchemist (finding random ingredients), A lawman (clearing out bandits, escorting people). I'm sure that these could be added relatively simply to an existing RPG. Sure it wouldn’t create a living breathing world, but it would help – and it would be feasible. Hell,GTA already offers a form of this in the guise of Ambulance, taxi, fire engine and gang missions.

 

[Allanon]

If success is always possible and straight-forward, it's less rewarding.

Frankly, from the gameplay point of view, I see no reason to introduce quests that aren't solvable, unless we are talking about specific simulation of world events. Think about what procedural content might bring into the gameplay, by enhancing and improving it, instead of concentrating on the notion of real time generated quests. From that perspective, meaningful game time is more important than quantity or realism.

I think it would be more rewarding to be making focused, perceptive enquiries, but come up with nothing, than to randomly talk to every NPC until you bump into a quest trigger - which is all too often how things proceed at present.

My personal opinion is that failure to complete a task can never be rewarding, especially taking into consideration that it might have been unsolvable in the first place.

What is rewarding / frustrating for the player is the process - not the goal. A player will become frustrated when he has no way of knowing if he's doing the right thing, or if he is getting anywhere.

Exactly. You might just be wasting your time by trying to solve a quest which has no solution...and you'll never know until after investing a couple of hours into investigation.
That might be quite frustrating.

Having things occasionally too hard / impossible / impractical to solve doesn't have to be a bad thing though - so long as the player is aware of this possibility, and he has a clear idea of ways to attempt to solve the problem.

I seem to believe that reward is a major part of role playing experience, and when you decide to delve into unrewarding activities just for the sake of it, it takes out the fun (entertainment). For example, in OB you have tens of dungeons all around the land. However, I couldn't make myself continue to explore more than 5 of them because I got no satisfactory rewards out of it. As much fun as it might be wandering around dark hallways, it just doesn't make sense unless I'll be rewarded with something more significant than a carrot and a repair hammer.

There is only one trigger, and until you find it, you are wasting your time.

That's the same with procedurally generated content as well. You walk around wasting your time until you hit that specific clue you were lacking. At least the designed quests point you in the right direction to conserve time and enhance the experience (Unless it's a quest game, where pixel hunting is supposed to be fun).

For a merchant, he can fairly simply be set up to answer any of the following:...

And that's just for his wares! Just think about the overall number of answers he should be able to answer from all possible topics. :shock:

I think that involved dialogue is an important part of an RPG, but its absence from one quest does not mean that quest has bad gameplay. Procedural quests would just be one style of content, in the same way that combat is another style of content. Procedural quests are not better or worse than hand written quests - they are different.

Indeed, if you will be able to find a way to implement dynamic quest generation with static, dialogue based quests it would be great. I'm not sure it's possible.

Allow me to compare things to the film industry. How many of the following films would you describe as "fun"?

All of them. They were entertaining enough for me to get some "fun". If I didn't find the entertaining, I wouldn't watch them.

I'd say that all of the above films entertained me. I'd probably call some enjoyable. I doubt I'd use the word "fun" to describe any of them - and they're not all similar, by any means.

Then we simply have different meaning for the word "fun". Entertainment is fun for me

You are playing because you want to feel you have a role to play. You want to have to make important and influential decisions.

That's exactly the reason quests should be well thought compact adventures with high entertainment value and rewarding experiences. I don't need simulation of real life to the extent of dead ends and mundane tasks.

 

[Section8]

Ooh, discussion. I'll come play in the sandbox tomorrow, but I'm still recovering from a fairly intense week of debauchery. :D

 

[galsiah]

Wow, galsiah, those posts are making me feel like I'm listening to the St. Crispin's Day speech.

I might have got carried away .

I'll get this out of the way first:

All of them. They were entertaining enough for me to get some "fun". If I didn't find the entertaining, I wouldn't watch them.

Are you serious? Have you watched most of those films? (I'm not suggesting that you're inferior if you haven't - I'm sure you've seen films I haven't. I just don't understand your warped idea of "fun" )

I think it'd be reasonable (if pointless) to debate whether some of the films I mentioned are "fun" - e.g. Seven, Arlington Road...

However, some are simply not "fun" for any reasonable definition of that word. "The Remains of the Day" is in no sense fun. You don't walk out after seeing "The Deer Hunter" thinking: that was really fun.

If you think these films are "fun", then you have a very strange idea of that word - one that disagrees with the conventional meaning. Fun is amusing and playful. Watching Christopher Walken blow his brains out is not fun (for most people ).

Entertainment is broader than fun, and can encompass things which would not be called fun. Many parts of films are not fun in isolation, and wouldn't even be entertaining alone, but can contribute to the overall effect of the film.

It is useful to have a general word "entertainment", and a more specific word "fun", precisely because they do not mean exactly the same thing.

Then we simply have different meaning for the word "fun". Entertainment is fun for me.

I really think you are just plain wrong here.
If a film is tremendously moving, and leaves you in floods of tears, you have been entertained. You are most definitely not having "fun" in any usual sense of the word.

Similarly, many games offer more than a Mario style of fun. It is reasonable to draw the distinction - particularly when one feature is isolated by a critic as being "not fun", and therefore less worthy of inclusion. Not every feature of a game needs to be immediately and obviously fun. Anyone who thinks it does will come up with some very narrow games.

Games just need to provide a level of overall entertainment to give value to the player. A good game does not necessarily mean that the player will be grinning at the screen half the time.


Right, moving on:

Wait a second. Even in a scripted game, players typically don't randomly talk to all NPCs in hopes of finding the quest trigger. You look for the trigger where the trigger is likely to be found

True, but the important thing with scripted quests is usually that talking to NPC X triggers event Y. Even if the player has worked out the solution to the quest, or has guessed it, he has to find the trigger. Often he'll find it easily, but sometimes he won't be thinking like the quest designer - maybe he's found a short-cut the designer didn't think of. In that case he'll be frustrated as he needs to go back and talk to a load of useless NPCs, not because his character needs to, but because the quest needs triggering.

With a procedural need-fulfillment quest, things are different. There are no arbitrary quest stages or triggers. The player talks to NPCs until he has the information he needs (how much information will depend on the player, not quest triggers). He is then free to continue.

It's reason and purpose either way; one way you're trying to guess the designer's intent, the other you're trying to guess what the procedural algorithms did this time.

But you don't need to guess what the procedural algorithms did, because the procedural algorithms are just part of the way the world works. All those algorithms do is give NPCs needs and provide interactions which may fulfill those needs.

If an NPC wants to find another NPC, or some food, or to have a threat removed etc., there is no need for the player to follow any predefined trail generated by a procedural algorithm. He just needs to follow reasonable steps to achieve the goal - but those steps are up to him.

It is much more frustrating for a player to know exactly what he needs to do, but be unable to do it - because some arbitrary trigger has not been found -, than for a player to become stuck because he runs out of ideas.

The former sharply disconects the player from his character - since the player knows what he needs to do, but the game does not let the character realize until some trigger is activated. The latter connects the player and the character, since both are stumped for the same reasons - the character doesn't know what to do precisely because the player doesn't know what to do.

Out-of-character frustration is something to be avoided, but the same is not necessarily true of in-character frustration. Importantly, the second case also makes reasonable sense - perhaps this problem just has no easy solution, or the player needs to re-think things. The first case makes no sense, since the only obstacle is an arbitrary trigger which prevents the character knowing what the player already does.

And you could brute-force the quest the same way too, regardless of how it was generated.

With the example I gave above? How would you brute-force a system which allows you to ask over 10 question types, for any item of any merchant. If you've got only 10 question types, 10 merchants, and 100 item types, that is 10000 possible questions (this is probably a significant underestimate). No-one will brute force that.

Also, the main reason for brute forcing quests is when the player is attempting to find some arbitrary trigger he doesn't understand. This simply won't work for the type of case I'm describing: If the player asks "When did you last see NPC X?", he knows precisely what sort of answer he'll get. He will also have the opportunity to ask this of any NPC - there will be no chance to examine NPC topic lists for clues. The player will only ask questions he wants the answers to, since he'll only get straight-forward, standard answers.

It will be up to the player to piece together information. Brute forcing simply won't work, since the information will only be useful to the player if he knows why it is important. If he knows why some information would be important, there would be no need for brute force.

Of course, you could argue that there might be situations where a "brute force" technique were actually the technique that made sense - but would be boring for the player. For instance, the best way to find NPC X after obvious leads have been tried, might be to ask every NPC in the town "Have you seen X?". While this is different from brute force trigger searching, it still might be boring.

There are probably ways to design around this though - for instance, NPCs could have the potential to reply "I haven't seen X, but I heard that Y saw him", or "I usually see him around Z, perhaps you might try there"...
With relatively few options - perhaps about 10 - of simple information, finding an NPC could be made relatively simple in most circumstances, but still give the player an interesting (ish) problem to solve.

The failed actions are pre-determined to fail whether a designer's scripting or procedural algorithms generated the quest.

It's not the same situation at all. When you fail to find a trigger, you are failing to find some designer's narrow idea of what the next step should be. Perhaps you have a more clever idea, but that won't help you because you need to find the scripted trigger.

When you fail to find a way to satisfy some request from an NPC, you have had the opportunity to try everything you want (within the parameters of the game world). Any of your techniques might have worked, and there is nothing arbitrary about their failure - if they fail, it's because the world is like that, not because a designer didn't think of that solution.

This is a similar situation to failing to kill some creature that attacks you before it gets away. You are free to try any technique to attack it, and some techniques will work well at times and fail at other times. Whether the creature gets away will depend on many factors, including its stats, your stats, the terrain, what attacks you try, random chance...

Importantly though, your success / failure is always predetermined, assuming you act in the same way. Quite often, you will meet a situation where there is no way to be sure of killing the creature before it gets away. Whatever you try might be predetermined to fail. This might get annoying if it happened all the time, but it happening once or twice is not that annoying / frustrating.

Why not? Because it makes sense. There is no limitation (apart from game world rules) on what you can try. It just so happens that with your current skills and the creature's current stats, killing it before it escapes is impossible. It is the same situation with solving puzzles that emerge out of NPC need fulfillment. There is no limitation beyond the game rules on your actions. If you fail, it will be because your character did not have the skill to succeed, you didn't have the intelligence, or the task was impossible. There is nothing arbitrary about such a failure.

Usually I would guess that it'll be fairly clear to the player what tasks he'll be able to complete, and what he won't. I don't see the occasional failure as a problem - so long as the process is usually reasonable and somewhat interesting.

the flaw with this approach seems that it
1) does not guarantee compelling gameplay--most probably end up generating very simple one step quests (fetch me food/retrieve my sword)

I don't think this necessarily has to be a problem.

One way to try tackling this would be to have NPCs only ask the player to help solve problems once the NPC has tried and failed - or judged it too dangerous etc.

Therefore, you'd rarely get "Fetch me food" unless there were a food shortage. "Retrieve my sword" would be pretty unlikely in any case I'd have thought - what need does that correspond to? "Retrieve a sword" is more likely, but the NPC could probably handle that himself, so wouldn't need to ask.

I'd have thought that randomly generating quests such as "Retrieve <unique item>" might be troublesome in a dynamic world. There is no guarantee where that item will be, or that the player will have any way to find it. Such a quest is necessarily linear in one respect - the only way to achieve the goal is by finding that specific item.

The reason that having a goal of "satisfy general NPC need X" is attractive, is that there will usually be many ways to go about this. Finding one specifc item or person might well become impractical or impossible, but such problems as getting food or a type of item, removing danger, providing entertainment, providing shelter... usually admit many solutions. If one is impossible, another will usually work.

i am surprised no one has mentioned a far more effective way of generating random quests: come up with a set of basic quest templates, allow them to mix and match, and assign random actors to it.

This is worth a thought, but I'm not sure it'd end up being more effective.

One of the good features of "quests" that emerge out of a need to satisfy NPC needs, is that every part of these "quests" emerges from standard game mechanics and interactions. Things will automatically make good sense in the game world, since the problems arrise directly from game world rules. [whether the problems turn out to be interesting is a different question, of course]

If you have predefined entire quest templates, things could quickly get boring. You can solve this to a degree by mixing different sections from various quest templates, but that does run the risk of things not seeming convincing in many cases. If you construct things carefully, most / all the final quests might seem reasonable, but there is no guarantee of this - it would be up to the designers to make sure elements fit together reasonably. This would be a lot of work.

Again, you'd have a trade-off: the simpler and more generic you make the pieces, the more easily you'll be able to fit them together, but the less interesting the final quests might be. Make the pieces more individual, and they'll be more recognisable, and probably harder to fit with other pieces.

I'm not saying that you wouldn't get this to work well - perhaps it might be the best solution. However, I'd certainly prefer the idea of "quests" which emerge entirely from NPC needs (or higher level needs, such as faction needs / town needs... if these are modelled), rather than quests arranged by putting together a jigsaw of predefined pieces.

Having things emerge naturally just seems much more elegant. That certainly doesn't mean it'll be the most practical of course. Perhaps piecing together predefined sections would be more workable??

 

[Zomg]

I think there should be a distinction between different kinds of, "That would be a lot of work," because almost all of this stuff would be a lot of work. Procedural-emergent content has the potential to provide superlinearly more game per unit work than hand-crafted content or multi-templated HC variants. It has a higher action potential, though.

 

[galsiah]

That's true. I'd certainly prefer the "lots of work" in overcoming high action potential, rather than in painstakingly hand crafting many pieces. However, the handcrafted, or template combination approach is certainly safer from a development perpective. There's just no way to know how high that action potential is until you've invested a lot of time finding out.
Perhaps after a month of development you'd have a great emergent content generation system, or perhaps you'd spend three months and just end up with a mess of complications. Using combinations of handcrafted elements is less inspiring, but more predictable.

Perhaps it would be possible to use a top down (and therefore more predictable) model, but to gradually reduce the granularity - as far as practical - over the development of the game.

So you'd start by splitting quests into two or three sections, converting these to templates, and trying combinations. You'd refine - or generalize - the pieces until you reach a situation where you have passable final quests with many combinations (perhaps restricted, so that a type A first element must lead to a B or a C, but not a D or an E...).

Once you get to that point, you have a workable system - so things are safe in development terms. At that point you try breaking down the elements you have into smaller sections, refining, generalizing, combining...
If this works well, you proceed. If not, you stick with the previous version.

At some point you'd reach a low enough level so that the quest pieces could (hopefully) be tied in to individual NPC's needs / motivations etc. This way you'd be approaching a very versatile solution incrementally, and safely, rather than by going straight for the solution from the bottom up.

I doubt the above would work - things would probably get problematic before you got near the level of individual NPC needs etc. It's worth a (brief) thought though - maybe.

I'd always prefer the bottom up, emergent solution, but then I rarely get anything done, so perhaps it's not the wisest of policies .

 

[AlanC9]

Wow, galsiah, those posts are making me feel like I'm listening to the St. Crispin's Day speech.

Or how about procedurally generating story elements?

Ever heard of the concept of emergent narrative? Think of playing a game of Civilization - for any given game, I'd bet you could write a compelling two or three page mini-history of that world, a virtual Guns, Germs, and Steel. Stories are surprisingly self-organizing in procedural and reactive worlds - just try to play X-Com, a game with zero elements of literature, without a story popping out at you.

Sure, I've heard of emergent narrative. I just don't think it's ever going to be a substitute for the real thing. It can be good in its own way, but never the same.

I'm just questioning why there's so much resistance to the idea of conceptualizing RPG characters as characters, rather than simulated actors in a simulated world. Though maybe this topic should be in its own thread.

 

[AlanC9]

]True, but the important thing with scripted quests is usually that talking to NPC X triggers event Y. Even if the player has worked out the solution to the quest, or has guessed it, he has to find the trigger. Often he'll find it easily, but sometimes he won't be thinking like the quest designer - maybe he's found a short-cut the designer didn't think of. In that case he'll be frustrated as he needs to go back and talk to a load of useless NPCs, not because his character needs to, but because the quest needs triggering.

I still don't see it. If I really have guessed the correct solution to the quest, then I do know where the trigger is. Knowing who has the answer I'm looking for is part of the solution. If I don't know that, then it's the same thing as if I've misinterpreted the world.

Your argument might have some force if RPG designers created "arbitrary" triggers, but that hasn't been my experience. If you have examples, feel free to share.

 

[Section8]

So much to say, and as usual, Galsiah is all over it.

With regard to "safety" and the feasibility of implementing these sort of ideas in a commercial project, I honestly believe there are rewards to be had, though they may be too long term for many developers/publishers to consider. Ideally, once you have a robust and functional emergent system, it can be expanded upon and applied to future titles. Scripted content cannot.

Perhaps it is a safer option for this single game, to opt for scripts and templates, but if the long term plan is to make a continuation of the same series, or similar games then it makes a lot of sense to be thinking of more universal solutions, and creating tools that are powerful enough to exceed what is required of them here and now.

It's like buying a house instead of renting your whole life.

When it comes to generating "quests", I honestly believe that quests should be the first thing to go out the window, when considered in their current incarnation. Quests are basically a checklist, or occasionally a flowchart, and are generally little more than a contrived way for the player to advance the game. It's a pretty weak spine to be propping something up with.

I much prefer the idea that a character has needs, which are fulfilled by a variety of actions, one of those being player intervention. It gets rid of the inevitable checklist where the player is well aware that they're jumping through hoops for a reward and substitutes a need for actual motivation beyond completism, which in RPG term, is fucking brilliant.

For instance, does the average player really give a fuck about half the things they do? If it wasn't presented within such an obvious reward system, would the player ever really bother? The typical scripted quest model attempts to bring some kind of emotional involvement to the table, but it tends to be shallow and non-persistent.

When actions are tied directly into game dynamics, player actions will always have direct and indirect consequences to their actions, from which they can derive their own motivations and emotional connections. If the starving farmer dies because the player turned a blind eye, then that's a fairly simple consequence that might play on the player's heart strings (or tickle their funny bone) but it might also have cascade effects on the guy's family, his hometown, etc. that are all visible to the player.

By virtue of simplicity, being able to set foot in a household where a widow weeps, a child keeps asking if his daddy is going to be alright, may be a much more powerful evocative tool than being able to talk to the widow who says "Woe betide me! That foolish husband of mine never should have left to fight the orc army! Who will provide for us now. Now, leave me with my grief!"

Oops, straying from my point of rigid quest structure being done away with. In it's place, the player should be able to make a note of lines of conversation, in much the same way Space Rangers 2 lets the player send fragments of info to their journal bar. I can't emphasise enough how a informal style of questing would contribute to RPGs.

When it comes to dialogue, obviously the sort of scope being bandied around in this thread would require far too many interrogations and responses to be applied to a reasonable dialogue tree method, which is a disappointment, but I actually believe the shortcoming of systems that have typically employed keyword and wiki systems fail for other reasons than their core design.

The biggest flaw I see, is a genericism of NPC responses, when it would be simple to procedurally generate some distinct answers to player inquiry. For instance, take Morrowind:

Player: my (read: your) trade
NPC: I'm a farmer. <insert wikipedia entry on farmers>

...when the same system could easily incorporate more individual responses.

Player: my (read: your) trade
NPC: I'm a farmer. <insert NPCs "opinion" on his trade>

...where the opinion reflects the farmer's needs with regard to his job. Maybe business is poor - "It's pretty rough going right now. Crop yields are lower than we expected." Maybe he's just fed up with the whole deal - "I never wanted to do this for a living. I always wanted to be...a LUMBERJACK!" etc.

Basically, you're killing two birds with one stone. You're taking steps to make otherwise generic NPCs distinct, and you're also making the NPC's needs partway visible to the player. It may not be as compelling as decent dialogue trees, but I think the the game would amply compensate in other areas.

Something else the would be essential, and I don't think anyone has really thouched on it yet, is the idea of community needs. For instance, a typical NPC might work to satisfy his family in addition to himself. Leader types might work to manage the needs of their social groups (guilds, etc.) or on a broader scale, municipal admin type stuff, and so on upward. It would be a necessity to define various spheres of influence. From the personal, to the deific.

And that sort of leads to the inevitable conclusion, that the biggest task in designing and implementing these sort of systems would be controlling the scope of the systems. how much, or how little creates an acceptable closed system that the player can interact within?

More thoughts later.

 

[Vultok]

I still don't see it. If I really have guessed the correct solution to the quest, then I do know where the trigger is. Knowing who has the answer I'm looking for is part of the solution. If I don't know that, then it's the same thing as if I've misinterpreted the world.

Your argument might have some force if RPG designers created "arbitrary" triggers, but that hasn't been my experience. If you have examples, feel free to share.

Happens to me a lot. More power to you if you understand exactly how the developers implemented some specific quest, but I recall having issues with ToEE at the least. This sort of thing often happens when you're conceptually aware of how the quest should be solved, but you're not quite sure how the developer broke it up into pieces. If you've never run into the lazy "overarching quest broken up into a series of fed-exes" game design approach, then you won't know what I'm talking about.

Off the top of my head, I believe I ran into this problem in ToEE, Fallouts, and PS:T. Try looking at those for some examples.

 

[Stark]

The reason that having a goal of "satisfy general NPC need X" is attractive, is that there will usually be many ways to go about this. Finding one specifc item or person might well become impractical or impossible, but such problems as getting food or a type of item, removing danger, providing entertainment, providing shelter... usually admit many solutions. If one is impossible, another will usually work.

think of me as the devil's advocate. i'll be saying negative things about your idea not because i dislike it, but simply because i want to get discussion going, and i enjoy this thread.

I believe generating "satisfy general NPC need X" quest out of NPC needs would generally be single step quest.

99% of them would not be very compelling in terms of quest structure. this is necessary so, because in order to generate more intricate quest structure the interactions between NPCs and the world would have to be simulated at an increasingly exponential detail level, which is computationally unattractive and effort vs return diminishes very quickly.

One of the good features of "quests" that emerge out of a need to satisfy NPC needs, is that every part of these "quests" emerges from standard game mechanics and interactions. Things will automatically make good sense in the game world, since the problems arrise directly from game world rules. [whether the problems turn out to be interesting is a different question, of course]

the more you allow NPC and world interactions, the more it can breakdown. imagine after 20 hours of gameplay player arrive at most cities deserted: NPCs killed off each other, wandered off to forest to search for food, etc.

to prevent this, you ended up simulating more world rules, which in turn breaks down in more unexpected manner... and you coded more rules to prevent that, and...and so on. I think you get what i mean.

If you have predefined entire quest templates, things could quickly get boring. You can solve this to a degree by mixing different sections from various quest templates, but that does run the risk of things not seeming convincing in many cases. If you construct things carefully, most / all the final quests might seem reasonable, but there is no guarantee of this - it would be up to the designers to make sure elements fit together reasonably. This would be a lot of work.

I agree. there's the challenge of creating sufficient number of base templates, and allow them to interact with each other in a convincing manner to generate sufficient number of quests that do not feel like a repeat of one another to the player. still, i believe it is far more implementable than your alternative.

Again, you'd have a trade-off: the simpler and more generic you make the pieces, the more easily you'll be able to fit them together, but the less interesting the final quests might be. Make the pieces more individual, and they'll be more recognisable, and probably harder to fit with other pieces.

agree. however the same argument applies to NPC need generating quests too. see above.

 

[muds_animal_friend]

- Are there any procedurally generated concepts that you feel... would greatly enhance your experience with RPGs?

No matter how sophisticated the algorithms behind content generation systems they remain mere labour saving devices and nothing more. Such systems’ output replaces superior human authored content allowing the world builders to increase scale due to lowered resource costs. As an end user I shouldn't care how the content is generated but rather what kind of content - in terms of both quality and quantity - is delivered.

For example, in concept the sheer scale of The Elder Scrolls games always excited me but in actual execution the games tend to fall flat on their arses. Why? Because the content ultimately proved dull. Did it matter how the content was generated? No, Daggerfall was mostly built by machine and Morrowind by man. Framing the debate in terms of systems design and engineering is to focus on how the game is built and not on what kind of game should be built. It's very obvious that Oblivion fell into this trap at some stage during its development. Bethesda was so focused on designing a system to create a 'challenge' relevant to a PC's level that they utterly lost sight of levelling-up’s fundamental purpose in the first place.

Ever heard of the concept of emergent narrative? Think of playing a game of Civilization - for any given game, I'd bet you could write a compelling two or three page mini-history of that world, a virtual Guns, Germs, and Steel. Stories are surprisingly self-organizing in procedural and reactive worlds - just try to play X-Com, a game with zero elements of literature, without a story popping out at you.

This, I believe, is the way forward for CRPGs (or rather what may replace CRPGs), the concepts are much lower level than NPC AI or algorithmic quest generation and provide huge range of possibilities within a completely fixed - and thus robust - system.

 

[Section8]

the more you allow NPC and world interactions, the more it can breakdown. imagine after 20 hours of gameplay player arrive at most cities deserted: NPCs killed off each other, wandered off to forest to search for food, etc.

to prevent this, you ended up simulating more world rules, which in turn breaks down in more unexpected manner... and you coded more rules to prevent that, and...and so on. I think you get what i mean.

It's just a matter of prototyping. Since everything is being done programmatically, you could set up time-accelerated simulations of the entire world with tweakable variables to fine tune the systems. All that's really needed are events to emulate the impact a player might have, and I think any developer would jump at an opportunity to be the guy running tests.

"Hmm, I wonder what happens if I suddenly kill all of the farmers." It's a game in itself. :D

Otherwise you could test every negative feedback system you have to determine the breaking point, and find ways to cap the players ability to impact them just below that.

 

[Allanon]

However, some are simply not "fun" for any reasonable definition of that word.

As I've said, my definition is different and thrives from the fact I'm not natural English speaker. For me it holds a somewhat different meaning, closer to "entertainment" in English, just as I warned you in my last post. Too bad you've decided to concentrate on this nuance, instead of debating other points in my post. Quantity over quality again?

Basically, you're killing two birds with one stone. You're taking steps to make otherwise generic NPCs distinct, and you're also making the NPC's needs partway visible to the player. It may not be as compelling as decent dialogue trees, but I think the the game would amply compensate in other areas.

Even though you idea is better to barebones wiki, it's still lacking in comparison. Fallout 2 and PS:T are good example of what you're be missing. The absence of humor alone makes it all not worthwhile. Perhaps the world will feel more realistic, but I don't think it will be enough to compensate. The more I think about it, the less compelling it looks on it's own. Unless it will be possible to integrate both pre designed quests (along with the witty dialogues) and the generated ones, I don't think it will work for RPGs. Other genres might fit well though.

99% of them would not be very compelling in terms of quest structure. this is necessary so, because in order to generate more intricate quest structure the interactions between NPCs and the world would have to be simulated at an increasingly exponential detail level, which is computationally unattractive and effort vs return diminishes very quickly.

Yes, I was thinking if it's even practical as well. I think it was galsiah who told that it is. However, the level of interactivity between npcs should be so high, we might need an extension cards for AI, just like PhysX does for physics.

 

[DarkSign]

Quote:
If you have predefined entire quest templates, things could quickly get boring. You can solve this to a degree by mixing different sections from various quest templates, but that does run the risk of things not seeming convincing in many cases. If you construct things carefully, most / all the final quests might seem reasonable, but there is no guarantee of this - it would be up to the designers to make sure elements fit together reasonably. This would be a lot of work.


I agree. there's the challenge of creating sufficient number of base templates, and allow them to interact with each other in a convincing manner to generate sufficient number of quests that do not feel like a repeat of one another to the player. still, i believe it is far more implementable than your alternative.

Quote:
Again, you'd have a trade-off: the simpler and more generic you make the pieces, the more easily you'll be able to fit them together, but the less interesting the final quests might be. Make the pieces more individual, and they'll be more recognisable, and probably harder to fit with other pieces.


agree. however the same argument applies to NPC need generating quests too. see above.

In creating such a system there are several resources.

First there are the basic plot structures:
http://nosferatu.cas.usf.edu/lis/lis6585/class/modplots.html

Also, in the past Ive run across pages and pages of lists of types of plots - e.g. brother gets revenge on sister, father sacrifices life for child after ...etc. you get the picture.

Thern there was this old post in the Codex from the designer of Gearhead (which I posted first btw):
http://www.rpgcodex.com/phbb/viewtopic.php?t=11564

Another interesting link on automatic plot generation:
http://nil.fdi.ucm.es/nilweb/papers/2004-gervas-story.pdf

If these and other resources could be worked into a system tied to NPC characteristics perhaps something unique could come out of it.

Like the Gearhead post suggests, if given a chance, players will create their own meaning to the story - again suggested before in this thread. So at some point there will be a viable procedure for doing this