Procedural vs Handcrafted

I think the main difference is "top grossing" vs "best selling" - the business insider list is mostly subscription and stuff. Top sales doesn't include WoW for instance, despite grossing between 8 and 12 billion dollars. It also includes arcade games, which is why space invaders tops the list (inflation adjusted dollars).

That said... any of these lists are totally depressing. Gamers looking at game revenue figures is even worse than film guys looking at ticket sales, at least Gone with the Wind was in black and white... that counts as art ...right?

 

Weren't you originally making a D2 Hack n Slash style game? Would have been cool if you had finished that. D3 is a superb game. I play it often both alone and with my girlfriend and cousin. It's just extremely well done and a lot of fun. I do get burnt out on it from time to time and need a break but I always end up going back to it.

 

Plot, even if the whole thing is emergent.

Might as well say it was slot machines or poker.

 

I donno man, I feel like there's really something kind of different in that situation. I think games like Civ, or the Sims, or Mount and Blade, we are the ones telling the story in those games. It's just kind of different from traditional 'plot' like in mass effect.

 

Gone with the Wind was in color. How depressed are you NOW?

 

Might have to start another thread for this, but there's so much competition in that segment right? It would of been a million times easier, but that's the problem in itself.

 

It still has plot points, or rather you experience plot point that tie into overarching story structures, regardless. You experience a story even if you are playing a character in it basically. Yet again, being told the story is irrelevant, only experiencing it.

 

I'm not certain enough to full on debate the point, but I do think that there's something different that goes into this kind of narrative. From the developers perspective it involves different kinds of tools, techniques and approaches. From the player side, I think that you'd find that these two kinds of story telling appeal to different audiences for different reasons.

Granted, there are many shades of grey and it might not be easy to precisely draw a line (experience is subjective, etc) but I think there is something very different happening psychologically in those two types of 'plot driven narrative'.

 

The same could be said for a psychological thriller and an action story though, and frankly it doesn't matter how the developer sees it. Creating a film versus an animation involve two completely different processes, but the fundamental language of storytelling is unaltered in each case.

If anything the biggest difference in the feel of the two types of stories is in how they affect gameplay. In something like Civ, a new story beat changes the conflicts you need to address, which changes what you need to do in the moment and how you go about doing things for the foreseeable future. Meanwhile, most to all rigid story games create absolutely no change to the core gameplay for the entire game. They are a chain of "go do" where there is never a change to what 'do' is, and this is as true of Mass Effect as it is Super Meat Boy.

 

That's a good point, but Mass Effect has more similarity to either an Action movie or a Psychological Thriller than it does to Civilization. The player in Mass Effect is, essentially a passive consumer of the plot, much like in film, radio, or choose your own adventure books.

Are the player interactions in an MMO the same kind of 'plot driven narrative'?

It's kind of silly not to admit that bringing all the open endedness and possibility in player interaction doesn't fundamentally change the nature of story telling, especially when compared to film, or mass effect. Are intra guild politicing in WoW just plot points?

At some point, the amount of interactivity, responsiveness and possibility starts to create a different kind of experience from more traditional story telling. I don't know where exactly, (it may be different for everyone) but at some point the experience really begins to change.

 

So I can safely assume you've never been to an improv show? The possibilities don't matter. The only thing that matters is what you experience, because that is the story. Surprise! It's not the telling of a story that's important. What is important is consuming it. Even in traditional storytelling, people can come away with different stories if they failed to pick up on plot points (or pick up on plot points that may not have even been there).

http://tvtropes.org/pmwiki/pmwiki.php/Main/WildMassGuessing Pick something you know. Does any of it gel with what you thought the story was? For the record, I'm terribly sorry if you spend the next five hours on the site.

 

I love the improv analogy, but are you the audience of the improv show or are you an improv actor who is building the story on the fly?

Although every game tries to make us feel like we're the actor, we'd both agree that this isn't actually the case. In Firewatch, the player is most certainly the audience. In Mass Effect the player may feel like he has choices, but I think we'd both agree that he really doesn't, or at least, he doesn't have more influence on the story than the reader of a choose your own adventure does.

On the opposite end of the spectrum is Mount and Blade, a game that I think both of us love. In this game, you really are an actor. You can create your own plot points, and you can quite literally mold the story of the world.

I think that at some point, as the game is more and more emergent, we shift from being audience members to being much more like actors in an improv troop. The improv analogy is really great because in the extreme case the developer literally acts as the audience, yelling out nouns, providing locations and situations. The players take those locations and situations and compose the story.

 

Amen! I'm a huge fan of the Dragon Age series, especially Origins which allowed surprisingly deep interactions with the companion NPCs, and dialog choices that have real impact later. But their mechanics do two things that really irritate me:

1. Exactly what you say; they force you to relive a cut scene before a boss battle in which you're very likely to fail several times before you finally succeed. And they won't let you save the game during a battle when you reach what seems to be a milestone (e.g., "Okay, all the minions are gone...now let's attack that arch-demon!"). I don't mind the latter, if the intent was to make a loss condition have a bigger cost, but they should at least skip the cut scene on re-fights.
2. They force cut scenes before boss battles that have your character doing overtly stupid things. Classic example: I enter a long narrow room with demon goo all over the place. "Hmmm....must be something nasty ahead!" So I carefully position my party members in a battle formation and do a game save. The rogue creeps ahead, intending to make a sneak attack to start the fight and lead at least some of the minions into the ambush. Noooooo! As I step forward, the game engine plays a cut scene in which the whole party, with me in the lead, boldly and slowly steps forward in plain sight to confront the arch-demon, as its minions encircle us. It's as if the designers assumed that by the time my character has fought his way through 40 levels, he won't have figured out yet that rooms full of demon-goo often contain....demons! A well-crafted game should reward players for good tactical play, not punish them for it in the name of game balance.

Overall, I think cut scenes should be used very selectively (intro before character generation, and epilogue, for example). They should never assume what the player will choose to do in combat, and they should be easily skippable on re-play. Basically, I think cut scenes should almost never be connected with combat, but only with narrative interludes.

 

So the story of tetris isn't about putin and friends trying to commercially take over the world?

 

But is the actor not also an observer? Does acting upon a story somehow invalidate that you are also bearing witness to the story? I say no. The point of view of the seats and the stage is irrelevant. The storyteller experiences the story just as the audience does. This is not an either/or relationship, but an and. Understanding storytelling is not about the act of telling a story, but about understanding how people experience story.


I just realized this is an argument for the complete dismantlement of the forth wall... Totally fine by me.

 

I've thought about that, too. It seems possible, though, that NMS will evolve to allow more interaction, and the players will end up writing a more complex narrative than any single creative studio could ever create. A game like NMS could end up being a microcosm of the fabric of human history. If you think about it, evolution placed Homo Sapiens here on earth with no carefully written quest line or story -- we are in a simple survival, exploration, and world-building game. Yet we have crafted millions of individual stories, from individual struggles to global epics, that are told at family gatherings for a few decades, or that inspire entire civilizations for centuries.

One critical factor in that, for No Man's Sky, will be how those stories are chronicled and retold. How does a great battle, or a war-averting peace alliance salvaged on the brink of conflict, become known throughout the galaxy and become a legend in that game universe? Will it happen in game, or on the Internet outside the game, or not at all?

If an MMO can establish a story-retelling ecosystem among its players, the developers have effectively crowdsourced their writing team.

 

Deep observation. Having done some improve myself, I am intrigued by this question, although admittedly it's drifting a bit off-topic from the OP.

 

The problem is that, as far as I know, you won't be even able to hook up with other players intentionally in game, and you would only run into other players by accident, and probably very, very rarely given the size of the universe there. So you're essentially just a lone spectator to the procedurally generated universe.

I really don't know why they didn't allow for people to play together, it seems such a critical and essential piece of the fun considering there's no real single-player missions. You're essentially a lone, wandering fps camera with a ship. There's no real possibility of creating a decent story from what I can see.

I think they sort of lost the plot and got carried away with the novelty of what they could do with the procedural generation. I hope they uncover some 'brilliant new gaming experience' but from what I can see, it won't happen.

 

I just watched a vid of Sean Murray talking about the game. I always love it when I hear people say things along the lines of "the players can do whatever they want in the game, some of them will like to stay at home or go fishing ..."

Oh, they can, can they? But why the hell would I want to stay at home or go fishing in a procedurally generated lake with procedurally generated fish? I mean, reality isn't that bad is it?

Being a fan of procedural generation, I really admire what they did on that account, but if you're going to break some game design rules you have to have something pretty good instead. But I don't think they really replaced the standard gaming experience with anything except perhaps novelty, for a time.

 

Some countries require fishing license even for that. So even that depends...

Um... What rules?

And if you don't do anything novel... You will be copying standard gaming experience. So what's wrong with that?

 

When I said 'rules' I am talking about standard game design, which generally speaking calls for a 'designed' experience including pacing, objectives, plot (yes, even if it's so small you can barely see it), progression and so on.

I'm NOT saying that these 'rules' can't be broken, all rules are there to be broken

But if you're going to break them, what will you replace them with? I think you misunderstood when I said 'novelty', I meant that the only fun thing in the game it seems to me will be the novelty of seeing some new plants and animals or something. There's nothing 'novel' about their game design, it just seems lacking IMO.

Anyway, I'm not a hater, I really, really admire their procedural tech and I will no doubt play it if only to witness the best of procedural generation in action. But for me it seems like after a short while the experience will become empty.

 

I can't help but think of Eve here. I personally really hated Eve gameplay, but I got a kick out of reading articles about huge dramas or crazy wars. Some of them made it to major mainstream publications like the new york times:

http://www.nytimes.com/2009/06/15/business/15views.html?_r=0

 

That's the thing, if No Mans Sky had the ability to hook up with other players en masse, you'd have some possibility there, but it seems like it's essentially a single player tech demo.

I also read some of those stories and enjoyed them, and heard there's a book in the works based on some of the huge battles that went down.

 

Twenty bucks says it's the second coming of Spore.

 

Those Will Wright talks... guy makes an interactive doll house and thinks he's the king of the world.

 

What really bugs me is why don't these people apply the procedural tech to something more manageable? Why don't they make a nice Mass Effect style game or something, and use the procedural stuff to make asset creation manageable?

It seems like whenever you implement procedural in your software you have to go on some never ending exercise to recreate evolution.

 

I think its because the general consensus is that strong traditional narrative and procedural generation don't mix very well. The advantage to procedural is generally: ability to generate tons of filler.

So people think it lends itself better to 'open world' style, that requires crap tons of filler.

 

Or plate tectonics, glaciation, and erosion. If you can get into emergent systems, that's where the money's at (it's like two bucks, but still).

 

I mean, why don't they use this procedural to make assets for the game (not runtime) since the graphics of the game are pretty good IMO? Use it for creating the ship/space station assets, integrate it into level editors and such? You've got to admit, if you had the ability to generate ships, planets, asteroids, even animals (possibly people?) and so on with the press of a key, you would have a good chance of pulling off a sizeable openworld game, or at least a pretty big one.

But in any case I think their biggest mistake is lack of multiplayer or even co-op. Simply the ability to fly around and look at stuff with your friends would make the game orders of magnitude more compelling I think.

 

Who said they don't? I think AAA do generate stuff. Internally. But they never release tools and polish all those assets manually anyway so noone will ever notice that they're generated.

UMA? There's ton of those human generators actually. Yet creating truly different people would still require hand polishing. Generating human behaviour would require full pledged A.I. which we are yet to see though.

 

Well that's cool, so why don't they try and make a little bit more of a crafted game? I see their vids of flying through asteroid fields and landing on a planet and so on, and I think, how cool would it be if there were cities or people there, battles being fought, alliances being formed? But no, just endless animals munching plants, with a few drifting orbs appearing to shoot at you for no particular reason.

I'm not under the illusion that it would be easier, but if they spent all their time making this sort of multiplayer work rather than making sure the 10^26th planet won't break the procedural engine, I think they would be ultimately better off.

 

They're still not released, and with fleet battles I think they will have some kind of factions at least.
Right now I know that Elite: Dangerous has both planets, factions, battles being fought and multiplayer... But does that really change anything? It's still boring sandbox with nothing to do.

Why? Because nothing really changes other than numbers. There's no story, no progression, no uncommon events. Just plain quests that affect some numbers and planets to 'explore' and tons of battles to fight...

I wonder when someone will at least try to make procedural story generation in game?

 

I think people have been trying (and failing) at this for decades.

 

I've never seen any such experiment applied exactly to games. I've only seen general experiments and it's obvious they will fail: story is not linked in any way to actions you can really do which game(e.g. world) provides you, so there's just too many possibilities which either aren't connected with each other in A.I. "brain" or all connected with all so nothing can be chosen. Game would prove a good limiter and context provider.
I've so far only seen one game which used template stories (it was some russian idle browser game), so even that isn't used (and it would've been much better if it was polished and done instead of fixed procedural quests).

 

Well first, it doesn't seem like multiplayer is going to be there. I admit they probably have enough whiz bang stuff for people to probably stay interested until they can give the game a more compelling structure, but the question is, what structure?

Boring sandbox perfectly describes how I think the game will turn out. Elite can be boring, but don't forget it is a game designed for hardcore space sim fans, with a steep learning curve and a lot of 'stuff' to manage and control on your ship, and does a very good job of catering to them. All those people with HOTAS joysticks and game CABS or whatever got exactly what they wanted. Not to mention, Elite you can hook up with your friends, and there's stuff like competitive dogfighting which is plenty fun for space sim fans.

The thing is, I just don't know why people have to go all out with procedural and make some recreation of evolution, when that tech is exactly what's required to bring a 'traditional' large game within indie reach?

 

Because the results are so bad it never got anywhere. The only thing that comes close are strategy games or agent driven games like the often mentioned Mount and Blade. In terms of even semi dynamic or responsive dialog the best example was an experiment called Facade and I believe that the immense amount of work required to the scope of the result made others hesitate to follow in those footsteps (although, Facade is actually very, very impressive).

The reality is that 'quests' as we know them are still usually nothing more than fetch quests or 'kill x' quests. The only thing that makes them remotely interesting is all the hand crafting that developers put in to hide the fact that these quests are all the same.

 

The crucial thing is that procedural generation has to be married to artistic 'rules' in order to make something worthwhile. So a procedural story still has to have structure, the buildup and climax, all of that. Randomness is just that, random. I feel like it would be possible to make a procedural story but it would take recognition of acceptable structures and deviations.

I was very impressed with FaceFX 'procedural' animation system which I've mucked around a bit with recently. It does an analysis on the speech based on recognition of the structure of speech and what it means in terms of facial expression and body language. It's quite something. That's the sort of tools that are really useful for making games, because they find where the 'rules' are in order to take the workload off the developer, rather than simply generating random stuff.

If you applied those principles to the art of story-writing, finding the rules that underly compelling structure and pacing and so on, I believe it's quite possible to make procedural stories.

 

The industry doesn't even understand how to make a good story for a game in the first place. There is no way we are even remotely close enough to understanding how to develop systems that make stories, much less how to write for these systems to make what they churn out palatable.

 

^^^

I wonder if one of the reasons that as an industry we're so far away is that we're stuck using concepts of narrative and story from passive media.

But yeah, at the end of the day - the gaming industry is a decade away from good procedural story (at a minimum). Maybe longer, instead of trying to innovate on this trail, the gaming industry is trying to ape movies and create 'cinematic' experiences.

These are cool, and a lot of players love them. But ... man... it ain't what it could be.

 

It would probably take a decent writer a week to make stories that are way better than what most games have.

@frosted, not really, apeing movies is what Metal Gear does, and it's very far from the norm. Most of them just don't even bother except for a few explosion cinematics here and there.

 

While I doubt it: http://www.gamespot.com/articles/no-mans-sky-will-have-traditional-multiplayer-but-/1100-6420982/
So I'm asking to wait and see, that game isn't out of alpha yet.

While I'm not talking about dialogues at all, story is much simpler than dialogues (first comes action; then come words to describe it). While that game might be amazing, it's not limiting dialogues, and that's exactly the problem. You never know what user will type and it'll break (see microsoft chatbot), you never have enough voice overs to fill everything up, so it's obvious NPCs are avoiding anything that player types, they answer with generic phrases. You only need to generate dialogue, not try to understand what user typed. Simple Mass effect styled "4 answers" will make everything a lot of times easier. Can you generate 4 different answers based on generated question and try to evaluate how NPC will respond and generate next part of dialogue? While it's by no means easy already, that's possible.

When were you betrayed in procedural quest last time? That is exactly what gives flavour to "kill x" quest: you were sent by your competitor(relations<0), and he did it in order to backstab you(remove you=>have benefit). See how it changed immediately when I made two-step quest instead of one-step? Now you would need to make a lot of excuses like that and make it not two-step, but a lot of steps which would intertwine: you helped someone(debt++) => he intervenes with you being attacked by your competitor(repay debt by helping mitigate negative event) => competitor tries to burn his house(revenge: damage property) down, you need to track it back to who did it and punish him only to find... Everything was by him. Everything I said is logical, and every piece of logic can be described to computer, right?
The hardest part to describe to computer is when to stop and how not to repeat as much as possible...

Yes, of course! All of those are easily categorized and story can be generated based on those principles to not be all-random junk. A bit more after next quote.

I've seen papers where group of scientists analyzed 1000(or so) fairy tales. They came up with ~30-40 different types, all consisting of transposition of about 10 structural elements(villain, what villain does, hero, trial, etc) and their repeating(elements are same, objects are different, see below for objects).
Each structural element had about 10-40 different objects representing it:

hero: a boy, a girl, warrior, old man, etc;
villain: traitorous king, demon, greedy old granny, etc;
villain action: takes away something(item/human), kills someone, tricks someone, refuses to give something, etc;
hero trial: training, tricking to get something, helping someone, etc;
rewards for completing trial: magical artifact that will help later, skills that will help later, helper that will help later (and I think that's it, actually - there are no more reward types); Trial can be used to actually take away artifact/helper in order to "balance" story (so that it won't be easy for main hero).
climax: uses artifact/helper in order to beat villain. Happy end. Don't forget to repeat trial/action/something before climax several times using different trials/objects/etc.


Sorry, I just don't remember specific numbers, all numbers I've said are approximation, but quite close to what I saw. So basically any fairy tale is transposition of very limited number of elements!

 

It's not about the logic, it's about the presentation...

The game that does this best currently is Crusader Kings 2, as most of what you describe is simulated to some extent and considerably more. The problem is presentation, Crusader Kings can do this because the game borders on a giant excel sheet and your interactions are most frequently represented by numbers, attributes, or pre-written 'events'.

A grand strategy game can do this, as well as represent all the internal mechanics required to build a compelling system. A Mass Effect style game can't. It needs to rely on speech and detailed, compelling dialog.

We're just not at a point technologically where it's possible to do on a scale that's useful for game development.

If you want to try it, good luck, I look forward to the day when the Unity Game Design forum revolutionizes AI, story telling, and procedural content generation all at the same time!

 

Have anyone tried showing player a graph of what happened because of what? Like visual novel walkthroughs?
It's simple, doesn't require dialogues or any advanced things, will explain things so anyone can understand. Then make it beautiful and sparky. With arrows pointing from one event bubble into another with small description of why did someone do it?

P.S. I'm not talking AAA here. Just sketching prototype, which could theoretically later be polished into at least B grade game.

 

In that case. I believe @TonyLi had a procedural quest system that he had worked on that tried to build the underpinnings for various relationships.

If you're serious about it, then I would also invest the (possibly non trivial) amount of time required to learn enough of Crusader Kings 2 to really understand how their system worked (as it's probably the closest thing ever built to a true procedural drama simulation - and they were largely successful - read reviews or player 'after action reports' to see how well it ended up actually working). Mount and Blade / Sid Meier's Pirates is a solid secondary example that relies on core, genre, gameplay. Finally, you may want to look at Shadow of Mordor's nemesis system, as this might be the closest thing to an attempt at the AAA level.

It's worth noting that if you really want to do procedural story, it's really 100% about the characters.

The real key is turning the relationships themselves into game mechanics, along with enough supporting material to actually play out interesting and varied results. The only game that has ever done this full on is Crusader Kings, they achieved it by turning character's ambitions, families, and positions into full on mechanics. It also relied on huge amounts of abstraction. But again, read player accounts of the game, they ultimately succeeded.

 

There is a commom meme in theatre that the ancient Greeks were the last people to create an original play.

With fairy tales, it's important to remember that in some cultures these were allegory, intended to teach children the morals their parents wanted to impart. The emphasis was less on ars gratia artis and more on creating a verbal textbook in a time when most people were illiterate. I'm not arguing against your point, only saying there is a broader context.

The same old fetch/kill questlines ("go forth and obtain the Enchanted McGuffin of Doom from the Forest of Scariness so we may defeat the evil Overlord of Bossness") are well-worn partly because it is hard to think of something original, and marketing execs don't like to spend money solving hard problems when there is low-hanging fruit to harvest from well-trodden paths. But also, those questlines are still used after all these years because they touch a very deep nerve inside us. Many RPGs derive from AD&D, which derives heavily from Tolkein, who derived heavily from classical mythology from several cultures.

How is the World Eater in Skyrim all that different from the Midgard Serpent in Norse mythology? Not much. (I don't mean that as a criticism of Bethesda -- I'm a huge fan of their games.) If you're going to steal an idea, steal from the best sources, those that have endured for a thousand years.

All that being said, I agree with others that it would be nice of game companies (and film companies) that spend millions on special FX could spend a fraction of that to also hire a good writing team. I know there are fans out there who do care about story -- I am one of them.

 

The Hero with a Thousand Faces came out in the Forties, and even though every Hollywood film rips off the plot structure it outlines, Hollywood continues to produce garbage.

 

Who said that copying structure will automatically make it good? I said that there's understanding of what story is and how to make systems that make stories. Do they make good stories? Uhm... Nope, never seen good one yet.

 

I'm still working on the procedural quest generator. Someone needs to invent a few more hours in the day.

That's the idea of Love/Hate. Well, Love/Hate provides the infrastructure, and it's up to the game designer to define the game/story-specific stuff.

Regarding procedural story generation, here are some ideas I explored but for various reasons didn't implement in the quest generator. I think they still have merit.

Don't disregard the three act structure. Most drama, from Oedipus to Sense & Sensibility to Conan the Barbarian, follows this structure. It may be a good way structure procedural narrative, too. Not just because it provides goalposts for the algorithm to work toward, but it also provides a structure that players are familiar with.

Similarly, you can divide acts into sequences, sequences into scenes, and scenes into beats. You can generate appropriate beats to compose scenes, scenes to compose sequences, etc. in a bottom up manner.

Or you could define high level three act structure templates and then fill them in with appropriate details from the game world in a top down manner. (Decomposition, vs. composition in the previous paragraph.)

If you're interested in procedural story generation, you'll find lots of great papers searching for "procedural narrative" on Google Scholar. Gamasutra had a good article "Procedurally generating a narrative in Forest of Sleep" about generating fairy tales.

And as far as handcrafted writing goes, Cracked.com of all places had a good article "6 Things You Learn Writing Blockbuster Video Games". It'll give you an appreciation for some of the challenges game writers face.

 

This is how far I've gotten so far:

 

http://gdcvault.com/play/1023372/Practices-in-Procedural
Probably the closest to procedural storytelling.