Why haven't any games tried to tackle a procedurally generated story?

[TheMediator]

By this I mean we have a number of games that have procedurally generated levels, but none of these games seem to try to create a narrative that ebbs and flows as well. Generally the missions within the game take place inside a greater preset narrative.

For instance, in XCOM the general story beats of the experience of the game are preset but you have randomized missions thrown in here and there. People say "the gameplay is the story", and at times it is, but you still have preset cutscenes, preset mission dialogue, so on and so forth.

Another game that comes to mind on this topic is City of Heroes. There were procedurally generated stories in an essence but they were generally contained "one-shot" stories - some procedurally created gang might take over an abandoned complex, and you need to rescue some scientist and his research, blah blah blah. However, these one-shot instances weren't tied together into a cohesive story themselves - they were random disjointed events that happened across the city. The game DID have "campaigns" called Task Forces, but they were completely pre-generated affairs.

To me, this seems like the area that is the next leap forward. While you'll never get the quality of cutting edge pre-rendered cutscenes, and voice acting seems more difficult to pull off in such a system, a game that could continuously generate new stories seems like the pinnacle of infinite replayability... and yet I don't understand why we haven't seen attempts to tackle it. It is just that complicated?

[Afrospinach]

Probably because most people understand that we are not that impressed by randomness. It would be like an entire game made with skyrim radiant quests.

Also, randomness does not make infinite playability. I have never played the same map in any diablo twice. It actually gets excruciatingly boring. It is hard enough to write a decent story let alone make one random and engaging.

[Tryuk]

Probably because most people understand that we are not that impressed by randomness. It would be like an entire game made with skyrim radiant quests.

Also, randomness does not make infinite playability. I have never played the same map in any diablo twice. It actually gets excruciatingly boring. It is hard enough to write a decent story let alone make one random and engaging.

Procedural generation by definition is not random, it only appears to be.

Dwarf fortress has procedurally generated quests on the roguelike mode, and the lore of each world is procedurally generated by events. Sometimes (most of the time) it can be hilariously ridiculous though, but every now and then you get good stories of heroes killing megabeasts and the sort.

Edit: Honestly I don't see the reason for it other than filler content or stories set in a completely proc gen game (like Dwarf Fortress). Could be an interesting experiment though

[I Push Buttons]

You said it yourself, it will be shit quality compared to anything purposely written and acted.

I think the farthest we might go with regard to procedural generated stories are concrete stories set in a procedural generated world.

[TheMediator]

Edit: Honestly I don't see the reason for it other than filler content or stories set in a completely proc gen game (like Dwarf Fortress). Could be an interesting experiment though

I guess my point bringing this up was making a comparison to Dungeons and Dragons. Every time you play, unless you're following a preset campaign provided, you're playing with a new set of characters on a new adventure. In one sense that's sort of on the opposite end of the spectrum in that the experience that is crafted is a one-shot experience, but when you envision the "game" as the set of rules that each story is crafted within, then you're talking about a game with almost endless replayability. It's certainly harder for a machine to be as creative with each generated world as a human person coming up with a new campaign every X period of time, but it feels like that's sort of the future, a sort of "imagination engine" if you will. Again, probably difficult to pull it off in a fun way, but it seems like very little effort is being made in that way from the end products we get.

[Faroth]

Storybricks was going in the direction with EQ Next striving to design a game where player activity affects NPCs because the AI has certain preferences and reacts to the changes in the world based on how it impacts their preferences. An MMO where an NPC may give different quests based on your past interactions together (and not give them to others or even be hostile to you based on said interactions) and change those quests based on what's going on around them, each with their own daily needs and routines is extremely ambitious, but it's what SOE & Storybricks were trying to do. Too bad it got gutted before even the ground floor of such ambition could be seen.

We're probably still really far off for AI to be complex enough for a game to have NPCs that react to/learn from events and their environments and react to new problems (and attempt to resolve them) in a way similar to procedurally generated worlds. Not only does the AI have to get that complex, but then our games, systems, and data bandwidth has to handle it on a massive scale of thousands of NPCs all functioning at this level, all interacting, with all things constantly changing.

[Poppincaps]

The only reason to have a procedurally generated story is to increase replayability but how replayable a game is is so very subjective that I feel it shouldn't be a main focus for a game.

Games like Dark Souls for instance, facilitates replayabilty, but you still get virtually the full experience with just a single playthrough. A lot of effort is put into making something procedurally generated and that effort could be better used into polishing up the story and writing. It works for games with a short and solid gameplay loop like Binding of Isaac but it is not something that can work in every game.

Now a story that fully takes into account your choices. That's what I really want.

[Coldhearth]

It's an interesting concept. A game with a procedural story could be amazing if done right. However, I feel like it would just be a lot of random story lines that made little to no sense. Maybe when AI gets a little better in the coming years...

[May90]

Since the stories will be generated not by experienced storytellers, but by generic scripts, I don't see them coming out good. Would likely be just random set of events and characters, which will get boring fast.

Imagine a book written by computer. It makes up random characters out of a set of traits, generates dialogues from a few templates, scans through the story templates it has and chooses an appropriate one... Would this book be fun to read? Doubt it.

[Soulwind]

Since the stories will be generated not by experienced storytellers, but by generic scripts, I don't see them coming out good. Would likely be just random set of events and characters, which will get boring fast.

Imagine a book written by computer. It makes up random characters out of a set of traits, generates dialogues from a few templates, scans through the story templates it has and chooses an appropriate one... Would this book be fun to read? Doubt it.

An experienced storyteller could write each "piece" in a way that can start after and end before any other.

It would be extremely difficult and time consuming, and the result wouldn't be any better than a normal story written by the same person in the same amount of time, but hey, it'd be a cute gimmick for an indie game.

[zealo]

Much for the same reason you don't have your programmer write the lore and narrative.

It simply cannot stand up to what having a dedicated writer can put out.

[SpikeySquad]

A procedurally generated story makes me think of:

A woman/man/witch/monster walks/runs/skips down the hill/road until in the distance they see a city/rainbow/elephant.

That is a rather basic example but in my opinion this would be much weaker than intentionally written side quests (which in many games can be rather boring as well!)

However like others have touched on so far in the future we will see developments. I am curious.

[Xeenith]

on a certain point of view, we could say that comic books have some sort of procedural stories... same heroes, same villains, same cities, yet people continue to buy them years after years. (yes there are great arc of stories that change stuff then big reset button for next filler ^^)
on a mmo, you could have lots of filler stuff made with this system, while the big story is done by real writers, it could link daily quest or the like in some sort of chains of quest a bit more linked to each others, rather than "today go get me 20 of [this item]"
you mention City of Heroes and it was a lot like that : main story with a real lore and writing, and "random generated quest from random npc" as fillers, and it was still much better than most fedex quest we have on most mmo today

[May90]

An experienced storyteller could write each "piece" in a way that can start after and end before any other.

It would be extremely difficult and time consuming, and the result wouldn't be any better than a normal story written by the same person in the same amount of time, but hey, it'd be a cute gimmick for an indie game.

The problem is, writing stories is very effort and time consuming. One wouldn't spend maybe months, if not years, writing a complex story, knowing that the majority of consumers will only see a small part of it. If the story in a game is procedurallly generated, then a player playing through the game just once will only see a part of the whole. If to see all possible stories one has to play through the game, say, 10 times, then it just isn't worthy it, both financially, and just from the perspective of satisfaction.

Small story branches are okay, like in The Witcher 2, where one decision affects in what direction the story develops in the 2nd and 3rd acts, up to having completely different locations, characters, etc. A lot of branches generally aren't worth it.

[Soulwind]

The problem is, writing stories is very effort and time consuming. One wouldn't spend maybe months, if not years, writing a complex story, knowing that the majority of consumers will only see a small part of it. If the story in a game is procedurallly generated, then a player playing through the game just once will only see a part of the whole. If to see all possible stories one has to play through the game, say, 10 times, then it just isn't worthy it, both financially, and just from the perspective of satisfaction.

Small story branches are okay, like in The Witcher 2, where one decision affects in what direction the story develops in the 2nd and 3rd acts, up to having completely different locations, characters, etc. A lot of branches generally aren't worth it.

Choose your own adventure books aren't the best novels, but there's a lot of people who like them. They're limited because of the size of the book, a digital enviroment doesn't have that problem.

It's just a gimmick, it's not like a game built around that would be terrible because of it, it just wouldn't be any better than a game with a normal A->B->C story that would take a lot less time and effort to make.

[PixelFox]

Madlibs: The MMO.

[Deleted]

Neverwinter online has a huge amount of player made maps and dungeons with good stories attached to them. Portal 2 has the infinite testing centre.

If you want a game to have loads of new and unique stories and levels, then design your game to support player created content.

[Tackhisis]

Storybricks was going in the direction with EQ Next striving to design a game where player activity affects NPCs because the AI has certain preferences and reacts to the changes in the world based on how it impacts their preferences.

But that was just a marketing speak. Many developers promise such things, but in the end it is still made manually with complex scripts.

[Faroth]

But that was just a marketing speak. Many developers promise such things, but in the end it is still made manually with complex scripts.

Again, technological limits, but what Storybricks had as the foundation was a good starting point unlike what other games, particularly MMOs, use. It was the right start that never had a chance to grow.

[ro9ue]

Storybricks was going in the direction with EQ Next striving to design a game where player activity affects NPCs because the AI has certain preferences and reacts to the changes in the world based on how it impacts their preferences. An MMO where an NPC may give different quests based on your past interactions together (and not give them to others or even be hostile to you based on said interactions) and change those quests based on what's going on around them, each with their own daily needs and routines is extremely ambitious, but it's what SOE & Storybricks were trying to do. Too bad it got gutted before even the ground floor of such ambition could be seen.

We're probably still really far off for AI to be complex enough for a game to have NPCs that react to/learn from events and their environments and react to new problems (and attempt to resolve them) in a way similar to procedurally generated worlds. Not only does the AI have to get that complex, but then our games, systems, and data bandwidth has to handle it on a massive scale of thousands of NPCs all functioning at this level, all interacting, with all things constantly changing.

One of the reasons I was so sad to see EQN die.

But there's still hope, other developers (prominent ones) are interested in stuff like this as well. This may be of interest to this thread - and one of my favorite GDC talks of all time: