Hacker News new | ask | show | jobs
by socalgal2 356 days ago
> I disagree with his indictment of the video game landscape as being narratively poor.

I think he would say they are narratively poor by his defintion that the narrative must be generated by the game/player combo and not just pre-programmed. People love "The Last of Us" for it's narrative but that narritive is something that can arguably be conveyed via book or movie. Crawford wanted something where the narrative itself was generated.

And no, he wouldn't count the choices players make in the average game. Whether to get go west or east. Whether to get the a sword first or the arrow. He wanted the story and character dialog to change. Few if any games do that. Of course today with LLMs it's likely some games will soon / have already done it to some degree and will do better in the future.

Going back to his older work, you'd need to feed a context to the LLM about each characters motivations and then update that context based on player actions so that as the game progresses the way each NPC interacts with the player, and other NPCs, changes in a way that's consistent with each character's intrisict motivations and their interactions with others.

1 comments

People come up with complex shared narratives in multiplayer sandbox games like Minecraft/Roblox/Kenshi/etc. all the time.

In the single player realm, there are games like Dwarf Fortress, Caves of Qud, Kingdom Come: Deliverance, etc.

Point is, the landscape of what "narrative" means in video games today is broad and deep. If none of those are even remotely like what Crawford thinks is "right" - and he's not able to design a game that meets his standards himself - I'd argue his definition of "right" might just not be workable in the first place.

There's a kind of people who want video games to have all of the possibility, depth, and meaning of real life. A game where you could do anything, be anyone, but still have consequences matter and be far reaching (like "Roy: A Life Well Lived" in Rick & Morty). Well, that exists, it's called real life, but you're not going to recreate it on a computer screen.

> People come up with complex shared narratives in multiplayer sandbox games like Minecraft/Roblox/Kenshi/etc. all the time.

That sounds no different from comming up with narratives on a playground. That's not a designed narrative, that's people making up a narrative where none exists. Hey, the jungle gym in our secret base! The swings are space ship! The ground is lava!

I think we can all imagine what he wanted to build even if he failed to come close to building it. He wanted to make a story machine where you could play the game and converse with the characters in a free flowing way yet still have the game provide a setting and conflict. Imagine talking to characters in the holodeck on Star Trek. Ideally where, over the course of the game, the dialog and interactions are designed in real time within the constraints of the setting. And, the way you treat characters influences how they react. Be a dick to the bartender, all his connections are harder to get consessions from. Be nice to one romantic partner, get snubbed by another, etc... And not just by canned responses. Tell one character a piece of info and it gets leaked to their closest contacts who then change their behavior/dialog based on this new knowledge.

That describes Dwarf Fortress. I guess you don't have an text interface to talk to people but that feels like a no true Scotsman requirement.

The difference is that Dwarf Fortress is not fully opaque to the player. You see a large part of the world so can see the consequences of your actions. And that is what makes it fun.

An opaque world simulator is frustrating and tedious. Eventually someone just min-maxes a way to break it which makes it just a tedious cookie clicker. That's why D&D has a game master that modifies the world's design to align with the players so that everyone has a good time. That is the key difference. An illusion of rigid structure, which requires some actual structure, with opaque flexibility. A simulator is not a fun game. A world where god cheats to make you have more fun time is a fun game. A pre-defined narrative or narrative tress is one such cheat, but not the only one.

I'm glad you're trying to explain the difference.

I've been in too many conversations where this topic comes up, and it's very disheartening to me. Gamers insist there are plenty of great narrative games, and every example they give is basically a branching story with bunch of flags that gate which branches can be taken. If I give the Holodeck as a counter-example, well that's just too pie-in-the-sky.

These conversations remind me a lot of Paul Graham's Blub Paradox: "Blub is good enough for him, because he thinks in Blub." Current SotA narrative games are good enough for most gamers, because all they've played are branching story games.

The complex emergent narratives people love in Dwarf Fortress or Caves of Qud aren’t branching stories though?

If the argument is “let’s build games that tell character stories as complex as Last of Us with the world building techniques of Dwarf Fortress”, it might be worth considering why 1) it wouldn’t be feasible and/or 2) it wouldn’t lead to a game that’s as fun/compelling as a more narratively linear counterpart.

The narrative in Last of Us is compelling because it’s tight and focused and authored by the authors to be told in a very specific way.

The “emergent narrative out of raw world simulation” argument sounds like filming thousands of hours of security camera footage and expecting Citizen Kane to come out of it.

I mean again, I would be pretty receptive to a “this game had the right idea, we should push it further” argument, but the “I’m a visionary game designer who hasn’t shipped a game that embodies my vision AND the thousands of other game designers are clueless, but everyone will see how clever I was in 3 centuries” argument rings pretty hollow.

For what it’s worth, Chris Crawford’s tool (http://storytron.com/) is a branching story editor, and one that seems much less powerful than what is used in the industry (eg articy).

Crawford has no involvement with the Storytron 2 software at https://www.storytron.com/. It was developed by me based on some ideas Crawford outlined on his site and in his Discord group (https://discord.gg/Rc4dq7HwZz).

Crawford's latest software that he developed to illustrate his latest approach for interactive storytelling are called Slubber99 (https://www.erasmatazz.com/library/Slubberdegullion/Slubber9...) and Gullion99 (https://www.erasmatazz.com/library/Slubberdegullion/Gullion9...).

Storytron 1 & 2 history can be found at this link - https://www.ifwiki.org/Storytron

You haven’t played dwarf fortress or caves of qud and it shows. Great example of armchair designers insisting they know better than all the gamers and game creators immersed in the art.