I have spent a long time thinking about how to combine player freedom with engaging story without compromise. That has been generally regarded as an unsolved problem or something that would require artificial general intelligence.

I think it can be done, but it would take a lot of work with many years of incremental progress. The point is that I have patiently waited for that type of progress in games for the last 30 years and it seems I may have to do it myself.

I have collected a lot resources from people with similar ambitions. I think I have a solution but it consists of MANY interlocking components. I may do a proof of concept, but that would be something without any sort of graphics. But the goal of this is to show how it can be done in order to have more immersive open world adventure games with a completely dynamic system for story, characters and relationships.

I tried to explain some of this in a couple of blog posts here: https://blog.jonas.liljegren.org/systemic-story-game-development/

Forget about story. There is no such thing as “complete freedom without compromise” in any game. It can't be done. It wouldn't be a game if it could be done.

Imagine you're playing Grand Theft Auto V. Only, instead of hanging around in San Andreas, you want to go to see the Tokyo Tower. Or you want to see the emperor penguins in Antarctica. Or you want to join A Buddhist monastery in Tibet. Or you want to start a family and raise a kid, and your kid goes to school, and the school puts on a school play, and you attend that play, because that's what good parents do. Who is going to make the model of Tokyo Tower? Who is going to model the behavior of the penguins? Who is going to provide the acting for your child's play?

OK, maybe you didn't mean geographical freedom. OK, let's set the entire game on a tiny desert island with just you one one other shipwrecked survivor. All you can do is talk to her. So you ask her about her visit to Tokyo, and the time she went on an expedition to Antarctica, or the time she went to see her child's school play. Oops, turns out that her entire lifetime of experience (including second-hand experience from the books she read and games she played and movies she watched) is every bit as big as modelling the entire surface of the Earth. And that's the past. Ask her to make up a story. Give her a specific list of elements the story must include. Play Dungeons and Dragons with her.

Again, that's just the “complete freedom without compromise” part. That's the easy part. We basically already know how to do it, although the fact that it requires filling your computer with actual human brains makes it both unethical and impractical. The hard part is doing all that and still telling a satisfying story. Because even the greatest fiction writer in the world can't write a convincing fan-fic of one of the protagonists from Grand Theft Auto V suddenly going to in Antarctica to look at the penguins. No amount of brains in jars is going to solve that problem.

@a light breeze I'm trying to explain something with text. I know how to do more than the games of today. I just trying to figure out a way to describe it. The first part of it was an article about NethHack. Games can be better than today without having to do EVERYTHING.

I think that a AAA studio could use systemic story arcs in games in a way that would eventually surpass the things they author manually. It would take many years of development. …

But if I would do something similar, I would do it more like a rougelike. But the first iterations wouldn't even have a grid or even any graphics. Neither would it have dialogue. But it would be a component that could eventually be used in other games.

You would of course have to limit the content. Find something that would be interesting and enjoyable to play and replay even with very little content. But the system part would mean that you could continue to add more parts to it.

I would start with a couple of templates of detective stories of the type that gets recycled in countless TV series. And I would start with summaries of scenes where you do your choices. You don't have to come up with actual dialogue for the aggressive, sneaky, flattering, etc, choices. And I would generate just a couple of locations that you could jump between. But the point is that the consequences would be the result of a combination of system simulations and story templates, and that you could apply this to any type of adventure game. You could let the characters speak in simlish and topic icons, or try to stitch together text from a library of dialogue or generate the text from neural networks, filtered by a lot of logic checks.

So you're not really aiming at “without compromise”. You're aiming at “with better compromises”. That's an achievable goal.

However, a focus on a “systemic” approach is exactly the wrong way to go about this. The way to get better stories is to create more hand-crafted content. The way to get more freedom is also to create more hand-crafted content.

From another angle, “player freedom” is not absolute, objective and infinite; it means only that in the limited time and the limited number of games player spend on the game they don't feel constrained by technical limitations, lack of variety of content, heavy-handed narrative constraints, etc. They don't expect to do anything and go anywhere, because they trust the game to produce problems for them that they like to solve.

To return to a light breeze's examples, players in GTA V intend to drive around, talk to characters, commit crimes, etc. rather than to visit Japan or see penguins; they paid for a lovingly crafted city with plenty of cars and roads and they want to extract all the criminally adventurous fun they can out of it. On the other hand, if a game promises the whole world it needs the Tokio Tower and the penguins, and if it promises extensive meaningful conversations it needs simulated people.