-not as hard to create as you might think, doesn't require human-level AI at all. see examples of "The Sims" games. or dwarf fortress. [or, perhaps, Rain World]
-real difficulties [discrete VS continuous for computer legibility and simulation simplicity; first person games are usually continuous but simulation games are easiest to make with a grid system and other discrete stuff][unstable chaotic systems, will it just "blow up" or die out? Well, can always use some kind of "rubber banding", but still][just easier/more fruitful to implement multiplayer instead of even moderately advanced AI][combinatorial and meta-combinatorial possibilities to program/test?]["cursed problems", see this talk. In particular, near the end he mentions very relevant problems in The Sims (creative expression VS goal progression system) and the way that Sim City VS Civilization games are kinda like that as well (one is more free expression, one is more goal focused)]
-ask: could you watch just NPCs do stuff for hours? i've watched AI on super smash, RTS games, and The Sims of course should be mentioned
-no scripted story, not even choose-your-own-adventure style, that is still scripted
-no "player is the key to the universe" stuff. No NPCs that sit around until player comes. No "quest holders" who never recruit anyone except the player. No player exceptionalism.
-not just fake NPCs. Though some is fine, for crowds etc.
-if you google "simulation games", you'll find lots of stuff. But most will be lifeless aside from the player. Yes I do have to check out some more...
-dwarf fortress is about the only example? or the best example? [Too bad it's ugly as hell, somehow even uglier than minecraft! But it still counts as an example]
[could talk about some games and then point out how they fit and don't fit. Like Far Cry 2 I think is mostly NPCs that stand and wait for player and do nothing otherwise.]
[both Elite:Dangerous and Star Citizen talked a lot about NPCs and a simulated economy, seemed like this kind of idea. i don't think they ended up making such dynamic persistent NPCs though.]
[link to my criticism of that "immersive sim" video essay.]
[can i set some criteria for what i want (without being too picky about my personal taste) and see if ANY games meet those criteria? Would first person/ continuous variables be the only problem? Use this to have evidence about whether such a thing is possible/practical...though, of course, I've managed to do plenty of stuff just fine on my own in Unity, so I very much do not see any technological issue][first person RTS is one close genre]
the distinction between sandbox vs theme park games seems to basically be what I'm talking about. I'm more interested in the do-your-own-thing sandbox genre than the pre-written theme park genre. Distinction discussed at start of this talk, though the talk is about MMO games. Also talks about relevant things like overlapping extrinsic and intrinsic motivations, as well as needing the game to be kind of too difficult to mathematically solve so that it doesn't become simple and easy and boring.
Examples to check out
I'll try to see what's out there. Mostly I'll look for stuff that has some dynamic world simulation, and stuff with multiple AI characters.
Unfortunately, just searching for "simulation game" brings up a ton of the following, which I'll ignore:
- vehicle simulation games
- factory games
- games that use online multiplayer instead of AI [admittedly this is a sensible design choice!]
and a lot of the following which I'm not sure are actually very dynamic (though I'll still check some that might be):
- management and tycoon games
For sure to check out these specific games:
- Nivalis ["cyberpunk first-person life simulation game Nivalis"][a 'slice of life simulation in a cyberpunk world']["Grow your business, manage restaurants and nightclubs, make friends and enemies, buy and decorate apartments, go fishing and maybe even find love in Nivalis, the city that stretches from the ocean to the clouds.]
- Dwarf Fortress
- "The Sims" Games
- the spiritual successor to Daggerfall i heard about [found where i read about it, here and here], from the original creators or something. googling that now, must be "The Wayward Realms" [official website][steam]
- Crusader Kings 2
- Rain World [the AI is done in a bit more of a simulation-like way, with some enemies having abilities that are about equivalent to the player, and with an intelligence that is supposed to make them seem about as intelligent as the player]
"Factory Town" seems to be the natural evolution of the factory genre to include AI workers. I did wonder if the genre would evolve towards simulating more like this. But also as usual it's not a 1st person game, and very limited to just the factory building genre still. I'm not sure their AI does much besides pathing, carrying items. And I think the player just spawns workers into existence, out of thin air [unlike my game/sim, where you recruit NPCs].
Oxygen Not Included
older games, just to list here:
- Majesty: The Fantasy Kingdom Sim
In some ways what i'm interested in might not count as a "game". Games tend to give goals to the player and such. I'm more interested in the experience of the dynamic simulated world. This is neither the scripted story many games have, nor even what you would find in a game with no story, like chess. Contrast with what this guy says about games, that they give you clarified goals/values. I have no interest in clarified goals. My interest is in "what is it like to be in this situation?", right? Like "what would it actually be like to be a pirate?" for example. Scripted stories can try to answer those questions, but following a story is always going to feel different, because a real pirate is not following a story, right? There will unfortunately always be some limit there [being a pirate isn't like a safe simulated experience either], and some interesting questions remain [we want to make it juicy, so do we need to reduce entropy somewhat?][and limits of generative art, lack of novelty, lack of meaningful art design, how much needs to be hand crafted]. Though a pirate will have goals, they are also defined by the pirate themselves. This has two sides, first a lack of direction, but also a lack of restriction. There are more goals than can be pursued. I do like the multiple goal endings of Alpha Centauri, that's pretty good. Still, it is a bit restrictive and redundant. I'd prefer the game be designed so that such achievements are possible, but not as something the simulation tells you and ceases. I don't want that narrowing (and distracting) of attention in the experience.
A contrast can also be made in terms of "means-ends chains". A game gives you a known end to the chain. But a different interesting experience is to be faced with an unknown to explore:
- what are the people "up to" in this world? what's going on?
- what might happen if i try this or that?
i think both of the clear inspirations i had each was about one of these things. One was a particular character that seemed shady and mysterious. Another was messing around with the stuff in the world, and the reaction of some other character being part of that [partly being the whole point for me].
Just as much about goals of others in the world as it is about your own.
No guarantees of "oh, the game designers made it so this is all part of the plan" or "there is a [set] way" or "it's all contrived" or anything. No static, controlled world, with fake evidence of change, but nothing will happen unexpectedly until you are at the right checkpoint in the story.