Critique of MUDs, RPGs, DnD, MMORPGs, and immersive interactive adventures

Thanks for the attempt.
Good read. But i still don't understand what RPG is about. I'll probably never get it, i'm afraid.

However, i would like to know what kind of job the Dungeon Master in DnD has. To me that's all just a myth, and i never saw or played such game.
So, is he kind of story teller, coming up with quests or refining some outlines given by the game?

And i wonder, if he's missing from computer RPGs, why was there no attempt to make one player the DM in multiplayer?

A friend showed me LLMs nowadays can come up with entire quests, considering given faction backgrounds, and giving dialogue and everything. So this could be done now, dynamically.
I'm afraid we'll see that soon, but personally i don't want it.

JoeJ said:
However, i would like to know what kind of job the Dungeon Master in DnD has. To me that's all just a myth, and i never saw or played such game. So, is he kind of story teller, coming up with quests or refining some outlines given by the game?

The game, DnD, is just a rulebook, and some basic lore (various common fantasy races and their stats and characteristics, various item types and their stats, an the rules for the game mechanics, etc.). The DM comes up with the scenario (or chooses an existing one), the concrete setting, backstory, etc. The players each design a character at the start, give them personality traits, etc. The DM then gives an introduction to the setting, and explains some lore maybe, and then it starts. Players choose what actions their character should take, and the DM then tells the consequence of those actions. Sometimes, if it's a “soft” kind of thing, like how an NPC should respond to a question the player asks it, he can just freely make it up, trying to keep it in line with his setting & lore he came up with. Usually, a good DM comes up with more layers of setting and backstory than he lets the players know. So it really is like an interactive novel. Whenever it's a “hard” kind of thing, with numbers and stats, etc., then a dice roll is made to decide the outcome according to stats, etc. And this loop simply repeats until everyone is dead, or until the adventure is beaten.

An example I heard from an old acquaintance was that in one campaign, there was a big dragon end-boss, and instead of fighting the dragon, someone cast gravity magic on it and lifted it across a cliff, and then cast paralysis on it. When the levitation got disspelled, the dragon just fell to its death based on its weight and size. This is the kind of weird creative freedom and depth that you have in DnD. If the player is really smart, and gets a lucky dice roll, he can even kill a dragon solo. Other times, things are more conventional. But even then, you have to interact with the world, its lore, and its people, to find clues as to how to beat the campaign, or even to find out what the actual problem is that you have to solve. For example you might start out in a city and learn that the lands are plagued by decay, but nobody even knows where it comes from, or of what kind it is. And then you might have an entire story arc to even find out what causes it, and then another story arc to beat whatever bad guy is responsible for it. Or maybe it's not even a bad guy behind it all, maybe some dragon died of a curse and its corpse is contaminating the lands, and the mission might be to safely get to the core of the corruption and cleanse it without dying in the process. Maybe there are multiple endings, etc.

So it's something where anything goes, and it's like reading a novel where you don't know the entire plot or setting ahead of time, but with the difference that you are the protagonist and you can make your own choices, and the book adapts to those choices. The DM is basically the author of the interactive book.

JoeJ said:
And i wonder, if he's missing from computer RPGs, why was there no attempt to make one player the DM in multiplayer?

Because the players can freely choose whatever comes to mind, they could suddenly say they want to dance or sing when faced with an enemy, and that's a valid choice. Probably gets them killed, but there might be an off chance of a dice roll for which the DM might allow the enemies to fall into confusion or even be charmed by that, depending on the corresponding stat or talent the player's character has. The DM is only there to interpret prompts of the players. He is basically the world simulation and the narrator. So if one had a DM, then that would only be because he would have to interpret every action each player takes. Maybe some actions could be automated, but creative prompts that require reasoning and contextual understanding and (potentially hidden) background knowledge can only really be handled by a human. And at that point, since the players can freely decide to do or make things that were not hard-coded into the game by the rules, one also cannot represent these things properly by a computer. Like if I decided to use fire and wind magic to deplete oxygen in a room, AFAIK such a thing is not part of the ruleset, and the DM can then interpret that as he sees fit.

So if you introduced a DM, he would basically have to reply in human words to shape the narrative, and at that point, the game wouldn't be able to keep up with the creativity of the DM anymore, and at that point, you might as well have a group chat that lets you track character sheets and that lets you do dice rolls.

JoeJ said:
A friend showed me LLMs nowadays can come up with entire quests, considering given faction backgrounds, and giving dialogue and everything. So this could be done now, dynamically. I'm afraid we'll see that soon, but personally i don't want it.

Yeah, I think that's what the official DnD is devolving into. There was a huge drama over an executive decision to go into AI-generated official lore etc., and an AI-driven game session app is not far off, I think. But IMO it completely misses the point of DnD, because I don't think an AI can hold long-term detailed context like the backstory and plot that it made up, and then refer back to that and update it as people make prompts. It requires lots of nuance, heart, and insight to be a good story teller or a good DM. An AI is not good for that, but maybe it can pull off a mediocre or bad version, idk.

JoeJ said:
Good read. But i still don't understand what RPG is about. I'll probably never get it, i'm afraid.

Well, I think RPG became many things to many people, and I'm just saying what I see it as. To me, it's the attempt of establishing an interactive adventure world that you can adventure in with your pals. But to computerise that, of course you lose out on the real-time creative direction of the DM as the narrator. So you have to concretise, systemise, put things into numbers and algorithms. And you have to hard-code things, and the game cannot invent new things on the fly. Which means there has to be a trade-off between narrative depth and technical feasibility. But computers can do things that a DM can never do, such as keeping track of lots of state, performing system-based simulations on the world in the background, etc. A computer can simulate entire economies, and generate very plausible reactions to at small and large scale, at least for actions that directly interact with systems such as the economy etc. It can track wildlife and monster populations and their behaviour effortlessly, which a DM would find very tedious and he would also gain little from it. So it would become a less narrative world than a book, but it would become more thorough, more detailed, more consistent with itself, etc.

Sadly, almost everything after DnD basically devolved into combat & levelling simulators, with only a few rare games pulling off actual depth and meaning to the world.

Interesting take, but I think it pre-supposes a bunch of goals and ideals that may or may not be true for a particular game.

There's a bunch I disagree with that isn't really relevant to your debate, for example, I think role playing games have been around for almost all of humanity. Role playing has been a way children explore the world, and we see it in stories dating back to antiquity as they take on roles of nurturer as a child plays with her doll, roles of adventurers as children explore near their homes, imaginative roles of saving the world and living out stories of Odysseus or Gilgamesh or their favorite Marvel superhero, or even kids playing tag in the schoolyard, but they are still people taking up roles and playing games with them. Even so that kind of thing doesn't affect what I think is your core thesis:

RmbRT said:
The take-away is: to achieve immersion and a social experience, you need to design every part of the game to encourage or demand social interaction, while also adding stakes to everything so that everything seems emotionally relevant, either because of danger, or because of necessity, or hardship/tedium, etc. Keep things meaningful by preventing them from becoming devalued and too easily replaceable, be it items, characters, or locations. Add harsh dangers that require preparation and careful strategising and cooperation to mitigate, and don't pander to the player too much all the time.

I don't think they are as critical as you state.

Why is immersion the goal? Immersion is extremely recent in games, for millennia games could be social with little more than stones, carved blocks, or cards, or even pure imagination. Some video games, really just a small subset of all video games, strive for immersion,. Most focus on whatever it takes to have fun, immersive or not. Less with role playing, but games like Hearthstone are hardly immersive despite the millions of players.

The social experience is similar: why is it the goal? It is certainly popular in many games, and humans love social interactions, but many popular games are single player with no social experience at all. Multiplayer is certainly a popular aspect, but if it were essential and core then single player games, be they BOTW or Hogwarts Legacy wouldn't have sold 30M+ copies if a social element were required.

Games being emotionally relevant is similar. Games like Go, Chess, Backgammon, these aren't games of emotion but they have lasted through the centuries as people place themselves in the role of battlefield leader and logician. Go through the popular tabletop games and while players get emotional, the game itself is not driven by emotion.

The concepts of harsh dangers, hardship in their story, they are present for many popular games but by no means are they universal. In many games they are a burden to avoid. Go replay some of the RPG's of the 80s like the original Final Fantasy and you'll discover the tedious grind was a barrier to players finishing the game. Others were notoriously difficult, I played Battletoads with my friends and we had a love-hate relationship as we memorized the levels and had to play together in the hopes that at least one of us would survive through the speeder bike level, and the other would respawn quickly enough so we have a backup player, again another severe barrier to the game rather than a key requirement.

Even if you stick with the BIG MMORPGs, WoW, RuneScape, FFXIV, GW2, SWTOR, and similar, I can argue that those features weren't core. They are extremely fun games, and the designers clearly focused on whatever is fun to their fan base. While some fans get immersed and dive deep into the lore, the stories, the fan fictions around the game, that doesn't necessarily immersion in the games themselves, and certainly not by all players. For social, while some fans focus on the social aspects and build a community of in-game friendships or bring their real-world friendships into the game world, the are also built so you can join with a group of strangers, play for a quick objective with emotionless utilitarian communication, and never interact with them again. Similarly for emotion, yes there are some players who seek emotional connection, there are many players, particularly top tier players, who are grinding for stats and have massive spreadsheets, tables, and calculators looking only for numerically optimal values that have nothing to do with an emotionally compelling story.

I'm not saying games don't use those things, and there are games that focus around those elements. Many of the currently popular games tend to incorporate them, but certainly not all of them. Games like BOTW or TOTK were incredibly powerful role playing games despite having no social aspects, and for top-tier players the grind is a painful slog. They are features that many games incorporate, but by no means are they universal nor are they required.

Game designers know about them, and they may decide to explore aspects of them as part of their fun. That doesn't mean they're essential for games, and game designers can completely ignore any of those elements and find the fun in other ways.

Yes, game designers need to know about them and study them. But I don't think they are as essential as you make them out to be.

It's true, I have not considered children's games. And yes, I was giving advice towards my own ideals, and the way I see things and what I think went wrong, etc. Others have different tastes maybe, and therefore will desire or build differing things.

Why immersion? Because I want to go on an adventure in a different world for a while when I play a game. I don't play for leveling up, for leaderboard rankings, or for competitiveness with other players. Games like Go, chess, etc. are not really role playing games, and they have different ideals. I'm strictly speaking about DnD-derived role playing games, which I think should be about offering that adventure experience, as if reading a novel, but interactive, instead of passive. I explicitly also seek out video games, and not traditional or tabletop games. And for a computer game, the immersion factor is simply part of the appeal of the medium. I'm sick of everything becoming a causal commodity to consume. For DnD, people had to physically meet up and spend an entire afternoon + travel time on it regularly. That is a hassle, but it also means the thing becomes an established part of one's schedule, and one dedicates time slots for it. The pains the players go through to play it, means that they hold it in regard and aren't too casual about it. Only people who were serious about wanting an adventure would even play for more than an afternoon or two. And campaigns can take months or years.

Why the social experience? Because that's simply what I lack. I don't really find joy in playing alone anymore. I want to unwind together with other likeminded people and have a good time, and ideally do that in an adventure world together with them. And since I don't really have social contacts, I think it would be great to have a place to play with others who are in a similar situation.

Other people with different kinds of connections or needs will see games and game genres differently, obviously. But everyone can always only speak from his own perspective, because that's the only thing everyone really experiences. I live a life where I don't really have many joys, and also don't really have social activities. Many of these games have become pathological in their design, luring in players by evoking an expectation that never gets fulfilled, and turn them into hostages of the grind and sunk cost. People start out with a sense of wonder and expectation and stay out of a sense of obligation that the game imposes on them.

Many people are highly discontent with the way MMOs evolved over the years, because they desire that initial dream that they evoked in them. Meanwhile, others don't really find issue with the state of things at all. Obviously, my writings are not for those who don't take issue with the state of things, or desire fundamentally different things. But those with the same yearning, will probably agree with my words and take inspiration from them.

But it's always been like that. Different characters appreciate different art, different genres, different music, etc.

I'm basically on the search to clearly identify what it is that I personally want out of a video game, and over the years it became clearer and clearer. I think the best name for my take on the entire genre is “Adventure world game”. Like kids love minecraft for its creative sandbox, I want something that is just a world for adventuring in there. And I think that is what made DnD so great, and DnD being the origin of computer RPGs, I think it is valid to try to look at how a computer can best be leveraged to create such a thing, and what the computer-specific version of it would looke like. I do like fantasy novels, but I would rather share that experience with others than just consume it by myself. And that's what computer adventure worlds can offer, theoretically. And that's what I'm trying to lay the theoretical foundations for.

RmbRT said:
The DM is basically the author of the interactive book.

I've asked so many people… over many years… to help me out on understanding RPGs. What's the magic? The unfullfilled expactations? I never quite understood.

But you did it. :O I think it clicked on me finally. I get it! Thanks, man, this really was helpful! \:D/

So i understand now why there is a desire for dynamic stories and why it's so difficult to deliver the RPG people dream about.

The story about the dragon beaten by gravity magic reminds of stories from the game NetHack, where such things can happen a lot. That's from the Rogue genere afaik, so maybe this genre is missing from your initial list.

Personally i'm not a RPG player, but i always felt somehow related to the (unkown) dream about it.
And indeed, i want similar things. Most people do.
Recently i had an eureka moment, which might get me closer to that goal.
It's a simple shift in perspective which changes everything for me regarding game design, solving so many problems just by looking at things from a different angle.

The toy problem is, say we want to add more depth to simple FPS game. The obvious solution is: Make an RTS to control the NPCs. So they are not just guards spawning around the player, but they have an actual purpose and function in the world. E.g. they mine for resources, drive them back to a base, manufacture weapons and ammo out of it, which are used for a great war between two factions.
The corresponding RPG thingy could be NPCs having a daily routine of going to work, maybe. But RTS is better i think, because it's simple and gives a meaningful world simualtion where NPC actions actually matter and are not just facade designed around the player.

So that's good, but my complaint about this idea always was: ‘The FPS player lacks oversight. He has no large scale top down view to understand the processes happening in a RTS sim, so it's pointless to implement this idea. The player simply won't recognize the additional depth.’
And i was stuck at this, until recently. The new idea is: ‘Ignore the player, but design the game for the NPCs instead.’
This is great i think. It also solves the other big problem i see, actually the question: ‘What can i do in the game, beside shooting stuff?’
I solve it by ignoring it. Instead i only think about ‘What can the NPCs do in the world and with each other?’
This is much easier. I'm not limited by player input limitations for example.
Now i hope, by moving focus to NPCs, i can make an interesting and rich simualtion. And then, by playing it, takign the role of one NPC initially, i will notice and enjoy the depth, assuming it will just work. I also expect some emerging options rising from the simulation complexity. And so far i can avoid what i want to avoid the most: Dialogue. Because traditional programs can't chat.

So that's a good perspective and starting point i think.
The question now maybe is: ‘How can i transform the kinda boring and predictable war simulation of an RTS into diverse, unexpected, and more interesting adventures?’
Maybe it works by adding monsters, treasures, super weapons, and such RPG things to the world?
NPCs might find them eventually before the player does, though.
But i request parity of players and NPCs. Same rights to everyone! ;D

Dang it, accidentally refreshed the page after writing half an essay… I'll leave out some points that I'm too lazy to rewrite, lol.

JoeJ said:
This is great i think. It also solves the other big problem i see, actually the question: ‘What can i do in the game, beside shooting stuff?’ I solve it by ignoring it. Instead i only think about ‘What can the NPCs do in the world and with each other?’ This is much easier. I'm not limited by player input limitations for example.

That is a great way to put it into words. And right on point with the player input, and also that you do not have to visualise it, either.

I think the realisation of having additional, possibly obscured, layers of depth or something is great, and should not just apply to FPSs or RPGs, but to any kind of world building, even to novels. For example in a good novel, you might have a childhood friend who wants to become a merchant, and then later on in adulthood, some scene takes place in one of his company's stores, and even if most readers would not make the conscious connection, the fact that this hidden layer of causality and simulation and plot lines exists, will definitely improve the general coherence of the story, and to those readers who pick up on the connection, it is very satisfying, because it strongly signals that the story takes itself seriously and that the page-by-page plot isn't just throwaway stuff, but that the world remembers what happened. The world takes its own circumstances seriously, and also takes plot / player actions seriously. Obvious expected consequences to circumstances should also be observable, at least up to a certain level of detail. You don't want to make a particle simulator, and you don't need a fluid simulation or anything like that, that's overkill.

An RTS would probably benefit from something grand-scale like in Europa Universalis or something, where you have macroscopic and long-term things going on, and the case-by-case skirmishes are just a more detailed, but locally constrained snippet of that. The FPS focuses on individual battles, an RTS on individual wars, but only if you have something like a global geopolitical / economical layer, do even the wars have context. Otherwise, they are all completely isolated and also hold no intrinsic meaning beyond the battle of wits, because there is no consequence or outcome that propagates across matches. There could also be political factions, national alliances, etc., which get affected by the individual match, and also have each individual match be affected by the coarser-grained world state in return. The causality should go both ways up and down the layers of depth / abstraction.

JoeJ said:
So that's good, but my complaint about this idea always was: ‘The FPS player lacks oversight. He has no large scale top down view to understand the processes happening in a RTS sim, so it's pointless to implement this idea. The player simply won't recognize the additional depth.’ And i was stuck at this, until recently. The new idea is: ‘Ignore the player, but design the game for the NPCs instead.’

I think that's also the thing: if you add additional layers of depth and consequences to everything, maybe most players will not pick up on most or even all of the ways in which this concretely expresses itself on the surface. But if you leave it out, you definitely do take away all meaning that a player could otherwise have ascribed to a place, a battle, a war, etc. So, it's something the player optionally experiences, but if you take it away, and the player becomes fully aware of, and fully understands all the depth there is to the world, then he will immediately start to think of the game world in terms of mechanics, and predictable outcomes (even if randomness plays into it). The same goes for other kinds of things in a game which are initially fun because they have some sense of mystery to them, and the player can simply go by their look and feel and enjoy them: as soon as he understands the formulas etc., it all just turns into a numbers game, and you can't un-see or un-know that knowlede, and will never be able to enjoy it the same way as before again.

JoeJ said:
And so far i can avoid what i want to avoid the most: Dialogue. Because traditional programs can't chat.

Yes. This is in the realm of novels, and other hand-curated works. It is not something that belongs in a computer-simulated sandbox. Now, you can still have some kind of interactions and scenarios that play out inside a sandbox, but that would require that the simulation is already aware of semantic annotations for everything. Which is why in that old thread that I linked, I talked about terrain generation that first generates a semantic representation of the world, and then generates an immutable landscape that matches that. And the layered collective psyche thing and retroactive causality that gets made up as soon as a finer-grained resolution of a coarse-grained simulation layer is required. Basically instead of requiring an AI that extracts meaning and semantics from the world, make the world semantics-first, and retroactively fit the landscape to those semantics. Since you then already start with all the contextual information you would need for scenarios, and already have all the annotation you need a priori, it becomes trivial to then let the world act out that meaning in a fitting way.

JoeJ said:
The question now maybe is: ‘How can i transform the kinda boring and predictable war simulation of an RTS into diverse, unexpected, and more interesting adventures?’ Maybe it works by adding monsters, treasures, super weapons, and such RPG things to the world? NPCs might find them eventually before the player does, though.

Tough question. To a certain extent, yes, it might add variation, but too much of it is also bad. And for an RTS, it could come in many forms, such as having locations or territories or individual units that persist across matches, or consequences of actions that persist across matches.

I'm super tired so I hopefully wasn't too incoherent and rambly (stayed up through the night, reading)…

RmbRT said:
An RTS would probably benefit from something grand-scale like in Europa Universalis or something, where you have macroscopic and long-term things going on, and the case-by-case skirmishes are just a more detailed, but locally constrained snippet of that.

Yes, but this comes with the risk of making the player feeling pointless. He may realize, he can change some things locally, but it wont affect who wins the big war.
So i think winning the war should not be the goal of the player, who is on his own adventure. The war and RTS stuff is meant to build up a dynamic and itneractive background. The goal of the RTS AI might not be to win, but to keep the war going on forever at equlibirium, so this background remains constant.
My goal is, once this simulation is running, i can spawn into the world and play. I can join the fight or avoid it, i can sabotage supply lines to weaken guards, to steal resources or tresures, etc. There is already a lot i can do without having designed a single quest or level yet.
Basically i want some way the game plays itself, and i can join the fun which already exists when i come.

RmbRT said:
The same goes for other kinds of things in a game which are initially fun because they have some sense of mystery to them, and the player can simply go by their look and feel and enjoy them: as soon as he understands the formulas etc., it all just turns into a numbers game, and you can't un-see or un-know that knowlede, and will never be able to enjoy it the same way as before again.

That's maybe just bad design then. Ideally it's the other way around: After you realize the higher level mechanics, you can use this knowledge to affect things to your advantage even on this higher level. E.g. there might be the chance (but no hard requirement) to win the big war from the example above. If it's not a numbers game decided by pure luck, but a challenging optimization problem the player can affect, playing on the higher level should be as fun as playing on the low level. Maybe even more rewarding due to long time investment.

Btw, the use of dice is something i would remove for video games, replacing it with something skill based like shooting, beating a level, winning a race, etc. If i see dice in a game, like recently in Baldurs Gate 3, i think that's something for board games and rage quit.
It's hard to befriend RPGs as an action player… ; )

I talked about terrain generation that first generates a semantic representation of the world, and then generates an immutable landscape that matches that. And the layered collective psyche thing and retroactive causality that gets made up as soon as a finer-grained resolution of a coarse-grained simulation layer is required.

Hehe. Whenever i try to come up with something to model speech, my brain instantly blocks and refuses to work on that. I'm not the one who will figure out anything on this subject, i guess. <:)

Oh, okay, so even all that talk about RTS was fundamentally still about an FPS wargame? I guess it is somewhat true that since the game is an FPS wargame, you don't want to make the war end, lol. But even then, if you had mostly conquered everything, there could be rebel factions & political decay, breaking up the realm into pieces again. And if there is an economical side to things, there could be battles that are fought economically, like the defender and the conquerer having an asymmetric setup, with different types of units and vehicles being used (like we see in current wars, with all the drones and missiles and stuff — I'm not really good with that kind of geopolitical analysis, though).

And yeah, the “meaninglessness” of the player is something that should not crush ambition. Therefore, the world should also offer smaller-scale things to look forward to. And the higher level simulation should not just have one dimension to it, so that maybe you “lose” on the territory side, but have a “win” in some other dimension such as political, economical, morale, or something. Maybe one side is larger and winning battles but taking heavy troop losses and morale sucks because at home the people are poor or starving. Maybe you lose at the front, or are in a stalemate, but the player guards the supply lines and infrastructure at home, while also going on missions to disrupt things deep behind enemy lines. Maybe that starts to wear down the enemy despite him having the advantage on paper.

JoeJ said:
That's maybe just bad design then. Ideally it's the other way around: After you realize the higher level mechanics, you can use this knowledge to affect things to your advantage even on this higher level. E.g. there might be the chance (but no hard requirement) to win the big war from the example above. If it's not a numbers game decided by pure luck, but a challenging optimization problem the player can affect, playing on the higher level should be as fun as playing on the low level. Maybe even more rewarding due to long time investment.

Yes, do not make it luck-based, but the player should be playing for the fun, not for some formula. So, I recommend making it so that you cannot get the exact formula and numbers for everything, therefore always remain somewhat uncertain about your understanding. Do not give the player all the concrete numbers he needs to turn the game into a spreadsheet simulator like they do with Eve Online. But of course you should be able to gain an intuition of roughly what the effects of an action are. Just make sure that some level of mystery stays there, no matter how small.

JoeJ said:
Btw, the use of dice is something i would remove for video games, replacing it with something skill based like shooting, beating a level, winning a race, etc. If i see dice in a game, like recently in Baldurs Gate 3, i think that's something for board games and rage quit. It's hard to befriend RPGs as an action player… ; )

This is my big pet peeve with roguelikes. They are basically reskinned slot machines. Everything is completely up to the random generator, and your skill basically has no effect on how well a run goes.

JoeJ said:
Hehe. Whenever i try to come up with something to model speech, my brain instantly blocks and refuses to work on that. I'm not the one who will figure out anything on this subject, i guess. <:)

I wasn't really referring to speech. By semantics generation, I meant: first generate a map of very coarse-grained “cells”, and each cell can contain one village, a few resource deposits, and a terrain type. Then, you generate the connections between the cells. You decide whether there can be a road between towns of neighbouring cells, and also the travel difficulty / distance of those roads. Only after you already have the economy-simulation-relevant data about the terrain, you start to actually generate a concrete terrain that fits those parameters. And you can make that multi-layered, so that you also have even larger cells that contain the economy-simulation-relevant data for entire regions, entire kingdoms, entire contintents. And each lower layer gets generated in a way that satisfies the constraints decided by the higher layer. You start with statistics, and then collapse them into more and more concrete data points that fulfill those statistics. If the actual terrain comes last, but you already have the full statistical analysis & semantical tagging of all relevent points of interest, etc., then you can easily run an economic & political simulation at a higher level and retroactively make up the actual detailed state & events that occur at the finer levels of detail.

And since you no longer really have bottom-up emergent behaviour, but top-down scenarios etc., instead, and you don't need to acquire context & meaning from raw data, you can easily select dialogues etc. from a pre-defined set of hardcoded lines. And the actions at the finest granularity level just propagate upwards into the higher level statistics to generate large-scale outcomes. The statistical layers could also contain not just a statistical summary, but also some kind of event log or similar, which adds some concrete memories to the system, so that at a lower level of the simulation, things can refer back to what happened and seem causally related to things that happened. Of course you can then also add some actual causality simulation on the lowest level, so that the surface-level behaviour looks more convincing.

RmbRT said:
Oh, okay, so even all that talk about RTS was fundamentally still about an FPS wargame?

No. I mean it can be, but it does not have to. Imagine a love story, but they live in different countries, and those two countries are at war. Then the war adds some complexity, action and violence to the story, but it's just background and not the primary focus.

Instead of the RTS war, we could also simulate an economy, e.g. framing. Just somthing that is easy to simulate on all scales, keeps the NPCs busy with doing something meaningful, which isn't centered around the player as usual.

But i don't want it to be an instant treat to the player. Rather it should be a potential, which he can just observe. It fills the world with life.
After soem time the player should find ways to manipulate the system, or contributing to it. He may be attacked if he isn't carful. The system may even make out most of moment to moment gameplay.

But it's not necessarily related to the primary goal or main story. Maybe better not, because if we do this, the whole system feels again like an inscenation for the player. Smoke and mirrors. Fakery. Tricks. The stuff we have seen often enough.

Thus, the player could start as a neutral outsider as usual, but also as a soldier of war. In the latter case i would motivate him to desert or revolt, instead executing orders from his general.

And yeah, the “meaninglessness” of the player is something that should not crush ambition.

That's my primary problem with RPGs. I feel meaningless long before i could discover something like a goal.
Often i fell meaningless right from the start. Often i have to kill people although i don't know the reason yet. Often i could go anywhere, but i lack motivation to choose a direction since north looks the same as south.
Quit and refund, and trying another game.
Meanwhile i have actually stopped to try new games just to keep updated. I have no more hope they would make any game for me, and i end up playing the same old games again and again.
I feel we are at the peak (or bottom?) of a new video game crash. It's actually hard to keep motivated.
My personal best bet to get out of this is character simulation. Proper robotics instead lame animation. I think this can makes everything new and fresh, but that's another topic.

RmbRT said:
Therefore, the world should also offer smaller-scale things to look forward to.

Does not work for me. Adding additional smaller things means more things, so i will feel even more pointless. Granted.
Modern games focus on so many little things, and they are proud on all those details.
But it doesn't work. It's better to focus on just one thing imo.
For example, if we add both the war and the farming, pluse some more alternatives, we can brag about a real rich and detailed simualtion, and a game with great depth. Great. But then neither the war nor the farming feels important anymore. Nothing does. Maybe because we have no dungeon master who guides our attention amd focus. So we have to be more narrow, contradicting the dream.
I agree the dream is ‘choose your own adventure’. But well - be careful about what you wish for - i might say.

RmbRT said:
Do not give the player all the concrete numbers he needs to turn the game into a spreadsheet simulator like they do with Eve Online.

Imo this is another fundamental flaw of the RPG concept. Stats and numbers should not be used this way in a video game at all, same as dice.
Super Mario is the perfect video game. There are no numbers, but you can see all the formulas precisely, e.g. player movement following laws of physics, but enemy movement at constant velocity with simple and predictable behavior. Nothing is hidden from you - no secrets or mysterious mechanics. You can basically predict all the futrue frames of the game in your mind. And that's exactly what you need to do to master it. Your ability of prediction is no problem here - it's essential to play.
That's where action games are miles ahead of RPGs, and that's what i mean with skill based gameplay.

RmbRT said:
You decide whether there can be a road between towns of neighbouring cells, and also the travel difficulty / distance of those roads. Only after you already have the economy-simulation-relevant data about the terrain, you start to actually generate a concrete terrain that fits those parameters.

Ah ok. That i understand, and i agree. Also remembering the talk of back then.
The RTS idea made me think, maybe the AI game of building bases could be used here to build a world as well.
I think, in procedural generation, we want to explore the concept of ‘growth’ rather than 'instant creation'. Could add depth and fight the boredom.

In general i have the impression that games like NetHack, Dwarf Fortress, or now Caves of Qud, they seemingly implement all those ideas we discuss here, or most of them.
I sure could learn a lot here. But ugh - those insanely complex interfaces! I just can't play them. <:)

JoeJ said:
No. I mean it can be, but it does not have to. Imagine a love story, but they live in different countries, and those two countries are at war. Then the war adds some complexity, action and violence to the story, but it's just background and not the primary focus. Instead of the RTS war, we could also simulate an economy, e.g. framing. Just somthing that is easy to simulate on all scales, keeps the NPCs busy with doing something meaningful, which isn't centered around the player as usual.

Definitely agree. I just couldn't find good examples like that. In fact, you could even have the opposite where you have the war instead, but international love or conflicts of interests of noble families (as their bloodlines usually span multiple countries) as something on the side which spices things up. Something that's mostly minor and in the background, but at decisive moments, leads to small or big changes that deviate from what the core system would demand. For example a noble who suddenly deserts or revolts or hinders supply lines because of some familial ties or lover on the enemy side. Or maybe some noble's domain is much poorer or more impacted by the war than other territories, and he leaks intel etc. in return for aid. Something that in a pure war game would never be considered. And to some players, it's just an interesting or annoying random event, but to others, it's really exciting maybe.

JoeJ said:
But i don't want it to be an instant treat to the player. Rather it should be a potential, which he can just observe.

Definitely. At this point I'm not even sure whether I like presenting hard numbers to the player at all. Like character stat sheets. In DnD, you need to have that, because the system needs to know the concrete numbers, and there is just no way of hiding those numbers from the player in a table-top game. But in a videogame, why would you need to display the exact HP percentage? Or the exact strength and agility values you have? And why show the exact damage number that you deal with an attack? I think handing all the details and mechanics and numeric values to the player on a silver platter (or in a table) is actually depriving him of a lot of interaction with the game. Because if it wasn't all presented in a spreadsheet, then he would have to put in effort to find out what's what, and what has which exact outcome. He needs to count how many hits certain attacks need, or how long a battle with a certain build needs against an enemy, instead of just running the numbers ahead of time, skipping all the trial and error and exploration. This is something where video games actually have a big competitive edge on tabletop / pen & paper games.

JoeJ said:
That's my primary problem with RPGs. I feel meaningless long before i could discover something like a goal. Often i fell meaningless right from the start. Often i have to kill people although i don't know the reason yet. Often i could go anywhere, but i lack motivation to choose a direction since north looks the same as south. Quit and refund, and trying another game. Meanwhile i have actually stopped to try new games just to keep updated. I have no more hope they would make any game for me, and i end up playing the same old games again and again. I feel we are at the peak (or bottom?) of a new video game crash. It's actually hard to keep motivated.

Yeah. That is because these games all mistake the mechanics as gameplay. They think that RPGs are about combat, leveling up, grinding. But actually, combat is a plot device, leveling up is also a narrative device to achieve character progression, and grinding is also a plot device for something in that area (filler? stretching the excitement a bit thinner so that it can be appreciated more when it occurs? something like that). And yes, movies, novels, games, they all went more and more creatively bankrupt. And also the whole internet pretty much died, most people out there are now “content creators”, instead of personalities who share their works and inspirations with you or something. It all got optimised, streamlined, standardised, and the soul was driven out of everything. Everything is afraid to offend anyone, and trying to pander to everyone, instead of standing for itself, and letting its character and merits speak for themselves.

JoeJ said:
Imo this is another fundamental flaw of the RPG concept. Stats and numbers should not be used this way in a video game at all, same as dice. Super Mario is the perfect video game. There are no numbers, but you can see all the formulas precisely, e.g. player movement following laws of physics, but enemy movement at constant velocity with simple and predictable behavior. Nothing is hidden from you - no secrets or mysterious mechanics. You can basically predict all the futrue frames of the game in your mind. And that's exactly what you need to do to master it. Your ability of prediction is no problem here - it's essential to play. That's where action games are miles ahead of RPGs, and that's what i mean with skill based gameplay.

I fully agree regarding numbers, and think that definitely the fundamentals should definitely be like that, but I think having slightly obscured things happen in the background is also a valid design choice that can add additional depth and a bit of mystery that waits to be explored or ignored. For example in the romance game with the war in the background, you might not understand exactly how your choices will affect the war or the economy, or maybe sometimes those choices have an obvious direct outcome, but an unforeseen secondary outcome. That kind of thing keeps things fresh. But you should not obscure the fundamental mechanics of the core gameplay. But also should not let the core gameplay devolve into spreadsheets, either, in the mind of the player. Only if the player really wants to, should he through effort be able to turn it into spreadsheets. Like I said when trying to optimise builds, let him count how much time and how many attacks it takes to defeat an enemy with a certain build, instead of showing the damage per second value & hitpoint number directly in the status window.

JoeJ said:
Ah ok. That i understand, and i agree. Also remembering the talk of back then. The RTS idea made me think, maybe the AI game of building bases could be used here to build a world as well. I think, in procedural generation, we want to explore the concept of ‘growth’ rather than 'instant creation'. Could add depth and fight the boredom.

Yes, I think everything we learned from RTS games is fundamentally applicable to worldbuilding and world simulation. I don't know what you mean by growth vs. instant creation. Do you mean evolving state, vs. constant state? In my system I envision, you generate an initial state, and then that state evolves on the detailed level: villagers get born or die, or gather resources or fight, villages get built in the predefined cell slots, road slots between cells get built or destroyed, etc. And then this informs the higher-tiered simulation level's behaviour and choices, which then get reflected back into the actions and appearance in the lower tier. But as soon as you allow the player to construct arbitrary roads, and place houses literally anywhere he wants, suddenly, the higher simulation tier would crumble, as it would need to extract meaning and analyse the surface level, and then adapt to that. This would require true intelligence.

JoeJ said:
In general i have the impression that games like NetHack, Dwarf Fortress, or now Caves of Qud, they seemingly implement all those ideas we discuss here, or most of them. I sure could learn a lot here. But ugh - those insanely complex interfaces! I just can't play them. <:) Like Quote Reply

They do that, but I think they do it exactly the wrong way, the opposite way. Notice how an RTS AI does not simulate an agenda for each unit, but simulates the overarching intent of an entire faction, and derives actions for units top-down. It does not think in terms of what interests a unit has, but what buildings need to be built and what needs to be produced or acquired, etc. In Dwarf fortress, individual dwarves and individual cats etc. are simulated and the cats random-walk and end up drunk on beer puddles. That's useless, bottom-up simulation of details that don't really hold much meaning. AFAIK, there is no top-down “governance” or something being imposed on the individual actors in their simulation, and instead they bet everything on “emergent” bottom-up behaviour by having a simulation that is as detailed as possible. I suggest we focus on RTS-like top-down simulation for the overall large-scale momentum of things, and then add a layer of bottom-up behaviour that gives some individuality to each town, each villager, etc. But that fine-grained individuality should not override, or interfere with, or replace the grand scheme of things, instead it should cosmetically supplement it to make it more interesting to also look at the small scale, as the player interacts with the world on the small scale level. Basically have the detailed per-actor simulation be something that fills in the gaps, fills in the words on the page, but have the top-down RTS simulation of the world dictate the general plot line of the novel.

I also assume it is much more brittle and computationally expensive and harder to build to only have bottom-up simulation and entirely rely on emergent behaviour. To fine-tune a complex interdependent system is almost impossible. But the RTS-like top-down simulation is conceptually simple, and works on multiple layers of statistical simplification, and can at any point simply squash entire villages that are far away into just a simple numbers simulation, and then on-demand generate a detailed state out of the statistic again. And of course, to maintain continuity at the surface level between those squashings / regenerations, the lower level needs to be saved before squashing, and upon regeneration, load and retroactively make up plausible changes that occured to the detailed layer since the checkpoint in a way that fits the squashed statistics.

And with RTS-like stuff, you can ensure that there will be meaning to it, because the RTS layer holds a primitive layer of meaning and overarching intent. The extreme of bottom-up behaviour would be a physics/particle simulation and expecting something meaningful and coherent to come out from it. Kind of like AI simulates brain cells and expects intellect to emerge from that, instead of trying to codify understanding and rules of thought.

And to some players, it's just an interesting or annoying random event, but to others, it's really exciting maybe.

This really is one key to game design, and the point I was making earlier.

The designer needs to understand both options and more besides. The designer needs to consider what the players will find more fun. The choice will depend on the game.

In one game no numbers and no health bars. In another health bars alone. In another game always showing numbers, in another showing tiers of health and damage, and still another perhaps also with color codes revealing effects that have particular strengths and weaknesses, such as normally white, yellow if the target is weak against the attack, gray if they are particularly defended against it. In another breaking down types of damage, maybe revealing blunt, pierce, fire, ice, poison, and others as a stack of color-coded numbers. In another it may be a UI toggle to show or hide. The designer chooses, and it modifies the feel of the game, all the options are preferred by different groups of players for different games.

There is no right or wrong, no best option, it is a design choice. The choice in one game players find essential, in another game the same players feel is a distraction.

Basically everything so far in the thread has been a design or player preference choice, not an essential element. Each one has popular games that are counter-examples, where the supposedly “essential” elements weren't present and would likely have harmed the game if included.

The social element essential in one popular game missing in other popular games. The immersion element in one game, another popular game entirely symbolic with no immersion. All are valid design elements that a good designer can leverage, both through including or excluding them.

RmbRT said:
but international love or conflicts of interests of noble families

Possible, but that's not a interactive action game background, but rather something Hollywood would do, requiring lots of story, cutscenes, and all that AAA stuff i want to avoid.
But yes, it would be interesting to see how personal relations of kings and aristocrats affects the ongoing war.
Though, i would tend to depict them as selfish, shortsighted, and rich assholes, instead showing their love.
That's much more realistic, and would make the endgame, where you can finish them off finally, much more rewarding. :/
Lookign at current real world politics, i sure would enjoy a french reveolution 2.0 simulator.

RmbRT said:
But in a videogame, why would you need to display the exact HP percentage? Or the exact strength and agility values you have?

Sure. We can show real wounds on characters isntead HP points, and muscles instead a number for strength.
Tabletop is not the template. It's the enemy we want to beat. ; )

RmbRT said:
(or in a table) is actually depriving him of a lot of interaction with the game.

Exactly. I have to interrupt the game, to go to some shitty GUI page observing my character stats.
What's wrong with them? Why do they do this? Playing the game should not feel like adjusting gfx settings?

Ofc. that's just me, an action player demanding immersion. Other people enjoy stats, management sims, etc., and they should get it.

This is something where video games actually have a big competitive edge on tabletop / pen & paper games.

Yes, and action games do it the way you proposed - players just figure out how much hits it takes to kill enemy of type X. And it works.
RPG always looks like a step backwards to me, regarding combat.
It's like: ‘ugh - i can't implement a dungeon master to capture the true magic of RPG, but i can implement those stat sheets and rules of dice. So let's do that and sell the RPG promise based on this.’ ???
No. I think we have combat in video games to serve our primate instincts. It's not about numbers, decreasing immersion breaking health bars levitating over enemies heads, with sprinkling magic stars particle effects.
It's about split second decisions, reflexes, fast movements, resulting in hit or miss depending on precise player actions.
And since ARPGs actually do this properly, i don't see why they still carry those bolt on stats around as well.

RmbRT said:
I don't know what you mean by growth vs. instant creation.

The default example of ‘procedural generation’ is: Mix some procedural noise functions to give you a pattern which looks like a rocky surface. You may feel creative by extending the basic apprioach with domain warping, using nice color gradients, etc.
But the outcome is always uniform and featureless. You can generate features, but they appear in uniform distributions at locations where they don't make any deeper sense. So in practice you end up minimizing the appearence of features.
This is whay proc gen is considered 'boring', and rightfully so.

A growdth base approach can do better. E.g. growing trees can analyuze their environment to avoid growing into a wall, or they can grow along the wall for climbing plants. And we could use the same principle to grow cracks into rocks, or river networks.

And it shoudl work very well to grow villages or economy, micause in teh real world those are usually processes of growdth. Small villages become towns, street networks grow to connect them, etc.

It's probably better than the ‘default’ approach, which could be: Generate Voronoi cells from an intial distribution of random points, make a house in each cell, connect some cells with street paths, and so on.

Personally i think of this world generation process as an offline process, gerenating a static world which is then enriched with manually made levels and shipped with the game (like e.g. Assasins creed or Star Citizen). Thus i can consider much more expensive generation methods than games which do it on client like Minecraft, or even dynamically during the game like Dwarf Fortress.

However, there is always a way of compromise. E.g., you probably remember Arresseras project here with the procedurals planets showing landscapes with proper erosion, generated at runtime while you fly towards them. This is not just about visuals imo. I want to play on those planets, i want to explore them. Which is not the case with Minecraft, where the landscape is just boring.
Imo, that's a real opportunity for indie devs. But they miss out, since their off the shelf engines don't serve them the tech on a silver platter yet.
Those engines turned game development into just content creation. And imo that's the coffin nail for an industry struggling with creative burnout.

RmbRT said:
But as soon as you allow the player to construct arbitrary roads, and place houses literally anywhere he wants, suddenly, the higher simulation tier would crumble, as it would need to extract meaning and analyse the surface level, and then adapt to that. This would require true intelligence.

I don't see the need for intelligence. See the example of ant simulations meantiened last time. If you change soemthing, e.g. putting soem food at a new place, new paths emerge to harvest it. It's simple logic at the low vevel, producign seemingly complex resutl on the high level. But there is no intelligence, but rather just time spend to converge to a new optimized state.

In the real world we see both: China plans a new city for millions, and it looks like a huge simple grid eventually. In Europe we see cities grown over centuries, which look more organic and chaotic, less organized, but also more interesting.

There are many realted algorithms useful for such things. Spannign trees or shortest paths, erosion or ant simulations, cellular automata, etc. Notice how all of them solve some form of optimization problem over time. The final result is optimal, but the algorithms are not intelligent. Intelligence is only needed once when writing those algorithms.

So i think you can be more optimistic in this regard.
But the real problem is to generate detailed and playable geometry from such higher level representations.
That's why we have blocky Dwarf Frotress, but not yet an Assassins Creed which does the same thing at AAA 3D fidelity.
It's possible and an opportunity, but we have not yet found a way to make it easy.
Personally i have a tool which can model large scenes using SDF primitives, like the Dreams game does.
This way there is no need to work with meshes, and CSG operations become trivial.
I have not yet used it for more than the terrain, still need to work on procedural architicetiure at some point...
But it's clear that the soft and imperfect appearance of SDF modeling lacks the precision of CAD modeling.
And for anything that's made by humans using modern technology, CAD is actually what we want. But sadly that's the opposite of ‘easy’.
Btw, generative AI has the same problem here: It's output is equally soft and imperfect.
And i doubt AI will be able to operate CAD programs anytime soon.
(hehe, those AI nerds will never find an actual need for their ‘tool’. All it can do is turning people into skillless idiots. : )

RmbRT said:
In Dwarf fortress, individual dwarves and individual cats etc. are simulated and the cats random-walk and end up drunk on beer puddles. That's useless, bottom-up simulation of details that don't really hold much meaning.

Ok, but the ant simualtion is the conter exampel of this, where higher leavel meaning emerges out of simple rules at the bottom level.
Just give the cat another objective than getting drunk ; )

Bottom up is as important and useful as top down, i think.
But - hmm… why have i never worked on something which does both at the same time? hmmm…

frob said:
Basically everything so far in the thread has been a design or player preference choice, not an essential element. Each one has popular games that are counter-examples, where the supposedly “essential” elements weren't present and would likely have harmed the game if included.

You're right of course. Or, you should be right. Or, you was right, 15 years ago.
But now the situation is different. We have one Dark Souls clone per month, and we even see AA RPGs receiving a lot of praise, which is nice.

But the counter example to all this would be a game for me, one that i like. But it does not exist, and it does not happen anymore.

And i'm not alone. The number of ‘Why modern games suck’ or ‘gaming is dead’ messages is constantly increasing. We see mass layoffs and studios closing, more than before. Ignoring the growth in China, the industry is actually shrinking. The golden years are over, and the disappointment dominates over the excitement.

So the altitude has to change. From pointing out what works, to investigating what doesn't, and how to do better eventually.

I'm sure this is happening behind the curtains. I'm also optimistic the industry will change, coming up with new things, rising again.
But currently i don't see that yet. Currently it's rolling downhill at full speed, like the rest of the planet does as well.
It's actually so bad, i consider starting from scratch after a crash landing might be better than holding up former design principles. : /

Have you heard about that Youtuber, who is going to buy the Commodore brand, probably inteding to release new retro computers?
Totally crazy, but who knows. Affordable HW would be a good first step. People try all kinds of stuff. Something will work… : )

Yeah the growth type or maybe convergence type generation is quite useful for something like a town layout or something. But then if the towns grows into a city, you can't just get a completely different result suddenly, for the sake of temporal consistency of the world. And for the sake of being able to simulate it at multiple layers of abstractions, you also don't want a town to grow across simulation cell boundaries or something, so that its statistics strictly count towards one cell, etc. when you squash it. And when you un-squash it, you need to get the same state back. And when you add a road between the towns of two cells, that should also have a deterministic path, even if one of the towns grows or shrinks or disappears. And I don't want to save the full height map of each region, and all the assets like trees and rocks that get spawned in there, I just want to remember whether there is a road, whether there is a town, etc., and how many people work there on which things, and how long / difficult the transport routes are to neighbouring cells, so that I can run a month's worth of economic simulation with just a few simple formulas, and not having to run a swarm behaviour algorithm, or actually trace the road length and elevation changes of the road between the neighbouring cells' towns, of a road that got generated via ants or something. So I am in favour of trading off realism / organic growth with a reverse causality type of thing where I generate plausible parameters ahead of time and then generate a road and town placement that fits all the pre-decided metrics. But I'm starting to repeat myself. I just think my approach is less computationally expensive, even if it is probably more complicated to pull off well. But it lets me simulate far-away regions properly even if I never generated any terrain there, allowing me to do unlimited-scale simulations without issues and without offline preparation phase. So I can drop into an entirely new world's starter village and already simulate entire continents of political and economic data with almost no effort. And only when I actually want to enter a town or a region, do I generate its appearance, its inhabitants, etc. And then I just need to make sure that if I come back later, it re-generates the same appearance again, and remembers concrete changes that the player already witnessed. All this without gigabytes of offline generation, or hours of offline computation. It all would take just a few milliseconds if written well. At least that's my intuition for it.

And I wasn't really considering the problem of “how do I generate a city layout". I was more aiming at “given I know I want to have a city, but do not want to generate the city before a player ever comes close to it, how can I accurately simulate its behaviour and relationships with the rest of the world?”. I can look into how to solve the actual generation separately. Because ideally, I want a system that works at global scale, but does not ship with large amounts of baked data, and does not take lots of memory or computation to run on a client machine. And it needs to allow for new worlds without waiting for long. Ffaraway regions and other continents should not cost time until you actually set foot on them, while still “existing” already in terms of the mechanics / world simulation. And concretely instantiating them and simulating them on a more detailed level should still yield very similar results to the “squashed” simulation.

If such an approach is found, it would let me simulate entire worlds even on a smart phone (just for example — not to actually condone smartphone games…), and let me “zoom in” and out arbitrarily. And it would be lightweight enough to be just a minor workload in the background of an RPG or FPS or similar. And the ability to zoom out without a noticeable loss of accuracy also means that the visual layer is mostly isolated from the world behaviour. Which means I could have a text-based game, a tilebased 2D game, or a fully 3D game with the same world simulation. Meaning I can massively vary the production value without an actual change in overall world behaviour. Only the on-screen visible parts would change, but the world would still fundamentally be the same, and actions would still have the same outcomes. And it wouldn't really matter whether I had a voxel-style 3D world or a realistic 3D world.

Which would mean if I started out with a text-based game, I could focus on building the mechanics, and later on switch out the visual layer with a 2D or 3D version without touching the world simulation part. Which means I can treat the visualisation as disposable and focus entirely on building the actual game mechanics first.

Anyway, it's always a matter of personal preference of the programmer, and also of his personal experience and insight and personality which determine what approach he likes best, or which seems most intuitive for him. So I don't think we'll ever really come to a full agreement on the matter. Maybe, if any of us ever get around to actually building a game that tries to fulfill these ambitions, we'd be able to argue about who pulled it off better, lol. And terrain generation doesn't exist in a vacuum, so all the secondary and tertiary goals I consider in the back of my head also make me choose that route, while you might have entirely different secondary and tertiary goals, meaning, we aren't even trying to solve the exact same problem.

@frob It is true that none of these things are universal, but if you want to seek out something that does not exist yet, you need to identify a concrete path towards that by reasoning forwards and backwards, identifying rules and constraints and possible options in the process. That is why we are making all these rules about what we do or do not want from the kind of game that we want to see. Maybe we're just two weirdos who want something that nobody else wants. But we are not satisfied by what exists. Maybe most people are satisfied by what already exists, and to them, all these rules do not apply.

Given the current trends I observe, I personally wouldn't trust that anyone will magically build what I desire. Across the entire software industry (games or not), everything gets more and more disappointing as time goes on. And I don't see a turning point in sight, and I also don't see people really talking about it productively (only complaining, if at all). The age of dreams and visionaries seems to be ending, so I think it is important to have opinionated people with clear dreams and visions, if we want to have any prospects of a revitalisation of any aspect of society. The only idealists I see besides woke activists are the rust fetishists with their deranged “safety" masochism, so that's not going to lead us anywhere, either. Nobody really seems to have unique dreams and visions anymore.

There's lots more I would have liked to reply to, but this is already way too long again, and probably a chore to read…

The thread so far was great, though, and helped me clarify my own understanding and also consider a few new perspectives. All this depth just for one tiny aspect of the game design. There is so much more to tackle, like concrete RPG mechanics, the fundamentals of the game economy (like, professions, crafting, NPC merchants & craftsmen, spell systems, combat systems, stats, levelling, skills, item types, where does gold come from, etc.). There was also this thread that I chanced across recently: MUD/RPG economics (so ancient, it is from an entirely different era). There was lots of debating about money circulation, whether monsters drop money, the importants of crafters, and whether monsters should drop anything other than a corpse that you can harvest materials from, etc. It was pretty interesting, and was what prompted me to make this thread initially. I think MUDs are also great for solo projects because they don't require any visual or graphical assets, only textual descriptions. And you can always take a text-based game and upgrade it to a graphical game afterwards.

RmbRT said:
Which means I could have a text-based game, a tilebased 2D game, or a fully 3D game with the same world simulation. Meaning I can massively vary the production value without an actual change in overall world behaviour.

Ok, i get your vision.
I don't intend to propose ‘better ways’ to get there, i only picked up the claim ‘true intelligence required’, then proposing ways to overcome the obstacle if needed.

Ofc. my own (offline) work on proc. gen is something entirely different. So it's never meant to be outsmarting, or some form of competition. ; )

RmbRT said:
Maybe we're just two weirdos who want something that nobody else wants.

Pretty much all people i know are like that, it's not just the two of us. It's the whole generation, from my perspective.
I would not talk so negative otherwise, which maybe i shouldn't.

I'll have a 2-month vacation soon (forgot to take my vacations last year), and then another month near the end of the year, so I hope I can get around to building a proof of concept for that kind of stuff. Probably text-based, but maybe also something like a mix of graphics and text. It would be great to have something tangible. But then again, I also didn't really get any compiler work done this year…

I'd like to at least get the system done where you can zoom in and out of a simulation's level of details where it then comes up with retroactive causality for the various actors. That system could also be used for novel writing, actually, to keep track of stuff. I guess the closer it gets to a novel world building system, the closer it also gets to a system that makes a great game world.

Other than that, I got lots of ideas for an RPG that would only really be realisable in a fully 3D game. Sucks to have such a high entry barrier to actually prototyping most ideas one has. One such idea is that you can customise parameters of your spells, such as cast speed vs. energy consumption / power, cast speed vs. mana efficiency, power vs. mana invested, and then stuff like area of effect vs. projectile size or projectile speed, etc. Like the same fireball skill could be used as quick / rapidfire fire bullets, or a large-scale siege meteor, etc. And then you could maybe select some parameters ahead of time, and others could be controlled by how long you press the skill button for, or something. And then you could theoretically take a minute to channel the spell to make it huge at good efficiency, or channel it for shorter with the same power, but consume much more mana. Stuff like that would add tremendous versatility and depth to even basic things. The same would go for archers, you could have a power shot that channels for a long time, and trade off channeling time, range, damage, velocity, etc. Or area heals of different intensity and area and cost. The more power or the shorter the cast time, the more expensive the spell becomes, but your stats would also limit the extremes, for example your maximum mana pool, your maximum mana expenditure per second, and also the mana dissipation rate while channeling (so if you channeled mana into a spell at the rate you regenerate, you can pour infinite mana into a spell, but the mana stored in the spell also dissipates over time, based on a corresponding attribute). This would let a mage play as siege unit, as assault unit, as suppression fire, etc. And you could also have some kind of long-preparation one-shot kill strategy where you mask your preparation time and then send one attack that you channeled all your power into for minutes, maybe with a very small and dense projectile, trading off AOE for damage, etc..

JoeJ said:
I don't intend to propose ‘better ways’ to get there, i only picked up the claim ‘true intelligence required’, then proposing ways to overcome the obstacle if needed. Ofc. my own (offline) work on proc. gen is something entirely different. So it's never meant to be outsmarting, or some form of competition. ; )

I didn't want to kill the fun, but I think we reached a depth where if we go any deeper into it, we completely lose each other. I don't want to turn this into a tedious discussion, is all. I hope I can manage to refrain from going into more detail for my idea regarding world simulation, but feel free to elaborate on your idea, and the purposes / vision you have for it. If you do, I will try not to convince you of my ways any longer, haha. After all, we already have a lot of common ground. I feel like I did most of the talking.

JoeJ said:
Pretty much all people i know are like that, it's not just the two of us. It's the whole generation, from my perspective. I would not talk so negative otherwise, which maybe i shouldn't.

One of the most quintessentially German proverbs is “Da kann man nicht meckern” (one can't complain about that), which is the highest and most earnest praise you will realistically encounter from a German. I think the culture around not complaining and criticising that's prevalent in the anglosphere is just hindering progress and leading to pandering and flattery. So what if you hate the state of things? Pointing out faults is the first step to making the world a better place. Harsh comparisons against the ideal do not devalue, but actually point out aspects for improvement, and also constantly keep the ideal mentally present. If you lose sight of what ought to be, all progress becomes impossible, and only confused fumbling can ensue.

RmbRT said:
One such idea is that you can customise parameters of your spells, such as cast speed vs. energy consumption / power, cast speed vs. mana efficiency, power vs. mana invested, and then stuff like area of effect vs. projectile size or projectile speed, etc.

There was an indie game, +10 years ago, made by some industry veteran in Unity. Iirc, it was ‘The Inner Circle’. Did not receive a lot of attention, but it's extremely good. The setting is game dev. You play a cancelled game on some server, which transformed from FPS into RPG, or something like that. And a primary mechanic is that you can hack enemies. You see a code class and it's fields representing character properties, like the ones you just said. And you can change the values. Mostly you use it to turn enemies into allies. For example, you hack a flying enemy to become friendly, then you can fly on its back to reach places you could not reach otherwise. This really makes you feel smart and is super satisfying. Awesome game. :D

RmbRT said:
which is the highest and most earnest praise you will realistically encounter from a German.

That German must be pretty old, otherwise he'd say ‘geiler Scheiss!’, or something like that.
Well, i'm from Austria, which is a slightly different mentality. But if you say this explains all my rant, i'll take that.
And maybe you're right about such regional habits. I need to observe… ; )

lol, I think “geiler Scheiß” is something fairly recent. Maybe I was just lucky and my parents were raised by more conservative people and they themselves also did not live with the moral progressives too much later on. I think the prussians and their virtues are still held in such high regard because they were obnoxiously insistent on the ideal.

The hacking game seems unique. No idea whether I would enjoy it, but definitely made interesting choices. There is this sci-fi story, called “World trigger”, where people have interdimensional territory battles or something, and the combatants get trained in the handling of energy weapons that take their energy from some mystical energy that accumulates in the body, so they have to be very conservative, creative and efficient with their usage. They use some kind of hand-held device that lets them channel that energy, and there are multiple types. There are for example energy blade weapons that let the user freely change the shape of their blade or even extend the blade from arbitrary places in the body, allowing for highly creative close combat. There are short-range and mid-range guns that for example let you either create a powerful bombardment, or precise, redirectable shots (that can shoot around corners etc.), or sniper type rifles that can also turn into siege weapons based on how long you charge a shot (although most do not have the right physique for that type of shenanigan). I think that kind of thing would be quite interesting for an FPS enthusiast. It's basically the same thing as what I proposed for fantasy RPGs with magic, but for a sci-fi action FPS setting. Of course, you can't copy that 1:1 because it's probably infeasible to trace crooked bullet paths ahead of time in a real-time combat situation. Would work really well for a turn-based game, though. But even for a real-time game, there should be some kind of equivalent. Blasts that charge longer, but shoot faster, straighter shots, blasts that charge longer, but create bigger explosions, some kind of garden hose weapon that can spray a cone of bullets of varying narrowness, etc.

And regarding RPGs and their economy, I think if money itself is actually some kind of consumable XP crystal dropped by monsters (or humans), and every item has to be crafted by a crafter from raw materials, instead of just transmuting money into items at a merchant, that would probably lead to a much more cooperative playstyle, especially if there are no NPC crafters. Then players have to create a division of labour, but the craftsmen would get levels by selling wares, while warriors would get levels from battle. If the game also has permadeath, then you also have less egoism, as people don't pour years of effort into a character, and instead are always reliant on the collective strength of the other players (for access to high level crafting techniques, for access to high level materials, etc.). Instead of everyone just looking after his own interests and everyone's progress being tied to his own character. And then you also no longer need a per-character story line, where every player saves the world over and over with each character, even though it was already saved.

P.S.: However, I am actually still uncertain how much I like the concept of levelling up, actually. Yes, it faciliates strength progression, but in a real fantasy novel, you don't have or need levels. I didn't see Frodo or Gandalf ever getting a level-up notification… An FPS can also get by just fine without you turning into a superhuman.

RmbRT said:
P.S.: However, I am actually still uncertain how much I like the concept of levelling up, actually. Yes, it faciliates strength progression, but in a real fantasy novel, you don't have or need levels. I didn't see Frodo or Gandalf ever getting a level-up notification… An FPS can also get by just fine without you turning into a superhuman.

To me this is a very big reason of why modern games fail on me.
The problem is, idk which advantages levelling up will give me. Should i invest in strength or charm? Should i get the levitation or the agility ability?
I can't decide, becasue i don't know yet. This causes me serious uncertainty and stress. In practice i postpone the decision, and then i end up playing the whole game without any extra skills, abilities, and what not, and thus i have a harder time and more frustration, on top of the uncertainty and stress.
Oh, and there also is the need to go through boring GUI pages of stats and skill trees, which i just hate, and i want to ignore this stuff.

This is very, very bad to players like me. But still, you can hardly find a game these days without all this ‘bolt on complexity crap’.

All this was done so much better in the past. E.g. the Power Ups in Super Mario. You collect them and you have it. Simple and just works. And it's temporary until the next death, so you don't even need to introduce a concept of permadeath, which sounds more like torture than fun.
Later they extended the concept. If you already have a power up and collect a new one, it goes on top of the screen into a hud element.
You can click a button to release it. It then falls down from teh top edge hud of the screen, and you have to catch it up. If you miss it, it's gone. Sometimes you could not reach it, so you have to be careful when to drop it.
So they increased complexity, but they avoided interruptions, and even made it a minigame semlessly integrated and adding to the main game. Good design.

Then 3D came up, and imo Super Mario 64 was a total failure due to shitty camera controlls.
Quake took the lead. Same acceleration based movement, same simple power up system.
Later Half Life 2 added the ability to pick up objects, throwign them, or stacking them. Which Super Mario already had long before.
At this point the 3D action game had successfully catched up with the 2D masterpieces from before.

It was the peak of gaming history. But it did not work for consoles and their shitty gamepads. You could not control your view directly, so you could not aim.
But console makers had no balls and refused to make better controllers. They were afraid their players would not like a trackball or a touchpad.
So game designers had to improvise, working around an unsolvable problem.
They came up with cover mechanics to slow the games down, so there is more time to aim, but less action and fun.
It was not great. So tried to add depth to the game, introducing RPG mechanics like skill trees. Interrupting gameply with video totorials to explain how all those overcomplicated mechnics at all work.
Hell, they even made the first level entirely about explaining all those barely working mechincs. No fun at all, and worse: Players like me have forgotten everything from those annoying tutorials already 5 seconds later.

It was a disaster. And while it became better, the medium has never recovered from all those failed attempts to work around gamepad limitations. It died a slow death, and it also explains why they make no games for older gamers, who grow up with the medium, and still want to play. The reson is: Only young people play bad modern games, because only young people don't know about the older games which were so much better.

But it's not jsut about gamepads. It was not enough. We needed another devil beside that. IT was NVidia, introducing raytracing coming with a big price hike on GPUs. Nobody askid for this, and worse: Nobody was asked about teh flexibility we woudl need to make the technology actually useful. But there it was, and it was the future as dictated. Crypto mininig bade it even worse, just at the right time. And the result: GPUs are unaffordable, and the PC platform has a cosntantly shrinking userbase exactly since 2019. They transfor to cheaper consoles. I was right all along - NV is killing the PC platform. They don't even care, since it's all about AI.

No PC platform means no more Mouse at all, but also no more developers. Becasue without a home-computer, kids can not learn to code while they are still young, which is needed, because gamedev is hard.

So that's in a nutshell how the gaming industry is killing itself.
The future is vibe coding and playing on server side 100tf GPUs, because nobody can code, and nobody can afford the HW needed to run the inefficient ‘future of graphics’.
But nobody sees it coming. Like lemmings, they tell us ‘don't buy 8GB GPUs - you need 16 to run the latest and greatest!’.
They do not realize the rising conflict of interests between gamedevs and HW industry, and it will be too late when they finally do.

You build your own ASIC cpu and compilers?
That's crazy. A waste of time.
But you're right. So i wish you luck. Something will work… ; )