Randomized Object Stats, Randomized Quests?

[guest509]

  What I think you might be saying is that procedural architecture and spawns create re-playability, exploration and surprise. Hard coded Items and Monsters create the strategy and balance (assuming an adequate drop rate).

  I guess completely random stat Items and Monsters would create a sort of identify heavy game. The ID game is my most hated aspect of roguelikes. It's like a frustrating chore you must do in order to play...but that's just me.

  As a bit of a compromise I was thinking of a game that had binary mutants, like in Ancient Greece. Bull+man (minotaur) goat+man (satyr). Each part of the beast has hard coded stats and abilities. For example Bull would have a charge attack and bonus hit points while the goat would have the ability to eat anything it wants.

  So instead of learning the monsters and weapons, you'd learn the components of monsters and weapons.

"The Massive-Robo-Lizard-Man is killed by your Full Metal Jacketed Explosive Tipped Energy Javelin?"

  That may be a bit over the top...but then again....

  The real issue would end up being balance. There is bound to be monsters and weapons that break the game.

[guest509]

  For modern weapons you could create them procedurally.

  Muzzle Velocity
  Ammo Type
  Accuracy
  Penetration (this is actually based on velocity, range, gauge, bullet density and shape)
  Flesh Damage (based on velocity, gauge and the shape and mash/shatter characteristics of the bullet)
  Rate of Fire
  etc...

  The real issue would be communicating this with the player so they know how to use this weapon. Also, thematically, naming gets to be an issue. Heavy Auto Shotgun and Penetrating Energy Caster says quite a bit about what you are wielding, but being able to describe the weapon in ways the player understands without having to stare at the stats puts limits on what is possible. At least from a fun perspective.

[AgingMinotaur]

I guess completely random stat Items and Monsters would create a sort of identify heavy game. The ID game is my most hated aspect of roguelikes. It's like a frustrating chore you must do in order to play...but that's just me.

If properly done, a (pseudo)random world might also make id-ing less important. In fact, setting up a traditional id subgame might be difficult in such a game. It would be much more important to find flexible ways to convey this information to the player, rather than finding ways of keeping it hidden.

Of course, the player would still get to know the game more and more. But instead of going: "There are six of this potion in my inventory, so it's got to be healing", an accomplished player might go: "In this world, feline predators are super fast forest dwellers, so I need some strategy to counter that before I enter the jungle."

As a bit of a compromise I was thinking of a game that had binary mutants, like in Ancient Greece. Bull+man (minotaur) goat+man (satyr). Each part of the beast has hard coded stats and abilities. For example Bull would have a charge attack and bonus hit points while the goat would have the ability to eat anything it wants.

My abandoned game Squirm does this to a certain extent. Centaurs have one horse part and one human (sometimes animal) part, tweaking the individual's stats. So one described as "the head of a baboon, attached to a horse's body with four legs" is much faster and stronger than "the head of a horse, attached to the body of a child". At later levels, centaurs spawn with stronger body parts. In the early game, one also encounters monstra biformii (brown C's), which similarly consist of two animal parts.

I didn't develop very much content for these systems, but for instance, chimerae with bird parts are flagged as flying. And of course I had grandiose plans Setting up the system was trivial (some classes of monsters/items/features pick and inherit random subclasses when the game is initiated, others have every singly individual pick their unique subclasses when they're spawned). As you already said, getting good and balanced content in place is always the difficult part.

But it can be done!

As always,
Minotauros

[requerent]

 What I think you might be saying is that procedural architecture and spawns create re-playability, exploration and surprise. Hard coded Items and Monsters create the strategy and balance (assuming an adequate drop rate).

Well, the idea is that you don't randomly generate monsters, you randomly generate incomplete paradigms and archetypes, and then you pseudo-randomly generate monsters, filling in the blanks of the archetype as you go along.

What I find lacking in roguelikes is a complete lack of indication as to what is going to happen. If you have tablets, tomes and cave-drawings that gave you information about what you may face next, you don't have to make it a game of surprise. In a game where we generate everything, we don't want players to die frequently, but to appreciate the intricate landscape and what not.

The real issue would end up being balance. There is bound to be monsters and weapons that break the game.

Genetic Algorithms!!! You can test any randomly generated creation using background cycles with GAs. You don't really just set bounds, fill in parameters, and expect for the game to be playable? I mean, you could have a set of base stats/qualities and give each generated monster 20 extra points, or so, to pump into stats/abilities- but that does assume that all the skills are somehow sensible and balanced... Beam attacks, for example, in ToME are ridiculous. If a set of 8 monsters spawned with beam attacks you'd have a problem...

[Holsety]

[...]
What I find lacking in roguelikes is a complete lack of indication as to what is going to happen. If you have tablets, tomes and cave-drawings that gave you information about what you may face next, you don't have to make it a game of surprise.

NPPangband has mystical wall writings that give you hints about monsters on your current floor (they disintegrate when you read them too often).

Brogue might have some subtle indicators as to what is going to happen (ie you see a locked door; you know there's going to be something key-based going on on this floor).

IVAN drops indicators to the upcoming first boss battle by spawning those goddamn plants every so and so on that floor.

That's three examples I can remember where a roguelike gives you an indicator of what's coming up, while still remaining close to Rogue.

ADOM gives clear indicators once you step foot on the world map.
Legerdemain gives indicators, but that's really pretty much a barely-roguelike story-driven RPG.
Sword in Hand gives you mission briefings...

Most roguelikes give you one indicator; there's this floor you gotta go to, see? And there's this thing there, see? It'd be really swell if you could bring it back to the surface. (Or something along those lines usually  )

Any MORE indicators you drop and you're rapidly going towards story-driven gameplay.

I don't really see a solution though... All my imagination comes up with is "You see a mysterious sign, conveniently left by some previous adventurer. It tells you there might be grues ahead in the darkness."

[...]
 In a game where we generate everything, we don't want players to die frequently, but to appreciate the intricate landscape and what not.

I disagree!

First off, I like my games to kill me. I want them to try to kill me as brutally and quickly as possible, while remaining fair. I'd rather have a game that tries to kill me like Nethack than a game that tries to kill me like Angband.
(Ie angband not giving you any challenge untill it decides to kill you with a monster that has more speed than you)

Second, I'd appreciate the intricate landscapes if there were any intricate landscapes to appreciate! Most games either take rooms-with-corridors or big-empty-plain-with-some-trees (whoop dee doo).*

Only roguelikes with real landscapes to appreciate I know of are Brogue and Incursion (especially incursion).

* I'd like to add here I avoid playing roguelikes with tiles/graphics.

[requerent]

[...]
What I find lacking in roguelikes is a complete lack of indication as to what is going to happen. If you have tablets, tomes and cave-drawings that gave you information about what you may face next, you don't have to make it a game of surprise.

...

Most roguelikes give you one indicator; there's this floor you gotta go to, see? And there's this thing there, see? It'd be really swell if you could bring it back to the surface. (Or something along those lines usually  )

Any MORE indicators you drop and you're rapidly going towards story-driven gameplay.

I don't really see a solution though... All my imagination comes up with is "You see a mysterious sign, conveniently left by some previous adventurer. It tells you there might be grues ahead in the darkness."

I'm taking issue with the idea of randomly generated creatures (more or less what the thread is about). If you see a goblin and you've played the game previously, you know what to expect from a goblin. Even if there is a bit of random stat-fudging.

If you have genuinely randomly generated monster archetypes and paradigms then the player is incapable of having any idea of what to expect without being given some degree of previous information.

Anyhow- you typically don't get those indicators in the game, but from playing it. Those indicators should somehow be integrated with the UI in a meaningful way. My first run-through in Brogue I didn't know that the Key was specifically for the level it was found on. I didn't know that the only spawned potion was necessary to get to the door (or see it). It wouldn't be unreasonable for the Key itself to provide some form of indication. Brogue's philosophy, in regards to mechanics and UI, is transparency, but knowing the Key is on the same level isn't hinted at in the least. The Key is only a hint that there is a treasure room on the floor when you've already familiarized yourself with the game.

Ivan is one of the more Mature RLs, IMO. It does so much so wonderfully...

That said- markers or indicators don't create story. If your RNG whips up a fire-breathing race of lizards, having some swampy pools and scorch marks in the area may be ample. Similarly, if you have a Key your character might make the observation that it is made of similar materials as other constructions seen on your current depth. Etc.

There's also nothing wrong with being story-driven. IVAN, again- is wonderful in that regards.

[...]
 In a game where we generate everything, we don't want players to die frequently, but to appreciate the intricate landscape and what not.

I disagree!

First off, I like my games to kill me. I want them to try to kill me as brutally and quickly as possible, while remaining fair. I'd rather have a game that tries to kill me like Nethack than a game that tries to kill me like Angband.
(Ie angband not giving you any challenge untill it decides to kill you with a monster that has more speed than you)

Second, I'd appreciate the intricate landscapes if there were any intricate landscapes to appreciate! Most games either take rooms-with-corridors or big-empty-plain-with-some-trees (whoop dee doo).*

Only roguelikes with real landscapes to appreciate I know of are Brogue and Incursion (especially incursion).

* I'd like to add here I avoid playing roguelikes with tiles/graphics.

Thread context. If we put a lot of care into monster generation, then we would like to give the player a decent amount of opportunity to explore what that uniquely generated world has to offer. I'm not saying that the player shouldn't die, but they should live long enough to engage the sense of wonder we're likely trying to produce.

[guest509]

  Yes. Communication to the player of what he's facing is the issue. With tiles and binary monsters you can do paper doll monsters. Top is a lions head, bottom a spider's body and feet. That sort of thing. Show it graphically. Can you do that with text? It would look a bit wrong I think.

  With ascii you can use color, capitalization and letter. You might also use bold and itallic in some way. Or, like Brogue, you can list monsters in sight on the side of the HUD. By the name (Slimy Bitch Mommy) you can discern stats and behavior from knowledge gained through play and tropes. No need to communicate too much graphically. The above monster example would be a capitol 'B', have a viscus quality and give birth to canine minions. :-)

[Holsety]

I'm taking issue with the idea of randomly generated creatures (more or less what the thread is about). If you see a goblin and you've played the game previously, you know what to expect from a goblin. Even if there is a bit of random stat-fudging.

If you have genuinely randomly generated monster archetypes and paradigms then the player is incapable of having any idea of what to expect without being given some degree of previous information.

Anyhow- you typically don't get those indicators in the game, but from playing it. Those indicators should somehow be integrated with the UI in a meaningful way. My first run-through in Brogue I didn't know that the Key was specifically for the level it was found on. I didn't know that the only spawned potion was necessary to get to the door (or see it). It wouldn't be unreasonable for the Key itself to provide some form of indication. Brogue's philosophy, in regards to mechanics and UI, is transparency, but knowing the Key is on the same level isn't hinted at in the least. The Key is only a hint that there is a treasure room on the floor when you've already familiarized yourself with the game.

Ivan is one of the more Mature RLs, IMO. It does so much so wonderfully...

That said- markers or indicators don't create story. If your RNG whips up a fire-breathing race of lizards, having some swampy pools and scorch marks in the area may be ample. Similarly, if you have a Key your character might make the observation that it is made of similar materials as other constructions seen on your current depth. Etc.

There's also nothing wrong with being story-driven. IVAN, again- is wonderful in that regards.

Ah, that explains so much. Then what about God of Change?
It shows the full bestiary on screen, in a list. Every so and so turns the evil God may choose to add/remove properties to the monsters (territorial/friendly/fire-breathing/telepathic/undead etc) and your bestiary gets updated. That way the player knows what to expect.

Then again, it wouldn't do for a game with more monsters, eh?

  Yes. Communication to the player of what he's facing is the issue. With tiles and binary monsters you can do paper doll monsters. Top is a lions head, bottom a spider's body and feet. That sort of thing. Show it graphically. Can you do that with text? It would look a bit wrong I think.

Kind of like how Crawl handles the tiles for player characters. Would probably be the cleanest way to do it.

[yam655]

I've thought about making a game where the player has no idea what sort of monster they're facing.

"Hey, look. I think that's a... oh my god, what is that thing?!"

This is a type of gameplay more reasonable for ASCII-based graphics than for tiles. You want to depict a location on-screen without giving an indicator as to what it is.

My idea is that, (1) the player could identify certain aspects of the creatures depending on what is being done, then (2) visit local bars/taverns and gossip about what it might be.

You wouldn't even need to indicate what is being seen, just keep a counter of the amount of time observed.

And, of course, some of the gossip would be wrong. Better still, the wrong gossip could actually favor monsters the player has seen in earlier games. The wrong information would then play to the players' expectations.

You wouldn't even need to provide a computer-generated name. You can make it random and still make the player name it.

Then about random quests...

Even simple fetch quests are about narrative. A story-driven game has quests that relate to each other within a broader framework. The problem with quests outside of a narrative framework is that you're left continually dealing with short-term issues which act as distractions from the main plot.

Quests are side-plots that shouldn't feel tacked on to the main story arc. There should be reasons the character wants to go on optional quests -- outside of just being a fan of completeness.

For me, it is better to have compelling game play and no quests than it is to have boring forced quests.

Quests should never feel like grinding. In some games they can feel that way.

Cheers,
Steven Black

[Leaf]

There is a tabletop RPG plot generation system called "Mythic GME" that I really, really like.  It tracks rising tension and throws in plot twists and other interesting things.  It's pretty awesome to use with random solo-play stuff like the 1st Edition DMG, Dungeon Bash, etc.  :3

The problem is, it requires a great deal of human intuition to interpret the results.  I haven't been able to figure out how to adapt it to a computer program without the results just turning into a weird mess.  D:

I think the human intuition is the key thing.  The computer just can't provide it.  So with sandbox-style games that offer no random quest system but /do/ offer a high degree of simulation, the player kind of makes up their own plot/quests, using their own intuition, and the game is Fun.  But if you take the same game and add in some kind of technically (in the programming sense) cool but still obviously machine-generated quest/plot to it, the player looks at the quest/plot that isn't really very different from the last 20 quest/plots, finds that it is Lame, and gets bored.

If there was some sneaky way to get the game to underhandedly "read" things from the player's human intuition as play progresses and recycle it back into a plot generator, that would be cool, but I see too much possibility of the results going weird to make it robust. :s

[requerent]

There is a tabletop RPG plot generation system called "Mythic GME" that I really, really like.  It tracks rising tension and throws in plot twists and other interesting things.  It's pretty awesome to use with random solo-play stuff like the 1st Edition DMG, Dungeon Bash, etc.  :3

The problem is, it requires a great deal of human intuition to interpret the results.  I haven't been able to figure out how to adapt it to a computer program without the results just turning into a weird mess.  D:

I think the human intuition is the key thing.  The computer just can't provide it.  So with sandbox-style games that offer no random quest system but /do/ offer a high degree of simulation, the player kind of makes up their own plot/quests, using their own intuition, and the game is Fun.  But if you take the same game and add in some kind of technically (in the programming sense) cool but still obviously machine-generated quest/plot to it, the player looks at the quest/plot that isn't really very different from the last 20 quest/plots, finds that it is Lame, and gets bored.

If there was some sneaky way to get the game to underhandedly "read" things from the player's human intuition as play progresses and recycle it back into a plot generator, that would be cool, but I see too much possibility of the results going weird to make it robust. :s

That could just be an issue of flow management. If quests are chained together with interesting degrees of branching factors and result in visible changes to the simulation, then the quests don't have to be singular atoms.

IE- If there is a quest to kill the dark lord by the year's end and you fail, the simulation could adapt to become an undead warzone. It just depends on how you want to approach the problem. A variety of quests with different scales will keep people interested.

[AgingMinotaur]

There is a tabletop RPG plot generation system called "Mythic GME" that I really, really like. [...] The problem is, it requires a great deal of human intuition to interpret the results.  I haven't been able to figure out how to adapt it to a computer program without the results just turning into a weird mess.  D:

That could just be an issue of flow management. If quests are chained together with interesting degrees of branching factors and result in visible changes to the simulation, then the quests don't have to be singular atoms.

IE- If there is a quest to kill the dark lord by the year's end and you fail, the simulation could adapt to become an undead warzone. It just depends on how you want to approach the problem. A variety of quests with different scales will keep people interested.

Easier said than done. Just writing/tweaking the content for a scenario like this would take weeks, if not months. You'd need dozens of branching scenarios, some small and some big … throw stuff like insanely subtle&complex AI into the cauldron, and you might have a nice game in ten years' time.

I think one could contrive a system that takes human intuition into account, not by "intelligently" filling in the gaps, but by leaving stuff to the imagination much like a pen and paper "random plot generator" does. Let the player connect the dots that the algorithm is spitting out. Unfortunately, I haven't really a clue how such a system could be implemented in a computer game. A possible starting point could be a structure inspired by the "Hero's Journey" archetype[1] or Vladimir Propp, or some pattern used in Tarot or other divination. In any case, it'd probably require more wits than established know-how to pull it off.

As always,
Minotauros

[1] I think a RL based on "Hero's Journey", with randomly chosen twists for every step in the narrative, could work well. It's not an idea I see myself using for a game, though.