Killing Mister Cavern?

[Koumei]

Sure, DF is impressive...

It's a mess, I think is what you're after. It should also be on the flyers and posters for Autism Awareness Week, seeing as its creator is the fucking definition of an autist.

[Lokathor]

But if you're also autistic, it's flippin sweet.

[Endovior]

I shrug.

Dwarf Fortress basically demonstrates the limits of what's possible in a computer simulation, presuming that the programmer of said simulation maintains an insane obsession over the course of years.

That 'mess' is serious the best it gets, which says something about the possibilities inherent in simulations to begin with.

[Koumei]

Apparently, I'm insufficiently autistic then. Or maybe it's my hatred of dorfs. Or only autist males like it or something.

Even changing it to replace dorfs with moeblins didn't really make it enjoyable.

[Murtak]

Dwarf Fortress is a great example of procedurally generated worlds, which is far better than randomly generated worlds. It still does not compare to actual humans though, if only because a human narrator can provide drama, emphasis and detail at will or just make shit up on the fly, without any rules or guidance whatsoever. Procedurally generated content is great for creating a sandbox for you to play in, because humans suck at creating an entire world or even a village from scratch. So ideally you want procedurally generated content and on top of that content actual humans add the plot, adding, removing or linking the pregenerated parts as they see fit. Classic random encounter tables are a basic version of this. The DM would roll a die, see that 2 gorgons will ambush the party and then describe this ambush. Perhaps he would add detail, wait a little until the party enters the forest or say "screw this, we are just outside the city" and roll again.

There is no reason you could not do something similar for plots and quests or possible even for campaigns. You will need a sufficiently smart generator of course. The better the DM (or players) is at filling in the details the worse the generator can be. And I can easily imagine abolishing the DM for many or even all parts of the game and give control to everyone at the table. But what I can not imagine is scripted combat (or conversations for that matter). Randomly generated opponents acting on scripts basic enough to handle with pen and paper? That just has to be boring as shit.

As a DM when my random encounter calls for a dozen kobolds with bows I will probably insert a kobold with a tower shield and a choke point. Perhaps I will have the fight take place in a location that hinders the players movements. I will take advantage of the players' mistakes, I may let the kobolds run away at the start of the battle, after the first volley or right in the middle of combat. I may give the ambush away or have one of the kobolds make a tactical mistake during combat. Heck I may have one of them shoot their leader in the back of the head if the dungeon description mentions rival factions. Scripted fights take this away from me or from whoever has to run combat.

I may be missing something, but as far as I can see scripting games will turn them into bad computer games. Sure, we can depart from the scripts at any time, which is a vast improvement over computer games, but at that point we are not scripting anymore. Basically scripting is strictly inferior to actually designing content. Designing content is hard, and scripting is hard only for whoever has to design the scripting engine. So anytime you want to have a lot of content but can't afford to design all of it scripting is a good idea. It certainly is great to get a rough sketch you can then improve on. But scripting is a crutch. It may be necessary, but it should not be the goal.

Short version: I am all for better random generators, for encounters, quests, dungeons, campaigns, worlds, whatever. I am all for basic combat scripts. But they should be guidelines on how to run the game, nothing more. If you run a single combat or quest straight from the random script you are probably doing it wrong.

[Saxony]

I shrug.

Dwarf Fortress basically demonstrates the limits of what's possible in a computer simulation....

Current limit. Just nitpicking, you already knew based off your "until artificial intelligence" comment.

Even so, I wouldn't peg Dwarf Fortress as the current limit. It runs on household computers and is made by a small, underpaid development team. Saying DF is the current limit of anything is arrogant to the extreme.

[PhoneLobster]

A computer has no creativity.

Neither do a lot of Mister Caverns. As an example let me suggest whoever the dumb human bastards who wrote every bioware game storyline ever are. Because they write such boring tripe filled plots that seriously a random generator WOULD be better.

This is fundamentally a problem that cannot be solved with existing technology. Computers cannot go outside the strict procedures they've been given. Until and unless actual artificial intelligence is developed, this will absolutely remain the case.

Well. Frankly. NO.

Sure a computer doesn't understand what the heck it is doing. But the "problem" of dwarf fortresses procedural content generation making "incoherent" story lines lacking clear interwoven narrative associations and stuff is merely a problem of the focus of the scripting.

Frank already gave some vague outlines of how to procedurally generate adventures and they DON'T start with thousands of years of procedural dwarf fortress simulation.

They start with scripting rules about beginnings, middles, and ends. They focus on things like and actors, events, and outcomes. Things like the names of guys in the village and the layout of buildings and ruins are generated as secondary products of the more important material related to the story being generated, not the other way around as with DF.

You CAN write scripting like this, it WILL be "dumb" in that the computer WON'T understand what it's doing when it executes the script, but it certainly is RIDICULOUS to say "this can't be done with scripting".

I mean you could start writing a simple set of tables to script some basic story for a gaming session just like that...

Actor Generation Tables
(Every time you need an actor roll 1d6 on each table to determine the actor type)
Main Actor Type
1) Princess
2) Dragon
3) Wizard
4) Troll
5) Unicorn
6) Peasant

Descriptor 1
1) Rampaging
2) Naked
3) Fiery
4) Undead
5) Evil
6) Gentle

Descriptor 2
1) Suspicious
2) Beautiful
3) Giant
4) Mutant
5) Insectivorous
6) Conservative

Place Generator
(Roll 1d6 for a place and 1d6 on the special feature table for the place)
Place
1) Ruins
2) Village
3) Castle
4) Cave
5) Sewers
6) Dungeon

Special Feature
1) Lava Pits
2) Open Air Hot Spring Bathes
3) Spiked Pits
4) Giant Wheel
5) No Safety Railings
6) Everything is on Fire

Story
(Roll a story off the table below and roll actors and places as required)

Story
1) [Actor 1] in [place 1] has lost their precious keepsake amulet and needs you to return it from [place 2] where they last had it, but unfortunately [place 2] is the home of many angry and dangerous [Actor 2s] led by an infamous [actor 3]
2-6) blah blah, other entries.

And you generate a story bam...

A Suspicious Fiery Princess in The Village of No Safety Railings has lost her precious keepsake amulet and need you to return it from The Ruins That Are On Fire, unfortunately The Ruins That Are On Fire are the home of many angry and dangerous Evil Insectivorous Peasants lead by an infamous Giant Naked Wizard

There, how easy was that. And that was using an incredibly SIMPLE example, with STUPID options both on it's list of possible results AND in the end result. And it still made plenty of sense and sounds frankly a lot more interesting than many plot lines I've heard for an adventure. Certainly better than anything Bioware's human story creators have ever shat out.

[Koumei]

A Suspicious Fiery Princess in The Village of No Safety Railings has lost her precious keepsake amulet and need you to return it from The Ruins That Are On Fire, unfortunately The Ruins That Are On Fire are the home of many angry and dangerous Evil Insectivorous Peasants lead by an infamous Giant Naked Wizard

I think I might steal that for my game.

[RobbyPants]

Best. Madlibs. EVAR!

[hogarth]

I think people are vastly overestimating the quality of RPG plots. They literally do read as mad-libs with the loosest justification stringing together the murder and looting.

Agreed (with some exceptions).

I mean, Mister Cavern can make a simple werewolf attack into a encounter full of dramatic tension and powerful emotional overtones that resonate throughout the adventure and even the campaign, but for most people it's a CR 3 roadbump that might net you a Amulet of Slightly Increasing Your AC and temporarily sates the Smash Brothers urges and keeps them from taking over the group during the talky parts of the adventure.

And there are games that play to this level (e.g. the "Castle Ravenloft" and "Arkham Horror" board games, the flash games "Monster's Den" and "Monster Slayers", etc.). They don't really have the sitting-around-and-bullshitting feel of RPGs, though.

That being said, automating the bulk of the MC duties just makes sense.

That depends. You could argue that it makes sense to automate the game of "Snakes & Ladders", but then the whole game would boil down to pressing a button and flashing the message "PLAYER ONE WINS".

In my opinion, the most memorable moments in RPGs come from when something unexpected happens, and that's difficult to automate (or at least it's difficult to automate it so that something is unexpected after the first time seeing it).

[Username17]

I reviewed Mythic this morning. It works on a few levels.

Cool. And thank you.

A fundamental one that comes up often is the binary decision tree, and reading its description does not make for a compelling sell of the mechanic involved. Basically, you can ask the universe a yes/no question with odds determined by a Chaos factor (starting at a medium value and increasing with failures and decreasing with successes) modified by group fiat.

Random Events also pop up, based on the result of the binary question-related dieroll - they're more likely with more Chaos. This leads to a focus (new NPC appears, surprise movement towards/away from goal, positive and negative events, etc), an action (a d100 list of verbs) and a subject (a d100 list of nouns). The group interprets the results like a Rorschach test and implements them. This part seems interesting, in that you're interactively taking random crap and trying to make it maintain some kind of logic, which to me feels like what a good Mr Cavern does anyway.

There are some neat ways of maintaining some semblance of order: maintaining lists of NPCs so that you can hit the memory banks as Random Events or the plot imply, and Threads, which is the game's method of keeping track of plots. There's the character Threads (find my father) as well as the plot threads (recover the MacGuffin).

Plots and games move ahead expressly by Scenes - the Scene is set by previous events and character decisions (or if the 1st scene and no one has any great ideas, something from the Random Events results), but then possibly modified by being Interrupted or Altered in unexpected ways (the possibility of which is determined by the Chaos).

I'd have to see it in play, but I don't think its that bad a shot at pushing Mr Cavern aside.

It still doesn't answer basic stuff though: what opposition shows up and how to run them (Mythic itself plays like The Window, so concrete numbers are only really window dressing on narrative descriptions), for example.

That sounds like a pretty good place to start, actually. My personal tastes would run towards something that was a bit more structured and had more meaningfully fixed difficulties. I mean, my group did a seal victory on Atlach Nacha in Arkham Horror. And that means something, because doing that is legitimately difficult. I would like it if going on an adventure to stop the Efreet Sultan from doing something was genuinely difficult and doing so was as inherently meaningful as defeating Demogorgon in AD&D.

But yeah, I like the Threads and the Random Events tables. I think those could go really far.

-Username17

[mean_liar]

I mean you could start writing a simple set of tables to script some basic story for a gaming session just like that...

This is basically how Mythic's Random Events work, and with a table full of creative folk (or really even just one or two), you've managed to add another player to the game to replace Mr Cavern.

I still say that the largest hurdle isn't plot generation, its, "how to make sure that the antagonists in this scene are appropriate and what do they do".

I really like a bidding mechanism where you have some way of having the players choose the intensity of opposition without directly choosing the exact opposition: they can call for a "difficult" encounter (with commensurate rewards) but won't know what that actually entails until you consult the Chart O Fate.

Implementing that would require some effort: you'd have to have a list of enemies organized by level/CR/whatevs (which in turn requires a meaningful way of measuring their relative difficulty), as well as thematic tags on them (subterranean, arcane, fire, etc) so that you can not only generate appropriately-scaled opposition, but ensure that they're also thematically-appropriate.

Some of that can be short-handed in play (ie, a random result for a gang of goblins as a single entry rather than expecting to dice up each gobbo), but you're still looking at a codified list of everything you expect to ever encounter as a pre-req to using the system, and those short-hands are actually more work on the front end since you have to write them up.


EDIT:

That sounds like a pretty good place to start, actually. My personal tastes would run towards something that was a bit more structured and had more meaningfully fixed difficulties. I mean, my group did a seal victory on Atlach Nacha in Arkham Horror. And that means something, because doing that is legitimately difficult. I would like it if going on an adventure to stop the Efreet Sultan from doing something was genuinely difficult and doing so was as inherently meaningful as defeating Demogorgon in AD&D.

See my bidding mechanism idea. This especially works in games that have 'loot' of some kind since you can tie that to encounter/scene difficulty, though loot is such an abstract concept that the loot could easily be "advance a character thread" (giving you clues on your dad's murderer or some-such).

But yeah, I like the Threads and the Random Events tables. I think those could go really far.

Yeah... again, I really like how Mythic does things with the scene-setting and pushing stuff along, but implementing it outside its Window-like mechanisms for tactical combat buffs remains an unaddressed and difficult hurdle.

I definitely am going to incorporate Mythic stuff into my Heartbreaker.

[TheFlatline]

I still say that the largest hurdle isn't plot generation, its, "how to make sure that the antagonists in this scene are appropriate and what do they do".

Actually, if you follow the Hero's Journey style of storylines, it's incredibly easy to generate a plotline randomly.

You start out with overarching categories (sci-fi, gritty, low power, whatever) which narrow down your choice fields in each stage of the journey.

OR, you work backwards, and generate the BBEG and work backwards from there. This might function better. Spycraft does something vaguely similar to this in allowing you to create the mastermind opponent, and then using his stats and his goals, building an organization that the heroes investigate to track down the mastermind and stop his ultimate plan of evil.

Treating an antagonist as an individual at the head of an organization or organized effort would provide *much* more structure, and be more coherent, than going chronologically from the beginning of the story to the end from the heroes' perspective. This also retains a large degree of freedom that randomly generated encounters might not, since there is no set path from beginning to end for the heroes to follow.

[K]

And there are games that play to this level (e.g. the "Castle Ravenloft" and "Arkham Horror" board games, the flash games "Monster's Den" and "Monster Slayers", etc.). They don't really have the sitting-around-and-bullshitting feel of RPGs, though.

I don't know how you play Arkham Horror, but among my friends there is often sambuca and cake and bullshitting galore.

In my opinion, the most memorable moments in RPGs come from when something unexpected happens, and that's difficult to automate (or at least it's difficult to automate it so that something is unexpected after the first time seeing it).

Oh, that's the easiest thing in the world to automate. Random tables for things like monsters and battlefields and pregenerated scripts means sometimes uncertain things are bound to happen. I mean, I was once playing 2e DnD and when I used a charge on my Wand of Wonder everything in a 60' radius turned into stone.... and we were on a pirate ship so when the bulk of it turned into stone hijinks ensued as the ship tore itself in half.

That was just with a random item. Now imagine how scripts on monsters will lead to enemies fireballing their own troops or setting the forest on fire or occasionally concentrating fire onto someone. The lack of MC means that things can go in your favor or against you without players feeling like the MC is going easy on the players or singling someone out for punishment.

And that's a big deal. The feeling of accomplishment you get when beating something objectively hard is so much greater than when the MC is tweaking the difficulty or handing out blanket immunity to his girlfriend. I mean, even when I lost a game of Arkham Horror I had a good time.

On the topic of plot surprises, the big reveal you often see in a pre-published adventure really could be a roll off a table in most cases. I mean, the BBEG might get a special table where something crazy happens ranging from something as common as "More Troops!" and more troops to something as wild as Framed! where the BBEG turns out to be a wrongly accused Team Good hero who is now intent on trying to kill you because you broke into his house and killed his retainers.

The beauty of nested tables is that years might go by before you see any particular result so real surprise is built into the game. For example, if you have all six of the MMs that 3e produced and several hundred monsters scattered in various supplements, what are the odds you'd ever fight a Summoning Ooze if the monster of the week was chosen randomly?

Heck, the system even works for prepublshed adventures because you could have new tables for just that adventure if you really want themed adventures to break up the randomness.

[hogarth]

I don't know how you play Arkham Horror, but among my friends there is often sambuca and cake and bullshitting galore.

Fair enough. I was thinking of "in-character" bullshitting, but it doesn't make any difference. I've never played Arkham Horror, by the way.

In my opinion, the most memorable moments in RPGs come from when something unexpected happens, and that's difficult to automate (or at least it's difficult to automate it so that something is unexpected after the first time seeing it).

Oh, that's the easiest thing in the world to automate. Random tables for things like monsters and battlefields and pregenerated scripts means sometimes uncertain things are bound to happen. I mean, I was once playing 2e DnD and when I used a charge on my Wand of Wonder everything in a 60' radius turned into stone.... and we were on a pirate ship so when the bulk of it turned into stone hijinks ensued as the ship tore itself in half.

"Unexpected" is probably the wrong word to use; if I bought a lottery ticket, winning would be "unexpected" but I'd clearly realise that it's a possibility beforehand.

The Wand of Wonder (or anything else based on a table) is a perfect example of what I wouldn't want in an encounter table. Assuming you had a copy of the DMG, you'd know exactly what the possibilities are each time you used it, so turning everything to stone is not really any more surprising than rolling a 57 on d%.

Furthermore, it's overstuffed with results that are unusual and goofy. So how do you make a random encounter table? Do you stuff it with prosaic encounters, e.g. for every "you find a genie lamp" encounter you have 100 "you're attacked by bandits" encounters? Or do you have every encounter be "genie lamp" crazy and practically degenerate into playing a game of Toon? I don't know if there's a happy medium or not.

[JonSetanta]

I knew a gamer that created an encounter table concept with a bell curve years ago, way back in the early 90s of AD&D.

Roll 2d6.
There's 12 results for encounters.
The common encounters are near the middle while uncommon to rare are lower and higher.

[hogarth]

I knew a gamer that created an encounter table concept with a bell curve years ago, way back in the early 90s of AD&D.

Roll 2d6.
There's 12 results for encounters.
The common encounters are near the middle while uncommon to rare are lower and higher.

Sure. But I'd worry about encouraging the MMO mentality of "let's grind through these boring, common encounters 50 times so that we can get a rare, interesting encounter once!"

Like I said, there might be a happy medium.

[JonSetanta]

Sure. But I'd worry about encouraging the MMO mentality of "let's grind through these boring, common encounters 50 times so that we can get a rare, interesting encounter once!"

Like I said, there might be a happy medium.

Get ready for the instance raid. My 2-hour is cooled.

[Sashi]

It's a false equivalency to say common=boring. "common" could just mean "appropriate for level" while a roll of 2 on 2d6 means it'a a level -3 encounter and a 12 is level +4

[K]

The trick is not to make static tables, but to make nested tables. So, instead of Bandits! as an encounter with everything predetermined, you'd roll off the area for Forest and get Bandit, then you'd roll off the monster chart for the Mirewoods forest and get Serpentfolk, and then you'd roll off the Random chart and get something like Airship.

So instead of saying "ho hum, we are going to the Mirewoods so we are probably going to run into Human Bandits carrying boring ass bows because 60% of the encounters are like that" you instead have a wild airship battle with Serpentfolk pirates.

Not only that, but the chance of getting the same encounter twice is really quite hard.

You can even have a mechanic where parties can seek out certain kinds of encounters with their abilities like divinations or spy networks. I mean, if Airship encounters happen in the Mirewoods because a certain rare magic wood grows there and pirates looking to build their own airships are known to have bases there, then the players might want to steal an airship to complete some other quest goal.

[mean_liar]

I think you're overcomplicating matters, considering how arbitrary that last scenario is.

You might as well leave most of the world as a blank slate and fill it in entirely randomly as you go with just-so rationalizations. Coming up with nested tables for every location or situation when the results are basically arbitrary seems pointless when you can just make it up and go with it, then write it down for later:

"Okay, the Mirewoods are full of serpent airships. Next?"

[hogarth]

Not only that, but the chance of getting the same encounter twice is really quite hard.

As long as you don't end up living in Arbitrary-Land where you fight four flightless birds guarding three wands and a guillotine on Tuesday and 500 halflings on the moon on Wednesday (i.e. the "Wand of Wonder" approach).