Bay 12 Games Forum

Didn't you have that really good artifact thread a few months ago?

Oh man, for a moment I thought this meant Neonivek had found a really valuable artifact-level piece of thread in his fortress and I was all "whaaaat?"

But this is cool too.

I think you're over-valuing the interpretation phase.

It's certainly useful, but I don't think it's entirely necessary.

Example:

A tells a joke, very badly. B responds by mocking him, for his failure. A responds to this, by developing a grudge against B, which eventually grows into overt violence.

Nobody has failed to interpret anything, the conversation has just gone horribly wrong, as an effect of A and B making poor choices about their responses.

True, there's no misinterpretation there.  As I said, there are some situations for which the model isn't terribly useful.  That said, we could still use it here, if we liked:

A's offense: the failed joke
B's interpretation: A is dumb and worthy of scorn
B's reaction: to mock A
B's offense: the mockery
A's interpretation: B is mean.
A's response: develop a grudge.

And so on.  Nobody was misinterpreted, so the model is perhaps unnecessary here, but there was interpretation going on.  It sounds like we could use a way to skip or streamline the interpretation step if it wasn't needed.

An interpretation phase would certainly add a lot of depth to the simulation, so I'll make a pass at simplifying that.

Storing knowledge data for every person in the entire world is not feasible, but storing a limited amount of knowledge data for each civilization or site is not utterly beyond consideration. Data on how common the knowledge is, and an intelligence or education level for each actor who interacts with it, would be sufficient for most purposes, though it does allow for dwarves "forgetting" things that they knew before, etc.

Good idea.  I touched upon that briefly in the first post (see the "more about knowledge" spoilerblock), but I agree that that can be simplified a lot more.  How about we give each class a smallish set of "known data" that it definitely knows, and for everything else we can just randomly determine whether it knows it or not?

I'll edit the OP with some ideas for this.

This assumes that knowledge tracking is important or even useful to interpretation.

It's generally safe to assume that an actor does not have empirical knowledge of any other actor's motivations, unless he is some kind of psychic. As such, actors will always have to guess at the motives of others, based on their opinion of them.

Dwarves already track friend/enemy stats, and civilizations are going to need to track how familiar/strange they are to each other anyway.

These emotional variables have a much more meaningful impact on negotiations and relationships, and take a lot less space to store.

Well yeah, of course emotional variables are vital.  They're pretty much the core of the Interpretation step, although looking back I realize that I never actually made that clear.  "Urist is my friend" is a datum, and often often the only datum that Logem will consider during any given interpretation.  It's only when the emotional variables are ambiguous, or there's some serious conflicting evidence, that the drama happens.

Because yeah, nobody has "empirical knowledge of any other actor's motivations".  That's the point.  They have to use hunches and biased data in their interpretations, and if they get bad data out of an interpretation, it'll cause more bad interpretations down the road.

Example:

A meets with the ambassador named B, from a newly discovered nation.

A and B are both suspicious types, so they distrust strangers. Unless each can convince the other of his goodwill, one of them will launch a preemptive strike.

A series of social skills are tested.

They use negotiation to hammer out a trade agreement, without offending each other, but they remain suspicious.

A decides to lighten the mood with some humor, but he botches the job and makes an insensitive joke about B's height. B takes this as confirmation of A's ill will, and unless things change B's armies will be marching at dawn.

Perfect example of a bad interpretation.  B takes the facts "A made an offensive joke" and "A is potentially hostile" and makes the reasonable but false deduction that "A is hostile".

more example stuff

Knowledge, as important as it is, pales in comparison to things like trust and hatred, when it comes to matters of negotiation.

In fact, facts can influence negotiations indirectly, by impacting trust. If spies from one country had found out about the other's invasion plans, then the negotiations would have been even more tense, from the start.

Okay, there's a really neat thing at work here which I don't think you quite realized.  Trust and hatred aren't separate from knowledge; they're pieces of knowledge.  Knowledge of invasion plans--that's also a piece of knowledge.  Everything, every bit and burp, everything is a piece of knowledge.  A datum.   The only things at stake are 1. whether you know the datum, and 2. whether the datum is worth knowing.  Emotional data is definitely worth knowing, and everybody knows their own emotions (on some level), so yeah, it's going to have a big impact.

My current adventurer, the dwarf Obok Meatgod, is pretty awesome.

About midway through his adventuring career he found his way to the mythical fortress of Headshoots (yes, that Headshoots) and picked up some adamantine armour and enough adamantine weapons to fill a dwarven backpack.

Of course, he never actually uses those weapons.  Adamantine kills things too quickly.  He prefers to take his time, get to know his opponent before he kills it.  He uses a cheapo copper war hammer he picked up in a human town.  It takes him forever to kill anything.  But that's okay.  That's better.  That's terrifying.

He's gotten a bit of a reputation for killing giants, you see.  Everywhere he goes, somebody wants him to kill a giant for them.  He does it with gusto.  They're so big that his little tiny hammer can barely hurt them at all.  But please note--the operative word here is "barely".

He once spent a week visiting a cave in which there lived seven giants.  Each day, he would come in, track down each of the giants, and beat it to the point of unconsciousness.  Since he was using his tiny hammer, 90% of his hits did nothing at all.  Of those that did hit, 90% only caused light gray wounds.  Each day, he beat each of the giants down by about a full colour-change over their entire body.  Their unconsciousness wasn't from organ failure or body explosion or anything, just pain and sustained light bleeding and exhaustion and nausea.

Of course, it being impossible to dish out the wounds perfectly evenly, each of them ended up with a few more broken bones each day.  Not too much in any one sitting, but over the course of the week they started to add up.  Each time I fought them, they were just a bit more crippled than before.  Slowly, steadily, I battered them into submission.

By the fourth day, the giants turned and ran away when I tried to engage them.  By the fifth day, none of them could run any more because they all had multiple broken legs.  By the sixth day, they were reduced to just flailing weakly with one or two limbs.  On the last and final day, none of them lifted a finger against me as I slowly and purposefully battered through their stupidly thick skulls and grayed out their brains at last.

Maybe it was the spinal injuries talking.  Maybe it was the fact that every bone and joint in their bodies were broken to the point of unusableness.  But I like to think that--just maybe--I had beaten them into a state of despair so profound that, when death finally came to them, they welcomed it with open arms.

Oh, heh.  I think I clicked 'Quote' when I meant to click 'Edit'.

On the subject of Call of Cthulhu it would be interesting to see creatures who could spread Sphere Taint and/or cause insanity by their presence.

You might be able to simulate that with a creature that secretes an invisible inhaled poison.  There might be some insanity-like effects available, but if not you could use necrosis as a stand-in for taint.

If I remember correctly, random beasts can generate random poisons.  That means it's possible that you'd see something that does exactly what I described.  Unfortunately it'll probably look like a giant yellow marine ocelot.  Ah well, can't win 'em all.

Also, I'm going to steal that idea for another of my mods.  The taint aura thing, not the ocelot.

Unfortunately, ramps need to have a solid wall on their up side to be usable.

That is:

___
__#^__   works, but

___
___^__   doesn't.

Legend:
_ = floor
# = wall
^ = ramp

Your dwarves won't be able to get in or out of their houses.


EDIT: Ah, but you could use bridges in combination with ramps!

___.._
_____#^__

. = bridge


It seems to me that the distinction between a weapon on a chain and one on a stick has something to do with the velocity modifier (in the raws, at least).  The chain lets you build up more momentum with each attack, at the cost of a slower attack speed (it takes time to build that momentum) and a bit less accuracy (it's hard to change direction in mid-swing).

To me, it looks like the way combat damage is figured is roughly weight x velocity / striking area, which is then modified by skill rolls that determine what you hit and with what fraction of the striking area.  With an edged weapon, you then use the penetration size to find out how deep it goes.

That means you could model a spiked flail (or a morningstar on a chain, or whatever) by, say:

[ITEM_WEAPON:ITEM_WEAPON_BALL_CHAIN_SPIKED]
[NAME:ball on a chain:balls on chains]
[ADJ:spiked]
[WEIGHT:50]
[SKILL:MACE]
[TWO_HANDED:47500]
[MINIMUM_SIZE:42500]
[MATERIAL_SIZE:4]
[ATTACK:EDGE:10:500:bash:bashes:NO_SUB:2500]
That is, an edged weapon with a high weight (heavy ball), high velocity (on a chain), small striking area (spikes) and small penetration size (little spikes).  Unfortunately I don't think there's any way to make a weapon attack slower except to make it heavier.


Okay?  It doesn't matter what the actual names for things are.  You don't kill people with names.  We have numbers.  Numbers are what you kill people with.

Oh, I know well that my implementation is unwieldy to the point of unusability.  I just threw it together to get something out there before I forgot.  It'll need to be rethought a bunch of times before it ever sees a line of code.

It's a great idea. The main problem - in fact, probably the problem that leads to dwarves being omniscient - is that data is not as small as you think it is.

Actually, it is as small as I think it is.  I think it's hugely, ridiculously large. 

We can reduce it a bit, though, if we limit the number of people who are allowed to know things.

However, if the game only does this for 'important' entities (which Toady actually mentioned it already does now), this would probably only really slow down world gen a little.

Yeah, exactly.  Like Toady said, we can't keep track of everybody at once.  Luckily, according to the podcast he's already thinking about how to decide whether somebody is important or not.  I think we can leave that part to him to solve.

Just for fun, I went through the legends and the history file for one of my medium-sized worlds.  It brought up about 40,000 historical figures and 600 rulers.  If we assume that most of the interesting people in history are rulers, we have on the order of 1000 interesting figures per world.  I could be off by a factor of 2 or 3, but that should still give you an idea of the numbers we're dealing with.

Yeah, I have to agree about the concerns.  Think about what happens when 50 dwarves go party in your statue gardens.  They talk, and.... chug chug chug... FPS drops to 0 while the game processes new data, metadata, responses, etc... for all the dwarfs.  Ouch.

Oh, aye.  Interactions will probably be relatively limited among your dwarves in Dwarf Mode, which is a shame.  Legends mode can afford to have more detail, since it only deals with a few people at a time, but in dwarf mode pretty much everybody's all talking at once.  I imagine that only a few dwarves in your fortress will have the capacity for real knowledge.

One small quibble.  If X hears a conversation about turtles, and hates turtles, he might get an unhappy thought.  But if X hears Y talking about turtles, and Y also dislikes turtles, Y is probably ranting about 'those damn turtles' again, and he should get a happy thought.  So not just subject matter, but the attitude of the speaker should effect who gets good and bad thoughts for a given conversation.  (And dwarfs should talk about things they hate).

Ooh, good point.  Fixng that would add a bit of complexity, but I think it'd be worthwhile.

Nice, and C might even simply be playing them against each other, like dark elves with dwarves and elves in warhammer fantasy.

Yeah, the War of the Beard was actually a major inspiration for some of this.

One thing I'm surprised nobody mentioned is how borked metaknowedge creation is right now.  The way I have it, the vast majority of all data out there will be metadata about other metadata.  I'll have to do some thinking about how to limit it safely without losing too much useful information.

I'll try to update the OP when I can, to take your suggestions into account, to add any more ideas I get, and to clarify some things that I think need it.  Until then, thanks for all the comments and kudos!