Carrnage
Human
- Location
- New Zealand
actually the QA costs when not paid by the developers is usually paid by the community
see:
1e
2e
2.5e
General Exalted Thread
see:
1e
2e
2.5e
- Location
- Ottawa
No.
Seriously, I have no clue how you got that impression.
Not so. Writing a style that works for eight splats is more work than writing a style that works for one splat, but it's not eight times as much work. Maybe one and a half times, or twice.
Balancing is not the biggest part of writing, and splats are similar enough that balancing a style for one splat does a lot to balance it for others as well.
Jon Chung
=][=
I disagree, from experience. There's a mutual exclusion problem where different splats have different enough meta requirements that, say, costing something one way breaks it for the other splat as an example, and you can't solve that without making splat-specific costs and splat-specific effects. At this point, you might as well just make splat-specific styles.
Like, say I make a CMA perfect defense. Do I cost it competitive to Solar Seven Shadow Evasion, or DB Body of [Element] Defense? If I do the former, every DB in the goddamn world will be using this one style, if I do it the other way, I've just wasted Solar Bob's 8XP. I can solve this by writing one charm for Solar Bob and one charm for DB Steve. If I do this every time something like that comes up, why am I not writing a Solar iaijutsu style and a DB iaijutsu style separately?
Note that if I do make separate styles, I can make them meaningfully thematically distinct. Solar iaijutsu will be fast as light, with blades as sharp as the edge of a perfect shadow. DB will be full of storm thematics and lightning imagery. I can't do this if I want to mash them together into a dumb one-size-fits-all block, can I?
This is an outrageous assertion. Back it up.
Last edited:
azoicennead
Overly-Affectionate Zombie Squad
- Pronouns
- He/They
From you saying "the game is getting X amount of resources for QA, regardless of how much it needs" and Exalted being Exalted.
Aaron Peori
Rest in Peace
Claiming this when no other splats but Solars exist is incredibly presumptuous. Martial Arts and Evocations so far are balanced as Solar Charms because that's all the have to be balanced against. We'll see if this keeps up when we have Dragonblooded, Sidereal, Lunar, Abyssal, Infernal, Exigent, Liminal, Getimanian, Fair Folk and general Spirit Charms.
Of course, at the current release rate, we'll find this out when we're all hyper-evolved cyberuploads floating in subspace colonies.
- Location
- Ottawa
A Charm that's broken is often broken for everyone, on account of being flat-out unfair. And a Charm that combines outrageously with options from one splat will often do the same with options from another splat; if Solar Athletics letting you fly breaks it, Lunar shapeshifting letting you fly probably will too.
So a Charm that's not broken for Lunars is less likely to be broken for Sidereals than one that is. Etc.
- Location
- Virginia
The solution 3E came up with, which I think will probably work fine, is that MA charms have "tiered" effects where a given charm sometimes has weaker mechanics for Terrestrials (and, presumably, Terrestrial-alikes) and/or stronger mechanics for Solars (and allegedly Sidereals). Not every charm has these but enough do that each style is clearly weaker for a DB than a Solar.
I think this is probably the 80/20 rule sweet spot for maximizing charm reusability while minimizing splat balance issues.
edit: examples:
Silver-Voiced Nightingale Style's first charm lets you use your voice as a light thrown weapon. Except if you are a Solar (or otherwise get Mastery benefits, which is the actual keyword) it is a light artifact thrown weapon.
Righteous Devil Style's charm Azure Abacus Meditation lets you ignore soak against enemies outside of cover, or some amount of soak for enemies in different degrees of cover. (Another charm of the style lets you negate cover.) But for DBs, this charm always treats enemies as having at least light cover, so they never get to fully ignore soak.
Last edited:
Jon Chung
=][=
Note that here we're right back to combinatorial hell and having to care about every possible splat when writing effects.
I agree with caveats (this only works for binary capabilities!), but address the PD example.
Last edited:
Graypairofsocks
What is Grammar?
- Location
- earth
Something that might reduce the work in Balancing out martial arts (or other charm shared stuff) between splats, would be to try to develop all splats on the same powerlevel.
- Location
- Virginia
No we're not. You more or less have three tiers of MA charm power, and you give splats access to the tier that is appropriate. (And this requires writing much fewer than 3x as many charms.)
I mean, yes, in theory, you could have a problem from combinations, but as I said this is the 80/20 solution.
Last edited:
Aaron Peori
Rest in Peace
So your solution to "It's easier to write a separate martial arts style for each splat" is "No, what you do, is you write the same style, but each Charm works differently for each splat! That different than writing a separate style for each splat because reasons."
- Location
- Virginia
please try reading what I actually wrote kthx
Jon Chung
=][=
Yes, I have three tiers of MA charms, and I still need to vet all effects against each splat's meta within a given tier. At best, this does not remove combinatorial hell, it merely reduces the temperature of our roasting because instead of ten splats, I need to check against four, assuming we end up with ~12 splats and they are evenly distributed across the three tiers. I would really rather not do this at all.
Is a four-splat combinatorial hell instead of ten really an 80/20 solution in your view?
- Location
- Virginia
edit: derp nevermind
Given that the entire point of combinatorial hell is the fact that the number of interactions is quadratic in the number of charms, and given the fact that 4^2 is 16 while 10^2 is 100... actually yes! This does seem like something pretty close to an 80/20 solution!
Last edited:
Jon Chung
=][=
We have a minimum of two new splats. Assuming all prior splats with access to martial arts retain it, that is twelve splats. Assuming the number of splats per tier is evenly distributed, we have four splats per tier.
Note that we're doing this from the comparison point of only needing to check any given splat's charms against their own charmset.
- Location
- Virginia
That's your preferred starting point, sure. But I'm looking at cross-splat Martial Arts - which, for the reasons Sanctaphrax and Omicron have mentioned, provide value and fun! even though they are also costly - and saying, "hmm, is there a way to really cut down a lot on the cost of this while retaining most of the things that are fun". And that's exactly what this tiering solution does - it provides most of the benefit while dramatically reducing the cost. Not to zero! But still a lot less.
Last edited:
samdamandias
Apprentice Plane Wizard
So, due to a game I'm running drawing heavily on Exalted, I find myself in need of Styles related to voidships and voidfighters.
So far, I've only come up with one:
Five Dragon Formation Style
Blah blah fluff about how the Ten Thousand Dragons must work as one. Five Dragon Formation is not the flying style of heroes. It is not taught to the elite pilots who can take on battle-moons in a singleship. Five Dragon Formation has a very simple goal: Every pilot, every ship, comes back.
1:+1 to any maneuver another member of your flight has performed this turn (Basic skill, keeping up with the rest of the squad)
2:+1 to Defend Other (Intermediate skill of the style, keeping other people alive)
3: I can't think of what should go here. Maybe something about Onslaught penalties and wolfpack tactics?
So far, I've only come up with one:
Five Dragon Formation Style
Blah blah fluff about how the Ten Thousand Dragons must work as one. Five Dragon Formation is not the flying style of heroes. It is not taught to the elite pilots who can take on battle-moons in a singleship. Five Dragon Formation has a very simple goal: Every pilot, every ship, comes back.
1:+1 to any maneuver another member of your flight has performed this turn (Basic skill, keeping up with the rest of the squad)
2:+1 to Defend Other (Intermediate skill of the style, keeping other people alive)
3: I can't think of what should go here. Maybe something about Onslaught penalties and wolfpack tactics?
I played a Solar who had their Redemption before starting play. However, the Redemption did kill a huge chunk of her memories. She didn't know she was a former Abyssal, all she knew was that she knew more about the Deathknights than the other members of the circle and hated all the Deathlords but one on a deep and personal level. But her arc wasn't about "Who am I?", it was about "Who was I?" (This may in part have been exacerbated by the fact that she wasn't around for the First Age and didn't have an unreasonably high Past Lives score.)
My Demon PC resembled that remark!
Jon Chung
=][=
The only starting point, given that we are evaluating charmshare and judging it for being terribly wasteful.
I keep pointing out that you can do themed trees of combat charms without charmshare, and nobody appears to notice. The value and fun people seem to like comes from "having themed trees of combat charms", not "having charmshare", yeah? Especially since Exalted 3 has all of one splat out right now and people still seem to like martial arts just fine.
Why, therefore, defend charmshare this way, as if the two are by definition conjoined twins that may not be separated? Remove each splat's Ability trees and replace them with MA-style trees, enjoy.
Last edited:
- Location
- Ottawa
If we're talking 2e, CMAs probably shouldn't contain perfect defenses at all.
If we're talking 3e, it's not really a problem. Perfects are not much of a thing, there's no need to write around them, and there's always the Terrestrial and Mastery keywords if you want to have them available to some splats and not to others.
Jon Chung
=][=
This is actually a great example of charmshare impact - if "CMAs" were Solar Charms, there is no reason why you could not put them in. The only reason you can't do it now is because you can't put a Solar-efficiency PD in a tree that a DB or a Lunar or whatever can buy or you will break the game. Why, conceptually speaking, should this even matter?
Clearly, Jon, the solution is to reach for a coding-style solution.
So we begin by defining public interface SnakeStyle. This defines the basic contract of a so-called Snake Style.
From this, we then create two abstract styles, which implement sections of the Charmtech judged to be common for that class of being:
- public abstract style AbstractSpiritSnakeStyle implements SnakeStyle
- public abstract style AbstractExaltSnakeStyle implements SnakeStyle
- public abstract style AbstractCelestialExaltSnakeStyle extends AbstractExaltSnakeStyle
- public abstract style AbstractTerrestrialExaltSnakeStyle extends AbstractExaltSnakeStyle
- public style SpiritSnakeStyle extends AbstractSpiritSnakeStyle
- public style EnlightenedMortalSnakeStyle extends AbstractSpiritSnakeStyle
- public style DragonbloodedSnakeStyle extends AbstractTerrestrialExaltSnakeStyle
- public style ImmaculateSnakeStyle extends DragonbloodedSnakeStyle
- public style AlchemicalSnakeStyle extends AbstractCelestialExaltSnakeStyle
- public style LunarSnakeStyle extends AbstractCelestialExaltSnakeStyle
- public style SiderealSnakeStyle extends AbstractCelestialExaltSnakeStyle
- public style AbyssalSnakeStyle extends AbstractCelestialExaltSnakeStyle
- public style SolarSnakeStyle extends AbstractCelestialExaltSnakeStyle
- public style GspSnakeStyle extends AbstractCelestialExaltSnakeStyle
(this is not a serious suggestion - it's just an extended corporate Java joke)
- Location
- Denmark
- Pronouns
- She/Her
The Phoenixian
The Glacier Witch
- Location
- My own little world
Well you've made a serious mistake: You've forgotten to define CorporateJavaJoke before you extended it.
- Location
- Virginia
We've been over this; you're selectively ignoring the responses people have given. See:
And:
Charmshare, but with this tiering mechanism (which requires writing far, far less than 3x as many charms) gets us these for much reduced effort in both writing (relative to "no charmshare") and testing (relative to "no tiering").
Plenty of people have noticed, Jon; the argument is that, if choosing between the alternatives:
(a) Every one of X splats gets Y unique "martial arts" trees, approximately balanced for that splat,
and
(b) There are Y unique "martial arts" trees, shared between the X splats but approximately balanced,
it's not at all clear that (a) is less work. (a) clearly requires creating XY total trees. (b) requires creating X trees, and checking each tree against Y splats. Your thesis seems to be that "checking this tree against a splat" is necessarily more work than creating a tree from scratch, i.e., that the effort for (a) is on the order of XY, while the effort for (b) is some quantity greater than XY - some (edit: thanks, @grommile) f(X)*Y, where f(X) grows faster than X. The opposing thesis is that f(X) grows more slowly than X, i.e., that checking a tree against one splat is in general less effort than creating a new tree from scratch.
There's also:
(c) The X splats are meaningfully divided into three tiers, and there are Y trees, each "tweaked" for the tier of character using it.
... which, if the X splats really are meaningfully divided into three tiers, in principle cuts the labor down to a maximum of 3Y, that is, the number of unique martial arts, times the number of tiers that art needs to be rebuilt and rebalanced for. (In practice, it's argued, the work required is somewhere between Y and 3Y, because some Charms work fine for all three tiers.)
The points of argument seem to be:
-What actually is the growth rate of f(X)?, and
-Is "tiering" actually an accurate way to describe the breakdown of splats?
We can't answer that second question for 3e, yet, because ha ha Ex3 release schedule, but it's not inherently impossible for it to be true. But it's not at all obvious to me that O(f(X)) > O(X) is necessarily a true statement, and that seems to be the position your argument hangs on; indeed, in the abstract, it seems rather more likely that for many systems O(f(X)) < O(X), in which case (b) or (c) are less net work.
Last edited: