How to Design Spells

[Wulf Ratbane]

Re: CR Formulae

Not every adventuring group is alike, and it's not safe to assume too much about a PC party when you assign CR.

CR doesn't account for specific party composition in the slightest-- you only get one CR rating, and by definition it's tuned to the "average party of four PCs."

To some degree this happens, to be sure, but that's where the art comes in; the factors of experience, familiarity with the rules and flexibility. These things won't be genuinely modeled, in my opinion, by a formula.

Nor are they accurately modeled by a single CR rating, no matter how derived. The art you speak of is the purview of the DM-- but a single, fixed CR is the DM's starting point of reference.

DM judgement needs to be preserved as an overriding factor in CRs.

I think you just have it backwards. The CR system is an underlying factor in DM judgement. It can and must be consistent in order for the CR system to have general applicability to all the PCs and DMs out there.

DMs need a good guide on how, when and why to set a CR, adjust a CR and spot a CR that's sound for one group, but maybe not for theirs.

Replace "CR" with "EL" and I can agree with you.

As evidence, I hold up the separation of CR and EL in d20. A monster's CR is meant to be fixed, but the EL it appears under is supposed to be the flexible scale a DM has to get a sense of real control of the system. The malleability of ELs has gone underused, in my opinion, by adventure and game designers. An EL can convey a lot, suggest a bit more, and give more leeway to the interpretation of a CR. A monster's context can't render its CR invalid, for example. An EL is still only mildly more valuable to the average DM. Only a given party's DM can accurately assess the relationship between a monster, the encounter and her party.

Seems pretty clear, from this paragraph at least, that you understand the distinctions and the system. CR is a fixed statistic, and in the most perfect of worlds, subject only to mathematical formula. On the other hand, EL is a varied, subjective thing subject to the art of the DM.


Wulf

 

[Ashy]

Wulf,

If you get a chance, please drop me a reply to that email I sent ya - it speaks DIRECTLY to this sort of thing. I think that you will be interested.

EDIT: Spelling...

 

[Word]

CR doesn't account for specific party composition in the slightest-- you only get one CR rating, and by definition it's tuned to the "average party of four PCs."

This is the point I'm getting at. Because the monster and its application is measured by a single value in the CR system (that is, the CR itself), it must necessarily be a broad value. A CR isn't an especially precise measurement, for all of the above reasons. Attempting to strictly (which I infer from "mathematically") formulate CRs seems to be tilting at windmills. The CRs aren't, IMO, going to end up more precise and it seems very likely that the system could end up with a built-in bias if you believe, as I do, that the CR system is inescapably subjective to one extent or another.

It occurs to me now that, as 3rd party d20 publications continue to accumulate, the meaning of "average party of four PCs" becomes more and more diluted. What sort of assumptions are we making when we say "average party level?" Those assumptions are likely to biased, too. Which is fine, but should be acknowledged. Those biases would mean that the CR system does account for party make-up; whatever make-up, that is, which defines the average party.

My feeling, then, is that a mathematically derived CR is still going to be biased, but is going to be presented as unbiased. The current CR clearly and fairly wears its bias like a badge. I think that's a good thing, myself.

In regards to your corrections, the "DM" in the example above is whatever creator is crafting a monster for his/her game or product. You're right in that I should've been clearer. Likewise, when you say I should revise my "CR" references to "EL," I agree. What I'm trying to show there is that most of the difficulties with CR are solved with the CR/EL system by first discussing CRs alone and then addressing ELs as a solution. Sorry I was unclear.

Also, I meant to mention this before ...

There better be a "unified" mathematical formula at the heart of it all.

All evidence to the contrary. With the years since 3E was released, a CR formula would have to be considered the best-kept secret in the business. There is that hit points to CR chart, though. Regardless, it's evidence that attempting to devise a formula could quite possibly meet with failure (since the original designers could not, or chose not, to devise a forumla).

After all, what sort of monsters from which sources are you going to necessarily discount when you're crafting your formula? What criteria will you use to determine what monsters do or do not qualify for inclusion in your model? Won't a CR formula be handicapped if it makes certain creations or methods illegal in the rules? Won't it have to sacrifice freedom for the appearance of correct-ness? I'm genuinely curious what others think about this.

I haven't yet decided that a CR formula can't work, either. I'm willing to have my mind changed, but right now this continues to be my opinion. I don't want to be snide, and I'll assume that wasn't your intent other, but I was hoping for a discussion of your theories and alternatives, rather than your assessment of my understanding. (I hate having to guess at someone's tone in these sorts of conversational text exchanges.)

Since you didn't mention it, I'll assume Bad Axe is still Lombard. I used to work over there. I don't anymore.

word,
Will Hindmarch
Atlas Games

 

[Wulf Ratbane]

Hey Will, I wasn't being snide at all... at least not intentionally.

I still am not sure what you mean when you say that the CR system is broad and imprecise. An ogre is CR2 (for example). That seems incredibly precise to me. CR2 is CR2. I am not sure how "2" is in any way "a broad value."

The only wiggle room you have is in the DM's jugdement of what CR2 actually means. Usually that'll be interpreted using the EL system.

You are correct that some monsters won't fit into any kind of mathematical CR formula. The same thing was true of the spells. But I don't think that the existence of exceptions invalidates the general usefulness of a mathematical guideline.

All evidence to the contrary. With the years since 3E was released, a CR formula would have to be considered the best-kept secret in the business. There is that hit points to CR chart, though. Regardless, it's evidence that attempting to devise a formula could quite possibly meet with failure (since the original designers could not, or chose not, to devise a forumla).

Aside from the "dice of damage according to spell level" table in the DMG, there isn't a lot of "official" support behind the design framework of the spells, either. But it is still clearly there. It's not a secret, just not spelled out for anyone. If you line up any of Monte's BOEM products and look at the ways that he tweaks new spells, you can see that he operates by some fairly common design rules (aside from his annoying habit of making all his spells Transmutation, even when they clearly shouldn't be... but that's another matter...)

Yet I was fairly successful in pulling out those design rules to create spell templates. No, they don't cover every spell, but there is a design framework there, and that's useful material worthy of publication (and sharing with others).

There are other examples of things that are not explicitly spelled out by WOTC, but for which a mathematical formula exists. The XP chart has a mathematical formula, as do the BAB and save progressions. I suppose you just don't see them widely trumpeted because they are not particularly useful (though they are often mentioned on the d20 lists by folks attempting to end-run the OGL...)

Back to CR-- I think the discussion you are looking for is ultimately going to have to come from mearls. I am not putting together a formula for CR, though the exercise seems like fun, and I think the end result has more applicability than you're willing to concede. Mike has already done some work in this regard, I'm happy to see what he comes up with-- checking out his work in The Monster's Handbook in the short term and waiting for his "grand unified theory" in the long term.

Regarding your Lombard question-- sorry I missed it. Bad Axe is largely an "internet staff" in several locations; as the founder I had the luxury of listing Lombard as our primary location (and our state of incorporation).

Wulf

 

[DDK]

I mean no offence...

...but, if someone is writing d20 material and needs a guide like that then, quite frankly, they shouldn't be writing d20 material...

 

[mearls]

Re: I mean no offence...

...but, if someone is writing d20 material and needs a guide like that then, quite frankly, they shouldn't be writing d20 material...

Anyone who wants to write d20 material and who doesn't think they need a guide like this shouldn't be writing d20 material. They have no grasp on the real issues they'll face in preparing and presenting manuscripts, which probably makes up the majority of the real grunt effort in working as a game designer.

 

[mearls]

Math Is Our Friend

There is no reason why mathematical models cannot be used to determine a creature's CR. Let's take a look at the problem.

A creature's CR is defined in relation to the 11 core classes. A creature of CR X should drain 20% of the resources available to 4 level X player characters equipped with items as per table 5-1 from the DMG.

Now, the problem here is that not all level X characters are created equal. Players who make sub-optimal choices are going to have weaker than normal characters. For instance, I ran one game where a player kept taking levels in random classes that meshed together very poorly (fighter/bard/rogue/sorcerer, with rogue taken after level 1) and who make bad choices with feats and skills. A human fighter with Skill Focus, Alertness, and Exotic Weapon Proficiency (kukri) at level 1 is nowhere near as effective as one who takes Weapon Focus (longsword), Improved Initiative, and Dodge. However, we can assume that players will generally make half-decent skill and feat selections. Thus, it is possible to construct a typical character for each class from levels 1 to 20.

There are a limited number of ways in which a monster can interact with characters via the system, and vice versa. I call these interactions "touches." A monster touches a PC's AC when it makes an attack with its base attack bonus. If it hits, it touches the PC's hit points with its damage spread. You can graph these interactions across different characters and different levels. The surprising thing is that when you sit down and look at the system this way, there are very few areas in which monsters and PCs touch each other. At low levels, it's all about AC and hit points. At higher levels, it transitions over to saving throws.

Given these relationships, and given the definition of CR, we can create a series of profiles that yield what a given CR can do in terms of its touches against PCs. If you know that on average a CR 1 monster can inflict 3.8 points of damage per round, you can play around with the probabilities of attack and damage touches to build a wide range of monsters that drop into that 3.8 damage on average per round. Even AC comes into the equation when you plot out how long the monster can expect to survive combat.

Now, you might be thinking that this is all well and good, but what does it mean for the game?

A lot.

If you can describe CRs in more concrete terms, ranges of base attacks, damage, SR, basically categories of all these areas where monsters and adventurers touch each other via the game system, you can create an interlocking set of values.

Now, Will is right that the CR system is fuzzy. Try going against undead without a cleric. The barbarian's nifty Improved Critical feat is suddenly useless, as are the enchanter's spells and the rogue's sneak attack. Personally, where I in charge of 3.5E I would've dropped the blanket immunities that certain creatures receive. They really exacerbate the all groups are not created equal problem. If you have the Monster's Handbook, you can see that I come out and say that not all CRs are equal.

So what does this all mean?

Monsters and PCs interact in predictable ways in terms of the system. Given that CR is defined in terms of character level and resources, we can construct a range of interlinked values for damage, base attack bonus, AC, saves, and so on and graph the combat results those values generate in relation to the level of a typical party of adventurers.

I think a lot of game designers are afraid of math. There's a general sense that math and other scientific studies somehow run counter to the creative process. I think things couldn't be farther from the truth, and that a dose of mathematics and computer modeling would go a long way to building better RPG systems.

 

[jmucchiello]

Re: Math Is Our Friend

A creature's CR is defined in relation to the 11 core classes. A creature of CR X should drain 20% of the resources available to 4 level X player characters equipped with items as per table 5-1 from the DMG.

And 25 point buy ability scores as definied in the standard set 15, 14, 13, 12, 10, 8. At least that's the rumor.

Now, Will is right that the CR system is fuzzy. Try going against undead without a cleric. The barbarian's nifty Improved Critical feat is suddenly useless, as are the enchanter's spells and the rogue's sneak attack.

The lack of hit point bonus due to Con and Wizard BAB for most Undead really does balance this.

The only problem with your touch system is that it isn't very conducive to the spell system. What is the average damage / round of a Hold Monster spell? Or Haste, Fly, Bull's Strength (on the fighter)? In fact, in a given combat, what is the average 10th level sorcerer or wizard's contribution to the fight in terms of 20% of his abilities? If you can determine this, SR is easy to model.

 

[mearls]

Re: Re: Math Is Our Friend

And 25 point buy ability scores as definied in the standard set 15, 14, 13, 12, 10, 8. At least that's the rumor.The lack of hit point bonus due to Con and Wizard BAB for most Undead really does balance this.


OTOH, undead frequently get lots of special, binary (ie, save or die) attacks such as paralysis or a death gaze.


The only problem with your touch system is that it isn't very conducive to the spell system. What is the average damage / round of a Hold Monster spell? Or Haste, Fly, Bull's Strength (on the fighter)? In fact, in a given combat, what is the average 10th level sorcerer or wizard's contribution to the fight in terms of 20% of his abilities? If you can determine this, SR is easy to model.

Spells can be defined by the total levels worth of spells. For instance, a 5th level wizard who can cast 4 level 1, 3 level 2, and 2 level 3 spells has 16 levels of spells to use. 20% of that is 3.2, so a CR 5 encounter should have him use about 3 levels of spells - one fireball, a web and a magic missile, and so on.

Saves against binary abilities work a bit differently than AC and hit points. In essence, you need to factor in the number of binary effects the PCs (or monster) has against the chance of a successful save against each, with the total chance of a creature falling victim to one projected out over the length of combat. Basically, a creature has to have a good enough saving throw to, on average, resist the party's spells long enough to absorb 20% of their resources including the spells needed to bring it down. This goes back to our example above with the wizard's spells. The act of casting hold monster to take out our creature might represent enough resources that even with the monster going down on turn 1, it has fulfilled the requirements of the CR system by draining party resources.

For bull's strength and other buff spells, you simply note the spell as exhausted and give the appropriate character the bonus. Note that buff spells are some of the most efficient spells in the game. For a few spell levels, you can grant bonuses that come into play repeatedly. Bull's strength is perhaps the most efficient spell in all of D&D. Not necessarily the most powerful, but amongst the most efficient.

 

[DDK]

Anyone who wants to write d20 material and who doesn't think they need a guide like this shouldn't be writing d20 material. They have no grasp on the real issues they'll face in preparing and presenting manuscripts, which probably makes up the majority of the real grunt effort in working as a game designer.

*blink*

*Trying to believe what he just heard.*

*blink*blink*

*Still can't quite believe his ears... bangs on them to see if they'r working right*

*blink*

I can't win with logic like this... YOU DA MAN MEARLS! YOU DA MAN!