Secret Writings of the Ash Ock

Aequam memento rebus in arduis servare mentem

Previous Entry Share Next Entry
The Art of NPC Design
fedora
codrus
Fun evening of 'gaming' at Chez Ronin tonight. By which I mean, sitting around eating yummy garlic soup while talking about 3e, 4e, and lots of other tangentially related gaming topics.

Ray and I sat and ratholed on NPCs in D&D 3e. I actually don't think we disagree as much as Nicole thinks we do -- I think we're actually on the same page, we just differ in methodology. Nicole thinks we need to collaborate on a game; if we both agree on everything in the game, it'll clearly make a zillion dollars. :)



Anyway, onto NPCs. It mostly started as a discussion about NPC classes in 3e, and how lame (or not) they are. Since I worked out one viewpoint on class and level for my Four Worlds game, I thought I'd share it here. Some of this I know I've talked about a few times before.

First off, as I've discussed before, I think that NPC design in any game system is going to be a matter of triage. There's always going to be NPCs that are worth a significant investment in time, and NPCs that are 'window dressing'. Do I need a full stat block on every shopkeeper? No, and for a published product, it would probably be a huge waste of space.

Second, when designing game rules, it is useful to think of PCs and NPCs assymetrically when it comes to designing characters for play. A PC with limited special abilities is effectively a one-trick pony; playing such a character gets deadly dull after a while. That's why, for example, Clerics and Mages were far more interesting than fighters in old-school D&D; the only way for fighters to get special abilities was magic items. However, when designing NPCs, too many special abilities for every NPC makes designing and running them into a chore.

When I ran Four Worlds, I had a few years of 3.0 experience under my belt, so I decided to take a more rigorous approach to my NPC designs. I had some clear ideas of how to triage the NPCs, working within the framework set up by 3.5's classes and skills.

Complete No-Name shopkeeper: 10 in all stats except the main one for the skill, which would be a 13. Max ranks (by level) in the appropriate craft skill, and a Skill Focus feat. Effectively, that always means a +7 in the main skill at 1st level, which is good enough for a competent crafter.

Cannon Fodder: Since it usually isn't that much more work, I go with the Standard Ability array (Monster Manual, p. 290) and usually an NPC class, like Warrior. Feats to be chosen from a subset of available feats -- ones that factor completely into stat blocks, or ones that are relatively easy to judge in play. The latter should only be for things that makes the character distinctive or interesting. For example, the Lerdina Marches soldiers had the Phalanx feat, for bonus AC when fighting in formation. That was a signature ability for them, and was not a significant imposition when running them; it also gave the PCs an interesting angle on how to approach that opponent.*

Significant NPCs: Elite stat array (standardized, MM p. 290). Unrestricted choices of feats, but usually limited to no more than one or two special abilities. This would be a henchman that leads a group of cannon fodder, typically.

Critical NPC: 28 point-buy, same as PCs. Completely unrestricted skills and feats. Basically, this is a PC level character, so it needs to (at least) match PCs in power level and interest.

So, basically, the less important the character, the more standardized they were. The more the NPC followed a template. By designing templates, actually developing new NPCs in play was pretty easy. I'll follow through with this idea in 4e, I imagine. I'm still hoping the actual workload to develop PCs will be lessened in 4e of course; even with the work I had done, fleshing out NPCs and typing in the stat blocks was a pain in the ass. IMO, having better automated tools is the way to go here.

One of the points I seem to be a significant minority on is the use of NPC classes, especially NPCs that are higher than 1st level. I actually think there's some interesting strengths that can fall out of these classes.

An example, the Commoner. While I'd probably never make a Commoner higher than 6th level ( a 20th level commoner is silly), I can totally see higher level commoner. He's the reliable King's Man who has fought in every major battle for the last decade or two. He's neither a knight (Fighter) nor a trained men-at-arms (warrior) -- he's the peasant levy who has been lucky or determined enough to survive every battle he's been in. He *is* a hero, just not a hero on the level of the PCs.

The argument seems to be that a commoner shouldn't ever be a match for the PCs -- but I don't think that's right. D&D should be a world of extraordinary individuals, both PCs and NPCs. I wouldn't just toss a 6th level or a 10th level commoner NPC without an interesting backstory.

An aside, my old viewpoint towards player characters was that I'd accept just about any PC concept that provided an interesting story that I used to hook plot points into. That goes the same for NPCs.

And, let's be honest, a 6th level Commoner, even with a decent Con, isn't an exceptional character. He has the hit points of a Wizard, and unexceptional BAB or saves; he's like a Wizard without the spells. Even with an Elite Ability Array (let's face it, someone described this way is a named NPC), he's moderately dangerous to 1st level PCs, but would only be a danger to 2nd level PCs if he's well equipped.

Example Commoner stat block
6th level Commoner
Str 16 Dex 13 Con 14 Int 10 Wis 12 Cha 8 (abilities optimized for combat)
Attack +6 Damage 1d6+3 (Shortspear)
AC 14 (Padded Armor +2, Small Shield +1, Dex +1)
Skills: n/a to combat, things like Farmer.
Feats: n/a to combat, he's not a combat brute.
HP 27

I maxed the guy's strength using the Elite Array, but even with an exceptional strength and con, he's not that tough. I assumed equipment that he looted from a battlefield; a real peasant probably wouldn't even have the weapons he's got. His attack bonus is just a bit better than starting 1st level fighter. His hit points are about the equivalent of a 2nd or 3rd level PC fighter. But against 4 PCs, he's really no match -- he's going to get flanked and mauled. Two or three MM orcs would be tougher than this guy, they are capable of doing more damage in aggregate.

So here's what I'm saying: It is easy to come up with an interesting story for a higher level NPC classed character, and yet, they aren't so tough that they are going to crush a group of player characters -- not without giving them some unusual equipment. And if I put, say, a 12th level commoner in the game, he'd be someone whose name is known by everyone in the local area.

My other example I like to use is the higher level blacksmith. Crafting DCs are actually pretty low; it isn't hard for even someone with a +7 bonus to craft routine items. For masterwork items, with a DC of 20, it wouldn't be too hard to make reasonable progress with say, a bonus of +12. I assumed a "13" ability score for the 1st level crafter, so let's look at a 4th level Expert. 7 ranks of skill, +2 for ability score, +2 for skill focus and +2 for masterwork tools is already a +13. "Greater Skill Focus" isn't an official D20 feat, but is a pretty common extra. In any case, it doesn't take many levels to make a crafter capable of making masterwork items. I gave my best or more extraordinary crafters magical tools; not unreasonable for the King's personal swordmaker.

For someone significantly better than that, the best example I can think of would be some Japanese stories, where to be a highly skilled anything implies a centered character -- someone extraordinary. A highly skilled swordsman or swordmaker would also be extraordinarily skilled in other areas. Tea ceremonies, blah blah blah. :)

Anyway, there's lots that can be done to both make higher level NPCs interesting, yet also keep them focused enough to be playable.




* One of the side conversations that happened tonight was how many 'fiddly-bits' a game system should have to make it interesting. On one end of the scale is something like D&D 3.5 or GURPS -- complex systems with tons of fiddly bits. Useful for designing distinctive NPCs, characters and encounters, but it imposes a heavy burden on the GM. On the other extreme is something like D&D Cyclopedia or as we discussed tonight, SOTC. The SOTC rules are pretty simple, but after almost a year of playing, we've noticed a tendency for minions to start looking bland. Monkeys, Nazis, Chinamen, they all play out the same way.

The interesting design question for me is coming up with a broader set of guidelines for how much is too much. I'm thinking 1-2 things for minor NPCs and cannon fodder, 2-3 things for henchmen, 3-7 things for PCs and critical NPCs. So, for the soldiers with the Phalanx feat, that's their one interesting schtick. Without it, there'd have to be something else to make them distinctive. As an alternative, Ray tends to layer significant mini-games on top of the cyclopedia rules. e.g. last session, we had big worms moving on random patrol patterns that were too tough for us to kill, we had to sneak past them. In this case, the random worm movement would be the schtick. The bottom line for me is, pre-written feats provide good building blocks for GMs to work from. Without them, it is still possible to add them. I particularly like building block systems, but I'm a stickler for consistency. :)

  • 1
Er, I know I brought this up at least once before, but I'm a favor of examples and templates built on top of more generalized rules.

So, general rules: Commoner and Expert classes.
Simplified Template: Example that can be used for a 'typical' shopkeeper.

General rule: Generalized Fighter Character Class, Feats, and Skills.
Simplified Template: Player's Handbook II simplified characters (pre-picked skills and feat progressions.

In my opinion, True20 and Blue Rose did a great job with this pattern in their character generation and sample character archetypes.

And I guess I'd also add that this is a problem well understood in the programming world.

Edited at 2008-01-11 11:58 am (UTC)

You put an order of magnitude more effort into miscellaneous NPCs that I ever have, I tell you what...

It only seems like it. :)

All I really did was spend a little time early on figuring out what a 'typical' bonus would look like, so that if I had to improvise an NPC, I'd have reasonable numbers. Once I'd done the early design work, the actual amount of time in both game prep and improvised NPCs during a session was substantially reduced.

For the larger story of NPCs, one big strength of going to elite arrays was a reduction of time to make NPCs. Rolling to get a 'reasonable' set of ability scores takes a lot longer than just splatting out 15,14,13,12,10,8. :)

So, basically, I'm able to improvise very quickly, but from a position of knowledge. Yeah, I'm overthinking this, but part of that is me putting a designer hat on, rather than a gamemaster hat. :) In an ideal world, that sort of baseline would be provided by the designers in the actual books in the form of sample characters or character templates.


Edited at 2008-01-11 10:43 pm (UTC)

It only seems like it. :)
Mmm, no, it's pretty concretely spelled out there—tons and tons more effort than I've ever put in.

I, for example, would never conceive of actually generating ability scores for miscellaneous NPCs because that's wildly more work than necessary.

Even for a major NPC I'd never actually make sure it was done by rolling or point buy or anything. Madness.

I'd very likely arbitrarily assign HP according to how powerful the party is and focus on its bonuses to saves and the like (also merely "assugned"), and only generate an actual ability score never, I'd just make it up at the table if it ever came up—which it almost never does. If the PCs bust out a -Str or such spell, then it's -1 bonus for every two they steal. Done and done.

And don't get me started on the morass that is feats...

What I was saying is, it wasn't "tons and tons" of effort. A minute to pick a class and race, jot down a stat array, and choose x skills (always maxed ranks). Another couple of minutes to do calculations for saving throws, hit, and damage. For HP, I usually just picked average hit points.

The whole reason I went for 'point buy' or elite arrays was to reduce the time required to build NPCs. I wouldn't waste time rolling up village idiots while tryign to roll up a major npc. I never really considered giving ability scores an imposition, fwiw, particularly since I'd really only be doing full arrays for important NPCs anyway.

And, importantly, maybe your group doesn't use abilities like Trip, but mine certainly did, every melee round of every fight. Knowing how resistant some guy was to trip was always important. Yes, I could just 'wing' the numbers there too, but past a certain point, making everything a wild assed guess is counter productive, at least for me. If I used the numbers as calculated, characters were very consistent.

The big headaches for me when building NPCs were: equipment, spells, feats. Equipment was probably the toughest of the 3, since equipment doubles as loot. I totally agree with you on feats, particularly as you add more arbitrary rulebooks to the mess. I usually went to a very limited subset of feats for no-name NPCs: skill focus, weapon focus, that sort of thing. I never made a specific list, I just sort of worked from memory for quick and dirty feats.

An obvious argument for arbitrary numbers: Splatting down numbers based on the PC's numbers (without regard to NPC ability scores, class, skills, equipment, and spells) means that guys are always appropriate challenges to the PCs. A very MMO viewpoint -- you can have 'level 70' mobs that don't have equipment to be looted. Of course, if you don't know that a specific bonus comes from a spell (that can be dispelled), it means winging it when the players try to detect or dispel those buffs.

It does sound like 4e is going to try to make sure the numbers are within more limited ranges -- monsters aren't built, you just assign them numbers appropriate to the PC's level. I'm still waiting to see how that's really going to work.

Actually, the most lengthy part of the process of making important NPCs was dumping a statblock template into my HTML files and filling it out. When I had the prep time, I loved doing that, because it meant less work at the table...but it definitely took longer than designing the character.

Seems like tons and tons compared to what I do...

And, no, they very rarely if ever conduct attacks based on ability scores, like trip or, heavens forbid, grapple. I think we all know that monsters don't grapple them and they don't grapple monsters. ;)

But I still can't imagine making feats for no-name NPCs. If they have no name, they're certainly not worth the effort of, of all things, picking feats for 'em.

You're crazy, I says! Crazy! ;)

Crazy like a fox. ;)

Yeah, we all signed the "grapple accords", but the Druid in my game had a pet with improved trip, which was basically a free action with no downsides. :)

Actually, the thing I *liked* about picking feats for specific no-name NPCs/cannon fodder was that it let me make them more distinctive. I've brought up the phalanx fighters, for example, as one where choosing the feat had an upside when it came to adding flavor to the fights.

The gripping hand is: I'll do the extra work when I see the upside, when I don't see an upside, I'll ditch it and wing it. :) Edit: For example, the phalanx fighters came about when I was fiddling one afternoon with background bits for one of the core campaign locations and decided it would be interesting to know how the army fights. That led to statblocks....

Edited at 2008-01-12 01:16 am (UTC)

"Grapple accords." :D

Our group used to have the "Shatter detente."

  • 1
?

Log in