But I actually started to discover issues with D20 run time complexity. I kept forgetting to use NPC feats and abilities (and a couple times I even accidentally invented abilities in play).

Now I’m not sure if Cold Iron seems less complex because of my familiarity with it, but somehow, it feels like it has just as much depth as D20, with seemingly a lot less complexity. Now one factor is that it doesn’t have so many ways to do the same thing (look at how many different ways undead “drain” works for a simple example…).

Part of D20′s complexity actually turns out to be the challenge rating system. By having such a metric (and the metric actually having a significant factor of useability), and tying XP very directly to that metric, the prep is more complex. Now the good part is that a relatively innexperienced GM will probably do better at setting decent challenges with D20 than they might do with Cold Iron (which doesn’t have such a metric). On the other hand, the novice D20 GM may have no clue what to do when the CR rating is horribly wrong, and if that happens very rarely, he may never learn how to deal with it.

Of course the complexity in a given GM’s view is dependant on how the individual GM deals with each of the 4 categories of prep. For me, D20 prep is the most intensive of all the games I’ve run recently. Cold Iron is fairly light. Dogs in the Vinyard is even lighter (especially if you use towns others have created). Burning Wheel completely stumps me, and I have no idea how light or heavy the prep would be if I had some clue how to prep for it.

Frank

I know what you mean, though I want to disagree on a minor point to support a larger one: d20 NPCs are typically not all that complex for the degree of detail you get. Sure, it can be somewhat difficult to run a d20 high level character with a lot of resources (such as a bunch of spells), but not grossly so compared to options available. That’s not entirely true, but I’m generalizing for clarity.

However, d20 NPCs achieve this state by moving tons and tons of complexity into the prep side. Heck, just writing up the listing is an exercise in calculating and organizing a bunch of things that might be needed. Now, in d20, derived stats aren’t the only thing that contributes, but they are a big part of it. There are really only three options for the game designer, where the question of complicated derived options are concerned:

1. Simplify the game (i.e. remove the complexity, perhaps more than desired), by using only derived stats that can be quickly and easily calculated in play.

2. Keep the derived stats, but calculate them during prep instead of play (possibly impeding play time).

3. Abstract the complexity into some mechanic other than derived stats (perhaps leading to subtle, unwanted consequences due to the nature of the abstraction).

It is in this area that I think the drive to have better character sheets for games can yield the most useful results.

Note that I’m using “derived stats” in a very general sense here. Getting more spells when your D&D wizard gets another level is a form of derived abilities.

My problem with D20, and especially Arcana Evolved which doesn’t have the level of 3rd party support, is the complexity of NPCs. AE spell casters basically need to be built from scratch. I also kept needing monsters that the easiest way to get what I wanted was to advance a monster from the Monster Manual, but that then takes work to re-figure BAB, saves, skills, choose new feats, etc.

And sometimes picking treasure became such a hassle that I basically told the players: “There’s a bunch of treasure, in fact enough to get each PC up to what they should have by the wealth guidelines – tell me what items you want.”

As the PCs gained levels, the complexity went up, and the spell casters became more and more dominant.

Right now I’m struggling with how to prep for Burning Wheel. I’m about ready to can the game as being unplayable for me right now.

One thing that’s starting to be really frustrating is that D20 players can’t be bothered to learn anything else because it’s too much work. Meanwhile, D20 GMs are burning out due to the prep effort. I guess as long as there’s enough people willing to GM, the situation will never change.

Frank

Martin, I haven’t put out anything on my game in public yet. I’m trying to get the first draft written and playtested before digging too deep into that aspect. My gut tells me that some Forge ideas are helpful for me. Other Forge ideas I’m deliberately bucking against. So I want to see where I can go on my own before reviewing the game against those ideas.

The need to handle prep issues is really driving my design at the moment, and not just in that big gaping hole that “scaling” represents. I have several pieces that work well enough for play, but I’m not satisfied with the amount of prep required. One of the ways I’m trying to make prep easier is use some of the concepts from software design. In Frank’s terms, the main effect is that certain “organizational” prep support must be built into the game itself, rather than supplied by the players.

Frank: Thanks! Great post — it’ll be the subject of tomorrow’s minipost here.

Frank

Building a good scaling mechanism for character design is one of the two big gaping holes in my current game design. I have identified the problem, but I’m not happy with solutions thus far. I was going to cop out. Unfortunately, other parts of the design (“players play all the monsters”, and design many of them) indicates that I must find a solution, or find a way to bypass the issue. ;D

The best I have so far is that I’ve promoted some categories in the design from a purely conceptual state to having an effect in design. For example, I have the category “weapons” (referring to ability with weapons), which has whatever specific groups of weapons the GM wants for his particular campaign. Might be “blades, hafted, etc.” Might be “short sword, rapier, etc.” But the important thing was that “weapons” was only a way of talking about them, not something you recorded on the sheet. There are about 15 such conceptual categories in the current design.

Now characters can be passed around. The person who designed the initial character, first introduced the character in-game, added to the character, or play it next–they might all be different people. So you see the need for scaling. The important thing is to convey an idea of the character without necessarily pinning it down precisely. (We want those players that add to the character to make informed, but real choices.)

Say I design a generic orc grunt. Orcs use all kinds of weapons, and this guy hasn’t even been used in game yet. I specify that his “weapon” ability is 5 (competent, but nothing spectacular). This does *not* mean that he has a 5 in all weapons. You can’t directly derive his battle axe ability from that number. All it does is provide a marker for the next player to pick him up–so that they don’t do any reverse engineering from pure text or unrelated skills. It says that if the orc had all of his individual weapon skills written out, he’d roughly have a five in the things he bothers to use.

Player A picks up (a copy of) the generic orc, gives him a name, and decides this particular orc has a short sword. He might decide that the orc prefers the sword, and bump his skill up to 7 or 8. Or he be in a hurry and use the 5. Or maybe the orc picked the shortsword up in recent loot and only has a 3. Whatever, this makes a statement about the orc. If Player B later takes over, and decides the orc has a knife at his belt, then skill with that probably needs to average out with early short sword pick. (Either player could also bump the orc’s skill up or down, using “GM Points” to do so. But then this isn’t a generic orc anymore.)

As you can see from that example, I’m largely punting on the issue so far. I don’t want derived values. I do want a way to communicate information about a large number of skills without writing them all down. The fact that I want to leave some wiggle room actually helps in my particular design.

I’m also considering the idea of semi-derived values based on random die rolls. So perhaps Player A decides this orc has a short sword, then rolls againt the 5 to determine actual skill with it.

Back to the larger topic, I do believe that something that really address such prep issues has to be built into the game design from the start. This issue is holding up about a third of my game text, because I anticipate any decent solution necessarily changing other parts of the design.

I touched on the reasons you mentioned, but only broadly — your examples are excellent, and I couldn’t agree more that traditional RPGs can learn from their nimble indie cousins. Flag framing is an incredibly useful — and driftable — concept.

Al: I think that’s one of the big reasons why so many indie games focus on shorter play cycles. As I’ve gotten older, I’ve found that real life intrudes on long campaigns more and more often, and that’s really starting to change my POV.

Chris: Yep, we probably do see light/medium/heavy prep differently. Your points are excellent — thank you.

CJ: Care to share any details on your game-in-progress?

Frank: Burning Wheel is a tricky one, which is why I didn’t bring it up. I’m not sure which of my categories it fits into, if any.

You’re definitely onto something with your division of prep by type — I’d love to see that expanded. If you post about it on your blog, please let me know so I can link to it.

Roger: I’d forgotten about Universalis, the game that breaks so many tidy systems of categorization.

That brings to mind that there is value in indentifying the amount and types of prep necessary. Off hand, I can think of three separate types of prep:

+ creative – generating ideas, situation, etc.
+ mechanical – writing up NPC stats, determining obstacle DCs, etc.
+ organizational – organizing and neatening up notes, e-mailing notes to all the players, updating a game Wiki, etc.

Two minor notes on Burning Wheel: First, the derrived stats aren’t separately purchased, so the rules on simple NPCs is just providing some information to make it quicker to compute the derrived stats. Second, Burning Wheel has another thing that helps simplify prep and better use the simple NPCs. Burning Wheel does not require the full combat system be engaged for every conflict where combat skill is relevant. The GM may call for a simple opposed test, and this is especially worthwhile when the survival or injury of the participants is not relevant to the conflict.

Frank

Maybe I need to reword my “easy, obvious, and explicit” criteria, because the d20 bartender example was meant to be an example of a GM wanting the kind of scaling I mean, but not having it. So they make it up. And of course the problem is that the bartender might end up in combat, but the notes the GM has on him are worthless for combat. That is, you don’t have a simplified character, you have an incomplete character. (A purist might note that giving a few ability scores and a few skills, one can make an educated guess on the bartender’s class and level, and thus rough combat capabilities. This is true, but is exactly the kind of reverse engineering that isn’t easy, obvious, or explicit.)

Derived stats are pretty much the only way I have seen games attempt to address this issue at the design level. Toon does it, Hero system does it. GURPs does it. Sounds like Burning Wheel does it. Trouble is, derived stats don’t so much handle complexity as push it into another part of the design. Derived stats have other issues–which is why there is endless debate, for example, in Hero circles over the usefullness versus complexity cost of the derived values. For one thing, a lot of derived values make it hard to fiddle with a system without changing all the derived material.