Stats: 3d6

Thinking about character stats and methods, using this chart of cumulative probabilities on 3d6 for the high end. The chance of rolling a ...

18   = 0.5%

17+ = 1.9%

16+ = 4.6%

15+ = 9.3%

14+ = 16.2%

13+ = 25.9%

12+ = 37.5%

11+ = 50.0%

For straight 3d6, and 6 stat rolls per character, having a character with at least one score at:

18 = 2.7%; 

17+ = 10.6 %; 

15+ = 44.2%; 

13+ = 83.5%.

These numbers explain why I like the bonus system:

13-15 = +1

16-17 = +2

18 = +3

I feel this makes the best compromise between players' needs to play a distinct character, ease of use, and focus on player rather than character skills.

It justifies using 3d6, giving a special bonus to the truly rare rolls. (With a flat bonus at 13, you may as well roll d4 for the stat, giving the bonus on a 4 and penalty on a 1)

It makes sure a large majority of characters will have some bonus from stats, which goes a long way toward making players feel special. This is true even if +1 is just a token bonus to a d20 roll, overshadowed by the +2 or more that can be handed out based on player-skill choices. (With bonuses starting at 15, most characters will be unexceptional.)

What makes players obsess about stats is stuff like extra spells at low levels for high stats. A high stat should make a character 15-25% more effective at what he or she does best, not 100% more effective...

Digression: What the World Needs Now

Last night I was looking through old stacks of whatever self-written gaming material has survived from my teens and 20's. I realized a few things. For example, what the world needs now (from me, anyway) is not:

* A big, jam-packed dungeon written by a high school sophomore with encounters and dressings seemingly rolled at random from the charts in a certain book. Especially considering that none of the surviving maps have keys and none of the surviving keys have maps. Especially considering that my wife won't stop teasing me about that 2nd level female paladin chained to the wall wearing a "torn, revealing, dirty garment."

* Another hex map for a campaign world.

* Stats for the following "monsters": the whole angelic host of the pseudo-Dionysius the Areopagite; Slug-men; the Braintree (hm ... beaten to the punch by Neopets, anyway); the entries in Borges' Book of Imaginary Animals that the Monster Manual missed.

* Arcane ramblings about the genesis of whole hierarchies of made-up gods.

* The role-playing system I used in a number of memorable home-brew improv games, that, as neat as it is, is not really doing anything more or less than its 500 or so competitors.

What the world needs more is material that addresses real needs of gamers. I managed to dig up a few things that I hope to gussy up and present to the world in the coming months:

* The pretty neat d6-based random dungeon generator from my later years of DMing.

* A few good monsters, some of which I'm already including in the nearly done Varlets & Vermin pdf - because what the world really needs is lots more low-level monsters. A few of my adaptations also sparked the idea for another, small monstrous compendium based on the worlds of art.

* Project X - a fairly short and manageable release, based on another realization about what some old-school approaches to gaming in particular assume but don't always provide ...

Practice Break 1.5: Hard Act to Follow

Over the past weekend, this post at Ode to Black Dougal has brought forth a very instructive set of comments to a game ruling problem that itself is traced back through another layer of blogland to a Dragon Magazine article. What's cool for me is how the answers seem to line up with my analysis of solutions to game event resolution from a while back.

A referee is ruling on the party's chances for passive detection of a trailing halfling spy. Success at this will open up new areas of plot space, much the same as detecting a secret door opens up a new area of the dungeon. The question assumes B/X D&D as the starting point. That is, a system without a general character-based mechanic for passive perception. Let's look at the excellent answers, starting with the first advocate of each type of solution.

Stuart and others propose solutions based on player skill. Either the referee works recurrent references to the spy into area descriptions and waits for the players to notice. Or, more harshly on the players (and more easily on the ref), he or she judges that an expert spy would not be seen unless the players specifically look for him. I'll take these as falling within my "improvisational mode".

Clovis Cithog and others propose random solutions based on a comparison of character skill and player strategy with the spy's presumed skill, looking more like my "resolution dice" than "oracular dice," and looking in particular like the seed for a house-ruled detection system.

Jeff Rients cuts the Gordian Knot in "oracular dice" style and puts forward a simple 1/3 chance on d6. This reframes the question as "does it happen?" vs. "does the party do this?"

Alex Schroeder and Lord Gwydion offer up applications of existing rules, for secret door detection and hiding respectively. Various combinations of the above elements are also proposed.

But the most interesting of all to me is JB's answer. Its radical message - information should not be subject to resolution mechanics. If the referee wants the trailing spy to be found, and that plot space to be opened up, the spy should be found, by fiat. This echoes the "prepared" solution, non-random variety.

I confess a liking for the fiat solution because it exposes a general problem with information detection mechanics. If you meant that secret door to be found and the fungus chamber to be discovered then why leave all that preparation up to something as meaningless as a dice roll? JB's point is that the game consists of what the players do with the information at hand. So, the referee can take control over what information gets to the players. Presumably player-initiated information gathering would also be kosher under this system. But  important, game-branching information is not left to chance.

What keeps the fiat method from feeling like a one-way railroad? Player effort and choices, and the referee's responsiveness to them. In particular, to get a diverse party's players involved, the referee can fake a resolution mechanic. The party will always decipher the runes, but it's the cleric who knows the old church script they are written in. The spy will always be found, but it's the elf or ranger who noticed him. And if the party is deficient in anyone with the plausible skill, the referee may - again by fiat - reshape the adventure around this deficiency, and have the spy never be found.

Anyway, I'm really more happy with the diversity of these responses than with any one. Each approach has its good and bad sides, and the taste of the referee and players should be the ultimate arbiter of the halfling spy solution at any given table.

My meta-solution? The main question the referee should think about is, "Does the party need this information to move the game along?" If the answer is a clear "yes," give it to them, but don't make it look too easy. If the answer is a "no," let them surprise you and find it themselves. If "maybe," resolve the outcome through player skill, character-based dice, or oracular dice.

Practice Break 1: Strength-Related Actions

Before seeing what I can dredge up from my theory-dense essays of 14 years ago to supplement my theory-dense essays of today, I though I would take a break to work out some ways these ideas can inform the practice of house ruled 20th Century D&D. 

Last post I mentioned strength-related actions such as bending bars, lifting gates, and breaking down stuck doors as an example of how using the dice roll to decide the environment, rather than the character's effort, can overcome some problems with traditional ability and skill rolls. Strength tests are especially good to illustrate this because they are only applications of force against a resisting object. There is no secret insight or split-second decision for the character to make. Let's see how we can apply this to a strength test system.

Character factors: Should strength action chances go up with level? It doesn't seem they should, unless weightlifting and muscle development are part of a special character class concept.

Of course, the Strength ability should count. I'm not going to presume to know how your system uses 3d6 ability scores, but I'm partial to a general, elegant system of 13-15 = good (+1), 16-17 = exceptional (+2), and 18 = legendary (+3). In any case, I'm going to base any character-side bonuses on the standard bonus your system gives to such things as to-hit and damage from Strength, to make record keeping easier.

As we'll see below, it will be useful to express a character's Strength in terms of "the strength of n men." Figure an average modern man (or average woman tough enough to go adventuring) can deadlift about 200 pounds, and the world record is just shy of 1000. Then for simplicity's sake, erring on the He-Man rather than strictly realistic side, a +1 modifier corresponds to the strength of two men, +2 to three and +3 to four (or two men, who each possess the strength of two men).

Negative modifiers fractionate your strength, so at -1 you have the strength of half a man, a third at -2 and a quarter at -3.

Also, nobody's ever noticed this, but encumbrance really should matter when making some tests of strength. There's a reason why He-Man wears nothing but a Speedo, metal suspenders, and boots. If you're trying to pull iron bars, you're also trying to pull the weight of your iron gauntlets and vambraces; if you're trying to deadlift a stuck gate, you're also struggling under the burden of your backpack full of gold pieces. A reasonable penalty would be to shift the character down one modifier when heavily encumbered, so that +1 becomes average and average becomes 1/2 a man. For door opening, the load becomes part of your mass, so it shouldn't count there.

Depending on your campaign, small size characters like dwarves and halflings may take a general -1 modifier category to all these feats, due to their smaller mass and leverage.

Circumstance factors: This is where you roll to find out just how heavy, or strong, or stuck, the object you're trying to budge actually is. While all kinds of complicated weight and architectural calculations might come in here, really our "n men" system makes it simple. We only need to know two things: how many persons can apply themselves to the task at once, and how many man-strengths does it take to move? Whether for bars, gates, or difficult doors, a simple d6 table can be ginned up. You can roll on it, roll twice taking the highest or lowest according to your sense of the area's difficulty, choose a result on the spot, or work out each test's difficulty as part of adventure design. Just write two numbers next to each door, or zero if the door opens easily.

First roll: How much force is needed? Difficulty in man-strengths.

1: 1 man (ordinary stuck door, soft metal bars, well-oiled light gate)
2-3: 2 men (heavy stuck door, thin bronze bars, obstinate light gate)
4: 4 men (ordinary jammed door, thin iron bars, well-oiled heavy gate)
5: 6 men (heavy jammed door, medium iron bars, obstinate heavy gate)
6: d6+6 men (heavy sealed door, thick iron bars, jammed heavy gate)

The descriptions are only for scale, as it were, and can sometimes be rolled separately from true difficulty. Appearances can be deceiving; strong-looking obstacles can have a hidden weakness, weak-looking ones a hidden strength. Rust and corrosion add +1 to difficulty for doors and gates, -1 to difficulty for bars.

Second roll: How many persons can combine their strength in the effort? (For bending bars, always 1)
1: 1 person (narrow door or gate)
2-4: 2 persons (wide)
5-6: 3 persons (very wide)

Ingenious parties may figure out ways to increase the number of man-strengths applied to the problem through rope, rams, or leverage.

For doors, if the amount of force applied is one man-strength less than the difficulty, it actually will take 1d6 more tries - loud, echoing tries - to open the door, though this will have to be found by trial and error. Also for doors, if the amount of force applied is two men or more greater than the difficulty, the characters overshoot, and are rushed precipitously into whatever situation awaits on the other side of the door. Very strong characters may announce that they are applying a lower effort than maximum, to avoid overshooting.

Say the Dice

The final authority I want to discuss is dice, or random processes in general.

Games succeed when they meet a player's perfect point of challenge - when control and surprise are mixed exactly to his or her liking. Surprise in games can be achieved in three ways:


1.  Complex gameplay - enough to present a problem-solving surprise, even when all moves and pieces are completely visible and non-random. This is the approach of classic strategy games, like Go and chess. They manufacture surprise through the players' own emerging insights into game problems whose solutions weren't entirely obvious from the start.

2. Partial information - some things in the game are known to one player but not the others. Rock-paper-scissors carries this out at a very basic level; against a truly random opponent there is no reason to choose one secret outcome over others, and the game becomes sheer random guesswork. Poker is a better game for meaningful guesswork, as von Neumann realized when formulating game theory. The tension in poker is between making the optimal plays given your cards, and giving away their true value to other players through those plays. The potential for surprise is always there.

3. Random procedures - dice, cards, knuckle bones, yarrow sticks, what have you. The cheap and easy way to achieve surprise from time immemorial.

These three paths to surprise actually correspond to three game refereeing modes, if you make the insight of counting the referee as a player.

Following the path of complex gameplay, the referee lets him- or herself become surprised by the interaction of material made up on the fly with the creativity and initiative of the players. These kinds of interactions are deterministic but don't feel that way, because they come out of an unpredictable dialogue about what is possible. There is nothing truly random; what happens is whatever players make happen, vetoed by what the referee will allow, constrained by the limits of plausibility for everyone. This is the improvisational mode.

Following the path of partial information, the referee writes down contingencies beforehand for the consequences of player decisions, works out solutions to deterministic puzzles, and plans detailed encounters with opponents. This is the prepared mode. Player success here is a matter of figuring out, in a way, whether the referee has put the troll behind door 1 and the treasure behind door 2, or the other way around. Whether this game resembles the faux-random game of Rock, Paper, Scissors, or the more subtle guessing game of poker, depends on how adept the referee is at planting subtle clues, and how adept the other players are at finding them.

Following the path of randomness, the referee uses dice, but for two subtly different purposes. One use of dice I'll call the resolution dice mode. This happens when a referee knows the chances of something happening are between "no way" and "sure shot," but lacks something crucial to be able to judge it in a deterministic way. That missing element can be lack of knowledge. Let's take combat. Unless you are one of these guys, I don't think any sane referee would want to judge the blow-by-blow of combat in improvisational or prepared mode. Even if you came up with an answer based on years of sword combat experience, the players wouldn't buy it, and it wouldn't feel fair when they got hurt. There are just too many unpredictable factors in combat for your decisions to sound like anything more reasonable than kids playing "Bang! You're dead."

The missing element that propels referees into dice resolution can also be lack of patience with preparation, or lack of confidence with improvisation in other areas. Instead of going through the whole song and dance with moose heads, statue arms, and hinged busts of Shakespeare, sometimes you just want to roll a 1 and find a secret door.

Even more interesting from my point of view is the oracular dice mode. The twist here is that the referee is not using dice rolls to see how well the players succeed in their actions, but to determine the very makeup of the world around them as they discover it. Monster hit points ... rolled-up player ability scores ... wandering monsters ... and any other random tables for generating content on the fly ... all of these represent the oracular mode.

The oracular use of dice solves a nagging problem with both the improvisational and prepared approaches to setting up challenges for player characters. Often, a deterministic solution will depend on a match between a resource a character has, and a problem he or she encounters. Even if resolved randomly, game flavor demands that there are bonuses to resolution rolls from favorable match-ups, and penalties from unfavorable ones. At this point, the all-important sense of fairness can start to waver. Either the player complains they never meet a two-headed troll to slay with their sword +1, +5 vs. two-headed trolls; or the referee feels obliged to send a conga line of two-headed trolls to be carved into kebab; or there is some middle ground which feels awfully like a carefully plotted out, artifically fair middle ground, and not at all like real life.

The oracular solution, though, feels like real life: it's completely unknown to everyone just how often the character will run into a two-headed troll. And for that the absolute best thing is a random encounter table with a slot for two-headed trolls. (The other satisfactory solution is for the player to take part in improvising the  adventure and actively seek out favorable matches, asking in every tavern for the nearest two-headed troll lair; but even this should only increase the chances for a troll roll, not guarantee it every time.)

The reason oracular dice are so interesting is the possibility of using them to settle a number of situations usually sorted out with resolution dice rolls. For an example that got me thinking along this track, take a look at the language rules in version 0.5 of James Raggi's Lamentations of the Flame Princess ruleset. Every time you encounter a new language, you roll to see if you know it, the chances being determined by your Intelligence.

At first glance, this seems crazy; what do you mean I don't know whether or not I speak Finnish until I run into an actual Finn? But when you consider the alternative, it starts to sound crazy like a fox. The alternative is the two-headed troll problem. As the preparing referee I have to decide what language the all-powerful army speaks; you know, that army that can only be dealt with by parleying. I know what languages all the player characters speak. So do I screw them or give them a break? Why not just roll for it? That way, any outcome seems fair; blame it on the dice, you had your chances.

(As an aside, though, my preferred solution would have the roll be for what language the army speaks, and see if it matches the characters' known languages. Not knowing which languages you speak, for instance, deprives you of the ability to choose to travel to those areas or seek out those people. And then there's the munchkin factor, where players try to speak to as many people and read as many old manuscripts as they can to max out their language skills ...)

Here's another infamous example familiar to most of us. A certain game gives a character with a Super Duper strength a 50% to bend bars and a character with an Average strength only a 2% chance. Furthermore, this game allows only one try per character to bend bars, for fairly obvious reasons. So Super Duper Man blows his roll and Average Guy makes it. That doesn't make sense. Why not have the roll be oracular: determining the strength of the bars? That way both characters can bend weak bars (roll of 02) but only the muscleman can bend strong ones (roll of 47). Strength of bars, hiddenness of secret doors, complexity of trap mechanisms, and other "skill roll" problems could all benefit from taking a closer look at the oracular dice possibility.

Now, although this is a good wrapping up point, I'll just point out that all the above comments about referee decisions also hold true for rules. Because, as I explained last time, game rules are just referee decisions written down and standardized. This is true whether the rule tells you it always happens, it can't happen, or you have to roll for it.

Next up: I think I'll tie in some old essays I wrote to the discussion.

Says The Rules

Rules accumulate in a system where the player-referee conversation cannot by itself supply a sense of fairness. 

And by extension, in systems where the player-player conversation also falls down. The Legend of the Five Rings CCG, like Magic and so many collectible card games of the Class of 1994, started out with rules and card wordings that in many places had to be interpreted by good faith and good will. Issuing rulings was entirely a fan-supported effort, and by fan I mean Jeff Alexander. The atmosphere was more relaxed. Players at competitive levels were open to improvised stunts that a more serious game would have frowned upon. Like the two Dragon Clan players at the Day of Thunder championships who "settled their differences in the mountains" by playing their game on the tournament stand instead of their seats. Or the Scorpion Clan players who bribed the judges in-character with koku (L5R product points).

But the competitive environment evolved, with the game persisting far beyond the three-year ending point originally envisioned for its story arc. The gonzo atmosphere got separated from the game play, and spot rulings eventually weren't enough. Competitive players want to know that their questions will get the same answer no matter where they play.  The interpretations of rules lawyers get taken more seriously, no matter how strong the temptation to take them aside, casino-style, and tell them not to be such a wanker. So what previously was left to player agreement, then tournament judge decisions, moved to an official list of case-by-case rulings that got longer and longer.

When I moved from design team to rules editor for 2007's Samurai Edition, taking over from Jeff's 12 year reign as rules guru, one of my priorities was to clean up the morass of often inconsistent rulings that had accumulated over time, codifying them into a background rules document. From case law to principles, in other words. This move also created a more consistent environment, because it laid down rulings based on general rules rather than specific precedents. Fair? Very. But the door had finally closed on the carefree early days, when players had to figure out for themselves the implications of phrases like "The Samurai must not do whatever it is he was about to do."

So it goes with role-playing games. I use "game" here both in the small-scale sense of a particular referee's campaign, and in the large-scale sense of an evolving game system. RPGs, too, start to grow more rules wherever and whenever the sense of fairness starts to falter.

House rules. Let's assume that a game can start from a Golden Age of improvisation and negotiation over rule gaps in a spirit of mutual trust. As more and more spot rulings are made in this game, a sense of consistency and fairness dictates that repeated rulings be noted down. If the Dungeon Master rules that an unarmored character moves silently on a 1 in 6 chance one day, it's unfair if the next day's ruling is that he moves silently on a 25% chance modified by Dexterity. Fairness thus eventually demands the expansion of house rules - the equivalent of a local scene judge's consistent way of making rulings in a CCG or other complex game. This evolution is an ideal, but the truth is that even before house rules, all but the most freeform of games start out from a core of basic rules. These rules are simply the designer's idea of the basic stuff that should be consistent in the game in order to be fair. Everything added on serves the player's and GM's sense of what should be fair - opening some avenues in a consistent way, closing others that prove abusive.

Standardization. The next step of evolution, then, is when house rules are not enough to serve an expanding scene. This is the step that Gary Gygax explicitly took in the AD&D books, which filled in many of the gaps previously addressed by house rules, in the interest of greater portability of players between different campaigns. For some, fairness suffers if the house rules are arbitrary. (In my experience though, the profusion of rules in AD&D in fact led to a sort of house-ruling by subtraction. First off the plank in our campaigns: things like psionics, weapon vs. AC modifiers, and the grappling system.) In CCGs, this step is represented by the precedent-based rulings document.

As with CCGs, and as also noted by Gygax in AD&D, the growth of serious competitive play at conventions also created a push for standardizing house rules into game rules. Fairness has to depend more on rules if the players don't know each other socially and if groups are competing for some award.

Market-driven reasons also count. One way to sell books is to fill them full of new official rules you just can't do without.  Then there's demographics. It is hard to escape the involvement of new gamers, younger people, or the kind of difficult folks that nerdish hobbies seem to attract. Rules create a superficially easier way to create that sense of fairness without requiring much in the way of trust, life experience, or social skills.

All this has been noted before in various places in blog-land, but no matter how strong the rules, there still has to be some level of agreement on how far the rules can be bent, exploited, and munchkinized. I'll never forget the day one of our long-time players brought a female friend back from MIT to play in our campaign. We pretty much had an agreement to play straightforward heroic-style AD&D, but this young woman threw every cheesy trick imaginable, from the Command spell using "masturbate!" to creating hand grenades by having Magic Mouth + Fire Trap cast on clay balls. She was playing by the rules, and would have torn up a convention game. But she wasn't playing by our unwritten rules. That's also a matter of trust and fairness - even if our way of playing to the spirit of the game was one that competitive gamers would sneer at.

Consolidation. In CCGs and other complex games, the case law of competitive game rulings can accumulate and make it hard to learn and see the underlying principles, requiring a housecleaning and simplification. Likewise, the ever-mounting number of rules and options across the supplements of a long-lived roleplaying system can reach a point where they would most obviously be improved by consolidating the mechanics involved. This was most famously the promise of d20 D&D - to compact the mass of charts, tables, ascending and descending numbers, different dice rolls, and subsystems over 25 years into one core mechanic; to make monsters work like PCs; to get combat down to a science. Ironically, this simplification was quickly overwhelmed by the profusion of character-based features and cruft, and it turned out that giving monsters the same stats as PCs wasn't really a step toward simplification of play. But these faults were bolted onto a rock-solid mechanical chassis, one that Castles and Crusades in particular has taken advantage of while adopting a more stripped-down set of options.

Next in the series: Dice and fairness.

Says Who?

Welcome all. In this blog I'm going to keep a jotting of my thoughts about game design, in whatever game I'm having design thoughts about at the moment.

Right now my thoughts are all about D&D systems in various "old school" incarnations. In fact, my blogroll pretty much consists of writers I've found have interesting thoughts about roleplaying in that system. That may change as I move my thoughts on to other topics like board/card game design, wargames, and CCG.

I'll also use this space to post my own amateur productions for the public domain. Right now I am close to finishing a booklet of low-level monsters for Swords & Wizardry. Another longer-term project is a complete reworking of the D&D character classes and spell lists - an attempt to combine New School variety with Old School elegance. We'll see how that pans out.

---

So, old D&D. It's not about nostalgia for me. It's about the new insights into an old game, summed up most eloquently in Matt Finch's Primer. And with that, the possibility of players old and new knowing more, and playing better, than most of us ever did in 1982.

Key to understanding the different forms and flavors of D&D, or really any refereed game, is knowing who or what the system gives authority to. How do you determine what things happen, and whether plans translate to success in the game?

Referee and Players
One of Finch's key points about the old school ethos is that authority comes from "rulings, not rules." The examples of tactical play in the Primer in fact show an even more complex principle. It's not just ruling by referee fiat, or player say-so, but by an active conversation between referee and player.

While the referee's authority is final, the player moves things along, proposing and trying actions that generate new rulings from the referee. If the authority tips over too much in favor of the referee, the players don't feel like they are playing. You get bad DMing like the play example in the 1983 Mentzer edition of D&D, where the DM starts to tell the players what they do, think, and feel. But too much in favor of the player, and you get a walkover campaign with no real challenge - an infantile world of primary-process wish-fulfillment.

That's on the tactical level. But it also works on the strategic level. The old school love of open-ended "sandbox" settings and many-forking megadungeons - the distrust of pre-ordained plots - speaks to the same reliance on improvised communication between players and referee.

Even more relevant is the return to underwritten rather than overwritten adventure locations. Yes, crass practical reasons might underlie the appeal of one-page dungeons and one-sentence room descriptions. Grown-up gamers have less prep time on their hands. But the minimal approach to campaigning also gives room for the same kind of conversation between players and referee. Players say where they want to go. Players, not referees, get themselves into trouble, come up with plot material and adventures. Referees improvise dungeon features, creature tactics, villagers and taverns. There is no boxed text to read.

That's great! That's golden! Indeed. But underlying this is the assumption of trust - of fairness. Because the ultimate authority rests with the referee, players have to trust that their referee keeps both tactical and strategic play balanced between possibility and challenge. It's not just a trust in the referee's good will, but trust in ability to generate fair and believable content on the fly.

The referee in turn expects the players to respect his or her authority - to push the implicit limits of play, certainly, but not tiresomely, and never to challenge the referee's final "No."

That describes the improvisational system where referees, and secondarily players, have tactical and strategic authority over the game. Take this idea to the utmost, and you end up with a refereed Engle matrix game, or something like the wonderful Baron Munchausen, where all actions are resolved by proposal and counterproposal. But D&D is not so free-form. There are rules - for characters, combat, saving throws, magic, and (controversially) non-combat skills. Why are there rules?

Rules accumulate in a system where the player-referee conversation cannot by itself supply a sense of fairness. 

And - because I'm trying to keep these posts at a reasonable length - I'll explain what I mean by that in my next post.