Nick Thomas

Anatomy of an RPG

Monday, August 30th, 2004

This is the first issue of my column on RPG mechanics and RPG theory. Never in here will specific RPGs be discussed at length; this is a column about RPG mechanics as a whole. This month’s issue, not unfittingly, is a detailed anatomy of RPG mechanics themselves. This should prove useful to RPG designers by allowing them to split their mechanics up into bits. Things are always easier to do if you can split them up. It also drops some nice hints on mechanics…
All rules in RPGs can be divided into two categories; character description and action resolution. You need rules in each for the RPG to be playable. The cleaner the mechanics, the more separate they are. In D&D, for instance, the character description are hoplessly entangled so it’s hard to distinguish one from the other, wheras in GURPS, you can clearly divide character description and action resolution. The ideal RPG would have these two sections capable of operating independently. In the rest of the article, I will split these two categories into more categories and detail on each.

Character Description

Character description invloves describing characters using various descriptors (skills, for instance), and assigning values to them, usually numbers. A quick list of ways in which a descriptor can be described:

  • Quantative- a number. A quantity.
    • An integer.
    • A percentile value.
    • A range. Such as 1d8 (1-8). More on random number generators later.
  • Qualitative- anything that it not measurable as a quantity. A quality.
    • A concrete value, such as “female”.
    • An abstract value, such as “unlucky”.
    • A general value, such as “elf”.
    • An adjective, such as “strong” and “very strong”.

Now some discussion of distribution of these descriptors. RPGs tend to divide descriptors into attributes and skills (though most use different terminology). Attributes are descriptors which describe the character’s raw abilities, while skills are their training and experience. There are actually several generic categories:

  • Attributes- quantative descriptors representing raw abilities.
  • Skills- quantative descriptors representing training and experience.
  • Advantages, Disadvantages, and Quirks- qualitative descriptors representing almost anything.
  • It seems worth noting that D&D has categories for attribute-type quantative descriptors, attribute-type qualitative descriptors, skill-type quantative descriptors, skill-type qualitative descriptors, various assumed descriptors (see below), plus assorted odds and ends (such as Size), plus whatever the magic system is. This is a stark example of mediocre game design. RULE OF THUMB: If a game system needs a separate system for magic, you should probobaly ditch it.

Now some assorted descriptor anomalies and peculiarities:
Assumed Descriptors: Almost all RPGs do not list many descriptors, but simply “assume” them. Take D&D for instance:
Assumed Descriptors in D&D

  • Height, weight, others such as hair and eye color
  • Gender
  • Alignment (!)
  • Religion

It is my humble opinion that you should never, ever use these, because you can’t assume anything. According to this, dragon characters have hair colors, and worse, centipedes have religion. And that’s without even stepping out of the fantasy genre! These may be okay in genre-specific RPGs if carefully thought out, but never, ever in universal RPGs.
Adjectives as opposed to numbers: Many RPGs (such as FUDGE and The Mirror, couldn’t find a link, sorry), choose to use adjectives instead of numbers for their quantative values, on the grounds that they tell you more about the character. What these RPGs fail to do is work the adjectives into the game system seamlessly. They essentially use the adjectives as a subsititute for numbers, and in the case of The Mirror, as a thinly disguised range pretending to be an adjective. I have yet to see a system using well-designed adjective-based quantative values.

Action Resolution

Action resolution involves weighing various descriptors of PCs, NPCs, and outside forces to arrive at a conclusion which determines logically the outcome of a certain situation.
Data Processing: Action resolution is all about processing data. You get data from character description and then you do things with it with action resolution. A successful action resolution is one that provides realistic results and covers as many situations as possible with as much speed as possible. The features you look for in an action resolution system look eerily similar to those you look for in a computer:

  • Processing a wide variety of data.
  • Producing high-quality results.
  • Processing data quickly.

At this point, I will burden you with one of the design epiphanies I have had in recent months: You can gain a lot by thinking outside of the box. You can probobaly gain more by stepping out of the box and finding what is already there. For instance, talk to a programmer about RPG design and you will probobaly get some good insight on RPG design drawing ideas from, say, Object Oriented Programming. Talk to a philosopher about RPG design, you will get concepts drawn from philosophy. This advice may be less useful, though, because philosophy is rarely as concrete as programming. Therefore, I conclude that one can draw an infinite amount of inspiration for RPG mechanics by looking at existing logisitical schemes. Let’s take an example; our timekeeping system.
Timekeeping is about measuring numbers; therefore, we will be using the logic scheme used to divide up time to divide up quantative values. Instead of measuring values on one scale, we divide them up into seconds, minutes, hours, days, weeks, etc. This means we can now have a value of 6 be a value of 1 with a higher unit. Incidentially, this system is highly impractical, so we adapt it. We use base 1 units. Now we can shave some zeroes off of round numbers by using higher units. This is a nice rules fragment, if we can work it into the system; working-in is the main difficulty of adapting existing logic schemes.
Okay, that huge tangent complete, back to action resolution.
Randomization: Almost all RPGs have some way of introducing random factors to action resolution, usually using dice (and with good reason; dice are one of the more practical and powerful random number generators out there). Most systems that advertise themselves as diceless use cards or some less convenient random number generator. This silliness falls into the same category of using adjectives instead of numbers; unless you can make it work, don’t do it! Anyway, back to random numbers. There is an incredible amount of detail that I could go into on this (see John Kim’s dice mechanics article for said detail), but I won’t. For now I will just cite a few things I find interesting:

  • Ranges: Like those things in The Mirror where you have just a die roll for each of your descriptors. This is a very good method of randomization (aside from the cost of dice) if you can make it work, which is easier to pull off then the other schemes I have said that about. The nice thing about this is that if randomization uses any random number generation, everything is a range. So, if you can come up with a practical way to initiate the generation of any range in one step, all action resolution will be one-step. The trick is finding a number generator that can do this pratically.
  • Varying Degrees of Randomness: Some RPGs put more weight on the random number generators than others. In some RPGsD&D, this can vary based on the situation, which is more a result of poor design than intent. However, there is something very good in here. While there is clearly no determinable answer to how random your game should be, there are three answers, one of them inspired by the silliness of D&D:
    • Variable Randomness: The randomness varies logically by situation, much like D&D, except D&D varies randomly like an electrified hula dancer. What we want is an RPG where the degree of randomness is based on the degree is randomness that is logical. Someone let me know if you figure this one out.
    • Customizable Randomess: The GM can determine how random they want their game. This is a tough one, but I pulled off a similar thing in my RPG by letting the GM vary how precise they wanted to measure player abilities.
    • A Combination of Both: If the GM can vary the randomness on a case-by-case basis, then they can fit not only their preference, but realism also. This may be the best solution, and also the hardest to make workable. It does present all sorts of opportunities for imbalance. Suggestion: Character stats are based on a die and a number added to it, such as 1d4+6. Well, if the GM could vary this by taking 1 away from the number to be added and adding 2 to the sides of the die, then he could vary how much of it was random. For instance, 1d4+6 could become 1d6+5 if he wanted more randomness, or 1d2+7 if he wanted less randomness. This is impractical on several levels, but just an attempt to prove that you can come up with mechanics to accomplish this stuff.

Now, a graphic:

It turns out that this graphic is not only as cute as a button, but describes the layout for the rules scheme of almost any RPG. I used a few new terms here, so let’s take a look at them:

  • Generic Descriptors: Descriptors which apply to all characters, or at least most. Like ability scores in D&D.
  • Specific Descriptors: Descriptors which make a character unique, or at least less generic. Like skills in D&D (and feats, and class, and spells, and..)
  • Data Acquisition: Before it can process data, the action resolution system needs to get some data. What data does it use? Is it quantative or qualitative? What kind of quantative or qualitative?
  • Data Acquisition: Before it can process data, the action resolution system needs to get some data. What data does it use? Is it quantative or qualitative? What kind of quantative or qualitative?
  • Variables: The information about the outside world that the resolution system acquires. While descriptors are predefined, this is more interesting because you need to figure out how you want to describe the variables. It is generally a good idea to make it as similar as possible to the way the descriptors work, for easy compatability.
  • Factor Weighing: You need to figure out how the data acquired will interact. For instance, most systems use “the highest number wins” (and there is nothing wrong with this). Straightforward and modeled after reality are good traits to go for.
  • Random Number Generation: To add that element of uncertainty. The value of random number generation is debatable, but you should all of the aforementioned factors if you intend on using it.

In conclusion, there are an obscene amount of ways in which you can design your RPG, and no shortcuts, but working on one piece at a time and keeping your goals in mind while doing so will probobaly end you up with a good RPG. Also, study up on RPG mechanics. The better you understand how RPGs work, the more easily you will be able to build them.

This article by Nick, who thinks he may have made an RPG for everyone: Don’t be afraid to tell him if he’s wrong. You can contact him at

What Truly Defines a Good RPG?

Tuesday, July 27th, 2004

Despite the constant effort to turn out good RPGs, rarely is the question asked, “What is a good RPG?”. You would think this would be a central concern, since nothing can be more important in making a good RPG than knowing exactly what you are trying to accomplish. This article will *attempt* to answer that all-important question.

Ultimately, everyone expects something slightly different from an RPG. This may be why there are so many RPGs out there. Everyone who is competent enough to do so makes an RPG which will fit their preferences.

From this, we can deduce that it is impossible to create an RPG which will be good in everyone’s eyes. It is safe to say that it is impossible to please everyone, and therefore impossible to make an RPG everyone will enjoy, especially with many people having mutually exclusive desires in an RPG.

Despite this, I believe it is possible to make an RPG that will please most people. How, you ask? It takes some cleverness, to be sure. Despite everyone having different specifications, there are a number of points which most people can agree on (should be simple, should be realistic, should be balanced, etc). Therefore, make an RPG with all of these traits, and you have made a pretty awesome RPG. Of course, anyone who has ever made an RPG before will know how hard it is to make these traits coincide. I will now go on a brief tangent to elucidate on this point.

Fabien Ni?oles’ System Color: A typology of RPG mechanics states that all RPGs can be measured on the basis of Fluidity, Consistency and Immersivity. Fluidity is the speed of resolution, angle of the learning curve, adaptability, and general unobtrusiveness of the mechanics. Consistency is the game balance, realism, and general solidity of the mechanics. Immersiveness is an abstract concept not related to mechanics, so we will leave it out of the discussion. Anyway, so we have the criteria of solidity and unobtrusiveness. Making the rules lightweight and fewer adds to unobtrusiveness, but detracts from solidity. Making the rules more precise and comprehensive adds to solidity, but detracts from unobtrusivness. Thus, we are still stuck with mutually exclusive goals! Countermanding this rule is the challenge of the RPG designer; cramming in as much solidity as possible without pushing out unobtrusiveness, and vice versa. This requires clever and innovative mechanics. Thus, originality is a virtue in RPG design, not just because of copyright issues, but because creativity in mechanics design truly makes the RPG better.

Our tangent has wandered back towards a solution! But the question remains unanswered: what makes an RPG good? The answer is pretty obvious; see if you can figure it out.

The more people an RPG pleases, the better it is. There. We have a general, neutral, blanket statement defining a good RPG.

Now let’s take it one step further and figure out what will make the most people happy. In order to do this, we will refer to the GNS player model. For the benefit of those who have never heard of it, let me go on another tangent to summarize it. The GNS model described three types of players; Gamists, Narrativists and Simulationists. Gamists play the RPG like any other game; their sole goal is to gain wealth, power and fame. They see their character as a collection of powers optimized for the acquisition of more powers. They will take advantage of loopholes in the rules for power. Gamists are usually not picky about mechanics, but, as with everyone else, solidity and unobtrusiveness are virtues. Narrativists focus on the storytelling and roleplaying aspect of the game more so than others. They are the type who is least concerned with the mechanics. Ironically, they also tend to be the pickiest about their systems. Unobtrusiveness is an enormous concern for Narrativists, but solidity is also important, though never at the expense of unobtrusiveness. Narrativists also tend to be very picky about any rules which limit their actions (such as the combat system in D&D). Simulationist is a much more abstract and complex category. Simulationists essentially want to do something in the RPG that they can’t do in real life, such as have a gunfight with aliens and robots in a dark alleyway on a space station. This is a huge and diverse category, so it is impossible to generalize as to their rules preferences.

Okay, now we have the GNS model defined, we can continue with our discussion. In order to please as many people as possible, you have to make rules which will accommodate each of these three styles. You have to have mechanics which have enough solidity to them to put focus on to please Gamists. These rules also have to be unobtrusive and open-ended to please Gamists. Finally, the rules have to be able to handle extremely unusual situations and be unobtrusive and solid to please Simulationists. Not possible? In the immortal words of Johnny Depp, “Not impossible- just not *probable*”. Still, it’s quite a challenge. So how do we do it? A recipe for a really good RPG:
# Start by identifying your goals. If you can specify exactly what you want, it will be that much easier to achieve.
# “Borrow” things that worked in other RPGs. If it ain’t broke, don’t fix it.
# Where nothing works as well as you want it to, make up your own rules that other RPGs will want to borrow.
# Once you have perfected all of the things other RPGs do, add more features!
# Having achieved your objectives, make some harder ones and redesign the RPG to achieve them; an RPG is never completed; only abandoned.

Remember, we don’t truly need an RPG that works for everybody. It’s worked fine so far having an RPG for every group of roleplayers. But it would be nice to have a banner which everyone across the GNs model could unite under, wouldn’t it?

This article by Nick, who thinks he may have made an RPG for everyone: Don’t be afraid to tell him if he’s wrong. You can contact him at

CARPS – Customizable Abstract Roleplaying System Version 4.

Tuesday, July 27th, 2004

Well, here it is; version 4. of CARPS. In the making of this, I trashed every single rule. This is a completely different game system than 3.5. I think you will find it quite drool-worthy. A lot of the rules seem kind of crazy and pointless, but they are all there for a damn good reason; don’t question the RPG designer! At the bottom of the page I have added a justification for every rule. Finally, this is not for beginners. This is pretty advanced and makes no attempt to explain the basics of roleplaying. Newcomers would be better off with Version 3.5 or CARPS lite.


Concepts: Nick Thomas
Writing: Nick Thomas
Editing: Nick Thomas
Publishing: Nick Thomas
Ideas and Inspiration, with all due respect:
GURPS, FUDGE, Risus, Alternate Realitites, and of course SLUG.