So we’ve stated some design goals, and we’ve set a baseline with an in-depth review of Cyberpunk 2020. Now, it’s time to get into the weeds. As I stated in the design goals, I want to create a game inspired by Cyberpunk 2020. As such, most of these articles will revisit one or more mechanics from that game. That said, after considering the implications of these mechanics, I will more often than not rip them apart. Want to see us journey from nine stats and a d10-based resolution mechanic to three stats and a dice pool? Read on.

Core Mechanic

Before we get into how we’re describing characters quantitatively, we need to understand what these quantitative descriptors actually mean. Cyberpunk uses a relatively straightforward system where the result of a d10 is added with both the statistic and skill being used to make the roll. Roll targets are set by task difficulty. Below is a table from the corebook, with an additional column showing the probability of success for a character with an average stat (5) and an average class skill rank (4):

Difficulty	Target Number	Probability of success
Easy	10	90%
Average	15	50%
Difficult	20	10%
Very Difficult	25	5%
Nearly Impossible	30	1%

Works out pretty nicely. The question now should be whether this is a spread we want. There are a few things to note here. 90% is the highest possible probability, because a roll of ‘1’ always fails. On the other hand, no roll is actually impossible, because 10s explode. These both end up being adjudication issues more than math issues: One could easily argue that something with less than a 10% chance of failure shouldn’t require a roll, and additionally argue that something with no chance of success shouldn’t be allowed a roll. I will note that these facts aren’t spelled out in Cyberpunk, but at the time that the game was new, they weren’t spelled out anywhere.

The most important thing needed to determine if this probability spread is ‘fair’ is the definition of an ‘average’ skill roll. One thing that many people, myself included, have philosophical difficulty with is the fact that almost nothing an average person does has a 50% probability of failure. Now, characters in role-playing games aren’t doing everyday things, and the action which would be ‘average’ in a game session is likely to be under a lot more pressure than driving to work or checking math in a spreadsheet. A game which lays this out is more intuitive to run; the best example of a system which does this that I can think of is Torchbearer, where the target number starts at 0 or 1 and builds based on what factors you add.

One nice thing about this dice mechanic is that increases in stat and skill values are palpable. A character with an 8 in the stat and an 8 in the skill is going to succeed way more often at higher difficulty skills than the average character profiled above…a target 20 roll goes from 10% to 60% chance of success. That’s great from a gameplay perspective, but it leads us to ask: is there a spread of ability so large that one person is six times more likely to succeed at a task than another? And if so, is that more from skill or more from innate ability?

In order to simply address both of those issues (span of capability and differentiation of skill and attribute), we’re going to need a different die mechanic than straight d10. Not because d10 can’t do what we need, but because making it do those things will require a lot of documentation…the rules will bear the weight of the differentiation rather than having it captured in the dice.

If we start with this “average character” notion, there’s a similar mechanic which gives us more levers to pull: a dice pool. The d10 gives us a 90/50/10 split for Easy/Average/Difficult, while 3 dice seeking one success for an easy task, two for an average task, and three for a difficult task gives us 87.5/50/12.5 assuming your target is half of the faces on the die (4+ on a 6 sider, 6+ on a ten sider). This reduces the swing somewhat, but leaves us with the same center. If we assume that a skill up is an extra die, we get a roughly 19% increase for the average and difficult tasks, but only roughly 6% for the easy task. This built-in diminishing return nicely reflects reality, but the core increment says we’re probably working with half as many steps as the 2020 dice mechanic, where each point increase is a 10% increase in probability of success.

So let’s consider attributes as well. Cyberpunk provides some definition of what the attributes mean from 3 to 10, but I’d honestly dispute whether there’s that much range of innate capability in humans. Most physical abilities, be that how fast you can run or how much you can lift or how long you can hold your breath, are all trainable. This is true for mental skills as well…while some are born with better perception or mental acuity, practice and training are significantly more important than inborn differentiation in most cases. Let’s consider a slightly different approach for base attributes. If we incorporate an exploding dice mechanic, this means that the probabilities of success will go up a little bit, as now there’s a chance to gain extra successes when the maximum face of the die shows up. For our average roll, the increase in success probability is around 5%, which comes down to the probability that the player will roll a single success (a failure without explosions), have that success come up maximum (in this case 10 on a d10), and then have that explosion generate a success. Now, imagine that this was a d8 instead of a d10. The probability of rolling an explosion has just gotten higher, since there are fewer faces on the dice. So, if we set the target number of each die to a 50% probability, we can keep the raw chance of success the same while increasing the probability of exploding dice with each incrementing die. Under this system, d12 would be a weak human, d10 an average human, d8 a strong human, d6 an augmented human, and d4 a superhuman, perhaps an AI (mental) or a full-body conversion (physical). Using this system for attributes makes training primary, but directly increases the probability of going beyond what training would allow.

Pulling it all together, we’re going to keep the five difficulty tiers from 2020 and say they require between one and five successes from a pool of dice. The probability of success on a single die will be 50%. This means that the probability of getting two successes (average difficulty) on three dice is also 50%. Therefore, three dice can be an average skill, making two dice rudimentary and one die untrained. If the number of skill ranks goes up to five as well, this keeps the Nearly Impossible task fairly difficult without advantages or smaller dice. That said, these difficulties are bound by human ability…the skill system would need to have difficulties higher than 5, but tasks modeled by those difficulties would be either at the very edge of humanly possible (6) or beyond what is considered humanly possible (7 or 8). Explosions will make nearly impossible things possible, and also give a small chance to succeed at any task no matter how small the dice pool.

So we’re using a dice pool, and having attributes and skills impact it differently. This means that each roll will require the player to track three things: how many dice to roll, what kind of dice to roll, and how many successes are required to succeed at the roll. This means that how many attributes and skills there are, as well as how they’re organized, will need to be designed in such a way as to minimize the mental load.

Attributes

Categories of human capability in role-playing games are often informed by models of learning and cognition which seek to categorize the different ways in which the brain processes information and learns skills. Benjamin Bloom’s theory of Domains of Learning is often reflected in RPGs, dividing skills into cognitive, psychomotor, and affective; in plain English this translates roughly into mental, physical, and social.

Considering our level of granularity, is it necessary or even useful to split up innate attributes more? If we’re talking on the level of innate ability, it’s difficult to argue that there are non-correlated categories beyond these three. So for skill-driving attributes, let’s leave it at three.

Skills

By having three attributes and no more, there seems to be a lot of things left on the table. I’d argue, though, that pretty much every other human ability under the sun can be trained. Constitution, or Cool, depending on the system, is an example of a skill which is often treated as an attribute. The things that fall under Constitution, like pain tolerance, stress tolerance, and endurance, are all trainable. Elements which aren’t, like the strength of your immune system, are mostly tangential to a Cyberpunk game. So to start, there will be a couple things which other games treat as attributes which we’re going to treat as skills, namely Cool and Mobility (the equivalent of Movement Allowance).

So how are we going to build out the skill list? This may be biased on recent experience but I find that Torchbearer has a good quantity of skills at 25, in terms of both searchability and usefulness. One thing which Torchbearer does very well is ensure that every skill has direct applications ingame, and lists out what those applications are. To accomplish the same thing here, it becomes necessary to define what the game “verbs” look like, essentially what we expect characters to spend most of their time doing. Here, the Cyberpunk roles give us some insight. Just from the roles we get fighting, fixing, hacking, healing, reporting, influencing, driving, commanding, administrating, and acquiring. Some of these are extrapolated a bit (command from Cop and acquire from Fixer) but it does cover the basic ground of the Roles.

So going from the start, we take the verbs from the Roles, possibly splitting a couple up (Fighting can have a melee and a ranged component, fixing can be split into repairing and building), and adding the overarching skills, Cool and Mobility. What other overarching components can be modeled this way? Well, if we continue taking pages from Burning Wheel and Torchbearer, both social influence and money are modeled as skills. Money is likely to be its own thing, but let’s throw in a Reputation skill. Beyond this, it’s time to add some things which are missing from our beginning list, and maybe clarify some of the ambiguous ones.

Mental	Physical	Social
Repair Build Bypass Examine Heal Research Cool	Fight Shoot Sneak Drive Mobility Strength	Influence Bluff Lead Acquire Reputation

This is a good start. We see some bias towards mental in terms of skill quantities, but that’s all right. This skill list is fairly short, and there are clearly some of these skills which could drop down in many different ways. There are two different mechanics that are used to bulk up short, high-level skill lists, knowledge mechanics and specializations. We could, theoretically, combine these. Knowledge specializations could model both general knowledge as well as the specific application of the abovementioned skills. Mental specializations could be things like chemistry and computers, while physical specializations could be things like motorcycles or swords, and social specializations could be specific gangs, corporations, or other organizations. The difference between these specializations and those found in other games is that these specializations cover the entire span of skills under the attribute.

There are also a number of these skills which you will start with. One must learn how to drive or how to build something, but everyone knows how to run and lift things, everyone has some degree of endurance, and everyone has a reputation. These can all be pulled out…even though they’re treated like skills, they can be isolated as a bit different.

Other Statistics

There are a few items which don’t quite fit into the standard attribute/skill paradigm but are still important character descriptors. First, Reputation is an edge case. While using your reputation to get what you want fits into the mechanic we’ve built above (attribute determines die type, skill determines die quantity, success/fail is 50/50 on each die), earning reputation is more complicated. First, a ‘global’ reputation is likely very difficult to get. There’s two ways to address this, with successive levels of granularity. If we limit the sphere of Reputation to the campaign city, it gives a more realistic number. That said, one would still need to consider a different Reputation if the characters traveled, so we’re going into the second way anyway, which is having each character track multiple Reputations. This also gives us some stuff to play with, since we’re now officially pulling Reputation out of the skill list and treating it differently. For one, Reputation can now be either positive or negative, which addresses one of the issues with the Reputation system from Cyberpunk 2020. Second, each campaign can now track the Reputations which are important to them, without needing to define them ahead of time.

Another statistic which doesn’t fall into our framework at all involves money. Tracking currency dollar by dollar is a perfect example of when an absence of rules as opposed to a presence of one makes the players’ lives more difficult. Depicting money or wealth as a ‘Resources’ stat both simplifies this and gives guidance on what needs to really be tracked. While a very poor character would need to worry about every last ramen cup, even a moderately wealthy one would be able to buy meals without giving it a second thought. A good Resources mechanic should still do what a currency-as-inventory system does (tell you how much your character can spend and what they can afford) while making other tasks (cost of living, economic modifiers, rent-seeking) easier to model. Considering that, Resources should definitely be modeled as a statistic, and we can figure out how best to do that when going into more depth on gear and possessions.

We have a very basic set of core mechanics: Attributes determine die type, while skills determine how many dice you roll. Things like Reputation and Wealth may be tracked differently, but should broadly have the same set of magnitudes as other skills. Now, there’s still tweaking to be done. Are there more skills that can go under Social? Is the Repair/Build/Bypass trio enough detail, or do they need to be broken out into physical and electronic? Does the Knowledge Specializations mechanic need tiers or other fleshing out? These questions still have to be answered, but as more of the rules are developed, it becomes easier to answer them. For now, these broad strokes mechanics are enough to allow us to build other systems around this basic scaffold, like combat and character creation. Once the system becomes a little more complete, it’s easier to identify the gaps and figure out how to fill them.

---

This is the start to building the Cyberpunk Chimera concept into a real RPG. These brainstorming articles put all the mechanics into place, where they can then be cemented in with more directed writing, prototyping, and eventually playtesting. This is just the first real mechanics brainstorm, there’s still a lot to be done with both more mechanics and refining and rewriting the ones above. Still, every journey of a thousand miles begins with just one step, and every RPG starts with one core mechanic. As this System Hack moves forward, more and more pieces will come together, and hopefully start looking like a playable whole.

Want to check out more Cyberpunk Chimera? We’ve got all the articles tagged and easy to find.