My elegant 4d6 system

Roleplaying systems use many different resolution mechanics. My favorite one is this:

Roll four d6 and count the "flops", which is 1s and 2s:

• No flop means you succeed.
• One flop means you succeed at a cost.
• Two or more flops mean you fail.

If you succeed (with or without cost), the highest roll determines how much "progress" you make.

Example 1: You try to attack an orc with your sword. You roll 3, 3, 4, 5. You succeed (no flop) and you make 5 damage (progress in this case).

Example 2: You try to charm the guards. You roll 1, 2, 2, 5. You fail (three flops) and the 5 does not matter.

Example 3: You try to construct a trebuchet before the enemy army attacks, which requires 6 progress. You roll 2, 3, 4, 5. You succeed (one flop) at a cost and make 5 progress. GM decides that the cost is that the enemy army attacks now yet the trebuchet is not finished (5 of 6 progress).

Counting the flops is easy and quick. Watch me:

Why it is best

Using d6s a common choice because they are widely available. The first roleplaying session is usually the first occasion where people encounter other-sided dice. While this introduces an interesting factor, it is also a barrier initially. There is no real downside to avoiding it, so only d6 here.

Rolling more dice simply feels better. So any single-die system, like D&Ds d20, is worse from a tactile point of view. The remaining questions is how many dice? Too many is possible because it is impractical to roll a hundred dice. Seven is roughly the number of things one can recognize at an instant. My choice of 4 is somewhat subjective. The system should work with one die more or less should work as well. (Something to keep in mind: We can use 3 or 5 dice to make a test harder or easier. See below.)

After deciding to use a dice pool, the next question is "counting" vs "adding". Counting the 1s and 2s with four dice is something one can do instantly. Adding four dices is work in comparison. Since a game should not feel like work, counting it is.

Also subjective is the ternary result, the "success at cost" besides "success" or "fail". I have no argument that it this is better and I'm just going by gut feeling here. It does affect the next decision though.

Dice pools and counting successes is a well used mechanism (Exalted, World of Darkness, etc). Counting flops is unusual. The reason is that I want an asymmetric distribution for a natural behavior: With increasing skill, the variance decreases. In other words: A bad cook might deliver an extraordinary meal on rare occasions. A good cook delivers a meal at least good always. It is more critical to avoid mistakes than to achieve perfection once.

I still get the criticism that it feels wrong to count flops instead of successes. However, I believe that is only habit which you can unlearn quickly. As a player, when you roll, you know what you want. You have a plan. If one step in your plan goes wrong, you can improvise a little. If multiple things go wrong, only then you fail. This is what this system models. The dice are not a resource where more is better. Instead the dice represent all the things that could go wrong. If you try to charm the guards, think of four things that could go wrong:

1. You accidentally say something wrong.
2. They are too smart.
3. Someone else accidentally breaks your charm.
4. You picked the wrong guards.

Each one is represented by one die and rolling a flop means it goes wrong. Roll them all and see if your plan executes flawlessly.

Finally, many systems require a two step process: Do you hit? How much damage? Some systems compress it to one roll by declaring damage fixed or you always hit. That simplifies away some interesting results though. One could compensative with narrative skills, but I prefer the let the system do this work. Thanks to all the dice in our pool, we can extract more information from a single roll. After counting the flops, we can look at the other dice and pick the highest one, something between three and six. I call it "progress" in general. In a fight it would be "damage" probably. However, you can apply the same mechanic to negotiations, travel, research, farming, or quests. So a generic term makes more sense.

And this concludes the argument why any other dice system is worse. Some aspects are subjective but a lot is derived by objective logic.

Modifiers

An important point of roleplay is that characters are different. So we need ways to modify chances depending on who does what. There are two ways with this system: The number of dice and what counts as flops.

The "what" is usually the first choice. As an extreme example, Index Card RPG uses a single target number for anything you do in a room. It consists of all the external circumstances. The "who" is sometimes subject to arguments. It consists of everything a character internally brings to the table. A player might remember an additional aspect of their characters background which qualifies for an additional advantage. Looking at the two ways, reinterpreting what counts as flop is easy. On the other hand, changing the number of dice requires a reroll. That means we extend the rules with this:

• Under stressful circumstances, roll 5d6.
• Under relaxed circumstances, roll 3d6.
• If something is hard for a character, 3s are also flops.
• If something is easy for a character, only 1s are flops.

Modifiers do not stack. There is no way to roll 2d6 or 6d6. The effect would be so strong that you might as well not roll at all.

Items cannot have subtle effects. There are only few ways to adapt chances and all are noticeable.

• Make stuff easy or hard (interpret flops differently)
• Reroll one or more dice.
• Add or subtract progress.

There are more modifications with more complex extensions. For example, Savage Worlds has "exploding" and "wild" dice. Both add too much complexity for their effect size though. Not something for basic rules but it could be introduced with special magic items, so the complexity is optional.

The numbers

What are the actual probabilities. Under normal circumstances, here they are:

The default case means you have a 60% chance of success, though 40% of that is at cost. If something is easy for your character, the chance of failure drops from 40% to 13%. If something is hard instead, it rises to 69%. This feels right to me.

Let's consider stressful circumstances: You're on a battle field. Smoke is burning in your lungs and you can barely see anything. The ground is shaking. You must roll five dice to also overcome the additional difficulties.

The success rate for a hard task is halved to 3% and in 81% of the cases you will fail. Even an easy task will now fail with a 20% chance.

Finally, relaxed circumstances: Having fun in the inn. Happy music playing. No pressure. Only three dice.

For an easy task, with a 57% chance it will be a plain success and only 7% for a complete fail. A hard task still fails half the time but that is better than the 69% under normal circumstances.

The environment determines the circumstances. The character sheet determines the task difficulty. How does it balance out? The normal-normal case is 20/40/40%. A stressful-easy case is 40/40/20% so success is more likely. A relaxed-hard case is 13/38/50% so success is less likely. They do not balance out. In both cases, the character effect (hard or easy) is stronger than the environment effect (stressful or relaxed). Is this a good or a bad thing? That depends on the setting. How can characters make something easy (skills, gear, enchantments) or hard (curses, disabilities, conditions)?

Weaknesses

This system is not suitable for finely nuanced simulations. You cannot improve the odds by 3% somehow. There is nothing to model the difference between a bow and a crossbow, for example. On the other hand, the real life differences cannot be easily summarized as "+1 something" either. Effect combos, like stacking armor bonuses, are rare because the impact is maxed out quickly. I don't mind because there are other options: A magic fire sword could set the opponent on fire instead of "+3 fire damage".

The "success at cost" option is tough for GMs. While the "fail forward" style is helpful to craft interesting stories, it requires highly creative improvisation. PbtA games, which made this popular, also come with collaborative world building. It enables the GM to offload the improvisation to the players. For a GMs who depends on this and players who dislike it, this system does not work. However, changing that design choice to binary resolution requires changes to other aspects of the system.

This is not a full system. Well, you can use it like that. It is more complex than Lasers&Feelings, for example. However, the system I use in practice is larger. For example, if you fail you get a token. Tokens you can redeem to reroll a die and other things. This article is long enough already though.

What I like most about this is its elegance. The rules are brief. You could fit them on an index card. Yet, they implement a lot of mechanics: ternary results, two kind of modifiers, and progress.

I initially shared this idea on Reddit.