While creating Inept Sorcerers, I created a dice mechanic that I think I could generalize into its own system. Here's the outline.

  1. The player states an objective, like "my ninja is going to take out that sentry without raising the alarm."
  2. The player determines how this action maps to a Gambit - like a Dungeon World Move - which has a cost and some optional costs. For example, there can be a generic Gambit called "Defeat" which removes a target from play, and killing a sentry sounds like that Gambit.
  3. The player determines his Control, based on a skill on his sheet, and adds modifiers to it based on equipment, circumstances, or whatever.
  4. The total cost to execute the Gambit is determined. There might be some global modifiers - for example, "Quietly" might add +5 to the cost.
  5. The player starts accumulating Risk, by rolling d4. If it's not enough, he can continue to roll, going up a die size each time and stopping once he's rolled d12. The total of all dice rolled is the Risk budget.
  6. The player determines outcome.

If the Risk budget is less than the total cost, and he can keep rolling, do so. If he can't keep rolling, he can try to redo the action - like dropping "Quietly", or turning "Defeat" into the cheaper but less effective "Harm". But once you've introduced Risk, you can't "unroll" a die.

If the Risk budget exactly matches the total cost, the action succeeds as declared.

If the Risk budget exceeds the total cost, that excess must be accounted for. Subtract Control from excess Risk. If that's still not enough, an undesirable modifier comes into play and subtracts its cost from the remaining excess. This can also negate a positive modifier.

Here's a few examples of how this might play out.

We'll create a Gambit called "Defeat". The effect is to remove a target from play decisively - killing or permanently immobilizing a living target, for example. Defeat has a cost of 10 + the target's level. A typical sentry or other guard is a level 1 mook, while the final boss of a scenario might be level 10.

We'll create a modifier called "Quietly", which brings the cost to (10 + 1 + 5), or 16.

We'll give our ninja an "Assassination" skill of 5 - about average, on a 1-10 scale.

First scenario: the ninja rolls d4 (2), d6 (5), and d8 (8). This is not enough Risk to do the move, so he must roll a d10 or give up the action.

Second scenario: the ninja rolls d4 (2), d6 (5), d8 (7), and d10 (2). This is exactly 16, so the action proceeds as planned.

Third scenario: the ninja rolls d4 (2), d6 (5), d8 (8), and d10 (2). This is 17, one more than he needs. 1 Risk less 5 Control is nothing, so there's no further complications and he succeeds.

Fourth scenario: the ninja rolls d4 (4), d6 (6), d8 (5), d10 (8). This is 23, 7 more than he needs. His Control of 5 isn't enough to make that go away, and 2 points of Risk are left. The GM can offer the ninja a choice: lose the "Quietly" modifier (reducing Risk by 5, but making a noise), or take some other sort of drawback.

It feels a little more complicated than I'd like. I'm hoping that negotiating an action would be interesting enough to make that worthwhile. There's plenty of room for PC differentiation by giving them different Gambits - for example, a ninja might have an "Assassinate" Gambit of his own that has a lower cost than "Defeat", but only works from stealth, letting him do sentry-removal and backstabs more easily than the party fighter.

The system has a small box it lives in, in terms of how high target numbers should be. Things feel like they logically live on a 1-10 scale, including a hypothetical "character level", skills, and other things. There's not a lot of room for nuanced modifiers.

There's several dimensions for game state: things that influence Control (like skills or bonuses), Gambit and modifier costs, PC or class-specific Gambits, and so on.

If anyone has any thoughts or feedback, let me know.


Related Posts


Grand Adventure is the third and probably last try at a small light-hearted system using the baroque dice mechanic. It'll probably use the reward-token system f...