Degree of success, and accumulating it

Study note. Written per ../02-originality-policy.md: functional description in our own words, no quoted text, no reproduced tables or values.

Prompted by: the AGE system (Adventure Game Engine, Green Ronin), specifically its Drama Die — known as the Stunt Die or Dragon Die in other games on the same engine.

Problem

A binary pass/fail wastes most of the information a roll produces. Two characters both "succeed" and the fiction cannot distinguish a scrape from a triumph. A system wants a cheap way to read how well an action went, without adding a second roll or a lookup table.

A second, related problem: some tasks are not resolved by a single action. Repairing a wrecked mech, slicing a network, or building something over weeks is not one roll, but rolling repeatedly with no memory between rolls produces no arc.

Approach observed

The AGE engine rolls three six-sided dice, one visually distinct from the other two. That distinct die does two jobs at once:

• its face value reports the degree of success on a success;
• when any two of the three dice show the same value, it also generates a pool of points that buy discrete special effects.

So one throw yields hit-or-miss, a magnitude, and an occasional bonus economy. The magnitude is read from a die face, independent of how much the roll beat its target by.

Cost

• Requires a visually distinguishable die, which is a component dependency.
• Reads magnitude from a die face rather than from performance, so a marginal success and a crushing one can report identical degree. Magnitude and quality are decoupled by design; that is a deliberate choice, not an oversight, but it is a choice.
• The bonus economy is a second subsystem with its own spend list to learn.

Convergence

Degree of success is strongly convergent across the tabletop tradition and is a genre convention rather than anyone's property. Independent solutions include: banded percentile results, tiered outcome ranges on a single roll, symbol-based dice reporting success and complication separately, and margin arithmetic where the amount by which a target was beaten is itself the magnitude.

Accumulating progress toward a threshold across multiple rolls is likewise convergent — segmented progress trackers and countdown mechanisms appear independently in many modern systems.

Both concepts are therefore safe ground to build on. The safe ground, not the safe expression: our terms, our numbers, our thresholds.

For C.A.T.S.

Take: the principle that one roll should report magnitude, not merely pass/fail, and that magnitude can accumulate toward a threshold to model extended work.

Reject: reading magnitude from a dedicated die. Two constraints already rule it out — one-roll resolution forbids reading a second independent result, and the one-set dice constraint means the two d10s are already committed if 2d10 wins OQ-20.

Diverge: C.A.T.S. reads magnitude from Margin — the amount by which the roll beat its Target Number. This is a materially different mechanic:

The coupling to skill is the whole point for C.A.T.S. The brief's Edge economy is already Margin-derived, so magnitude, Edges, damage, and progress all flow from one number. That is a tighter integration than the source has, and it is forced by our own constraints rather than borrowed from anywhere.

Numbers: none carried across. Our thresholds derive from measured expected Margin — see ../analysis/resolution-comparison.md.