Based on a suggestion from another Fading Suns GM, my preferred method of using the Victory Point system was with two d20s instead of one. The gist of the system (apart from accenting and wyrd mechanics explained at the link) was:

- Roll 2d20, keep the highest die that’s still a success.
- The roll is only a critical success if one die roll the target number and the other is also successful.
- The roll is only a critical fumble if one die rolls a 20 and the other is also a failure.

Doing this changes the success rate pretty drastically:

Target |
1d20 Success |
2d20 Success |

1 | 5.0% | 9.8% |

2 | 10.0% | 19.0% |

3 | 15.0% | 27.8% |

4 | 20.0% | 36.0% |

5 | 25.0% | 43.8% |

6 | 30.0% | 51.0% |

7 | 35.0% | 57.8% |

8 | 40.0% | 64.0% |

9 | 45.0% | 69.8% |

10 | 50.0% | 75.0% |

11 | 55.0% | 79.8% |

12 | 60.0% | 84.0% |

13 | 65.0% | 87.8% |

14 | 70.0% | 91.0% |

15 | 75.0% | 93.8% |

16 | 80.0% | 96.0% |

17 | 85.0% | 97.8% |

18 | 90.0% | 99.0% |

Just looking at the chance of success, it’s interesting how much it suddenly curves to look much more like a White Wolf-style dice pool mechanic than a percentile mechanic. Importantly, in my mind, this means that it’s not as drastically necessary for players to try to absolutely max out their skills to regularly succeed: in practice, a trait total of 10 is supposed to be pretty good for a starting character, and now that character has better than a 50/50 shot on rolls. It’s immersion-breaking in the extreme for the system to pretend that you have a good trait and then fail on it half the times it’s important, at least in my opinion.

Additionally, this method puts a curve on fumbles and criticals. In 1d20, you have a 5% chance of a crit and a 5% chance of a fumble, no matter what. In 2d20, the chance of crit goes from 0.3% at TN 1 to 9.3% at TN 19, while the chance of fumble does exactly the opposite. Effectively, the higher your TN, the bigger your chance to crit and the smaller your chance to fumble, which seems more logical.

The other interesting thing is what it does to expected success totals:

Target |
1d20 Avg. VP |
2d20 Avg. VP |

1 | 0.0 | 0.0 |

2 | 0.0 | 0.0 |

3 | 0.3 | 0.4 |

4 | 0.5 | 0.5 |

5 | 0.6 | 0.6 |

6 | 0.8 | 0.9 |

7 | 1.0 | 1.1 |

8 | 1.1 | 1.2 |

9 | 1.3 | 1.5 |

10 | 1.5 | 1.7 |

11 | 1.6 | 1.9 |

12 | 1.8 | 2.1 |

13 | 2.0 | 2.4 |

14 | 2.1 | 2.6 |

15 | 2.3 | 2.8 |

16 | 2.5 | 3.1 |

17 | 2.6 | 3.3 |

18 | 2.8 | 3.7 |

The chart above is the average number of victory points for a successful roll (not counting criticals). The numbers don’t look terribly different, save that the 2d20 is slightly higher. In practice, this is because, with 1d20, success VPs are completely flat: if you succeed on 1-10, you a successful roll has a 10% chance for each result. In other words, any time you succeed, you will roll less than half your best result half the time. Conversely, with 2d20, you have at least a 75% chance of rolling over the halfway mark (because if both dice are under the target number, you choose the larger result).

Old school game design looks at the 1d20 and declares it adequate: the higher your score, the higher the chance of success and the result of success. But looking at the raw numbers doesn’t cover the feel at the table, where excessive swinginess results in player disappointment. Over multiple rolls, a flat die result evens out, giving an advantage to the better character, but how often do characters make multiple rolls on the same skill outside of combat? In practice, a player may get once chance to shine with a given non-combat skill per session, and, with a flat die, the result of the roll can feel almost completely disconnected from the score. Using 2d20 to curve the result creates a situation where, even on a single roll, a higher score feels meaningful.