Nargott
Auto-translated
This is especially for those who don't understand why doubling the damage is acceptable with +3 luck, but not with +1 (asked on ICQ).
There is a concept in mathematics called variance (the square of the deviation from the average), which characterizes the stability of a random variable.
1) 50% luck with a 10% chance: D = (1.5-1.05)^2 = 0.45*0.45 = 0.20, increasing the chance to 50% reduces it to 0.06
2) 100% luck with a 10% chance: D = (2-1.1)^2 = 0.9*0.9 = 0.81
3) 100% luck with a 30% chance: D = (2-1.3)^2 = 0.7*0.7 = 0.49
4) 100% luck with a 40% chance: D = (2-1.4)^2 = 0.6*0.6 = 0.36
5) 100% luck with a 50% chance: D = (2-1.5)^2 = 0.5*0.5 = 0.25
The same variances after normalization (relative to the average damage):
1) 18%, decreases to 4% as the chance increases
2) 67%
3) 29%
4) 18%
5) 11%
What we get when taking Elven Luck (which gives 100%): it's a level 3 perk, i.e., in fact, we will have luck from +3 and above. Plus, the motivation to increase it to +5 (with an enhanced bonus).
Without Elven Luck (with 50% luck), the variance is low, and luck plays a consistently stable role.
For comparison, the maximum luck variance in TE (70% luck):
- with a 10% chance (1.7-1.07)^2 = 0.63*0.63 = 0.40, normalized 35%
In early versions of TE and in RPE (80% luck):
- with a 10% chance (1.8-1.08)^2 = 0.72*0.72 = 0.52, normalized 45%
Conclusion: a variance of no more than 40-50% is comfortable for many players.
Another comparison:
- Maximum damage range in G5 (and G5 WGE) = 40-160% (1:4) has a variance = 0.6*0.6 = 0.36 = 36%.
- Typical damage range in G5 = 75-125% (3:5) has a variance = 0.25*0.25 = 0.06 = 6%.
- Typical damage range in G5 WGE = 70-130% (7:13) has a variance 0.3*0.3 = 0.09 = 9%.
All these figures are significantly lower than the crazy variance of 67%. It would only be achieved with a range of 1-10.
I hope I have clearly argued my position regarding the stability of 50% and 100% luck.
P.S. And about negative 50% luck:
1) -1: D = (0.95-0.5)^2 = 0.2 = 22% (after normalization)
2) -2: 0.16 = 20%
3) -3: 0.12 = 17%
4) -4: 0.09 = 14%
5) -5: 0.06 = 11%
That is, it is clear that the variance is initially low everywhere, and there is no need to change anything.
P.P.S. The concept of variance in this review is deliberately simplified for ease of understanding. Instead, the term "square of the maximum deviation" is more correct, which serves as a measure of stability.