[UMC] More (not full) Deterministic Luck

[SolaEvoli]

I've read following two threads:

• http://forums.wesnoth.org/viewtopic.php?t=40729
  • Gentle approach to removing random
    Author: @optimother
• http://forums.wesnoth.org/viewtopic.php?t=40947
  • Biased Random Damage Mod
    Author: @Crow_T

EDIT1: Both topics resulted in one officially downloadable addon. Neither is free of issues which I am going to explain:

1st has the problem of being too deterministic. You can »save up« luck on a unit or %-evasion class for the more important attacks such as a leader or last-hit.
2nd has the problem of poisoning or slowing not being defined. it uses the default system.


I've also read the topic in which developers explain why they don't want a fully-deterministic game. The main argument against it is save-loading.
(( http://forums.wesnoth.org/viewtopic.php?t=21317 ))

• My answer: I would be happier with random chances ((+- 10% for the strikes)) and the system explained below.
  It is easier to remember the occasions when 3 units after each other, each elvish champions/heroes with 4-5 hits miss, just to die to the counter-attacks which always hit ((especially from orcish warriors)).
  
  I don't mind 1 random unit per turn having luck and another having no luck (per side). But 2 enemies which are lucky and 3 allies which are not are an experience users should be saved from.
  Especially when the individuals represent a platoon, occupying a whole hex at the size of a village.

I want to share a third approach:

• • Count together the % of hit chances.
  • Let the unit hit once per full 100%
  • Let the unit miss 1 hit for each (full) missing 100%
• • There may be 10..90% left, neither full for a hit or full for a miss.
  • Each target starts with 50% luck.
  • If 10 damage is inflicted to a 50 hp unit with 40% probability, it's 4 damage or 12.5% of the unit's hit-points (the luck that is added or subtracted from the bias).
• • If luck is below 0% it always hits, if it is above 100%, it always misses.
  • If it is between 0% and 100% (50% is default or after the unit was out of combat) it remains completely random.
  • But if the strike misses, the target's luck will decrease by 12.5% (after the check for miss/hit), else it will increase by that amount.
  • Whenever a unit's health is restored beyond it's maximum (or gradually after not fighting for the last 2 turns), the luck may restore itself back to 50%.
  • 2 turns (( without fighting! )) should be enough to catch or lose track of a unit if not finish it or being stopped by enemy interceptors..

The differences to the 1st other Thread I mentioned at the beginning are

• that there is still a randomness involved (( maximum 1 streak in 2 turns )).
• That the luck is placed on the target, not the attacker.
• It is 1 single value, not 1 per evasion % of different terrain types.

The differences to the 2nd other thread I mentioned at the beginning are:

• No problem to determinate wether poison or slow should take effect.
• No problem that rounding down values after resistances creates a bigger difference because of a lower resolution of the possible final damage-values table.

Additionally the attacker can have a separated luck-value for when he will hit -- the defender's luck will determinate if, the attacker's luck when a strike hits.

• Simplified example:
  0=miss
  1=hit
  0011: worst case; luck++ by 50%
  0101: second worst case; luck++ by 25%
  0110: average case; luck== unchanged.
  1001: another average case; luck== unchanged
  1010: second best case; luck-- by 25%
  1100: best case; luck-- by 50%

This requires more thoughts.

If somebody thinks a player could manage to keep track of all these values to his own advantage, please suggest a better solution for the WHEN-issue.