[Vote] (5-20) Withdraw From Melee Changes

[Stalker0]

Voting Instructions
Players, please cast your votes in the poll above. Vote "Yea" for every proposal you'd be okay with if it were implemented. Vote "Nay" if you'd be okay if these proposals weren't implemented. You can vote for any number of options.

All votes are public. If you wish, you can discuss your choice(s) in the thread below. You can change your vote as many times as you want until the poll closes.

VP Congress: Session 5, Proposal 20
Proposer: @Stalker

this is part 1 of a two part proposal. Ultimately either can be implemented independently but they can be thought of as the same concept.

Part 2 can be found here: https://forums.civfanatics.com/thre...ee-damage-reduction-for-certain-units.684425/

Proposal: Change the "Withdraw from Melee" mechanic to the following.

• The %chance is now 100% on the first melee attack against the turn that round. It is now 0% on all subsequent melee attacks.
• The %chance is no longer affected by speed.
• The unit will always move in the following manner. (note the picture below)
  • If Tile 1 (the tile right across from it) is open, retreat to Tile 1.
  • If Tile 2 (the hex counterclockwise from 1) is open, retreat to Tile 2.
  • If Tile 3 (the hex clockwise from 3) is open, retreat to Tile 3.
  • If Tiles 1,2,3 are all occupied or unavailable, the withdraw chance lowers to 0%.

Rationale: One of the major problems with this mechanic is its randomness. In a game that has very little randomness in combat, ESPECIALLY when it comes to positioning and movement, the randomness of this mechanic creates too many unacceptable scenarios. Sometimes the unit withdraws when you rather it hold, holds when you rather it withdraws. also its wirthdraw location can be a bit arbitrary at times.

Lastly the impact of speed is poorly understood and not documented in the tooltips, so its good to remove that as well.

This version creates a definitive withdraw, effectively immunity to damage from the first melee hit (assuming you have room to move). but to prevent abuse and silly scenarios where a unit can literally move across the map due to withdraws, all further withdraws are removed. Its once per turn period.

---

VP Congress: Session 5, Counterproposal 20a
Proposer: @Tekamthi
Discussion Thread: (5-25a) Counterproposal: limit withdraw from melee to 1 withdraw per turn, prioritize plot with least adjacent enemies

Background/context

The existing RNG basis of the withdraw from melee has a solid base of critics, however it has been around, relatively unchanged, for a ~decade. Less-is-more when it comes to changing long-standing, incumbent features. Furthermore, though much improved over vanilla, the AI remains somewhat simplistic in its analysis of the battlefield, and the uncertainty regarding whether a unit will withdraw on any given attack, or not, brings parity to AI & human decision making -- both are uncertain about a withdrawal in the status quo, vs. proposed changes this round that give human certainty about withdrawals, while AI remains effectively blind to them (without further AI work, which remains only vaguely touched on through all discussion). Finally, RNG has always been a part of civ series combat, and serves as an abstraction of the uncertainty of battle, the unpredictable factors of environment and human fallibility etc. RNG need not be eliminated completely to improve the withdraw from melee feature.

Proposal

• leave existing withdrawal RNG mechanism as-is as far as determining if a withdrawal occurs (for withdraw from melee only -- heavy charge and withering fire not included here), with following restriction:
• add a WithdrawalsPerTurn field to UnitPromotions table, and set all existing promotions that confer a withdraw-from-melee chance to WithdrawalsPerTurn = 1; restrict withdraw-from-melee function so that a unit can only withdraw x times per turn, corresponding to the sum of its WithdrawalsPerTurn value (in most cases this will be max 1 withdrawal per turn, with room for modding)
• referencing same graphic as other 5-25 proposals re: direction (the star-of-david shaped one)
  • 1) evaluate all plots 1, 2, and 3 (not just default to 1 first), and withdraw to clear plot with fewest adjacent enemies;
  • 2) If two clear plots have same low # adjacent enemies, evaluate adjacent friendly units, and withdraw to plot with most adjacent friendlies
  • 3) if two clear plots have same preferred number of adjacent enemies and adjacent friendlies, choose withdrawal plot randomly

---

VP Congress: Session 5, Counterproposal 20b
Proposer: @axatin
Discussion Thread: (5-25b) Counterproposal: Withdraw from Melee - Changed Withdraw Direction Logic

Proposal:
(same as in the original proposal)

Change the "Withdraw from Melee" mechanic to the following.

• The %chance is now 100% on the first melee attack against the turn that round. It is now 0% on all subsequent melee attacks.
• The %chance is no longer affected by speed.

(different to the original proposal)

• When a unit withdraws, the tile to which it moves is determined as follows:

1. If the tile behind the unit (Tile 1) is free, the unit always withdraws to it.
2. If Tile 1 is blocked, the unit retreats to either Tile 2 or Tile 3. If both of them are free, the tile with the lower number of adjacent enemy units is preferred.
3. If Tile 2 and Tile 3 are both free and have the same number of adjacent enemy units, a random decision is made.
4. If Tiles 1,2,3 are all occupied or unavailable, the withdraw chance lowers to 0%.

Rationale: As in the original proposal, the unit always tries to withdraw to the tile opposite to the attacker. If that tile is blocked, the unit withdraws diagonally while trying to avoid enemy units as well as possible. If both of the possible tiles in the diagonal direction have the same number of adjacent enemy units, a random decision is made. A random decision is better than a fixed pattern because a fixed pattern could be exploited by human players.

---

VP Congress: Session 5, Counterproposal 20c
Proposer: @Rekk
Discussion Thread: (5-25c) Withdraw Logic: 100% + enemy preference

Proposal:
(same as in the original proposal)

Change the "Withdraw from Melee" mechanic to the following.

• The %chance is now 100% on the first melee attack against the turn that round. It is now 0% on all subsequent melee attacks.
• The %chance is no longer affected by speed.

(different to the original proposal)

1. The unit retreats to plots 1, 2, or 3. The unoccupied, legal plot with the lowest number of adjacent enemy units is preferred.
2. If any number tiles have the lowest number of plots, withdraw to plot with most adjacent friendly units.
3. If there is still a tie, a random decision is made.

- If Tiles 1,2,3 are all occupied or unavailable, the withdraw chance lowers to 0%.

Differs from 5-25a:
- 100% fallback, once per turn
Differs from 5-25b:
- withdrawing straight back is not a priority
- includes adjacent friendly units

Rationale: The rules are still straight forward, and this rule set is the most favorable for the retreating unit. Random final fallback is less exploitable than an "always retreats to the left just because".

 

[Tekamthi]

I really like this one EXCEPT for the 1, 2, 3 logic. Its too arbitrary, when applied to various maps and different situations, it will still feel RNG even though its not. I'd prefer that we evaluate plots 1, 2, 3 via Map.PlotDistance(plotx, ploty, capitalx, capitaly) and thus prioritize plot with smallest distance to capital -- if two are even distance from cap then we can fallback to the numerical priority.

also consider evaluating plots for adjacency to enemies -- say 1 and 2 are both adjacent to enemy unit, you'd want a withdraw to prioritize plot 3, no? I know you expressed desire to keep this mechanism super simple, but imo the 1,2,3 arbitrary directions is too simple.

 

[axatin]

How is it currently determined in which direction a unit withdraws?

 

[azum4roll]

How is it currently determined in which direction a unit withdraws?

Prioritize opposite tile, then randomly pick left/right. I think it's better than the proposed order.

 

[axatin]

Suggestion how to determine withdraw direction:

1. If the tile behind the unit is free, the unit always withdraws to it. (It would be strange if a unit that's attacked frontally retreated diagonally.)
2. If that tile is blocked, the unit retreats to one of the two tiles next to it. The tile with the fewer amount of adjacent enemy units is preferred. (Withdrawing means trying to escape enemy units as well as possible.)
3. If both tiles have the same amount of adjacent enemy units, a random decision is made. (A fixed order like in the OP could be exploitable for human players.)

 

[axatin]

I sponsor this.