Changes needed to Bombardment Engagement Rules

[cracker]

I have posted the summary of this position paper here as the initial posting in this thread and invite you all to provide comments and any supporting examples that you may have:

The bombardment engagement system in CIV3 needs to be overhauled to include four significant changes that combine with the option of lethal bombardment to make the units more realistic factors in winning and enjoyable game play.

The software that resolves the outcomes of bombardment engagements needs to have four significant changes implemented:

1) Target selection needs to include ALL potential targets in a target square including, non-combat units, artillery units, air units, naval units, terrain improvements, city improvements, terrain not in cities, and production and food bins in cities. Including in recognizing all potential targets, should be a modification factor that recognizes that in large numbers of targets it is easier to score a hit than under circumstances when fewer targets are available in the same square.
2) Rate-of-Fire coding needs to be updated to randomly distribute the shots or bombs across the potential target list instead of simulating that a bombardment unit would be able to score multiple hits on the same unit in the same turn even when many other potential targets are present. This change taken in tandem with the lethality flag helps to achieve a more balanced and accurate implementation of the bombardment engagements.
3) Bombardment needs to be a special class of engagement that does not use the ground defense factor for units in a given square as the rule for assessing which unit gets engaged or hit first. Ground defense factor has little or nothing to do with the probability that a unit will sustain damage from hits by stand off weapons when the probability of a hit is randomly distributed across potential targets.
4) The experience level of the bombard unit should be taken into account in determining the success of the bombardment mission.

If properly considered and implemented, these four changes to bombardment engagement rules will correct current imbalances in the implementation of bombardment units in CIV3 and facilitate a more accurate as well as more flexible implementation of strategies that should include bombardment weapons.

There are other, unit-level, changes that may need to be implemented with individual features of individual units, but these four changes will address the systemic problems with the bombardment engagement system. These changes will also facilitate considerable long term flexibility if implemented in a manner that allows some or all of the factors to be adjusted through codes, flags and variables that can be set in the CIV3 editor to effect standard units as well as any user defined units that may be added at the individual user or scenario developer levels.

Each of the follow postings includes more detail and support for the proposed changes as listed above.

 

[cracker]

First, the code that manages a bombardment engagement should be fixed to be more than just a modified version of the one-on-one Firaxis slugfest that is such a fundamental part of other types of unit engagements. There is nothing wrong with the one on one slugfest for units of a common mode, but this one-on-one slugfest is not a valid approach to bombardment engagements and leads directly to a number of inaccuracies and dysfunctional elements in how the units are currently implemented.

When a bombardment unit fires at a game square it is not technically firing at any given unit in the square. Commanders do not call up their artillery units on the phone and say “OK, Johnny, I want you to take your howitzer and shoot the 5th infantryman from the left. Yea, the one with the yellow patch on his shoulder.” Artillery also does not target only the strongest unit first and ignore weaker units. (At least not until precision strike munitions become more prevalent.)

When you select a bombardment engagement WITH A TERRAIN SQUARE that happens to contain enemy units, then the engagement could hit any of the units in the square, or any of the civilian improvements in the square, or the terrain itself. The probability of scoring a hit is directly proportional to the number of targets in the square (intended targets or otherwise).

This issue is significant in implementation because the list of potential damaged items in a square will include every unit and every terrain feature in the square. Artillery pieces, workers, settlers, great leaders, combat units, terrain improvements, even the terrain itself. Firing artillery at terrain squares within your own territory should carry the risk of hitting the improvements and destroying the terrain even if you are only really aiming at the military units.

The probability of hitting any item on the potential bombardment damage list should be equal (or skewed uniformly into classes such as improvements, non combat units, and combat units) and then within those classes the hits should be resolved one-by-one randomly. The distribution of how the hits are allocated within classes could be altered with the discovery of a technology like “precision strike”.

A key point of bombardment potentially damaging all the targets within the targeted square should relate to the number of potential targets and the probability of getting a hit. Instead of raising the defensive values of improvements and civilians to make it harder to get a hit and destroy them by bombardment, the defensive implementation should skew the probability of a hit based on the number of potential targets in the square. As the number of potential targets goes up, then the probability of damaging, destroying or killing any particular target will go down (the safety in numbers effect), but the overall probability of scoring a hit and scoring at least some damage will go up. As units and improvements in a square are damaged and destroyed, then the probability of scoring subsequent hits should go down as there are fewer and fewer targets to hit.

 

[cracker]

A second problem in the current implementation relates to how Rate-of-Fire is resolved.

When a single bombardment unit engages the contents of a single piece of terrain, then the probability that the shots fired or bombs dropped by that unit will all hit the same unit or improvement twice or three times is somewhat related to the number of potential damage targets. It would be possible to wound a foot soldier and damage a road with two different bombs from the same bombing mission. The rate of fire command currently sets the number of potential hits that a bombardment unit can get in a single turn or single mission. If rate-of-fire is three, then the bombardment unit can get three hits if each hit resolves favorably when comparing the defense value factor for the target to the bombard strength of the attacker.

The problem with the current rate-of-fire implementation is that all the hits from a single round of attacks are resolved against a single target. If there are 60 potential targets in a square under bombardment by a unit with a rate of fire of three, and the first hit kills a target, then the other two bombs never do any damage.

A good example of the impact of this implementation comes when bombarding a massive stack of units, which should take a lot of damage from a bombard unit with a high rate of fire. One target in the square will absorb all the hit damage and effectively deflect and waste any excess hit power.

A real example from the V1.21 test scenarios included a stack of 24 barbarian conscript (2 hit point) horsemen engaged in jungle by bombers with bombard of 8 and rate of fire of 3. With lethal bombardment engaged and the current engagement rules, each bomber has a 99.75% chance of killing one horseman in every turn. To simulate the rate of fire engagement rules resolved the bombing hits individually, each of 12 bombers was defined with a rate of fire of 1 each to have the same total of 12 potential hits in the horde stack. In the first bombardment turn, there were 11 out of 12 hits scored on the 24 horsemen with no kills. In fact there would be only a 1 in 61 replays chance of killing just one of the horsemen. Reduced fatality rate is due to the more realistic implementation of bombing that eliminates the ability focus all the bombardment hits in a single attack onto a single unit in the attacking stack. The “strength in numbers” effect spreads out the impact of the bombardment in a random distribution across the potential targets.

 

[cracker]

A third problem with current bombardment implementation is that it recycles and improperly implements the “strongest defender” engagement sequencing rules from the one-on-one slugfest rules of individual land combat. This probably began as a simple recycling of the code sequence for land combat to form the basis for all bombardment combat.

What makes this improper is the defensive umbrella effect that should not be applied to bombardment defense when it is appropriate to land unit defense. A very strong land defender can come forward to meet the attack of a land unit, but a that same defender cannot run around magically and catch all the incoming bombardment rounds to keep them from falling on the heads of the other potential targets in the square.

The effect of this mistake in implementation only becomes apparent in the extremes which come out of the V1.21 test scenarios. In the test scenarios, an aggressive AI civ was selected to allow them to declare war on our test civilization, and then when the AI civ attacked toward a city with one of their spearman and an archer, I used the CIV 3 multi-tool editor to change the spearman into a mechanized infantry unit and the archer into a knight and then added 5 more knights and 1 more mech infantry to the attack stack. The attack stack was 4 tiles away from our city which contained 6 of the standard Firaxis strength WWII type bombers with bombard strength of 8 and rate of fire of 3, plus two fighters for recon, and three pikemen defenders. I then bombarded the advancing stack of units (which the AI obligingly continued to keep together and use in the attack).

The defensive strength of the advancing mechanized infantry unit was 18 modified up to 22.5 to account for the jungle and forest they were advancing through. The defensive strength of the knights would have been 3 modified up to 3.75 for the terrain. In the first turn, the 6 bombers dropped a total of 18 bombs on the advancing stack and scored only 1 hit (this was unlucky compared to the expected 4 hits statistical prediction). At the end of the turn the stack advanced and the units inside were hit by 6 more bomber sorties with 18 more bombs scoring 3 hits that reduced the strength of each mech infantry unit down to two hits each. In the mech infantry units withdrew in the next turn but the 6 knights hurled themselves against our city and took it after sustaining 3 casualties.

Since all 6 of the knights and their horses could not possibly have taken refuge inside the mech infantry units, some of the bombard hits should have fallen on the knights to break up their attack and turn them back. If the bombardment rules had been distributed across all the attacking units as they were stacked in the square, then we should have had 8 hits in the knights and 1 damaging hit in the mechanized units. At least two or three of the knights would have withdrawn at that point. The next turn we would have again dropped 18 bombs on the advancing stack but with only 5 or 6 advancing targets the hits would be more concentrated and in this turn 2 hits would land on the 2 mechanized units and 8 hits would be spread across the 4 knights. At most, 1 of the knights and one of the mechanized units would be left to attack our city defended by 3 defenders and we should be able to survive the frontal assault to further bomb the attacking units again in the next turn resulting in their withdrawal or destruction.

One of the primary missions of defensive bombard is to break up and disrupt advancing attack forces. With the implementation that relies on a recycled version of the Firaxis slugfest software code, the bombardment gets absorbed by the strong defensive units and has no impact on the shielded attacking units. Whenever the defensive strength of the shielding units is more than 2x the defensive strength of the attacking units, they continue to be the most powerful defender until they have sustained substantial bombard damage and as a result no damage befalls the attack force. In a land force engagement this interpretation would be a valid implementation, but in bombardment it forces us to virtually destroy the strong shielding defenders to strip them off the attack stack and gain access to disrupting the main attacking elements of the stack.

The “strong defender first” rule should not be used to resolve bombardment combat because it effectively simulates one of two or three possible interpretations of combat rules that would not be valid for damage falling on the defenders from incoming shells or bombs.

 

[cracker]

A fourth failing in the current implementation of bombardment relates to the failure to integrate the experience level of the unit somewhere into the combat success system. Every other combat unit has its level of experience used in attack or defense and can sustain promotions and rewards for its success in combat. These factors are built in for air units and bombard units but are omitted from the combat resolution system. When lethal bombardment is not engaged, then all of the artillery units can never actually achieve the equivalent of a combat victory. The bombardment air units can only achieve a victory in the rare circumstances when they engage and defeat another air unit (i.e. when a bomber destroys a fighter on a combat mission.)

The bombardment engagement software should recognize the experience level of the bombardment unit and use this factor to modify the success rate in the bombardment missions. Elite units should have a higher accuracy and hit success rate than veteran units and veteran units should have a higher accuracy than regular units. When an artillery unit is captured by enemy ground forces, the experience level should revert to the lowest combat success level of “conscript” since the captured weapon system would now be manned by the new inexperienced crews from the enemy. The accuracy modifiers for conscript, regular, and veteran bombardment units should be 50%, 63% and 79% when compared to Elite units of the same caliber and this formula uses a simple calculation based on the hit points of the experience levels which can be easily implement in the code.

 

[cracker]

Our other discussion thread on Artillery Bombardment got a little off track because the we lost focus in the quagmire of discussing how worthless Artillery was during WWI at dislodging troops from fortified positions.

This set of historical examples was valid in its assessment that artillery never really totally destroyed any enemy units but it failed to recognize the conditions under which the lack of destruction was valid. The examples also failed to recognize bombardment as a general class of “stand-off” attack strategies within CIV3 and other strategy games.

Some general observations that need to be set forward to better understand why the current CIV3 implementation of bombardment desperately needs to be updated include:

1) Real world examples of bombardment are used to target areas (or locations) and rely on the statistical probability that the barrage will be effective against assets located in those areas. It is not true that artillery does not kill units. It is, however, very true, that the chances of artillery killing units that are dispersed and dug into defensive terrain becomes very reduced.
2) The concept of a unit not being destroyed by bombardment in game play, just because “units” were not often destroyed in the real examples of WWI, is also an invalid argument because many of the “units” in WWI trench warfare took very high casualty rates due to heavy weapons fire and the units remained operational in the sense of still being a unit only because individual combat reinforcements and replacements were brought forward to replace the permanent casualties. The only way to simulate these replacement and permanent casualty effects in game play is to have units be destroyed in battle so the human and AI players must provide reinforcement by new units if they effectively want to retain full combat effectiveness.
3) Bombardment can be an effective set of tools even without firing a shot because its presence in the strategy matrix causes enemy troops to be dispersed to counter the potential effectiveness of artillery and aerial bombardment. Concentrations of weapons and troops increase the probability that artillery or aerial bombardment will be effective. The current CIV3 “units stacking phenomenon” is a primary example of where functional simulated artillery is necessary. Dropping bombs on one infantry man fortified in a grid square of forest would be almost impossible to score damage on that infantry man. If there were 100 infantry men in the same terrain then the probability that someone would get hit and killed by the bombard should go up significantly.

Recognizing the real world examples and striving to find a way to have bombardment play a balanced role in successful game play strategies has lead me to test a number of engagement scenarios between air and artillery units and their corresponding ground units, and the data from these engagement tests only serves to reinforce that change is necessary and required.

The changes that are needed go beyond just enabling the Lethal Bombardment capabilities of planes, ships, and artillery pieces but also require so code changes to fix the rules of bombardment engagement to produce a more functional result. In some ways, lethal bombardment added to an unbalanced implementation of bombardment only serves to make the situation a bit more unrealistic.

 

[cracker]

Another suggestion meant to mitigate some of the negative screaming by the WWI artillery buffs would be to add a coin toss to the death determination for a unit whenever lethal bombardment would be the expected outcome.

This option should be controlled by a flag or a percent outcome variable in the editor and apply to land units only.

If engaged, the flag would cause a random result to determine whether a unit should be destroyed or randomly displaced from its current location into an adjacent terrain square that is further away from the bombarding unit position AND further away from nearby cities of nationalities that the unit under bombardment is at war with.

The displacemnt or death option would simulate the disruptive effect of the artillery bombardment and make the effected units potentially more vulnerable to counterattack by other ground units.

Implementing this choice as a percentage slider effect would let users choose whether lethal bombardement would cause 100% death or 100% displacement in cases when lethal bombardment is engaged. (or any ratio of expected outcomes in between those two extremes.)

Lethal bombardment would then be engaged as the default option in the as shipped version of CIV3 with the option to either turn it off for selected units or adjust the coin toss outcome slider bar to effect the percentage lethality versu the percentage of disruptive displacement.

It would be fairly important to implement the displacement routine to generate displacemnt of the units randomly around the square of origin without giving any attacking units a strategic advantage by dispalcing them closer to a valuable attack objective. The intent here is to disrupt attacking or defending units without giving them "a free move" so to speak.

best regards to all,

..... cracker

 

[Jeff Yu]

1) Target selection needs to include ALL potential targets in a target square including, non-combat units, artillery units, air units, naval units, terrain improvements, city improvements, terrain not in cities, and production and food bins in cities. Including in recognizing all potential targets, should be a modification factor that recognizes that in large numbers of targets it is easier to score a hit than under circumstances when fewer targets are available in the same square.

I agree

2) Rate-of-Fire coding needs to be updated to randomly distribute the shots or bombs across the potential target list instead of simulating that a bombardment unit would be able to score multiple hits on the same unit in the same turn even when many other potential targets are present. This change taken in tandem with the lethality flag helps to achieve a more balanced and accurate implementation of the bombardment engagements.

Another great suggestion. I liked the way it was in Alpha Centauri, where artillery had a ROF of 1, but fired at every unit in the stack. Makes sense, since a stack clustered together has a greater chance of being hit since they're so close to each other. For that matter, I also liked how AC artillery kept units from healing. Perhaps too powerful when considering that a single artillery piece could keep an entire 100+ unit stack from healing, though. I think arty should be able to damage a number of units up to the ROF.

3) Bombardment needs to be a special class of engagement that does not use the ground defense factor for units in a given square as the rule for assessing which unit gets engaged or hit first. Ground defense factor has little or nothing to do with the probability that a unit will sustain damage from hits by stand off weapons when the probability of a hit is randomly distributed across potential targets.

I don't think that bombardment needs a special defense. It makes sense enough to me. The more armored you are, the less damage you take. Being actually hit is very unlikely, and casualties will most liikely be from explosive shrapnel. Even a swordsman with a shield will take less damage than an unarmored warrior. (Actually, a knight's armor probably gives more protection against bombardment than modern infantry's flak jackets, I would think)

4) The experience level of the bombard unit should be taken into account in determining the success of the bombardment mission.

I would also add that terrain should be considered, as well. In Alpha Centauri, artillery got bonus if firing from higher altitudes, and rougher terrain (forest, hills) ought to give better cover from bombardment.

Anyway, some great suggestions for artillery in the game, and thanks for explaining how some of the mechanics should work.

As for elite units having a combat bonus, that's something I really think should be in the editor for all units, not just arty. I'd rather mod my game to have units of all exp levels to have the same amount of hit points, with vets and elites having combat bonues and conscripts having combat negatives, isntead. If I could set it, I'd make the game more like AC, with ancient units having 10 hp, middle ages having 20 hp, and so on, with experienced units having combat bonuses.

I set my game so that all arty units have 1 defense. This makes them uncapturable, but I think it's a fair compromise for allowing arty units to fire back and forth to kill each other, and they're still just as helpless as before, except that one unit can't capture an entire undefended stack.

 

[cracker]

I will add that the artillery units have some really cool death animation sequences included in the standard CIV3 package so obviously someone at Firaxis was under the impression that catapults, cannons, and artillery would be able to be hit and destroyed.

The only ways you could ever see these events now would be if you deliberately pulled an artillery piece up next to barbarian village and left it undefended -or- to leave the weapon undefended in front of a civ that had not yet discovered or traded for the tech that enabled them to use the weapon (as if that is ever going to be the case with AI tech trading levels).

The animation for the WWI type artillery has wheels blown off and flying into the air and is really cool looking.

 

[simwiz2]

Originally posted by cracker
I will add that the artillery units have some really cool death animation sequences included in the standard CIV3 package so obviously someone at Firaxis was under the impression that catapults, cannons, and artillery would be able to be hit and destroyed.

Actually the death animation is used for when a less-advanced civ captures an arty unit that they can't use, it is destroyed. Like if a civ captures an artillery but they dont have replaceable parts, then they would be destroyed with the animation.

But i agree with you, bombard units should be vulnerable to bombardment, just like workers, planes, etc should.