Still focusing on getting functional Stand-Off Attack Rules

[cracker]

We need to continue a discussion that focuses attention on getting the bombardment engagement system in the “out-of-the-box” as patched version of CIV3 fixed to be a functional set of tools. These fixes need to be a near term patch without waiting for 6 more months for the XP version.

The key issues must emphasize having the correct features set to ON in the standard product and not just taking refuge in the fact that the feature can be turned on in the editor “if you want to.”

We need to acknowledge that there is a fairly strong lobbying group that has helped to render stand-off style engagements as fairly ineffective because this group of engagements has been saddled with the unfortunate label of “bombardment”. Just because these engagements are labeled as bombardment does not mean that they are strictly simulations of the ineffectiveness of WWI era artillery against entrenched defenders.

The current system of stand-off engagements (labeled bombardment) has tried to balance a limited number of variables to result in measured ineffectiveness for the stand-off attacker. A fundamental problem in the current implementation is that it relies on a deterministic system of target selection to reflect a statistical engagement process. What I mean by indicating that the engagement system is wrong to use the deterministic process is that when you fire a bombardment weapon at a stack of targets in a given terrain square, the current engagement system is certain to try and engage the strongest defender with zero chance of effecting any other units or features in the square. (note that in towns and cities this deterministic selection is modified to split the targeting between improvements, citizens, and defenders but it is still deterministic and not statistical).

The first key change that must be implemented is to expand the list of targets for stand-off engagements to include ALL units, improvements, citizens, and terrain features in a targeted terrain square. Everything should be a target and that is part of the risk of standoff attack methods.

Getting the potential target list correctly implemented will immediately have several significant and positive effects on stand-off type engagements. First, the expanded target list will include all the units that currently are exempt from being hit by the effects of bombardment weapons. Artillery and planes should be able to be hit by incoming attacks within the statistical limits of taking their chances among all the other potential targets. Workers, settlers, scouts, explorers, and other non-combatant units should be able to be damaged by stand-off attacks just like the citizens tilling the soil within a city or town. Fundamental in this potential target list is the concept of including the terrain square as number of potential targets in order to facilitate a statistical engagement process that recognizes that it is hard to get a hit against a single defending unit but relatively easy to get at least one hit against a stack of 40 or 50 targets in a single square.

By way of a simple example focused on potential target lists only, we can look at the case of a single spearman standing in the open on a roaded and irrigated plains square defending a worker and a catapult. In the current system, when we engage that the contents of that square with a standoff attack by a catapult, we are absolutely certain that the spearman will be targeted for attack and there is a 67% chance that the catapult will get a hit on the spearman. This is patently unrealistic. In the proposed, statistical system when we engage the targeted square we would have a potential target list of 8 targets (spearman, worker, catapult, road, irrigation, 3 x Terrain) so even without considering defensive strengths there would only be a 12.5% chance of getting a hit on the spearman but this would be balanced by knowing there would be a 62.5% chance of hitting something other than dirt. If this same example included 25 spearmen as might appear in an AI generated stack of doom (SoD) then the probability of hitting the strongest spearman would be only 3% but the overall probability of getting a hit would be 91% because there are just way to many targets in the square to not have an increased chance of getting a hit.

A second key impact of getting the potential target list properly expanded to include all potential targets will be a shift in the statistical approach that tries to force the standoff attack to miss the primary defender a significant number of times instead of allowing the system to miss the primary defender using random statistical failure. When more targets are recognized as being available, the chance of getting a hit of any kind will be higher but the chance of hitting and/or killing any specific target will be much lower, just as it should be.

After getting the potential target list properly expanded, the implementation of the impact of Rate of Fire should be distributed to allow multiple targets to be hit in a single engagement. This decision moves the process further towards reality and helps to spread out the damage from the standoff attack across all the potential targets to avoid the current overwhelming power against a single target. It would still be possible to get multiple hits on a single target in a single turn, but now the process would be statistical based on the number and strength of other potential targets. Instead on focusing on the destruction of a single unit as is currently implemented, the proposed stand-off attacks would focus on reducing the effectiveness of whole groups of attacking or defending forces.

Units which engage in stand-off attacks should have their hit success rate modified geometrically to correspond to their experience level. Every other combat unit in the game has some benefit associated with the level of experience of the unit. Bombardment engagements should be no different in that respect and elite bombardment units should be expected to be more successful in inflicting damage on the enemy units when compared to the success rate of their conscript, regular, or veteran peers of the same type.

All forms of standoff engagement should include the potential for lethality, but the impact of this decision will be substantially mitigated by the statistical redistribution of the attack power across all the potential targets. In stacks of 25 or 30 defenders (or advancing attackers), the chance of lethal engagement would be statistically spread across all the hit points of all the potential targets in the stack. While it would be possible to destroy an individual unit with a limited number of lethal stand-off attack engagements, it is far more likely to continue to require massed numbers of artillery and/or bombers to generate a lethal event. In a stack of 10 regular spearmen standing in the open on a roaded and irrigated plains square each spearman would have only a 1 in 15 chance of being the potential target for a catapult attack, and then the catapult would have only a 67% chance (4 vs 2) chance of hitting the spearman for an overall risk of 4.4%. If you fired three catapults at the stack that would be the minimum number of catapults required to potentially kill a spearman but there would only be a 1 in 1,139,000 chance of killing any of the units. You would have to have over 30 catapults lined up and firing on the stack of spearmen in order to have a 50/50 chance of killing just one spearman if the proposed rules were implemented.

A further enhancement to the lethal bombardment option could add displacement as a coin toss option for units that might otherwise be destroyed by the lethal effects of an incoming attack. The move it or loose it option would disrupt attacks and further expose the enemy forces to the selective effects of focused counter attacks and this should be exactly how non-precision stand off attacks should function

A key point in continuing to advocate for eliminating the deterministic engagement rules for stand-off attacks would be that some units do not get hit by bombardment because they get lucky combined with their defensive strength and health. Other units get hit and killed because they are unlucky combined with their weaker defensive strength and lower units health (morale).

Everything we attempt to do with units that use stand-off attack methods is just a series of band-aids and work-arounds until we get these three elements of the bombardment engagement system properly implemented to facilitate balanced usage of these types of engagement strategies.

The availability of appropriately functional bombardment units is key to supporting strategies that emphasize numerically limited military forces that use technology appropriately to disperse numerically excessive but obsolete armies that are overly concentrated into limited terrain features.

[cracker]

I think it helps to have a simple example that illustrates the difference between the current deterministic rules for stand-off engagements versus the proposed statistical engagement rules that would provide a more realistic set of engagement results whenever bombardment style attacks are used.

For simplicity let us use the example of a stack of attackers from an AI civ moving forward to attack a city that belongs to the human player. The attacking stack contains two regular longbowmen (4/1/1), two regular pikemen (1/3/1), one elite knight (4/3/2), and two cannons. The stack is advancing across plains or grassland terrain that has been roaded and irrigated.

To produce this example, we need to use our “standard, pre-approved, gamer’s random number generator” to produce a set of random numbers between 1 and 1024 that we can use to compare the impact of the two sets of engagement rules. This sequence of random numbers is randomly generated but we will use the same sequence in each of the different rules to see how the results would vary.

Our 1st test sequence of 20 numbers is: 674, 781, 510, 191, 101, 574, 222, 395, 144, 66, 96, 766, 407, 149, 46, 44, 575, 908, 310, 337, 139, 525, 237, 252, 425, 844, 420, 819, 191, 641

Our 2nd test sequence of 20 numbers is: 509, 5, 563, 453, 75, 652, 418, 735, 612, 763, 827, 836, 168, 238, 838, 119, 417, 1, 659, 803, 1019, 38, 854, 503, 653, 768, 422, 252, 75, 937

Under the current rules, this stack of attackers is only one target for bombardment purposes and that target is represented by the strongest defender (hits x D rating) so in this case only the elite knight could be engaged by the first bombardment attack. Effectively the target list contains only one unit that is predetermined by strength and then this one target changes as each subsequent bombardment attack is processed.

In our example, we will attack the stack with the equivalent era bombardment units of Cannons with bombardment strength of 8 and rate of fire of 1.

In the first sequence of bombardments we need to know that the bombard power of the cannon (8) compares to the defensive strength of the knight and pikemen at (3 plus 10%) to give an 8 out of 11.3 chance of getting a hit. Our random number would need to be less than 745 for the shot to be counted as a hit. When the cannons target a longbowman the hit ratio changes to 8 out of 9.1 because the longbowman only has a defensive strength of 1 plus 10%, so in those cases the random number would need to be less than 900 to count as a hit. The current defensive value of improvements would seem to be set at 16 so when the cannons target to destroy roads and irrigation the hit ratio would be 8 out of 25.6 or a cutoff comparison of 320 for the random number generator.



Observe that the hit rate for the bombardment units is relatively constant because nothing is implemented to take into account the number of potential targets including the dirt of the surrounding terrain. The cannon that is firing in the current example is always engaged in a one-on-one battle with a predetermined defender (or improvement). In the current case, death is a statistical certainty for the defender when lethal bombardment is engaged because the bombardment will always target the strongest defender unit until that unit is destroyed.

[cracker]

One of the primary missions for bombardment type units is to prevent or discourage the enemy from massing and organizing attacking forces. Bombardment does not have to destroy these units to cause them to be disrupted and/or dispersed. In the current implementation of bombardment attacks, this mission is virtually incapacitated and omitted because the bombardment strengths of the units have been artificially reduced to keep them from being too effective against individual units without taking into account conditions when the bombardment units should score a higher percentage of hits in concentrations of enemy units.

If we look more carefully at the units and other contents of the targeted terrain square we would develop a more complete list of potential targets:



With an accurate interpretation of the potential target list, there would be 11 possible targets for the initial bombardment attack to randomly select as the potential victim of the attack. We could use our second sequence of random numbers to choose the target from this list. Notice that the expanded target list includes all the traditional ground combat units as well as the cannon that was previously exempt from being targeted. The list also includes each of the individual improvements in the square as well a three targets to represent the chances of missing all the units and improvements in the square. The number of base chances to miss all the potential targets in a square should be set in a field in the general settings of the CIV3 editor and for the purpose of this example three chances have been used.

After each shot is fired in the stand-off engagement, the potential target list may need to be revised if any of the targets get destroyed by the previous shot. When the number of unit targets is large compared to the number of improvements and “Dirt”, then the chances of getting a hit will be higher. When there is only one unit in a square with 2 improvements and three based chances to miss, then the chances of getting a hit on the unit will be lower.

Please note that the bombard strengths, rates of fire, and defensive strengths of things like roads, civilians, and buildings has been severely distorted in the current engagement system in order to try and compensate for a lack of statistical engagement. These distortions only work in limited cases and create problems whenever there are very few or very many units located in the same square (which is most of the time).

[cracker]

Even with the distorted bombard strengths and distorted defensive values for the “other stuff”, let us look at an example of statistical target selection within the square that contains the attacking force that we wish to engage with a standoff attack.

Before you look to the example results in the table, remember that the point being emphasized here is not that it should take 52 volleys to destroy a group of attackers, but that when there are many targets, the chance of getting a hit should be higher and that all targets (intended and accidental) should have a chance of being hit and destroyed even if that chance is very low and one the order of only 1 in 15 or 1 in 20.



Notice that the first shot was just a random lucky shot that hit and destroyed the cannon in the stack but that over time you can clearly see that the number of volleys that impact harmlessly into the Dirt goes up as there are fewer and fewer military, civilian, and improvement targets to select and hit.

When implemented, this statistical bombardment approach will significantly improve the realism and flexibility of the stand off engagement system. Implementation will also probably require a general increase in the bombardment strengths and rates of fire for many bombardment capable units.

[Fitz]

Nice analysis cracker, and it certainly makes sense to me. The defense strengths of the artillery and improvements/buildings could be modded much more realistically.

I do see a small problem here though. Cities. When you drop a shell into a city, the chance of hitting the buildings vs the chance of missing them should be relatively standard, no matter how many units. The chance of hitting a unit though should still depend on your system of a miss.

For example, if 25% of the city has buildings that could be destroyed by an artillery shell, it doesn't really matter if there are 1 or 20 units in the city. 25% of the time you will hit a building, whether or not the hit actually destroys it. But in that remaining 75% of the time, you could certainly miss.

In fact, I also think it is a better way to treat improvemnts too. If you assume a base 3 for "miss" then add 1 for each improvement, and first see if it hits the improvement, then if the improvement misses you can see if there is a chance of hitting the units in the square.

That might be much harder a recode though, depending on the way it is implemented in the first place.

Back to artillery & building/improvement values, artillery should really have a much higher chance of damaging the target when it lands in the right vicinity. So a much better way of doing it would be a base 6-9 dirt/miss chances, with a doubling or tripling of the bombard strength. Similarly with determining the chance of a building/improvement getting destroyed, you can reduce the value of the defense to reflect the fact that if the building actually gets tagged by a shell, it will probably be destroyed. This of course makes stacking units way more dangerous, since there is little chance that a single unit will get hit (but almost garunteed to take damage if it does), whereas a large stack will definately be hit somewhere and almost certainly take damage.

Example, 9 dirt chances, 1 unit with defense 3 (pikeman), cannons with new and improved attack rating 16. 10% chance of the pikeman actually being targeted, ~84% chance it will take damage if it does. That's 8.4% chance of that unit being damaged and 8.4% chance of any damage being done overall.

Example 2, 9 chances of dirt, 9 units with defense 3 (pikemen), cannons with new and improved attack rating 16. 50% chance of any pikemen being targeted, ~5.6% chance of an individual pikeman being targeted. Still ~84% chance of damage to a targetted pikeman. That's now a 4.7% chance that an individual pikeman takes damage (down by about 1/2), but a 42% chance that at least one pikeman will take damage.

Regardless, the chances of Firaxis actually making this change? <1%

[Jon Shafer]

Good lord.

Everything looks great... except, as has been brought up in many other threads, Civ 3 has been intentionally "dumbed down" in order to appeal to more people. If that means that only 1 unit/building/population point can be attacked at one time by bombardment, then that's how it will be. Most people (the unseen masses that don't care enough about Civ to come by the forums) don't want to have to deal with all those factors (may lightning strike them dead) in their little game. A pity too.

[Cyclotron]

Can I get the clif's notes version?

Actually, all I really need is a longer attention span... I'll read all this eventually.

[wrylachlan]

Originally posted by Trip
Good lord.

Everything looks great... except, as has been brought up in many other threads, Civ 3 has been intentionally "dumbed down" in order to appeal to more people. If that means that only 1 unit/building/population point can be attacked at one time by bombardment, then that's how it will be. Most people (the unseen masses that don't care enough about Civ to come by the forums) don't want to have to deal with all those factors (may lightning strike them dead) in their little game. A pity too.

I actually don't think this is that complicated an idea. I could explain it to newbies in a single paragraph:

When a tile is bombarded the shot will hit, at random: any of the units in the tile, any improvements to the tile, or nothing. The greater the unit's defensive value the less chance it has of getting hit. Any improvement hit is automatically destroyed, and any unit hit may take damage based on its defensive rating versus the attackers bombard rating. For the full formula please see index page 57.


My feeling is that anyone who cares enough to know what the formula is, wouldn't mind a more complex formula.

And cracker - that's a great piece of work and I hope Firaxis reads this and offers you a job.

[Kataphraktoi]

...a job for a great post and idea?

[IthacaMike]

I vote no.

Using the first example if I fire a Catapult at a stack of defenders I don't think that the road net in the tile should be at risk, unless I have chosen to attack the tile's improvements.

Artillery fire is not just blasting away at a map grid and hoping you hit something. Even in WW I the shells were falling in and arround the enemy trenchs. Is just that the trenchs were empty when the shells were falling. After the barrage lifted it was a race to see who could get into position first. Normally the defenders won the race.

Mike G

[simwiz2]

this sounds like a really good idea. I especially like the idea of having a possibility of hitting your own terrain improvements. It makes warfare more realistic, and makes you think before mindlessly shooting off artillery shells in every direction, especially in your own territory.

[Capt Dizle]

Pearls cast amongst swine.

[cracker]

IthacaMike,

You unfortunately have been corrupted by exposure to the bad first attempt at implementation of stand-off attack that has been included in CIV3.

Remember that we are talking about a whole class of attack strategies here and not just some half-baked simulation of WWI artillery effects.

The reality is that bombardment aims at inflicting damage to targets in areas and as such one of the first things to get damaged is structures and improvements. You do not kill, maim, demoralize or otherwise disrupt the GI Joes in a farmhouse without doing substantial and almost irreversible damage to the farmhouse.

I will personally send you $1,000 greenback American dollars if you can fire a standard artillery shell at any rap star of your own choosing while that rap star is standing in the middle of any road in the world, if you can kill that rap star with the artillery shell without doing damage to the road.

(hint: the only way to do it is to fire a dud round horizontally at point blank range and splatter the target with the kinetic energy and even then there a good chance that the velocity of the splattering fragments of shell, bone, and gold teeth/jewelry will still tear up the pavement.)

The reality is that bombarding any type of a unit on or near a roadway is 10 to 20 times more likely to damage the road than it will do damage to the personnel as they dive for cover at the first hint of the screaming, whistling incoming rounds.

We had a standing joke at one point in time that the Geneva Conventions had outlawed the use of a number of weapons against enemy personnel. The list of weapons you could not use against people included artillery, tanks, flamethrowers, and explosive devices. This was because these weapons represented cruel and excessive force against human beings. The way the military forces of the world got around these prohibitions was to issue standing orders that these weapons could only be used against equipment assets and not against people. So we could not shoot artillery at people, we could only aim at the vehicles they were riding in, their helmets, boots, buttons, and belt buckles.

[wrylachlan]

Originally posted by IthacaMike
I vote no.

Using the first example if I fire a Catapult at a stack of defenders I don't think that the road net in the tile should be at risk, unless I have chosen to attack the tile's improvements.

Artillery fire is not just blasting away at a map grid and hoping you hit something. Even in WW I the shells were falling in and arround the enemy trenchs. Is just that the trenchs were empty when the shells were falling. After the barrage lifted it was a race to see who could get into position first. Normally the defenders won the race.

Mike G

I think you are partially right on this on. If we envision a tile being at least a 10 by 10 mile area (and I know this isn't realistic based on the size of the world, but its what Firaxis says...). Anyways, if we assume a 10x10 area, your troops could be nowhere near the road. And definitely nowhere near the irrigation.

The problem that cracker is working with, though is that there is no real mechanism to respresent the fact that the more troops you pack into a tile, the greater your chance of hitting something. As such his idea works really well.

My suggestion for a compromise would simply be not to hard-code the defensive values for improvements. If it makes realistic sense to Cracker for a missed shot to hit the road, he can give the road a lower defense. If you think it's absurd that a missed shot would hit the road, crank up the road's defense. But either way your chance of doing damage to units will increase the more units there are in a tile.

And as a corrallary to this idea, could collateral damage be represented by giving the tiles hit points? Every time a bombardment hits the dirt, the tile itself has a chance of taking damage, and once all it's hitpoints are gone the tile becomes unusable until a worker comes in and clears out all the unexpolded ordinance and rebuilds the farm-houses, etc...
Though this may be going too far down the complication road...

[Puupertti Ruma]

The system, in the form Cracker has introduced it, has couple of problems which are consequences of the 'real world'. Of course this is a game, so we should try to depict real world with rules of the game, so I have tried to give some ideas how to do this.

The biggest problem is that a tile is, as wrylachlan stated, an area of at least 100 square miles (10x10). And because of this following is true:

Destroying the whole road network, or irrigation in a tile of 100 square miles would take an aimed bombing raid on key junctions in road or railroad network or the important spots of irrigation system. To do this in an accident would be very unlikely.
It could of course be that the unit artillery is aiming at, would be standing at one these soft spots, and they would be destroyed in a bombard, but this is so unlikely (in 100 square miles there will be many important junctions) that in game terms the possibility can be excluded.

Second problem is in the stated possibility to hit nothing but dirt. Artillery is not used just to shower a tile with artillery fire, it is used to suppress enemy or to cause loss in the ranks of enemy (as catapults and cannons were mostly used). Aiming at an area of 100 square miles and hoping it would hit something is not an option in warfare. Of course there is a possibility that the raid doesn't have any effect, but this is already simulated in CivIII by giving a possibility where unit isn't hit.
Of course there are times when armies shower some area with artillery fire, not knowing if there are any enemy troops in there. But this is also for a reason, they are making sure that area that they are shooting at, is, or will be, empty from defenders. And I think this is already included in the no-hit possibility.

My proposal is that artillery would be considered as a unit that can do collateral damage. This way the Stacks of Doom, being main problem within wars of CivIII, would be eliminated. In CivII the collateral damage was with every unit, so large operations by AI would fail miserably because of stacking, which wasn't good thing at all. I'm not saying nothing too accurate at how the collateral damage should work, I leave the thinking for you guys.

A note about the lethal artillery: I oppose it at in everywhere but in the sea warfare. When at continent, there are millions of little fox holes where you can hide, so artillery can never wipe out a whole unit. The main purpose of artillery fire in land is to suppress. This is in Civ easily simulated by dropping hitpoints to one. Suppressed unit can't attack nor defend succesfully and often will be destroyed in doing so.
But in naval warfare, bombard (in other words fighters dive bombing battleships) is meant to sink ships, not to suppress. And when the ship has one hitpoint it's not sunken. Realising this, it is easy to understand a fighter squadrons taking out whole fleet with enough firepower.

And if we are waiting for Firaxis to do something about this problem, we always have to remember that when changing rules of engagement the AI have to be coded to use these new rules. But don't get frustrated, we are discussing these things because we like to discuss and to shine with our knowledge

[Jon Shafer]

Phew, long posts. I'll address this topic once I have more time to elaborate, but basically, I feel there should be some sort of balance. The nature of bombardment is that it's not always 100% accurate. With that being said, you are after all, aiming at something (unless it's blind bombardment, but you can always see units in Civ III, so that's not an issue in this game). So basically, it should be easier to hit the target you're aiming at (not necassarily the top unit on the stack). This should be the case with every attack.. it shouldn't always hit the top unit on the stack. Not very realistic, but hey, that's the Civ III battle system for you.

[IthacaMike]

Originally posted by cracker

The reality is that bombardment aims at inflicting damage to targets in areas and as such one of the first things to get damaged is structures and improvements. You do not kill, maim, demoralize or otherwise disrupt the GI Joes in a farmhouse without doing substantial and almost irreversible damage to the farmhouse.

You have generalized damage to _a_ farmhouse to damage to _all_ farmhouses.

Originally posted by cracker

I will personally send you $1,000 greenback American dollars if you can fire a standard artillery shell at any rap star of your own choosing while that rap star is standing in the middle of any road in the world, if you can kill that rap star with the artillery shell without doing damage to the road.

This most clearly demonstrates what I think is the mistake. Damage or destruction of a road within a tile is not the same as damage or destruction of the road net within the tile.

Localized colateral damage is not the same a systematic destruction of the infrastructure.

Mike G

[cracker]

Originally posted by IthacaMike

This most clearly demonstrates what I think is the mistake. Damage or destruction of a road within a tile is not the same as damage or destruction of the road net within the tile.

Localized colateral damage is not the same a systematic destruction of the infrastructure.

Mike,

Your argument is entirely why it is appropriate for the improvements to have a defensive strength value that enters into the calculation of combat success when it is compared to the bombardment power of the unit that is engaging in a standoff attack.

Your logic for arguing that the road network in a given terrain tile should not necessarily be destroyed by a single hit, only serves to provide stronger support for statistical engagement. The current system makes it impossible for you to hit the road network if there is even one meager defender in the square. You have to blast the defender into the MASH triage category before you have any chance of damaging the road network. You have to damage all the defenders to the maximum extent allowed before you can have any chance of hitting the improvements.

I would add further that the improvement damage sequence is predetermined to be Railroads - Roads+MorI - then finally Fortifications.

So try to come up with a valid and logical argument that supports an engagement sequnce that says "kill all the military people first, then damage the railroads, then remove the mines and irrigation, and finally destroy the fortifications while leaving all the artillery and non-combatants untouched."

I agree with you that standoff engagements are not just random potshots (in the big picture) but what we are striving for is some element of simulated effect that includes both the risks and the benefits of this type of engagement.

I think you would agree that firing artillery into a square with a transportation network has some chance of hitting and disabling the roads (hit a key bridge or intersection, cause debris to block the road, topple trees or telephone poles. etc.) even if that chance is remote.

Another key point is the nature of the event that is being simulated when roads are "destroyed". At the point in time in the game when artillery and planes are present, it takes an average of 4 bombing run turns to destroy undefended roads and railroads. The enemy workers can reconnect the road network in almost the same number of equivalent worker turns. An industrious worker (China, America, etc.) can rebuild the railroads in a square even faster than you can you can blow them with artillery or bombers.

[wrylachlan]

Originally posted by cracker

I agree with you that standoff engagements are not just random potshots (in the big picture) but what we are striving for is some element of simulated effect that includes both the risks and the benefits of this type of engagement.

Here's an idea:

Collateral Damage and Precision Bombarding

Collateral Damage would simulate the... well collateral damage. There is a percentage chance that units or improvements not specifically targeted by an attack will recieve collateral damage, the more units in a stack the greater the chance will be that one or more will be hit. Obviously this must be balanced so that no one shot can knock out a whole stack.

Precision Bombarding would be just that, when you PB a tile an option menu pops up and you can choose which member of a stack, or improvement you want to attack. There is no guarantee you will hit that unit or improvement, but that's who you are targeting.

PB does not take the place of Standard Bombardment, so after you PB the fortifications and cannons you can use SB (which is easier - less clicks) to finish off the units. PB could be toggled on and off so that if you don't want the "Advanced" = "Complicated" features to be available in a game, you don't have to have them.

I would also change the order of SB targetting to units, artillery, fortifications, improvements.