Dune Wars 1.9.4 Patch Feedback

but I still would like to give fealty the free unit bonus, there's a real void for that in the late game and it makes a lot of sense for it, gives you some room to actually use the draft ability with the extra free units, what do you think on it?

Click to expand...

Seems reasonable; drop the happiness, have free units and draft, and maybe increase the Civic Upkeep level?

I've never personally really used Draft much (because of playstyle, not because it's weak) so I don't have strong thoughts on the best way to balance this.

It seems like it could still fit in flavor; if you are loyal to the emperor, then you have levy troops that are cheaper to maintain (because their costs are borne by the minor houses and lords that owe you fealty).

Another thing; the improved transport AI (dropping off its troops immediately) seems to be working nicely. Great job.

Seems reasonable; drop the happiness, have free units and draft, and maybe increase the Civic Upkeep level?

It seems like it could still fit in flavor; if you are loyal to the emperor, then you have levy troops that are cheaper to maintain (because their costs are borne by the minor houses and lords that owe you fealty).

Click to expand...

Yep, my thoughts, though I'd let it keep the happiness as a bonus effect that separates it more from Trial by Combat.

I do see the issue of non-imperial teams using it, so it would be cool to tie or amplify it's effects in cities with imperial, but given there are no other civics with effects tied to a specific 'religion', it would leave it as sort of an odd man out. For now I'll let players rationalize it out however suits them .

I've never personally really used Draft much (because of playstyle, not because it's weak) so I don't have strong thoughts on the best way to balance this.

Click to expand...

Me neither, though if you've got you've back to the wall and need troops quickly I can see it's use. At any rate, combining it with more free troops will make it actually economically usable I think, which it isn't very much now. Before you'd bankrupt yourself if you used it much I'm afraid.

Another thing; the improved transport AI (dropping off its troops immediately) seems to be working nicely. Great job.

Click to expand...

I know! I've had the Ecaz of all things repeatedly invading me in a recent emperor difficulty game really keeping my back to the wall. They've declared war on me four times and are getting tougher to repulse each time. The second or third war they took a city and it took a while to get it back. It's been fun.

>>>Download 1.9.5 Beta1.1<<<

In addition to the discussed changes to unit costs, inflation, imperial fealty, and Atredies promos, this has a few small AI tweaks (including encouraging the AI to get proper city defense units, e.i. Defense Tactics, earlier), updated air combat code, and some small further balancing of the Handicap settings.

Air Combat

It's amazing what you find sometimes when you dig into the code. For instance, ground units intercepting aircraft dealt no additional damage if they have bonus combat% verse them, it was solely based on their interception rate. I've revamped it following using much the same logic as how air to air combat damage is calculated. I'd already given the air unit the ability to damage the intercepting unit, but it's much more in line with how all other combat is calculated in the game, taking promotion bonuses and ground into account.

It's very unlikely the intercepted air unit will destroy and intercepting unit (unless there is a large tech difference), but as I said for the previous beta release, for immersion and some gameplay reasons I want them to do some damage.

I also think I've got the experience issue reported by Jester up above handled, but let me know if any one still sees it happening.

Handicap Settings

I've made some minor adjustments to reduce the 'step' between each of the higher difficulty levels a little bit based on my test game observations, and reduced AI bonuses for inflation (which is going to be somewhat smaller in this release for everyone) and the 'per era' bonus in the higher difficulty levels as I don't think the AI needs as much of them and is likely to just run away with the game in the late eras if they are too high.

The goal is to keep a competitive environment throughout the game in relation to difficulty level picked of course. My observation was that by about the half way point of the game and beyond the AI bonuses were piling up too much even when playing a pretty successful game. We'll save that effect for Immortal and Diety where it's expected Emperor needs to be reasonably winnable with a good effort.

Yep, my thoughts, though I'd let it keep the happiness as a bonus effect that separates it more from Trial by Combat.

Click to expand...

That seems like too many things for a single civic?
I would prefer to make the military upkeep the main benefit, followed by draft. No real need for happiness too, I'm guessing.

I do see the issue of non-imperial teams using it, so it would be cool to tie or amplify it's effects in cities with imperial,

Click to expand...

Oh, I wouldn't try to actually tie it to the Imperial Religion. I was just meaning that there is a way in which we could imagine that civic might reduce unit maintenance. The effects of a civic should fit with its name.

Me neither, though if you've got you've back to the wall and need troops quickly I can see it's use

Click to expand...

I tend to use homeworld transport for that. But yes, I can see the use. Having a big army show up on your doorstep happens a lot more in Dune Wars than in vanilla.

I know! I've had the Ecaz of all things repeatedly invading me in a recent emperor difficulty game really keeping my back to the wall. They've declared war on me four times and are getting tougher to repulse each time. The second or third war they took a city and it took a while to get it back. It's been fun.

Click to expand...

Yes, this is working nicely, very fun. I think much better than vanilla Civ; diplomacy is really important here (avoid being everyone's worst enemy and getting dogpiled!) and use recon and point units to spot incoming threats.
Its also another advantage for big borders; if your culture borders are wide enough, the enemy might not be able to get into your territory and unload on the same turn.

For instance, ground units intercepting aircraft dealt no additional damage if they have bonus combat% verse them, it was solely based on their interception rate. I've revamped it following using much the same logic as how air to air combat damage is calculated. I'd already given the air unit the ability to damage the intercepting unit, but it's much more in line with how all other combat is calculated in the game, taking promotion bonuses and ground into account.

Click to expand...

Great. I'd always assumed strength mattered, a bit weird to find that it didn't. We should do some testing then to make sure that ground-based AA are good vs aircraft, but not overwhelming.

My observation was that by about the half way point of the game and beyond the AI bonuses were piling up too much even when playing a pretty successful game. We'll save that effect for Immortal and Diety where it's expected Emperor needs to be reasonably winnable with a good effort.

Click to expand...

Ok, this seems reasonable. I was also finding the AI bonuses in the midgame quite significant, but that was in part because I struggled to keep science very high because of high military costs.

I think the spy detection chances might be too high. The AI never seems to be using any espionage against me (though in part this is because I've only really tested the early and early-midgame, so it might just be they haven't accumulated enough EPs yet) and my spies seem to not last long in enemy territory without getting caught.

I wonder if we should increase the hammer cost of the various spy units combined with reducing the detection chances?
Spies are pretty cheap, so you could spam them like crazy, but that's not really much fun.

Great. I'd always assumed strength mattered, a bit weird to find that it didn't. We should do some testing then to make sure that ground-based AA are good vs aircraft, but not overwhelming.

Click to expand...

I've done some basic testing to make sure I'm getting results within expectations. Relative strength and damage/round calculations work very similar to any other combat, except it does also use the relative ratio of the units interception rates (or 20% if it's zero such as bombers) when calculating damage. Both air-to-air and air-to-ground use that same calculation, but if the defender is ground (which includes suspensor craft) the attacker's damage is halved, so the intercepting unit still has a significant defense advantage.

This also reminds me, I altered the way plot interception % works. In vanilla BTS, it's the highest current interception rate of any unit that can intercept at that plot. So if you've got three Rocket Troopers, it's 30% since they all have 30% interception (without promotions at least).

Afforess created a modcomp that changed this, which was included in RevDCM, and is therefore in DuneWars, which was a nice improvement, but goes a little overboard. What it does in the previous example is changes the odds where it's like you have to pass the interception test for each valid interceptor unit for the plot. So, in the previous example, instead of 30% interception rate, it's (1 - (.7^3) = 65.7% interception rate. For 5 Rocket Troopers it goes to 83.2%.

In practice, this makes AI stacks with very many AA defenders almost un-assaultable with air units, which was part of the reason I wanted the air unit to at least be able to deal some damage to the interceptor.

I've changed the math so that additional interceptor units still add to the interception rate, but not at an exponential rate as he did it. I won't explain the math, just give an example, using the same situation as before. The three rocket troopers end up with a 42% interception chance, five rocket troopers would return a 52% interception chance. I also want to add text to the plot help telling the player what the interception chance is in that plot, but haven't gotten to it yet.

Oh, I wouldn't try to actually tie it to the Imperial Religion. I was just meaning that there is a way in which we could imagine that civic might reduce unit maintenance. The effects of a civic should fit with its name.

Click to expand...

Me neither, like I said, it'd be horribly un-symmetric with all the other Civics. I just said it'd be cool, not necessarily good

That seems like too many things for a single civic?
I would prefer to make the military upkeep the main benefit, followed by draft. No real need for happiness too, I'm guessing.

Click to expand...

And now to contradict myself... what if it gives 1 happiness in cities with imperial instead

I tend to use homeworld transport for that. But yes, I can see the use. Having a big army show up on your doorstep happens a lot more in Dune Wars than in vanilla.

Click to expand...

Me too, as long as I've got the extra troops available from the homeworld, and the cash of course. Another interesting use is pulling up extra troops to garrison cities during an invasion, I find it can be difficult to amass an invasion and enough garrsions to sustain the assault.

As a side note, another AI project I want to do is tie the AI decision to bring in reinforcements to city danger. There is already a city AI function called AI_panic() that changes the city to producing defenders and hurries them if possible when under imminent danger, I'd like to tie it in right there.

Ok, this seems reasonable. I was also finding the AI bonuses in the midgame quite significant, but that was in part because I struggled to keep science very high because of high military costs.

Click to expand...

Yeah, it also goes back to the previous discussions on AI versus human ability to amass units given respective empire and economy size. I like the balance early on, but the inflation help later on (which is much less necessary now for them to keep up) was really skewing it too much against the human.

A real issue here is even though the AI does need extra units, the power rating is an important diplomacy and warplan indicator and is human blind, so if the human is too disadvantaged in keeping up in the arms race, he gets dog-piled more. Before at the monarch/emperor level, the human had fewer units but was consistently at higher techs, so had a higher power rating based on the more advanced units. That's less the case now, so keeping a sufficient power rating requires more parity of unit levels than before, and the AI can compete better at closer unit levels with more advanced units.

I think the spy detection chances might be too high. The AI never seems to be using any espionage against me (though in part this is because I've only really tested the early and early-midgame, so it might just be they haven't accumulated enough EPs yet) and my spies seem to not last long in enemy territory without getting caught.

I wonder if we should increase the hammer cost of the various spy units combined with reducing the detection chances?
Spies are pretty cheap, so you could spam them like crazy, but that's not really much fun.

Click to expand...

There are issues with the spy AI I've got to deal with when I can get the time, it was really designed around spies being able to do all available missions as long as they had the EPs. Where it is deciding what mission to do, I'm pretty sure it actually looks first for the one with the highest value (without asking if the unit is able to do it), THEN asks if it can be done, and if it can't, the spy just does nothing. I think this happens alot... , but I've got to study the code more to know for sure.

The whole base AI code is horribly antiquated since it was built on a much simpler system with, remember, no promotions, and everything has been built on very piecemeal onto that original simplistic system. People have added to the abilities and restrictions for spies, but no love for the AI.

Another issue is it's assumed that all missions are designed to hurt an enemy, so it uses even missions like political marriages only against civs they don't like/have warplans against, that kind of thing, thinking it is hurting them somehow

The whole system needs a top-down rewriting. Expensive spies would just be expensive units the AI wouldn't know how to use (they really mostly wander around until they find something in they're vicinity to do, I'm serious ), while decreasing interception chance would make espionage too easy for a human. These are balanced very carefully right now against possible ranges of espionage defense buildings, the possible promotions to increase it, and things like the counterespionage mission's effect on interception.

The reason you see a lot of spies getting caught in your territory is the AI really has them wandering around a lot in enemy territory without any real mission there, so they wander in circles till they get caught. And since the AI has a simple (iSpiesNeeded = iNumCities/3) instead of a real appraisal of need for spies, what it wants to use them for, etc. it happily builds another one to send off to get caught wandering around aimlessly. Mostly the only reason you see them occasionally perform a successful mission is that they have enough wandering around that by pure chance it's bound to happen . It's really not good .

As a side thought I just had, there should probably be some handicap tags for spy interception rate, make it easier for humans at lower levels, give bonuses to AI at higher levels.

my spies seem to not last long in enemy territory without getting caught.

Click to expand...

The Stealth promos (available to almost everybody) are the best for spies you intend to keep in enemy territory for very long, it gives a mild increase to mission success, but really helps with not getting caught when just being in enemy territory. Taking at least the first one is a good idea. The ratio of espionage spending you and the other civ has made against each other is also an important factor, if you've outspent them considerably you're spies will be safer and vice-versa if you are being outspent. Oh, and watch out for Atredies territory . I really need to get around to updating the DuneWars concepts page.

I'm pretty sure it actually looks first for the one with the highest value (without asking if the unit is able to do it), THEN asks if it can be done, and if it can't, the spy does nothing. So this probably happens alot...

Click to expand...

Ahh... DOH!

That would explain a lot.

Ok, it sounds like the first thing that is needed here is a total rebuilding of the espionage AI code; then worry later about tweaking the parameter values.

while decreasing interception chance would make espionage impossible for a human

Click to expand...

I'm not sure I understand this though.

I don't mind high detection chance when actually in an enemy city, but it seems too high a chance of detection when just wandering through enemy territory.
Is it possible to differentiate these probabilities in the code? Or is the detection chance just linked to the cities?

In practice, this makes AI stacks with very many AA defenders almost un-assaultable with air units

Click to expand...

This was the design intention, and the interception chances were re-designed around this. Air units have been very, very powerful through most of the mod history. They can instantly relocate across the map, and they are devastating against anything that doesn't have air protection.
Though most of the mod's history, bombers have been absolutely devastating against the AI, basically an "I win" button. Adopting Afforress's modcomp was a very deliberate way to make aircraft less of total domination units and more tactical, for damaging units that aren't in the stack, and for attacking desert units. For city assaults and big stacks, you should use siege units and fight them on the ground.

I think it is entirely appropriate that 5 missile troopers give an 83% interception chance of anything attacking that stack (they don't protect other nearby areas, just the one stack). [At least for the first one; each unit should still only get 1 successful interception per turn.]1

I've changed the math so that additional interceptor units still add to the interception rate, but not at an exponential rate as he did it. I won't explain the math, just give an example, using the same situation as before. The three rocket troopers end up with a 42% interception chance, five rocket troopers would return a 52% interception chance

Click to expand...

This feels far too low to me. Please explain the math, or revert to the old system.
It should not be a good idea to use aircraft against a stack with 5 AA units; it should be near-suicidal. Otherwise air units are too strong and AA units are too weak.
Remember; AA have to be *everywhere* you want protected, air units can choose to attack *anywhere* that isn't protected.
We also deliberately had ground-based AA as the answer, because the AI is better at using this than it is actual interceptors.

what if it gives 1 happiness in cities with imperial instead

Click to expand...

I'd prefer to keep the religions and civics separate.

As a side note, another AI project I want to do is tie the AI decision to bring in reinforcements to city danger. There is already a city AI function called AI_panic() that changes the city to producing defenders and hurries them if possible when under imminent danger, I'd like to tie it in right there.

Click to expand...

Sounds good. AFAIK, the AI is not good at using the homeworld units at all.
Could we make sure that there is an animation for the homeworld units arriving then? I can imagine a human player to be frustrated if they show up at a city and all of a sudden there are units there, and they don't know where they came from. So it would be excellent if the human player could observe some sign that the AI used Homeworld.

Yeah, it also goes back to the previous discussions on AI versus human ability to amass units given respective empire and economy size. I like the balance early on, but the inflation help later on (which is much less necessary now for them to keep up) was really skewing it too much against the human.

A real issue here is even though the AI does need extra units, the power rating is an important diplomacy and warplan indicator and is human blind, so if the human is too disadvantaged in keeping up in the arms race, he gets dog-piled more. Before at the monarch/emperor level, the human had fewer units but was consistently at higher techs, so had a higher power rating based on the more advanced units. That's less the case now, so keeping a sufficient power rating requires more parity of unit levels than before, and the AI can compete better at closer unit levels with more advanced units.

Click to expand...

This sounds sensible.

To clarify a bit more; I think that groups of ground-based units should have a high interception chance, but not necessarily a high chance of killing the unit. A bomber attacking a stack with lots of AA troopers should nearly always get intercepted and being significantly damaged, but it should only have a moderate chance of dying unless the missile trooper has promotions vs hornets.

I'm not sure I understand this though.

I don't mind high detection chance when actually in an enemy city, but it seems too high a chance of detection when just wandering through enemy territory.
Is it possible to differentiate these probabilities in the code? Or is the detection chance just linked to the cities?

Click to expand...

Doh, you read it before I fixed the typo--I of course meant it'd be too easy for humans with lower detection rates. I want you to have to take promotions to really feel very safe in enemy territory. It actually is much lower moving around than doing a mission, it's divided by 10 if I recall correctly, vanilla code. We could try dividing by 12 maybe, I'd be ok with that. And it doesn't so much care if you are in a city as much as a city's BFC, as any working plots for a city get the same bonus from espionage defense/counter spies as the city plot itself, which I think is fine. Remember, not all missions occur in cities.

This feels far too low to me. Please explain the math, or revert to the old system.
It should not be a good idea to use aircraft against a stack with 5 AA units; it should be near-suicidal. Otherwise air units are too strong and AA units are too weak.
Remember; AA have to be *everywhere* you want protected, air units can choose to attack *anywhere* that isn't protected.
We also deliberately had ground-based AA as the answer, because the AI is better at using this than it is actual interceptors.

Click to expand...

To clarify a bit more; I think that groups of ground-based units should have a high interception chance, but not necessarily a high chance of killing the unit. A bomber attacking a stack with lots of AA troopers should nearly always get intercepted and being significantly damaged, but it should only have a moderate chance of dying unless the missile trooper has promotions vs hornets.

Click to expand...

It's a good argument, though it does needs to be more transparent to the player that attacking the stack with lots of AA is not the best idea, which adding in the plot help should help with.

I'd actually rather support individual units having a little higher base evasion chance that increases with technology, and a smaller synergy between additional ones on a tile. This will make more sense to a player and help the AI more because they don't need as many to get decent air cover.

**Math Stuff Below**

• You might want to skip it for casual readers

The code I added also has great flexibility in how much it clamps the additional effect of extra interceptor units (just to clarify, it uses interceptor aircraft within range and AA in the tile, no changes there from Afforess's code). It starts with the full interception value of the best interceptor for the plot. Afforess's original code multiplies the non-intercept chance for each unit (for 30%, 1-.3 = .7) that qualifies as an interceptor for the tile (so 3 rocket troopers is a (.7^3 = 34.3%) non-intercept chance or 65.7% intercept chance.

All I did was take the intercept chance of each additional interceptor after the best interceptor unit and apply a power function to it before multiplying the non-intercept chance by it's inverse:

(1.0 - (30%^2 = 9%)) = .91 instead of the .7 that would have been multiplied

If I use a power of 1.5 instead, the 3 trooper example returns:

(1.0 - (30%^1.5 = 16.4%) = .83 (approx.)
(1.0 - (.7*(.83^2)) = 52% interception chance

If the rocket troopers base interception becomes 40%, then three of them return 66%. If Mongoose troopers (who should be more effective at this kind of thing) have a 50% chance, then three give you a 79% chance.

To mirror the technological progress on the ground, the dragonfly bomber should probably have a small 10% evasion chance (effectively multiplies the plot interception chance by its inverse, .9) so that they'll have a small increase in effeciency of bombing runs against more advanced enemy AA than using the antiquated firefly, and of course the superlight promotion would really represent a major technological breakthrough helping evasion chances.

Changing the power to be closer to 1, like say 1.3, would bring it even closer to the original function, while where the power = 1.0, we get the original.

**Done with the Math**

The only reason I'm really pressing this is I think the rate of increase from additional units feels too heavy handed, and I am probably not the only player who will think that's the case. So I agree with the mechanic, just not the exact application of it. Given this plan, we end up with similar numbers in the end (by increasing the base numbers), but with a little technological progression (which will make sense to players) and less 'punishing' feel to it (since this will more often benefit the AI).

but it should only have a moderate chance of dying unless the missile trooper has promotions vs hornets.

Click to expand...

This is a good point, I'll have to run some spread sheet analysis of different scenarios between different possible combat strength ratios, and maybe test in game with a little world builder help.

And it doesn't so much care if you are in a city as much as a city's BFC

Click to expand...

Ah, I was wondering if that was the case. Ok. Changing 10 to 12 seems reasonable. Its fine to be dangerous to sit in the city (so it isn't a no brainer to wait several turns to reduce the mission cost), but I think I could be able to get a spy to a city without too much difficulty.

though it does needs to be more transparent to the player that attacking the stack with lots of AA is not the best idea, which adding in the plot help should help with.

Click to expand...

I'm always in favor of more transparency.

I'd actually rather support individual units having a little higher base evasion chance that increases with technology, and a smaller synergy between additional ones on a tile.

Click to expand...

I disagree. Technology should primarily increase the strength of the AA units (and so the damage they do when they intercept), and it does this now as higher tech units have higher stength. You get higher interception chances in late game because you have more units, and so there are more AA units on the tile.
I see no reason for reducing the synergy of multiple units on a tile. The game functions in stacks; the AI works in stacks, stack effects are what matter, and a stack with lots of AA defense should be pretty suicidal to attack. You should be rewarded for bringing along serious AA protection.

**Math Stuff Below**

Click to expand...

Ok, simple enough. How about we use a coefficient of 1.3. So, 3 missile troopers at 30% have a 56.2% chance and 5 have a 72.6% chance (down from 65.7% and 83%, respectively with coefficient of 1).

The only reason I'm really pressing this is I think the rate of increase from additional units feels too heavy handed

Click to expand...

I think it should feel heavy handed. Attacking 5 AA units with aircraft should be damn near suicidal. You shouldn't want to do it. If this isn't the case, then air units (especially bombers) will always be able to dominate any stack, and so it will always be fairly trivial to decimate any enemy stack from the air and then finish them up with even technologically inferior ground units.
All of the playtest feedback we received and my own personal experience was that aircraft were dramatically too strong, until we implemented this change, and after that I have never seen any feedback that aircraft were too weak.

The only function of missile troopers is as stack guards against vehicles (and even then they tend to lose, for equivalent tech) and aircraft. So they need to be good at it.

Remember btw that suspensor units have a lower intercept chance than missile troopers do. So air units are still very nice as desert superiority weapons, but they should be highly ineffective at attacking a well-protected stack on land. That's just not what they're for. They have to have a weakness, otherwise they are unambiguously a better investment than building a big ground army.

This is a good point, I'll have to run some spread sheet analysis of different scenarios between different possible combat strength ratios, and maybe test in game with a little world builder help.

Click to expand...

This is where tech should come in; if you sent a basic hornet against mongoose troopers, it should be toast. But not because of a high intercept chance, but because of the strength differential.

Alright, we'll keep the intercept rate constant, and I'll check on the current probability outcomes for different air vs. ground unit scenarios.

Right now, intercept chance still plays a role in the damage/round of the attacker and defender in interceptions. It was of course the only thing effecting damage in the original BTS code for air-air and air-ground interceptions. BBAI/RevDCM added in a better air-air combat system, upon which I've based both scenarios now which kept it as a factor, so I've kept it so far, but I'm not convinced it's a good idea.

I thik would rather base it straight on current combat strength just like land combat for air-ground, as it just muddies the probabilities with something not intended to represent a combat strength (I'm rather certain it was used for connivence since only one side ever did damage, scales damage oddly when interception rates go up with technology like in vanilla too), while leaving it in as a factor for air-air since it represents to a degree the superiority of fighters in aerial combat over bombers.

And just to clarify on that, right now, an AA unit that takes interception increasing promotions will inflict more damage on an attacking plane than an unprompted unit. Another possible oddity, depending on how you look at it, is it means fighters will inflict more damage to intercepting ground units and take less damage from them then a bomber with identical strength otherwise. So the question is should either of these situations be valid, or should the interception rate not matter when air-ground fights occur?

Ok, simple enough. How about we use a coefficient of 1.3. So, 3 missile troopers at 30% have a 56.2% chance and 5 have a 72.6% chance (down from 65.7% and 83%, respectively with coefficient of 1).

Click to expand...

Done.

Its fine to be dangerous to sit in the city (so it isn't a no brainer to wait several turns to reduce the mission cost), but I think I could be able to get a spy to a city without too much difficulty.

Click to expand...

Well, sitting in the city is that same as sitting next to the city in terms of danger, though anything outside of a BFC is probably safer in the late game when espionage defense buildings become more prevalent.

I thik would rather base it straight on current combat strength just like land combat for air-ground, as it just muddies the probabilities with something not intended to represent a combat strength (I'm rather certain it was used for connivence since only one side ever did damage, scales damage oddly when interception rates go up with technology like in vanilla too), while leaving it in as a factor for air-air since it represents to a degree the superiority of fighters in aerial combat over bombers.

Click to expand...

I agree with this. Intercept chance should affect the chance to intercept; damage should be determined by relative strengths, with all appropriate modifiers. If I get an interception promotion on my fighter, that should not also boost its damage; I should need a strength promotion to boost damage.

Sounds good to me, the only balance item in there is that removing this from air-air will make fighters less effective against bombers than what we are used to seeing, and increasing their general air combat strength has other side effects like increasing their damage in airstrikes, neither of which I think we want to do.

An alternative I'm leaning towards is to check if the 'attacking' and 'defending' unit each have an intercept chance. In the case of bomber being intercepted by a fighter, give an advantage to the fighter equivalent to the current situation ignoring promotions, so it's just a constant. I agree intercept promotions shouldn't affect the fighter's fighting ability, but I don't want to through off the existing balance between fight-bomber air to air combat, this will accomplish that.

Ahriman, I put together a spread sheet calculating combat odds for air-ground fights (fortunately much easier to calculate than land battles since there are no first strikes), it was actually a kind of fun excercise (now you all know I'm crazy ). There are too many possible scenarios for me to give a cross section of them, so I'm attaching it if you want to look at some different scenarios yourself.

I found that in addition to dividing the attackers damage to the defender by 2 as I'd done, reducing the interceptors damage slightly was necessary to keep the odds of death within a reasonable limit across different scenarios (within the 9 max round window that's coded), we want this to be less lethal on average than air-air combat. This is done with the InterceptorDivisor set to 1.25 and is the best variable I think for balancing this.

The first image below is an example of an unpromoted Mongoose trooper versus Dragonfly(or Cielago) bomber calculation. After that you've got the same units, except the modifier if the mongoose trooper has combat1 and both AA missile promotions. The third being the promoted Mongoose trooper if on mesa or rough terrain.

It just takes the base strength and modifier for each unit. The 'Odds of Death' should be self-explanatory, and Avg Damage is a rough average assuming the unit survives the fight (so never reaches 100). The reference chart is just there to make it easy to remember the base combat and (for AA) hornet combat% modifiers without having to pull up the xml. It also assumes both units are at 100 HP.







Unless I've really missed something, the math is all solid, though ultimately we'll have to make sure in game results match the predicted values.

The function seems really nonlinear; 1.5:1 odds has 7% chance of death, 2:1 odds has 38.54% chance of death, 2.25:1 odds has 72.97% chance of death?
That seems very extreme.
I wonder if a more appropriate function might be ~10% chance of death at equal odds, 20% chance of death at 1.5:1 odds, 50% chance of death at 2:1 odds, 80% chance of death at 2.5:1 odds.

How much control do we have over the function, how many degrees of freedom?

That's Civ4 combat odds for you, the only real difference between this and the normal land combat is that there is a max rounds, no first strikes, and we're clamping the damage in a way that gives an artificial advantage to the interceptor and reduces the chance of anyone dying in the slotted number of rounds. If you set both divisors to 1 and rewrote it without an artificial max rounds, it would be almost exactly the same as land combat.

Now, that said, we aren't limited to using the same exact math. The biggest reason you see this kind of non-linearity is that as the combat ratio between the units moves from 50:50 towards one of the unit's favor, both the odds and damage/round move in his favor. So not only does the disadvantaged unit have less chance to win a round, he does less damage when he does, while the advantaged unit is both more likely to win each round and does more damage when he does.

The other half of this puzzle, especially in land combat where the rounds aren't limited (so someone always 'wins') is that not only does the advantaged unit do more damage more often, but he also has to win less rounds. An interesting oddity is that at 50:50, both units have a 50% chance of winning. As soon as it's 51:49, the advantaged unit's chance of winning jumps up around 60%. This is because at 50:50, both deal 20 damage, so need 5 rounds, but at 51:49 the damage is 21:19, so despite how small a difference this is, it means in effect the advantaged unit only needs to win 5 rounds still, while the disadvantaged one now needs 6 rounds.

What we can do is hold either iOdds or Damage/Round constant (or just less dependant) regardless of the ratio between the units and only effect one and see what we get. Looking at the examples above, the sudden jump between the second and third ones is because the Interceptors damage moved up just enough to bring the needed rounds down one, which has a BIG impact on the odds, like what I just described in the land combat case. If we make the interceptor damage always 15, and just let the combat ratio affect the odds of getting hits, we'll have a much smoother transition.