Fixing combat mechanics

[Alpha Centauri Bear]

Fellow mod lovers, I am ready to fix combat mechanics in my The Will to Power mod. However, I'd like to discuss it with others to pick better implementation. Therefore, separate topic.

First let me review what is wrong with vanilla implementation even though this was discussed like thousand times already.

1. Combat round odds are NOT as shown on TV (on odds display dialog).
This is regular bug that needs to be fixed. Oddly enough the wrong formula has been propagated here from Civ1, thought Civ2, and then SMACX without modifications. Even though right formula is as easy to implement. Either no one it all three teams can compute probabilities or maybe nobody just cared.

2. Introduction of multi round combat and HP/healing in Civ2 broke combat balance.

In Civ1 unit combat effectiveness was simply ratio of corresponding combat strength (attack/defense) to cost. Meaning if one side places fortified Phalanxes in plain field (cost 2, defense 3, effectiveness = 3/2) and other side sends Legions (cost 2, attack 3, effectiveness = 3/2) to crush them - both sides incur equal losses in unit cost and the side with higher production power will advance. It wouldn't matter if attacking side sends Catapults in battle instead (cost 4, attack 6, effectiveness = 6/4 = 3/2) - the effect would be same.
Overall strategy revolved around three things:
1) Increase sheer empire production power.
2) Discover more effective units. Chariot effectiveness = 4/4 = 1. Armor effectiveness = 10/8 = 1.25. Armor is slightly more effective than Chariot.
3) Use units properly to increase their effectiveness.
From protection side: use defense bonuses of fortification, terrain, structures, etc. Prepare defense on anticipated attack directions. Build fortress and defend key spots: hills and mountains as well as just important resources.
From invasion side: research defense pattern and attack weakest spot to pay as little units as possible for captured bases.
#3 was actually most strategical fun. Imagine fighting two enemies from opposite sides. Without enough power to wipe them both at once one could seal one border shut with limited number of well placed defenders. Then throw the rest of units against other enemy. Makes you feel a strategical mastermind.

Now let's see what it turned into in Civ2. Multi round combat skews winning odds toward strongest side very much. In Civ1 combat with 3:2 odds is won 60% of times. Meaning 40% of times stronger unit just dies. In Civ2 it is won 80% of times (94% with infamous combat probability bug O_O). Meaning stronger unit almost never dies. The mere loss of HPs can be healed to restore unit to full capacity = zero loss for stronger unit owner. In this game sending Catapults and not Legions against Phalanxes makes a whole lot of difference! This introduced an indestructible army phenomena. Large enough group of strong enough attackers captures city without losses -> heals in captured city -> repeats. With reinforcement it gets bigger and snowballs turning into absolutely unstoppable swarm. I recall one Civ 2 game when I captured whole world with couple of Frigates carrying like 5-6 Crusaders. As if this is not enough, SMACX emphasized weapon even stronger turning conquest into easiest and fastest victory among others by far. People have to actually willingly stop the conquest to enjoy ascent to transcendence. Otherwise, it is plainly not achievable. Sad story.

Back to our bug fixing business.
#1 above should be fixed easy-peasy.
#2 is not so straightforward. I consider few solutions here.

Solution 1
Adjust individual round odds so that cumulative multi round combat odds are as advertised. I am listing this one just for the sake of completeness. I don't believe it is feasible. Too complex math and too much of fine tuning prone to errors. Besides, it would be exceptionally difficult to explain players how it works.

Solution 2
Use one round combat. One unit lives another dies. This immediately restores proportional losses. Now one does not need concentrated fire to be able to kick out opponent units from well defended base one by one. Attacker still incur quite large losses due to defense bonuses but the chance to kill a defender with a single unit now is not zero anymore. Helps AI who doesn't know how what concentrated fire is. This also completely eliminates indestructible units of all kinds. For example it is not possible to hunt natives with a single Scout Patrol forever. With 3:2 odds it is bound to kill 1-2 natives on average and then die. Now player would need to think whether hunting is worth the cost building new unit.
Cool stuff is that bombardment and collateral damage still deliver partial damage. Thus effectively and proportionally reduces winning chance. Which is an intended design.

Solution 2a
Even if bombardment and collateral damage deliver partially damage they are not that often in game. Especially not in early game. Ignoring healing would be undesirable loss in functionality. So I was thinking to amend above solution with random damage to winner. That does force winner to retract for healing or continue with reduced winning chance. Player's choice. Strictly speaking that slightly tilt combat balance toward loser as throwing damaged winner into further battle reduces this unit effectiveness. However, this also benefits base defenders. They do not need to go anywhere to heal and they are healing faster. Attackers in turn need to rotate for healing or otherwise sacrifice themselves for cheap.

Solution 2b
Same as above but fixed damage instead of random. Probably somewhere around 50% configurable. Benefit is in adjustable configuration. The effect can be toned up or down.

Both 2a and 2b above have nice tactical twist in them: player need to decide whether to waste time healing unit to full strength or continue with lower success chances.
No more stupid and 100% safe needlejet raids against formers for mere purpose of free promotions. Now needlejet may just DIE. With above rules there won't be stupid free attacks anymore. One needs to weighs cost/benefits for EVERY attack.
 

[bvanevery]

No more stupid and 100% safe needlejet raids against formers for mere purpose of free promotions. Now needlejet may just DIE.

But that is not a good goal.  It's completely nonsensical.  A Former should just die when attacked by a Needlejet.  If you don't like that players can destroy Formers with Needlejets, I have to ask why.  Going after enemy infrastructure is very basic to warfare.  If you think a Needlejet promotes too easily when going after a weak unit, then you should change the promotion formula somehow.  Personally I think that is a non-problem, I do not see the big deal.

I really hate game designers who implement "It is always possible to fail, and it is always possible to succeed".  I just got done lambasting the designer of King of Dragon Pass and Six Ages because of that.  It introduces nonsensicality and gaslighting into a game.  You never know what reality actually is.  Did you win because you applied a skillful insight, or because you got lucky?  Ditto why did you lose?  Especially when the game doesn't even tell you how anything works, when it is formula opaque, it's severely annoying.  So annoying I'm not playing those 2 games now.

As for other things... who is "combat balance" for?  Is it for multiple human players trying to kill each other?  Is it for a single human player trying to beat AIs?  Those are really different requirements.  Much depends on what the AI is capable of doing, and how capable you are of tweaking the AI's behavior.  You wonder why buggy formulas didn't get changed.  But for all you know, you could fall down a rabbit hole where the AI suddenly doesn't work very well anymore, if you change it.

AI does actually mass units against me all the time.  I rely on splash damage to deal with it.  Don't assume your personal play style is the only way people fight in SMAC.

Some games, I do actually use artillery on targets.  If you don't, that's your bias of play style.  Sometimes artillery is the best option available.  Sometimes options are limited.

Winning, but getting badly wounded, isn't necessarily victory.  You could lose everything on a counterattack.  Sending units back to be healed somewhere is also not a completely free lunch.  You are losing time and momentum.  There's a big difference between taking a base and not quite taking a base.

I don't know how your Scouts are surviving Psi combat without taking wounds, unless they've leveled to Elite or something.  I find that Psi combat usually chews up the participants.  If you think it's too easy for a unit to be good at Psi combat, you can't fix that by changing your effectiveness odds around.  It doesn't take any weapons to do Psi combat.  That's the point, that's a basic part of the game design of SMAC.  The only production decision is whether someone is going to make another Scout, which costs nothing.

In the past I recall conversations where you were really against Planet Pearl collection.  You went to some length to change combat odds around, in ways that I proved didn't work.  It just changed the best way to game the fungus fights.  Defense became better than offense.  Consequently after several rounds of experimentation, I gave up on your proposed changes for that.  If you really hate Planet Pearls, I suggest you change the amount of wealth they produce, and not how the combat works with mindworms.

[dino]

I'm not sure it's in need of further fixing.
As far I'm concerned it's already fixed in your mod with reactor, collateral changes and increased defense ratings and the current system fixes "destroying an ironclad with a trireme" issue.
The changes proposed here have less naunce than original rules and the main advantage seem to be an easy formula to calculate accurate odds.

It's very hard to gain more than one tier of tech advantage and without multiround combat, tech advantage would be almost diminished, since 3 weaker units could be as effective as 2 stronger ones.

As for odds, just Att*HP/Def*HP while inaccurate, is enough of estimation for decission making.
Especially if player understands, that the further from 50%, the more uctual odds differ from estimation and lets be real, your mod won't be played by not experienced SMAC players.

If you really wanted, to calculate odds correctly for the purpose of presenting them to the player, you could with a tree algorithm.
Generate a probability tree, calculate probabilty of reaching each leaf and add them up. It's a second semester in computer science kind of excersise, but I don't think it'd improve the game experience much.

As for issue #2

With current system, you can achieve everything you want just by reducing numeric differences in units strenght: slightly smaller percentage bonuses for morale, base defenses and smaller differences between weapon/armor ratings and lets say 4/3 Psi odds, would bring actual odds of destroying a unit, closer to what you want, while keeping all the naunce of the current system.
I'm not sure, it's necessary beyond increasing armor values closer to weapon, which you already did, though.

[Alpha Centauri Bear]

First of all, as I said, everybody has different vision on what game they want to play. Therefore the discussion.

No more stupid and 100% safe needlejet raids against formers for mere purpose of free promotions. Now needlejet may just DIE.

But that is not a good goal.  It's completely nonsensical.  A Former should just die when attacked by a Needlejet. 

If you are judging game sense from real life point of view you should be in "Phalanx cannot beat the Tank" discussion. I am not pro or against it - I just don't care about real life mechanics implementation in games.

If you don't like that players can destroy Formers with Needlejets, I have to ask why.  Going after enemy infrastructure is very basic to warfare.  If you think a Needlejet promotes too easily when going after a weak unit, then you should change the promotion formula somehow.  Personally I think that is a non-problem, I do not see the big deal.

I really hate game designers who implement "It is always possible to fail, and it is always possible to succeed".  I just got done lambasting the designer of King of Dragon Pass and Six Ages because of that.  It introduces nonsensicality and gaslighting into a game.  You never know what reality actually is.  Did you win because you applied a skillful insight, or because you got lucky?  Ditto why did you lose?  Especially when the game doesn't even tell you how anything works, when it is formula opaque, it's severely annoying.  So annoying I'm not playing those 2 games now.

I never said I want to prohibit player doing things. On the contrary, I want to enrich their options. Going for enemy infrastructure is an essential part of this game which I very like.

"It is always possible to fail, and it is always possible to succeed". This is called randomization and it is an cornerstone part of all games in the world. What is wrong with it? It is still possible for needlejet to die attacking former in vanilla. I am just proposing different probability distribution. This change is not that significant either. It does not turns odds upside down. I was jokingly exaggerating the danger a little. In vanilla needlejet change to die is about 0.0001%. In my version it'll be ~10%. Noticeable but still not that much to discourage purposeful former hunting.

I don't care about promotion in this particular topic. I care about impunity. That adds mouse clicks without much thinking. Why just not have a special button in vanilla "automatically destroy all enemy formers in the 10 square radius around your bases"?

As for other things... who is "combat balance" for? Is it for multiple human players trying to kill each other?  Is it for a single human player trying to beat AIs?  Those are really different requirements.  Much depends on what the AI is capable of doing, and how capable you are of tweaking the AI's behavior.  You wonder why buggy formulas didn't get changed.  But for all you know, you could fall down a rabbit hole where the AI suddenly doesn't work very well anymore, if you change it.

Aid AI. Now even single unit attacking a stack is not futile.
Remove some more cakewalks for human for more challenge and less thoughtless mouse click.

AI does actually mass units against me all the time.  I rely on splash damage to deal with it.  Don't assume your personal play style is the only way people fight in SMAC.

How can I not assume it? I am playing it myself, you know. I am completely aware my changes may be disgusted by some part of humankind. Do you think I should stop doing this then?
You among all others have an opportunity to introduce your own vision into my changes. I believe this is quite democratic and the best I could do.

Some games, I do actually use artillery on targets.  If you don't, that's your bias of play style.  Sometimes artillery is the best option available.  Sometimes options are limited.

Never said I don't. Quite the opposite.

Winning, but getting badly wounded, isn't necessarily victory.  You could lose everything on a counterattack.  Sending units back to be healed somewhere is also not a completely free lunch.  You are losing time and momentum.  There's a big difference between taking a base and not quite taking a base.

I don't know how your Scouts are surviving Psi combat without taking wounds, unless they've leveled to Elite or something.  I find that Psi combat usually chews up the participants.  If you think it's too easy for a unit to be good at Psi combat, you can't fix that by changing your effectiveness odds around.  It doesn't take any weapons to do Psi combat.  That's the point, that's a basic part of the game design of SMAC.  The only production decision is whether someone is going to make another Scout, which costs nothing.

Yes, yes, any change in mechanics change the strategy and the play stile to some extent. I am aware of it. I do want to change it. The question is how and to what extent.

In the past I recall conversations where you were really against Planet Pearl collection.  You went to some length to change combat odds around, in ways that I proved didn't work.  It just changed the best way to game the fungus fights.  Defense became better than offense.  Consequently after several rounds of experimentation, I gave up on your proposed changes for that.  If you really hate Planet Pearls, I suggest you change the amount of wealth they produce, and not how the combat works with mindworms.

Yep. I did. Then I cooled down about this topic. Partially due to your arguments and testing as well. Thank you for contribution! I appreciate it.

[bvanevery]

What is wrong with it?

Probabilities should have a rational window of applicability.  If you've got way more stuff than you need, you should always succeed.  If you've brought way too little stuff with you, you should always fail.  In your alternate universe, if I sent a 3 year old bare handed child against a Marine with a M-16, you'd allow that child to win 10% of the time, 'cuz Random.  That's completely stupid.

It is still possible for needlejet to die attacking former in vanilla.

Screenshot please.  I don't believe you.  An unarmored Former has a -50% non-combat unit penalty.  I think you are getting confused about Air Superiority ground attack penalty for Interceptors.  A plain Needlejet without Air Superiority and a reasonable weapon on it, should always blow away a Former, even if it is sitting on Rocks and Fungus.  I think even a strength 2 Laser would work, but it should be certain with an Impact weapon.  Now if you don't train your Needlejets at all, and you are taking MORALE penalties from Wealth etc., and your faction has a MORALE penalty as well, maybe a Very Green Needlejet could fail.  I haven't tested that, because it's a pretty contrived situation.

I don't think Air Superiority rules are all that great BTW.  Unnecessarily tweaky.  I think they wanted to make a big difference between a Fighter and a Bomber.

In vanilla needlejet change to die is about 0.0001%. In my version it'll be ~10%.

Which is completely stupid.  Yes I think phalanxes killing tanks is completely stupid too.  Drives me up the wall.  I wouldn't play your mod with a system that stupid.

I don't care about promotion in this particular topic. I care about impunity. That adds mouse clicks without much thinking. Why just not have a special button in vanilla "automatically destroy all enemy formers in the 10 square radius around your bases"?

Needlejets have to be manufactured.  If the player is able and chooses to manufacture them, then they are allowed and should be allowed to use them effectively.  What are you trying to do, rubberband the player so they can't succeed?  If you think it's too easy to manufacture a Needlejet, then make them more expensive.  Making them nonsensically unreliable is lazy ass game design.  You as a designer think there's a problem with the player's resources, so you've said ok, let's just throw the player's resources away for no logically defensible reason.  Just because you haven't actually figured out how to fix the problem in the complex production system.

How can I not assume it? I am playing it myself, you know.

Because other people tell you what's up.  Because the mechanics exist in the original game, regardless of whether you use them frequently or not.

I am completely aware my changes may be disgusted by some part of humankind. Do you think I should stop doing this then?

You should respect how a lot of people actually play the game, and not seek to undermine systems that actually work.  Where that line is drawn is a matter of judgment, but dismissing fundamental mechanics as rare and irrelevant, shows a serious bias in your play style.  The antidote for that is to get feedback and also playtest.

Never said I don't. Quite the opposite.

Your concerns about splash damage are most confusing then.

Yes, yes, any change in mechanics change the strategy and the play stile to some extent. I am aware of it. I do want to change it. The question is how and to what extent.

Think hard about what it means to debuff Scout Psi combat.  You could end up turning the mindworm into the One True Unit of Death.

Yep. I did. Then I cooled down about this topic. Partially due to your arguments and testing as well. Thank you for contribution! I appreciate it.

Happy to help.

[Alpha Centauri Bear]

Thanks for thoughts, guys.
I am going to fix #1 anyway, since it is a bug.

I am still thinking about singe round vs. multi round combat, though. Not entirely convinced I want to implement it.

I agree with dino on research advantage emphasis thing. Although, I do not agree that is very hard to get more than 1 level ahead, though. Just mere inherent faction RESEARCH bonuses makes a difference between UoP and Believers as 1.2 / 0.8 = 1.5 advantage! Add to this different starting conditions, different neighbors composition, different land masses, etc. Even in relatively early game I often see strength 5 weapon against strength 2 armor. That is so big advantage even PD cannot fix. And this is just random fluctuation that leads to a faction extermination. It is like somebody rolled dies and whole faction is wiped out. It doesn't add more challenge to the game.

It would be nice if research advantage, economical development, politics, combat tactics, strategical unit placement - all of these contributed to the winning. In vanilla, unfortunately, only better weapon rules the world. This is in big part due to multi round combat. I am all for it but it seems to be adding too much skew.


Another problem is that such deterministic combat outcome make difficult to work with different bonuses. They become either absolutely decisively important or not important at all. Both extremities are not good. Example, weapon 4 against base defenders with armor 2. Without PD defenders have intrinsic base defense of 50% resulting in 4/3 attackers odds = ~90% winning chance. With such odds they take base almost without losses. Now we build sensor and PD for combined bonus of 100% + 25% turning attackers odds to 4/5 = ~10% winning chance. Boom! Now base is unpenetrable. Of course, I am exaggerating here. You still can take it with artillery and lots of units but you got the point. It is too much rounds in multi round combat that adds that instability. The problem is not that you just need to make all bonuses smaller. The problem is that previously with highly randomized outcome each bonus added to your success is a simple well understood linear fashion. Whereas, with highly deterministic outcome they are either not relevant at all or crucially important. And you cannot understand which is which without CALCULATOR! That is what I don't like.

I agree that single round combat may undermine technology advantage. Maybe we can find a compromise of how many rounds we should use? Like more than 1 but less then 10?

[dino]

Example, weapon 4 against base defenders with armor 2. Without PD defenders have intrinsic base defense of 50% resulting in 4/3 attackers odds = ~90% winning chance. With such odds they take base almost without losses. Now we build sensor and PD for combined bonus of 100% + 25% turning attackers odds to 4/5 = ~10% winning chance. Boom!

I've never actually do the math, if it's this bad, then you have a point, something should be done with it.

For starters I'd make units in a base with command center/naval yard/aerospace complex heal at double rate, not to 100% in a single turn.
That, combined with no splash damage from loosing a unit, should assure looses on both sides during prolonged combat situation involving constant streams of reinforcements.

[Alpha Centauri Bear]

Fine fine, guys. You'll have you multi round combat and impunity to kill formers. Sheesh. It's not that it is too important for me anyway.


I still plan to tone it a little. Or maybe just expose some switches in configuration so you can play with it and fine a best combination.

[Alpha Centauri Bear]

Example, weapon 4 against base defenders with armor 2. Without PD defenders have intrinsic base defense of 50% resulting in 4/3 attackers odds = ~90% winning chance. With such odds they take base almost without losses. Now we build sensor and PD for combined bonus of 100% + 25% turning attackers odds to 4/5 = ~10% winning chance. Boom!

I've never actually do the math, if it's this bad, then you have a point, something should be done with it.

Yep. Unfortunately. I also got your point about killing Ironclad with Trireme. So I agree not to revert to plain Civ1 battle odds. However, it may make sense to find some gold middle. You see roughly each combat adds a power on top of single round odds. So 4:3 round odds turns into 4:3 ^ 2 = 16:9 for 2 round combat and into 4:3 ^ 10 = 1048576/59049 = 18:1 for 10 round combat. 10 round combat is just way too much of emphasis. I am thinking something like 3 rounds or so. That still favors stronger unit and research advantage. For example, it turns 5:1 odds into 15:1 allowing killing much weaker units with impunity. However, it goes easier on more comparable strengths. Like 3:2 psi combat now would resolve with 2:1 success ratio instead of 4:1 in vanilla.

For starters I'd make units in a base with command center/naval yard/aerospace complex heal at double rate, not to 100% in a single turn.
That, combined with no splash damage from loosing a unit, should assure looses on both sides during prolonged combat situation involving constant streams of reinforcements.

Interesting. You are reading my thoughts. I also think 100% healing rate is too much. I didn't even plan to touch it any time soon, though. However, now I'll put it on my todo list.

[bvanevery]

I think what you should actually do, is make some conservative changes and then get others to playtest.  You are chasing a big web of numbers around and it's easy to ruin stuff that way.

I think a major problem with SMAC's combat system, is one unit has to die.  There are some exceptions: artillery, land vs. sea artillery duels, and disengagement of a Speeder or fast ship.  The emphasis on decisive outcomes makes the combat pretty unstable.

[Alpha Centauri Bear]

I think I'll do exactly that. Starting from conservative and then exposing switches for more disputable changes.

[Alpha Centauri Bear]

Some numbers for reference on what I am talking about. Hope I can bore you with them.

Numeric odds. Initials and for each N round combat.
Chart for the same. Cut some skyrocketed odds. I don't think anyone would bother for anything beyond 10:1. It is almost certain win.

It is educational to see how 2:1 odds got magnified by the number of rounds.

[bvanevery]

BTW my informal point of certainty, as reported by the current Odds Checker which I have on all the time, is 3:2.  Given such reported odds, I expect to win.

4:3, I might not win, but I'm likely to fight anyways.

[Alpha Centauri Bear]

You gave me idea, guys. I think I'll calculate and display exact battle winning odds. That would be my gift to everyone. No more needs for "inaccurate, but enough of estimation for decision making".
 

[bvanevery]

I think that would be helpful.  Especially since the game is gaslighting me about some of those odds.