Game feels completely different once again. ^^

Where do you see these Hull Weakness, Accuracy and Evasion bonuses? They don't seem to be listed in the defense descriptions...

The current system is producing results that are not within the devs plans. They can't be, they are too chaotic and result in several different wrong end stats (anything that results in truly indestructible ships is wrong).



I'd also wish the devs were more open, we'll see what happens.

Umm so flak gives most reliable universal defense by lowering hull weakness... What would be the point of taking shield if it increases my hull weakness? Full hp heal is fun but only if your ship lives to see the next round.



Gah I wish devs were more clear on the base mechanics and the design theory behind what they are trying to implement.

I actually thought it was hilariously funny (in a good way).

But, I think having any multiplicative bonuses on stackable modules is very dangerous.



First, a reminder about Hull Weakness. Hull weakness bonus. Hull Weakness is bad, we want low Hull Weakness.



damage taken = Hull weakness/(hull weakness + Defense)



So with 100 Hull Weakness and 100,200,300 Defense, we get damage taken scores of: 50%, 33%, 25% (i.e., 1/2, 1/3, 1/4).

However, with 200 Hull Weakness and 100,200,300 Defense, we get damage taken scores of 66%, 50%, 40% (i.e., 2/3, 2/4, 2/5)

So Hull Weakness reduces the effect of defenses.



Looking at this from the angle of Effective HP (EHP) makes this really clear. A ship with 100 Hull Defense has 100% more EHP per 100 units of Defense. A ship with 200 Hull Defense has 100% more EHP per 200! units of Defense. It effectively halves are EHP gain.



Now, Missile Defense 3 has a -4% Hullweakness bonus, Lase Defense 3 has a +15%. A ship with 10 of each of these defense modules, would have a Hull Weakness modifier of +110% (15*10 + -4*10). With a base ship Hull Weakness, our ship would have a Hull Weakness of 210 ((100%+110%) * 100). That would mean our ships defense is half as effective as they would be without the Hull Weakness modifiers.



Now, how can we make this hilarious? Well, by reducing Hull Weakness to 0.



Damage taken = 0/(0+Defense) = 0



(If you have 0 defense and 0 hull weakness, I'm not sure what happens 0/(0+0). It may cause your computer to explode)



So we make our ship gain infinite damage reduction. We can do that with 25 level 3 Missile defenses (100/4).



Now, the next problem is that it isn't JUST Missile Defense that causes a problem, nor is it that it is only negative Hull Weakness modifers are the issues. All passive culmulative multiplicative modifers and all defenses are potential problems.



What happens if you take defenses as a set, one of each type? Well, let's assume a ship with 500 HP, 100 Hull Weakness and adding sets of level 3 Defense modules (1 of each).









As you can see, we'll never get more than 6000 EHP because of how Hull Weakness is increasing with the addition of everyone one of those level 3 Shields. Now, this isn't to say that survivability isn't changing. In addition to increases in EHP, we also (now) have changes to shots deflected, absorption, InterceptionAccuracy, DeflectionPerTurn, and evasion.



That makes the problem complicated. I need to get access to the remainder of the combat mechanics to make any real guesses on how things play out.



Now, I might have convinced you about the dangers of Hull Weakness modifier stacking, but surely others aren't issues you say.

Sorry, they are. Let's look at the deflector, level 3, +3% Evasion Bonus & -9% Accuracy Bonus. Let's take our ship from above and give him 50% accuracy and 50% evasion and see what happens.





Now, given my understanding of how evasion interacts with evade, that's not a big deal. Sufficiently many shots will mean you just dodge of couple of the first ones, and get plastered by the rest. The accuracy though, that's going to cause problems. Of course, with sufficient shots, the EvadeDisorientation might even catch up to your really poor accuracy.



But you say, "Fine, Fine, so those are bad. But Healing, healing is good isn't it?"

I've not actually played around with the healing mechanics yet, but I think I can do a quick job.

Two bonus, BattleHealingPercent & BattleHealing. My guess is that BattleHealingPercent directly increases your ships HP after every battle round and BattleHealingPercent boosts your repair modules healing rate per round of battle. For the level 3 Shield, the bonus is only 3% for each of these. So what does this mean, well, if I'm right, this means with 33 Level 3 Shield, you'll completely heal all damage every combat round.



The point I'm trying to make is not that these are broken (they are, but that's not important). The point is that the bonuses are stacking in ways that weren't intended. Plus Math is fun.



(Also, the advanced posting interface in Chrome is useless compared to what you can do in Firefox on this forum.)

While this fixes invulnerability, I think that this effect still is quite overpowered. Especially since it kinda works the opposite way from diminishing returns. And then it suddenly stops. Adding caps imho always simply is poor design from a mathematical point of view. There's always a better formula that accomplishing the same without using these last-resort-techniques!



In this case it should be calculated like this:



Initial-Hull-Weakness*0.97^Number of Modules



So for 33 Modules it would be 100*0.97^33=36.6,

which still is well above 25% and for no amount of Modules could ever reach 0.

Just updated my post ^^"



In ShipDescriptor.xml

Easily fixed by what? I need to mod it in order to continue my Let's Play.

Indeed, it's a stupid mistake I did; by stacking 33+ Flak (and without shield) , you're able to reach a null Hullweakness, making your ship invincible. It can easily been fixed by adding a Min Value to the HullWeaknessBonus



[CODE][/CODE]



I've been focused on other datas and missed that! It will be fix before the public release of the current beta patch of course =)

thuvian wrote:
I finally got around to working on HullWeakness and it is indeed hilarious.

Let me present to you, the PTH Untouchable. Using Disharmony 1.1.14, I guarantee complete combat satisfaction regardless of the odds.







[Note:Youcandoasimilarabuseusing1.1.9]

[Note:Stackingshieldsandarmorisevenmorehilarious.]

What exactly are you trying to say with this? Does this ship avoid all damage? And if so, how? Does it come from one of the silly side-effects?

Meedoc commented on the Accuracy mechanic in another thread. I fear this entire thread would have been better served in a more popular forum. This one is pretty dead as it is.

Thread Reference Link: /#/endless-space/forum/27-general/thread/10388-evasiondisorientation

Meedoc wrote:


It's a hidden variable used during battle, which reduces Evade of a ship within a round for each avoided salvo. So, the more salvo is missing, the higher the chance is to hit.

So for instance, if a ship has 50% evade, and you shoot at it 4 salvos:

Beginning of the round.

The first Salvo misses: Evade is reduce by the value of disorientation and becomes 35%.

The second and third Salvo hit

The fourth misses: Evade is reduce by the value of disorientation and becomes 20%

End of the round: Evade is reset to 50%

So this addresses the questions of:

When does evasion count reset? Every round (12 times per combat)

How much evasion is "lost" per hit? EvasionDisorientation value. In 1.1.9 it was .05 (i.e., 5%), in 1.1.14 it is 0.15.



This does mean all my previous math was off. However, the conclusions are the same, we just need to scale it by 0.15 instead of accuracy per successful evade.

I finally got around to working on HullWeakness and it is indeed hilarious.

Let me present to you, the PTH Untouchable. Using Disharmony 1.1.14, I guarantee complete combat satisfaction regardless of the odds.







[Note:Youcandoasimilarabuseusing1.1.9]

[Note:Stackingshieldsandarmorisevenmorehilarious.]

Scrangos wrote:
If your using a single melee kinetics against 1cp, it should be hitting one in 20 (I think theres a 95% cap, dont recall where i read that though). It does have a 75% dmg penalty at LR, but if its hitting every time in melee thats pretty weird and would disprove the theory since it should only rarely hit.

I think my tests actually used 2 weapons modules rather then 1. that said, having learned a lot in this thread I think I'm going to rerun the tests and see what comes up- will post my results when I do. Ok reran the tests, here are the results:



My fleet has a single ship with 2 level 2 melee kinetics modules commanded by a pilot with Lethal Modder. I observed when both modules hit the enemies I'm fighting die in a single salvo- when one hits they lose about 60% of life.



Battle 1: long range phase: all 4 salvos miss with both weapons Medium Range: all 4 salvos miss both modules Melee phase: first 2 salvos hit with both modules, second 2 miss with both modules



Battle 2: long range phase: all 4 salvos miss with both weapons Medium Range: all 4 salvos miss both modules Melee phase: first salvo kills the only enemy ending the battle



Battle 3: long range phase: Enemy retreats- all three attacks before the battle ends miss with both modules



Battle 4: long range phase: first two salvos miss with both modules, 2nd two salvos both his with one module Medium Range: first two salvos miss with both modules, 3rd salvo hits with a single module, 4th salvo hits with both modules- all enemies destroyed and battle ends before melee phase begins



Battle 5: long range phase: Enemy retreats- all three salvos with with both modules



Battle 6: long range phase: first and 4th salvo miss with both weapons. 2nd & 3rd salvo both hit with both modulesMedium Range: first salvo hits with 1 module- my fleet is destroyed and battle ends.



SO totaling up these results:



Long Range phase: out of 22 salvos 18 miss with both modules, 2 hit with 1 module, and 2 hit with both modules

Medium Range phase: out of 13 salvos 10 miss with both modules, 2 hit with 1 module, and 1 hit with both modules

Melee Range phase: out of 5 salvos 2 miss with both modules, and 3 hit with both modules





Sample size is smaller then I'd like for the melee phase. That said, long and medium range seem pretty similar- so It could be my melee range results and my first round of tests were just improbable outliers.

Accuracy



Accuracy as implemented as shown by testing and reported formula seems to be:



Accuracy*(1+successful evades) - evasion (reset each round)



What this exactly means is hard to understand and describe. In the following I try to break it down base on my readings in the forum and wiki, the game files, and game play experience.



Let's start with a simplistic example. We'll ignore ranges, combat rounds, salvos, cards, empire bonuses, and hero bonuses; basically everything. We are also assuming there are not hard caps on accuracy and evasion.

Our example fight will be two destroyers with level 1 kinetic melee (50% accuracy) and 50% evasion.

The chance of hitting will be as follows. For a single level 1 kinetic melee weapon.



Accuracy*(1+successful evades) - evasion (reset each round)

50% *(1 + 0) - 50% = 0%



This should always result in a miss.





1 weapon = 1st shot = 0% chance of hit = 50%(1+0) - 50% = 0%

2 weapons = 1st shot = 0%, see above

2nd shot = 50% = 50%(1+1) - 50%

3 weapons = 1st shot = 0%, see above

2nd shot = 50% = 50%(1+1) - 50%, see above

3rd shot = IFF the 2nd shot was a hit, 50% (same as above), if the 2nd shot was a miss, 100% = 50%(1+2)-50%

4 weapons = see above for 3 weapons

4th shot = IFF the 2nd and 3rd shots were both hits, 50%, if the either was a miss, 100% = 50%(1+2)-50%



In this example, you always miss shot 1, you have a 50% chance of missing shot 2 and all succeeding shots, until you miss once more, in which case you will always hit after that.

In summary, for 50% accuracy and 50% evasion

1 weapon will always miss

2 weapons will hit once, 50% of the time.

3 weapons will hit once, 50% of the time and twice 50% of the time.

4 weapons will hit twice 50% of the time, hit 3 times 25% of the time, and 4 times 25% of the time.



With a 50% accuracy weapon versus a 50% evasion ship.

Your average damage will then be: (Number_of_weapons - 1.5)*weapon_damage.

Your minimum damage will be: (Number_of_weapons - 2)*weapon_damage.

Your maximum damage will be: (Number_of_weapons - 1)*weapon_damage.



At this point, you might ask, what about salvos? Well, from my testing, it seems that the game treats salvos as just a weapon damage modifier. Each weapon "shot" is 1 shot, regardless of the number of shots per salvo. So a level 1 kinetic weapon with 150 shots per salvo shoots once, likewise level 3 missile with 2 shots per salvo shoots once, etc. Accuracy hits are all or nothing. Either all 150 shots in a salvo "hit" or they "miss". However, weapons do retain the reload & fire statuses, so a short range weapon that reloads at speed 0 will fire 4 shots during each distance phase. In one dev post it was stated that it takes 1 "round" to reaquire a target after destroying that target, I can't confirm or deny this. However, I have noticed that ships can change targets between distance phases.



Okay, so we have an estimate of how accuracy works now. Accuracy and evasion seem pretty... unimportant so far. This is true as long as you are dealing with simple situations. What happens if you mix it up.



Well, with high accuracy weapons (such as the 100% accuracy long range missile) versus about anything you'll get similar sorts of math. There is a chance of a miss, but after that first miss the others are virtually guaranteed to hit without some massive evasion modifiers.



What about the other extreme then poor accuracy weapons?

Let's look at what happens with 30% accuracy weapons (short range kinetics) versus a 50% evasion destroyer.



Accuracy*(1+successful evades) - evasion (reset each round)



0.30 * (1 + evades) - .50



I'm skipping the math and showing the results.



Summary of Table

Shot # (Shot)- The particular shot in a sequence

Chance to Hit (Chance) - The chance for this particular shot to hit

Cumulative Average Hits (CAH) - The total number of hits thus far

Number of Evades (Evade) - Number of times the ship previously evaded a shot



Shot Evade Chance CAH

1 0 0% 100% 0

2 1 10% 10% 1, 90% 0

3 1 10% 10% 1, 90% 0 (happens 10% of the time)

3 2 40% 40% 1, 60% 0 (happens 90% of the time)

3 1|2 10%/40% 1% 2, 45% 1, 54% 0



And you can see that with the next round, we are up to 2x2x2 = 8 outcomes

Rather that work that out, let's look at the worst case scenario.

The worst case scenario is just missing until you get to 100% accuracy.



The formula

1 = 0.30(1+evades)- 0.50

1.50 = 0.30(1+evades)

5 = 1 + evades

4 = evades



What happens when we evade 4 times, well

0.30 * (1+4)- 0.50

1.5 - 0.50

100% accuracy.



With a 30% accuracy weapon versus a 50% evasion ship.

After the 4th miss, we hit the rest of the time. Therefore:

Your minimum damage will be: (Number_of_weapons - 4)*weapon_damage.

Your maximum damage will be: (Number_of_weapons - 1)*weapon_damage.



What about even more extreme situations, such as with a +100% Evasion Hero & +50% Evasion Battle Card?

Well, I don't know how the numbers are added, it could be multiplicative or additive, or some combination.

My guess, based on looking at the files, is that such a combination would result in +140% of base ship evasion.

We'll do the usual, assume the worst case scenario and see what happens.



Let's look at what happens with 30% accuracy weapons (short range kinetics) versus a 200% evasion destroyer (i.e., +300% evasion).

Skipping the math, how many misses before we start hitting at 100% accuracy...

0.30 * (1 + evade) - 2 = 9 evades.



With a 30% accuracy weapon versus a 200% evasion ship.

After the 9th miss, we hit the rest of the time. Therefore:

Your minimum damage will be: (Number_of_weapons - 9)*weapon_damage.

Your maximum damage will be: (Number_of_weapons - 6)*weapon_damage.





So what does it all mean?

In the worst case scenario example up to 9 shots could miss in a very extreme situation. However, after that everything will hit. Dev posts has stated that the number of evades is stored at the individual ship level and refreshed between rounds. That suggests that two ships shooting at the same target will hit more often than two ships shooting at different targets. However, in large scale combats where ships have many weapons and low evasion scores, you won't notice a difference. Realistically only destroyers and corvettes with their 50% base evasion rate will show much of an effect of evasion, the bigger ships are less dodgy and benefit less from stacked evasion bonuses. Little ships will also have less hit points and be much more likely to suffer from overkill, hence they will dodge more shots but still end up dead. In the end, Accuracy and Evasion contribute relatively little to the combat and should not be a majory factor in your battle strategy. Regardless, I find myself seeing the AI use the Defense Card Camouflage all the time and habitually nullify both with a Sabotage Card.





So why worry about accuracy? As a player, it makes little predictable impact. You can't really change it very much. However, my interest is in trying to produce a more effective targeting algorithm. Currently the "best" choice is spread fire, which is inefficient at best. A better approach would be a more intelligent concentrated fire that targets weaker ships.



My current idea of an algorithm.

1. Calculate Effective HP (EHP) of each ship. Each ship will have 3 EHP scores based on which defense is being targeted.

2. Calculate Effective Damage (ED) of each ship using accuracy, range, phase, etc.

3. Create the Defensive Target list by sorting each fleet based on targeting rules:

3a For Target the "Most Dangerous" sort by ED/EHP

3b For Target the "Weakest" sort by EHP

3. Create the Offensive List by sorting each fleet by the ED

4. Take the lowest ED ship (worst damage), compare it to the EHP of the enemy ships.

5. Have that ship target the ship highest on the Defensive Target list that it can kill, and update that ships EHP accordingly.

6. If this ship cannot kill any ship, target the top ship on the Defensive Target list

7. Repeat steps 4-7 for each ship



This would be much more effective than our current targeting system. However, it has the problem that it would destroy any point in arranging ships in defensive formations. The current system was devised to have somewhat meaningful ship formations (evidence is in the presence of the protect siege ships formation). It was an attempt to abstract out combat which includes support, offense and defense ships. You could imagine this sort of situation if both fleets had very poor sensors and always stayed in one big group. You wouldn't know which ships to shoot at, so you shoot at either: Nose-breaker: the first one, Guillotine: the first three, or Spread-fire: All of them equally.



In retrospect that's not particularly good. Why on earth would you shoot the ship that your opponent wants you to shoot first (Nose-breaker)? Why on earth would you shoot the top three ships that your opponent wants you to shoot first (Guillotine)? Shooting everyone equally is a decent option, although it leads to mixes where your big damage ship picks a little freighter to shoot while your little destroyer shoots the massive dreadnaught. Ideally you want to optimize fire.



However, optimizing fire means that you need to optimize fleet arrangements, which means a major revamp to the current system engagement rules. Especially since system engagement takes place in real time. The solution that SotS implemented was to have movement occur at the end of the turn and that all combat occurs during the start of the turn. You'd still need to work on formations. It seems that the devs had thought that there would be mixed fleets of siege instead of the battle fleets and support fleets that game play currently endorses. We've also got problems where each ship only attacks once per round. I also don't know if there is anyway to specify which fleet you are targeting, the best I figure is the top one is the current one to engage. A couple of times I've ended up with my siege fleet being engaged by an attack fleet, while my own attack fleet just sits around and watches the slaughter.



So in conclusion, I'm not sure.

and now they change everything. The evasion malus does sound interesting...

If your using a single melee kinetics against 1cp, it should be hitting one in 20 (I think theres a 95% cap, dont recall where i read that though). It does have a 75% dmg penalty at LR, but if its hitting every time in melee thats pretty weird and would disprove the theory since it should only rarely hit.

As an addendum to this, I tried to do some further testing. If you save before a fight and reload, if the same cards are used, the exact same outcome happens. This occurs even if you delay the fight by a turn. I believe that the randomseed is somehow saved in the gamestate and this makes testing awkward. I believe that if you have a fight it "uses" some of the random numbers, but I'd need more testing to evaluate this. However, this means that if you find the same outcome to a fight happening several times in a row (e.g., running a fight and finding 0 damage occurs at long range using short range kinetics), it is not because of the mechanics, but because you are getting the same random numbers.

A portion from WeaponModule.xml

[CODE]












CriticMultiplier="2.0" CriticChance="0.0"

InterceptionEvasion="0"

NumberPerSalve="30"

TurnBeforeReach="0" TurnToReload="0" Accuracy="0.5">

Kinetic

ShortRange



[/CODE]



I believe you are referring to Accuracy = 0.50 ? I believe that is the absolute accuracy of the weapon. This is also the number that appears in the ship builder. Multipliers from say the scout module, from SupportModuleDescriptor.xml



[CODE]

















[/CODE]

I believe this is a multiplier to accuracy. The wiki notes there are two type of accuracy multipliers, this would be one of them.

A portion of code from BattleDescriptor.xml

[CODE]



























....

[/CODE]



I believe that addresses your question?

Scrangos wrote:
Theres actually 4 rounds per turn, and it resets each round, but it builds (on the target) up per ship (one ship can have as many salvos as it has weapons).

I'm aware there are multiple rounds per phase. I was however mistaken on the term salvo- I was interpretting salvo as all of a ships attacks attacks per round- my mistake.

Like he said the weapon range things are just damage modifier, calling using the word accuracy anywhere is just confusing people. It used to be accuracy penalties in vanilla afaik. Yes its gonna miss most of the time since it cant stack accuracy. And long range anything (including kineticS) dont get any penalties at long range.





The battlephase penalties are purely damage multipliers, has nothing to do with accuracy.

My observations have been otherwise. Again when I load a single ship with a single kinetic module(with a lethal Modder hero so damage is noticable) it misses just about every time at long range(doing no damage) and hits just about every time in the melee phase.



My sample size so far was only about 4 battles, but in light of this disagreement I may expand that number.











On another entirely different note: anyone have any idea how the accuracy mods on the actual weapon modules(the ones that vary based on whether you use the long, medium, or melee range version of the module) work? Unlike every other accuracy effect in the game they are decimals ranging from .3 to 1.2 rather then percentile bonuses. Perhaps static multipliers to accuracy at some stage of the calculation? Well any insight would be useful.