Is defense overpowered?

[marcelo r. r.]

Any buff to atack is just walk in circles:
Germany AI will stomp everything again and win eastern front in 42.

 

[homerCCCP]

I feel like we aren't playing the same game. I have never played a single play-through where "manpower losses seem within believable range." Check out the screenshots here. This occurs every game I play. As you can see, as France, I captured Berlin and Rome easily because of the problem described in this thread. German casualties from France: 3.2 million. French casualties from Germany: 110k. Just about 30 to 1.

Had this number been 10 to 1, I would have suffered 300k casualties, which would have made me change my conscription law with valuable early game PP, and would have resulted in considerable equipment loses, reducing the number of divisions I could field.

I want the game to be extremely challenging in single player. I remember Hoi2-- it was difficult to win as France, and you certainly weren't capturing Rome and Berlin in 1940.

If the devs made attacks cause more manpower and equipment losses, would the single player experience improve? This is the question I'm hoping to will attract further debate. I believe that it might. I also don't see how making the change will hurt multi-player. If anyone has any specific reasons why the change would damage the multi-player experience, I'd appreciate hearing these reasons.

View attachment 957633View attachment 957634View attachment 957636

I see the issue and I agree.
However to achieve 30:1 you had much better templates and air superiority I guess, while AI was blindly attacking with worse and worse units. It would just feel bad if such behaviour would not result in high losses.
It would be nice to see how the loss ratio behaves in less min/maxed scenarios. Wouldn't take bad AI behavior as the baseline.

I like the idea in this thread of having two layers of units the AI uses, defensive and offensive ones. However, for sure it would only solve part of the problem.

And I know it is a big issue that turtling is just the outright best strategy for this game, as it leads players to avoid a lot of stuff this game actually has to offer. Anyway, I just don't see making stupid attacks as the solution, sry. This will just shift the game to other cheesing strategies...

 

[Ilyasviel]

Any buff to atack is just walk in circles:
Germany AI will stomp everything again and win eastern front in 42.

Germany should be able to stomp everything, to some degree. The USSR took 4.5m casualties in only 6 months during Operation Barbarossa. The main problem with the eastern front is how ahistorically weak the USSR is, and how they have 0 ability to recover after taking losses. In hoi4, it's easy to completely beat the USSR while inflicting less than 4.5m casualties on them in the entire war.

 

[American Traveler]

I see the issue and I agree.
However to achieve 30:1 you had much better templates and air superiority I guess, while AI was blindly attacking with worse and worse units. It would just feel bad if such behaviour would not result in high losses.
It would be nice to see how the loss ratio behaves in less min/maxed scenarios. Wouldn't take bad AI behavior as the baseline.

I like the idea in this thread of having two layers of units the AI uses, defensive and offensive ones. However, for sure it would only solve part of the problem.

And I know it is a big issue that turtling is just the outright best strategy for this game, as it leads players to avoid a lot of stuff this game actually has to offer. Anyway, I just don't see making stupid attacks as the solution, sry. This will just shift the game to other cheesing strategies...

I think stupid attacks should be punished at 10 to 1 instead of 30 to 1. I think this would improve the game.

To reply to your comment about base line data: This was a France game and I didn't join the British and so the British provided no air force because a grant of military access for some reason doesn't allow fighter plans or bombers to be based on the host country, just troops. Point is, I did not have aerial superiority.

I loaded up a previous save to take some screen shots to show the ratio at a different point in time when Germany had massive aerial supremacy. As you can see, it's completely red air. Germany has thousands of planes in the zone of contact and I have none.

Still, the kill ratio is 28 to 1 even with completely red air in the only zone of contact. (They were not attacking the maginot, just attacking over the Rhine, and in that zone there were no allied planes.)

Again, I don't argue that the AI is the bigger problem, but shouldn't this ratio be smaller? The historical question is-- if the Germans in WW1 or WW2 had launched exhaustive, poorly planned, frontal attacks on the western front, with complete aerial supremacy, would they have suffered 28 to 1 casualties?

Surely the answer is no. So given this answer, shouldn't the game punish horrible offensives at a ratio less than 30 to 1 or 28 to 1 ? Fixing the AI is another matter altogether. But shouldn't stupid, headlong assaults where one modern army attacks another modern army trade with the enemy at closer to 7 to 1 or 10 to 1?? I believe the answer is yes, and so I'm in favor of tweaking the game for attackers to increase the damage to strength ratio while leaving the damage to organisation ratio the same.

 

[Corpse Fool]

but shouldn't this ratio be smaller?

Based on what? The survivorship bias of historical battles?

 

[American Traveler]

Based on what? The survivorship bias of historical battles?

Just to be clear, are you saying that you believe this result, 28 to 1 casualties, with air superiority, is a realistic outcome?

On the first page I referenced the battle of the Somme, where the casualties didn't exceed 2 to 1. I don't know of any examples of a 30 to 1 ratio outside of colonial wars. In Napoleonic wars, WW1 and WW2, a 2 to 1 outcome was a rout.

I do believe a 7 to 1 or 10 to 1 ratio could be observed based on how insanely dumb the AI behaves in this game, but 28 to 1 or higher is too much, would you not agree?

 

[bitmode]

Going from 30 kills to 1 dead, to 120 kills and 4 dead, is still a 30:1 ratio. Increasing strength damage suffered per hit doesn't really 'solve the problem' of getting ratio'd. If anything, that being coupled with the worsened damage modifier the more damage you take means you'd probably be shifting through the brackets faster and having an even worse ratio.

I think that such loss rates for equipment would mean that we'd have to be focusing that much more on efficiently managing production. You wouldn't be able to get away with some of the less efficient options anymore. Though I guess this would mean that hospitals and maintenance companies would be more popular?

I'd also like to question the manner in which most of the historical examples of attacks are being performed, and how those compare to the conditions in game. People know that you would like to have at least a 3:1 advantage when you're attacking, but the width limitations and the cheapness of infantry makes that very difficult to accomplish.

The strength damage would hardly change across the board but mostly situationally.

Let me try it like this: by having certain strength and org damage dice, the game makes assumptions about the relation of the two; somewhat loosely based on historical observations.
If a unit took material losses of a certain size, an opponent of average ability was able to make them collapse, i.e. bring their organization to zero. "Average ability" here means that across all the different variables like force overmatch, air/naval support, cryptology, unit experience etc. they were doing okay.

If the losing side across the board is doing far from okay, the generally assumed relation of strength and org damage increasingly don't hold anymore. As a very basic example, if the same amount of strength damage is spread out over ten times the duration, it will likely affect org less, right?

Two existing examples in the game are Last Stand and the increased org dice for unbreakable armor. Both of these recognize that this is not your regular battle anymore and change the relation between strength and org. Stack enough negative modifiers like the AI loves to do and I'd say you end up in a similar situation. Someone upthread said soldiers would shoot their officers (= org damage) before following the AI's attack order. So as an implementation example, what if the attacking side gets the following scaling modifier if their stats get reduced below a certain threshold:

contemplating mutiny:
-30% strength damage received
+40% org damage received

 

[Cavalry]

Simple thing and often suggest idea would be that AI makes good templates.

Actually AI templates was set by human (modder), and template is not the cause of very high kill loss other than infantry die more than tank. The high ratio was caused by a 20w division continued to attack 80w fortified line.

One quick fix is use a mod like EAI to set AI infantry and tank template at 40w. That ensure AI always have at least 40w on attack and 40w in defense, if they have trouble stay and fight together as a mass.

Or simply add 1 tank battalion to AI infantry can subtantial lower the casualties.

 

[Corpse Fool]

Just to be clear, are you saying that you believe this result, 28 to 1 casualties, with air superiority, is a realistic outcome?

I'm not going to say its 'realistic', because the conditions which are producing that ratio in game are also pretty unrealistic. The reason I'm not really comfortable putting a cap on how extreme this can be is the same reason I mentioned survivorship bias. We aren't getting the casualty ratios of all the battles that didn't happen, because the people involved knew how absolutely insane it would be to attack under a variety of unfavourable conditions. The AI is all too happy to commit to a full front offensive infested with unfavourable conditions, and so I don't know if we could reliably measure how bad things really could be getting.

I don't see any armoured divisions in the enemy stacks, so presumably they've been sending what is essentially infantry at you. Judging by how low their strength is across the entire line, they've also been doing that across the entire line. A line situated on either side of "a big river". And as much as they are enjoying what is essentially uncontested air superiority... they aren't running CAS against you? Knowing how much the AI likes to partition their army groups, how many of those divisions are even operating under a cohesive chain of command? What do the designs of their templates even look like, how is their training level, are they allocating their factories correctly?

There's just... so much the AI is probably doing wrong that an officer worth even half their salt wouldn't be doing. If we had controllers of a more or less 'equal' and considerably higher intelligence than the AI, I wonder what the ratio would look like then? Or lets try to find a historical example that more closely matches the conditions being presented in the game, and see what sort of ratios they get?

I don't think the somme was a good example. Because while it did have some pretty horrific attrition rates... it could be argued that it wasn't a mistake. The things that result in the AI suffering 30 to 1, is the AI making mistake after mistake after mistake.

As a very basic example, if the same amount of strength damage is spread out over ten times the duration, it will likely affect org less, right?

This is why I still think the idea of a small amount of recovery applying inside combat would probably at the very least be interesting to play around with.

My mind goes to the way that naval damage perhaps works, where org damage is modified by your current/desired level of strength. A unit with more strength will lose less org, a unit with very little strength would de-org rapidly.

But I can already see some problems with the AI with either of those ideas. With the first, they would probably leave the low strength divisions in the combat and just grind them into pieces when if they left, a fresh reserve could take their place and they could go replenish. The second, is the problem the AI has with constantly forcing low strength templates against the enemy, when it hasn't worked the last hundred times they got sent.

 

[The_Tim]

This is why I still think the idea of a small amount of recovery applying inside combat would probably at the very least be interesting to play around with.

My mind goes to the way that naval damage perhaps works, where org damage is modified by your current/desired level of strength. A unit with more strength will lose less org, a unit with very little strength would de-org rapidly.

But I can already see some problems with the AI with either of those ideas. With the first, they would probably leave the low strength divisions in the combat and just grind them into pieces when if they left, a fresh reserve could take their place and they could go replenish. The second, is the problem the AI has with constantly forcing low strength templates against the enemy, when it hasn't worked the last hundred times they got sent.

Something that could be done with this idea could be this:

Have it so that if a Division is forced out via Org Damage, in essence shattered, scattered, and driven from the field that it SHOULD only occur after the division sustains large amounts of Strength Damage, but in return a division that's fighting to the last point of org should do so for longer than some cases in the game(for instance, attacking a heavy SA defensive line can have more defensive line infantry org break within hours)

However, add in a army-level setting, call it "Determination" or something of the sort that'd automatically have divisions disengage once they reach a threshold of org, as in IRL it was quite rare for a division to fight to the point of being basically shattered when under attack. This should make it so that the AI wouldn't fight until the bitter end, and therefore needlessly waste large amounts of manpower and material in combat

 

[HugsAndSnuggles]

I feel like we aren't playing the same game. I have never played a single play-through where "manpower losses seem within believable range." Check out the screenshots here. This occurs every game I play. As you can see, as France, I captured Berlin and Rome easily because of the problem described in this thread. German casualties from France: 3.2 million. French casualties from Germany: 110k. Just about 30 to 1.

You mean by grinding AI until it ran out of manpower or equipment (something that I specifically mentioned)? That is not a damage problem, it's AI problem (something that has been pointed out repeatedly since game launch, which was replied by the devs along the lines of "it's better than WW1 stalemate"). While there were improvements over the years (it used to be possible to throw Italians out of Ethiopia with two volunteer divisions, for example), AI is still hardly in the good place when it comes to division control.

 

[Kanitatlan]

I remember Hoi2-- it was difficult to win as France

I think it is important to remember the big change in combat from HOI1-3 to HOI4. Prior to HOI4 the optimal tactical behaviour was to attack with all of your divisions all of the time because attack was actually stronger than defence. It wasn't until HOI4 that I ever fought a defensive battle. This is why it was hard to win as a country on the strategic defensive in previous HOI versions. In HOI4 Paradox solved the problem of the AI actually being able to launch any sort of proper offensive even against another AI. Now we have defensive bonuses that make defending a suitable strategy.

But what I believe to be the root of the problem as you describe things... is that the AI is bad.

This is the core of the whole issue. The AI makes bad decisions and ultimately the issue being discussed in this thread could only be fully solved by developing a much more effective battle AI. This is likely to be extremely challenging.

Personally I would prefer it if Paradox began to address the AI issues by some of the following

• The AI frequently attacks with significantly under-strength divisions. This could be addressed either by having the AI more sensible evaluate their effectiveness (and not use them) or have understrength divisions have their combat width reduced in line with their level of understrength
• Change the over width penalties to be a diminishing returns effect rather than be actually weaker than fewer divisions. This would make it so that the AI need not be too concerned about combat width
• I would quite like the combat evaluation to be changed along lines I have proposed in the past which would make combat width of templates have no impact on combat (ie the same subunits would evaluate the same whatever divisional structure you built them into)
• Implement some sort of reserves feature for the defensive AI. Defence desperately needs more options. Currently the defensive behaviour is very Hitler/Stalin stand your ground no matter what.
• Some improvements in the offensive AI around the balance between advancing and redeploying troops. The current behaviour seems to be strongly driven by the full set of enemy units that the AI's force is in contact with. The effect of this can be beautifully demonstrated by the tactic of retreating from a breakthrough. If the AI breaks your line and injects significant troops then its behaviour is dramatically different if you rebuild your defensive one line of provinces back (ie out of contact) instead of trying to block the next line of provinces. Some time ago I had an amazingly successful defensive after the AI pushed back a section of line weakly held by mobile troops. I pulled back an extra province to the rear and the entire offensive against that part of the line evaporated and I was able to successfully counter-attack with just the forces that had just be pushed back. How nonsensical is that. There needs to be change in this area based on varying the way parts of the line are evaluated. I'm not sure quite what to propose but this is because I really don't know what the current algorithms are. This could do with discussion by player ignorance about the algorithms tends to make it peter out very quickly.

 

[HugsAndSnuggles]

I think it is important to remember the big change in combat from HOI1-3 to HOI4. Prior to HOI4 the optimal tactical behaviour was to attack with all of your divisions all of the time because attack was actually stronger than defence. It wasn't until HOI4 that I ever fought a defensive battle. This is why it was hard to win as a country on the strategic defensive in previous HOI versions. In HOI4 Paradox solved the problem of the AI actually being able to launch any sort of proper offensive even against another AI. Now we have defensive bonuses that make defending a suitable strategy.

Really? I was under the impression that "planning bonus" is the main new thing. Before we had entrenchment and terrain adjusters, but nothing to help the offensive in a similar way (unless you count deathstacking your entire army in a single tile, but combat width got rid of that). Or do you simply mean that defensive bonuses can now be stacked to a ridiculous degree?

 

[Herennius]

i guess one might start by fixing the battleplanner. that would first require acknowledging just how truly bad is has been broken, since 2016. but it's possible to do. it's meme tier terrible, and since the ai uses it, the ai is by extension as well. even if you use more complex plans than the ai uses, it doesn't work properly on multiple levels. i could go into another big rant about it, but i think for the moment is suffices to say that the battleplanner is completely pathetic and altering mechanics because it doesn't work is absolutely the wrong way to go about improving hoi 4

This. At least for me personally the battle planer with all of its issues contributes a lot to the impression the OP describes.

The AI is nothing else than the batte planer active all the time - and even if the battle planner has issues in every combat situation, they are far more pronounced when attacking. Defending is mainly holding a position, being passive, denying the enemy from achieving something, keeping the status quo. Even losses can be taken, as long of those of the aggressor are significantly higher. So regardless whether we talk about about AI-AI, AI. vs human or even MP (at least I have that theory/feeling, can't speak from own experience for the latter) - as long as the battle planner rules everything, defense has inevitably an extra advantage just because of it being a factor. Yes, the AI distributing divisions evenly and becoming vulnerable for a breakthrough at one point and then getting potentially encircled is a flaw - but it outright sacrificing manpower in constant fruitless attacks without achieving anything is worse by magnitude, as will sooner or later lead to even the better slightly working defensive aspect breaking down (because of lacking manpower and equipment weakening the units)

Ironically though - at least for me - even taking out the battle planner as factor on my side (which I consequently do) leads to the same situation because of the UI. Managing huge frontlines and division numbers without the battleplanner is a time consuming and stressy thing - and I just feel overwhelmed, when only thinking about having to manually order and manage a huge offensive operation (and yes - I'm not a really good player, what ironically leads to me having tons of fun even with un-modded HoI4... ). OTOH, defending such a huge frontline is still challenge, but feels at least possible. That said - I usually shy away from playing majors for the forementioned reason, but if you only give me the choice between playing Germany and SU in a historic setup, alone because of the UI/managment burden I will take the SU, because I know that my important first goal is mainly defending (the offensive mop-up when suceeding with the former task is another thing, as that operation will be done under much more favourable circumstances then)

 

[Kanitatlan]

Really? I was under the impression that "planning bonus" is the main new thing. Before we had entrenchment and terrain adjusters, but nothing to help the offensive in a similar way (unless you count death stacking your entire army in a single tile, but combat width got rid of that). Or do you simply mean that defensive bonuses can now be stacked to a ridiculous degree?

That's an easy mistake to make. The planning bonus is clearly a key part of the fix. Prior to HOI4 defensive bonuses weren't sufficient to overcome the advantage of making sure all your units were engaged AND there were issues with the multiple combat penalty being inadequate. You ended up with the optimum combat strategy for a theatre was to maximise the combat engagement of all your units whilst trying to minimise the enemy's combat engagement. HOI4 saw increased defensive bonuses balanced by the planning bonus. In HOI1 and HOI2 this was an absolute rule and in HOI3 a bit more subtle but still your best theatre strategy was maximise the ratio of divisional combat days between your units and the enemy's units. HOI4 brought in a situation where a theatre pure defensive strategy was viable.

 

[Gort11]

Speaking of planning bonuses, I reckon they're another thing the AI basically ignores. I frequently see AI Germany just constantly attacking over and over, when they'd do much better to build up their planning bonus between attacks.

 

[PorkchopSandwiches247]

Speaking of planning bonuses, I reckon they're another thing the AI basically ignores. I frequently see AI Germany just constantly attacking over and over, when they'd do much better to build up their planning bonus between attacks.

This is (generally)a good thing for German AI though, because their CAS do stupid amounts of org damage, which lets them push so effectively without giving the enemy time to breath.

I have no idea where Britains airforce ever is. I love being an Ally trying to hold Germany back, and Britains enormous pile of planes is nowhere to be seen.

Also their doctrine gives planning speed bonus. So its actually kind of better for them to attack wothout hitting full planning, because they can deorg opponents and cause them to lose all their entrenchment in retreat while their troops get at least some bonus.

 

[TheMeInTeam]

Speaking of planning bonuses, I reckon they're another thing the AI basically ignores. I frequently see AI Germany just constantly attacking over and over, when they'd do much better to build up their planning bonus between attacks.

ai doesn't ignore it entirely, i will see it sometimes wait then attack with full bonus (unless that's reduced through espionage etc). i think this is a function of that battle plan evaluation thing the game has. if it's green the ai will go regardless of planning bonus, but if it's red it will build up first. something like that. it's rudimentary but it isn't quite as bad as "always attack no matter what".

of course, that battle planner estimate is often bad, the ai does not use the battle planner in a sophisticated way (like trying to spearhead in province from 2 directions on 2 spots on line to fish for a pocket), and the battle planner itself doesn't work well...it breaks if you use it for anything sophisticated.

I have no idea where Britains airforce ever is. I love being an Ally trying to hold Germany back, and Britains enormous pile of planes is nowhere to be seen.

lots of places for ai to put planes, lots of chances to do something stupid.

fighting in allies faction is crappy and unfun for me, even if i'm fighting the axis. thus i wind up just using support anti-air and camouflage expert in most cases. the org and strength damage is not so daunting then.

 

[PorkchopSandwiches247]

I love creating a spearhead in one direction, and then the battle planner will still decide to attack a city over a river 90 degrees to the right of where I want them to head, so my division has huge combat disadvantages. Just exhilarating.

This is even more fun when I choose the ‘safe attack’ option that is only supposed to attack areas with low defensive factors.

 

[TheMeInTeam]

I love creating a spearhead in one direction, and then the battle planner will still decide to attack a city over a river 90 degrees to the right of where I want them to head, so my division has huge combat disadvantages. Just exhilarating.

spearheads will also participate in pinning attacks (which amazingly, somehow, a few here didn't consider bugged despite it being objectively bugged in the english language).

it is trivial to reproduce a bug where front lines to close pocket attack away from pocket (literally 180 degrees!) into provinces that don't have a front line at all! make a pocket with 1 province gap, delete all lines, put line on pocket for some of the troops, and order them to aggressively close the pocket. they will attack provinces that are not part of any order whatsoever. this can be reproduced in '36 starts by simply doing allow diplo and turning off ai.

in that same scenario, if you leave front lines on the non-pocket provinces but don't have any active orders, troops will nevertheless "reposition" through enemy territory to attempt to reach pocket provinces, often attacking 1 division vs several and of course not actually reaching it.

thus, in the same fell swoop, you have units ignoring orders to attack non-existent front lines, and units who have no attack orders nevertheless attacking. and the ai is supposed to work with that! even for the player, it is frequently a detrimental tool that increases inputs and induces an extra penalty to manual micro. you can't even set a front line along with your spearhead for supporting units to follow behind; it's too slow. an ai spearheading would probably just pocket itself, which might be why it doesn't even try them. that or they don't work because spearheads don't spearhead, so nobody bothered to make the ai use something that doesn't work?