HOI4 has completely lost its strategic nature

[Qswhisper]

It's worth pointing out that there are really two different problems identified by your statement.

1) The absolute meta of the game does not reward historical designs.
2) The AI plays in a way that does not fully reward historical designs.

Thanks for the in depth analysis, after thinking about this for a few hours I think the MIO system could potentially improve this in the current situation, as MIO could use a form of tech tree and randomness to achieve a similar gameplay to stellaris upgrading the galactic fleet

On the other hand, I still insist that designers should provide several sets of design templates that would be automatically generated based on tech, military doctrine and resources, and let Ai and new players use them to make the game process more focused on upgrading tech, improving military doctrine and having more resources. This is a relatively low cost method of improvement. Keep players engrossed in macro-strategy rather than micro-design. Improving the sense of immersion. Similar to Sid" Meier's design philosophy "A game is a series of interesting choices."

 

[bitmode]

Because there are literally meme builds from the actual war, there has to be wide latitude with what you can do with a designer in the game.

What can be done is one thing; what the game encourages by making it optimal is another. Gamers will largely do the optimal thing and it is on the game designer to bring that in line with the game's intent (grand strategy) and limitations (AI). Having the possibility for outplay is important but beyond a certain degree it becomes boring. If the AI is not ready for a subsystem, the subsystem is very likely not ready for the game.
The large mismatch in stats between your typical BBA and non-BBA fighter plane should not come down to "well Junkers historically had this prototype once and now here we are".

And then you factor in the fact that the designer has to work within the parameters of the other mechanics. And then you don't have infinite man-hours and money to get the right balance.

Perhaps it's time then to cut down on ambitions?

With the current mechanics, I'm not even sure it is possible to balance the game in a way to make historical designs solid choices in the designer.

True, which means it was premature to introduce those designers. As far as I know, hoi4 was the first version to allow designing divisions from the battalion up. That came after a lot of refinements in division stats and mechanics in previous iterations. I have not played them, but I am sure if hoi2 or hoi3 had had a division designer like this it would have turned out bad for those games.

 

[HugsAndSnuggles]

As far as I know, hoi4 was the first version to allow designing divisions from the battalion up. That came after a lot of refinements in division stats and mechanics in previous iterations. I have not played them, but I am sure if hoi2 or hoi3 had had a division designer like this it would have turned out bad for those games.

Actually, HOI3 had a pretty similar thing:

Although it only operated on brigade (now regimental) level; the way I remember it, resulting problems were exactly the same as with HOI4 designers.

Fact that you can make something as silly as a zero width division should speak for itself:

 

[TheMeInTeam]

Agree, but I was thinking today from the other side, maybe this reduces the difficulty of the game? Makes it easier for more newbies to play hoi4?

i doubt it. from what i can tell, both from games with friends and based on reddit posts etc, the vast majority of hoi 4's learning curve difficulty is understanding how the mechanics and tradeoffs work. this also seems to be what overwhelms new players; they're presented with tons of information on various different screens and need time/effort to process all of that.

when you see players in botched sp situations, the templates are frequently not optimal. however, any experienced player could still dumpster ai with what's there. instead player issues will be things like attacking in bad terrain, issue with supply management, or not understanding basics like attack/def/breakthrough.

i think the window of time where players actually understand how hoi 4 mechanics work, but would still struggle vs ai, is small. even if the ai were improved significantly.

Yeah the IA block u only if u give rules to you self (like make realistic tank/ship/airplane) and not use exploit...or better is only harder but at least u can beat, but is like YOU need limit you self...

Yeah, and...? I've already told Qswhisper that if you are playing meta or using exploits AI can't naturally handle

only functions as an argument if you conclude that playing the game is an exploit, generally. op mentions "meta" tank designs to crush ai. but its unit control is so bad that you can curb stomp majors with basic infantry divisions, sometimes at > 20:1 casualty ratios in your favor. tanks can technically do a little better...but only a little bit, because decent unit control leaves the player already taking very few casualties on offensive operations.

meanwhile, basic aspects of hoi 4's controls still don't work, and we have documentation here that many ai nation will build military factories or convert civs to mils and then not use them at all. it's very clear the ai's performance using the mechanics isn't a priority.

Fact that you can make something as silly as a zero width division should speak for itself:

this was possible for a while in hoi 4 as well, and 0 width support-only divisions were capable of paradrops too. i don't think anybody would be surprised to hear there are issues created by that. at least the stacking penalty prevented just loading up functionally infinite damage in a province before it was just patched out regardless.

 

[Dimmie_Dumm]

meanwhile, basic aspects of hoi 4's controls still don't work, and we have documentation here that many ai nation will build military factories or convert civs to mils and then not use them at all. it's very clear the ai's performance using the mechanics isn't a priority.

Since I first mentioned it 7 months ago, it hasn't changed much: on top of not producing anything, Finland now sits on an empty stockpile, prepared to repel the Soviets by sheer willpower.

Spoiler

I encourage everyone to upvote the bug report, since no ETA for a fix is given. And with bugs as severe as this one still open we can only guess how well the AI will handle the new and seemingly far more complicated system.

 

[Louella]

for every decision the AI makes, the player makes many more. The AI only evaluates production, construction, plans etc. at 00:00 each day. A player can make decisions at any hour of the ingame day.
the player can also cancel division attacks and re-try them to see if they can attack with a tactic that counters the AI's tactic.
Even if the AI's equipment designs and divisional templates were as good as the players, the player is likely to win. It's not an even playing field.

The AI shuffles production too much, to its detriment. Production efficiency might make the player think about keeping things in production, but the AI doesn't seem to care. So the AI ends up with factories with low efficiency, unable to produce enough stuff. The AI used to, maybe still does, keep switching production of ships, cancelling them, with the result that AI resistance to naval actions is weak.

The game mechanics where attacks/defending always picks the same 3 divisions that are low on ORG, rather than the fresh ones, leading to a bungled attack or a failed defence when there are unengaged fresh divisions, is also detrimental to the AI. The player can micromanage to mitigate this, the AI does not.

Sometimes the game is a frustrating, rather than enjoyable experience.

 

[Secret Master]

Having the possibility for outplay is important but beyond a certain degree it becomes boring. If the AI is not ready for a subsystem, the subsystem is very likely not ready for the game.
The large mismatch in stats between your typical BBA and non-BBA fighter plane should not come down to "well Junkers historically had this prototype once and now here we are".

I don't fully disagree with you, but in terms of outplaying the AI, I am reminded of a few things.

1) The old post/information from podcat about what difficulties players use tells me that if a lot of players are using lower difficulties, then the game is "difficult enough" for a lot of the player base. I know that's not the same as having an AI that can give all players a run for their money, but if I was in the position of seeing that large segments of the player base use lower difficulties, I'd think that these issues with the AI are not the most important thing.

2) There is a huge skill gap between certain groups of players. Nothing brought this home to me more than when I posted a video on my channel with a light-hearted attempt to win the war as the Soviets using the really old 40w infantry divisions I used back when the game was new. I received a number of comments on the video from people telling me that the 40w infantry division was still meta. I was flabbergasted. I haven't bothered to use a formation like that in years. I thought everyone had gotten the memo some time ago. But I was wrong. And that's from players who like the game enough to watch Youtube videos about it. I see the same thing on livestreams. Even with a tiny audience like mine, I still get questions about really basic stuff I'm doing. Sure, there are deeper mysteries that I get questions about (training cadres and my weird method of production lines), but sometimes I get questions from viewers about stuff like using the close support gun to save tungsten. And while I could be like "These players are idiots" the reality is that that they aren't. There's just a large skill gap, and some of those skills are highly context specific. Even a good player will sometimes just not have information on some parts of the game.

All that being said, there are gaps that really should be filled in AI use of these mechanics. The designers are clearly game-selling mechanics that players want. So, there are things the AI should be better about even if it's not meant to play at top-tier MP. I really shouldn't see AI Germany never building medium tanks when it receives tank designs from focuses for free. And maybe the planes don't have to be top-tier meta, but the AI could use some additional guidance on building them.

 

[TheMeInTeam]

2) There is a huge skill gap between certain groups of players. Nothing brought this home to me more than when I posted a video on my channel with a light-hearted attempt to win the war as the Soviets using the really old 40w infantry divisions I used back when the game was new. I received a number of comments on the video from people telling me that the 40w infantry division was still meta. I was flabbergasted. I haven't bothered to use a formation like that in years. I thought everyone had gotten the memo some time ago. But I was wrong. And that's from players who like the game enough to watch Youtube videos about it. I see the same thing on livestreams. Even with a tiny audience like mine, I still get questions about really basic stuff I'm doing. Sure, there are deeper mysteries that I get questions about (training cadres and my weird method of production lines), but sometimes I get questions from viewers about stuff like using the close support gun to save tungsten. And while I could be like "These players are idiots" the reality is that that they aren't. There's just a large skill gap, and some of those skills are highly context specific. Even a good player will sometimes just not have information on some parts of the game.

i've done coop with players who lose to poland while using germany. they just aren't there yet with the macro side of producing and fielding enough divisions.

you can also find a non-trivial % of reddit posters who seem to believe that stacking as much breakthrough as possible is the key to making openings in the enemy's line. to the extent that they'll see a tank design that's doing minimal damage and suggest adding more cost to add breakthrough, or see a 1k base breakthrough division and consider investing in more to be a reasonable decision (as sp advice!).

it's more a knowledge gap than skill. i guess those are related, but they're not entirely the same thing. there is some % of players who just don't have the capacity to make correct inferences from knowledge in a timely fashion, even if they have said knowledge. but most can, given varying amounts of effort, and simply don't have full knowledge of the tradeoffs they're making in the first place.

width is a good example of that. lots of parrots, not many people who will articulate why x width performs better and when. in the old world blues mod discord, i even had someone ask me "what i was smoking" when i pointed out that going from 10w to 20w with same support companies would reduce my damage dealt to enemies in most cases, not increase it. it's kind of incredible how long that argument lasted. when i showed examples, they fell back to "anything works against the ai". sort of true...but also just not acknowledging the blatantly wrong assertion or explaining how the same number of battalions with fewer support companies somehow does more damage. not even an attempt to invoke coordination as an exception on the off chance (rare for the mod) the ai actually has more than 2-3 divs in one spot for it to really start making a difference.

 

[Secret Master]

lots of parrots, not many people who will articulate why x width performs better and when.

Yeah, this is an old video game (and tabletop to boot) problem that isn't HOI specific.

There are people who "know the meta" but don't have a clue why it is the meta. And these are the kinds of players who can play the game well in a rote fashion, but the moment something goes off the rails, they are lost.

There are also players who are slaves to meta in the sense that they can't process contradictory information. Literally the entire reason I got into testing aircraft in the ways that I do stemmed from a situation years ago where I was trying to figure out why my results were not matching what other people were seeing in their games. Once I solved the "you need enough planes in the test to iron out RNG and detection" issue, I discovered that improving guns on your fighters actually helped in kd/ratios. Even after showing IC-normalized tests with thousands of aircraft, I still had people telling me I was stupid because the meta at the time was +5 engines, no guns.

The problem was that these people hadn't bothered to update their meta with small changes made in a patch that made gun upgrades more competitive. But even with data staring them in the face, they couldn't change their minds.

 

[Louella]

every tank, ship, and aircraft is a compromise design. It shouldn't be possible to design stuff that excels in every statistic, in HOI4s equipment designers. If the only compromise for a tank design is that it is ludicrously expensive, whilst having high speed, thick armour, and a big gun, then the designer doesn't work, imo.

 

[Dimmie_Dumm]

If the only compromise for a tank design is that it is ludicrously expensive, whilst having high speed, thick armour, and a big gun, then the designer doesn't work, imo.

Sounds like you're speaking of IS-7
What's wrong with that, though? For HoI4, IC is as fair of a balancing leverage as any other one. And also IRL certain other extraordinary things like KpZ-70 were discarded for the same reason of abnormal cost.

 

[LordWahu]

every tank, ship, and aircraft is a compromise design. It shouldn't be possible to design stuff that excels in every statistic, in HOI4s equipment designers. If the only compromise for a tank design is that it is ludicrously expensive, whilst having high speed, thick armour, and a big gun, then the designer doesn't work, imo.

I liked the idea that it would be reliability that would apply here: That you could have the equipment, but probably not for long

But there are too many ways to mitigate low reliability, and too many ways to keep reliability high enough, so that doesn't really work

There needs to be some other balancing factor here across the board

 

[Louella]

What's wrong with that, though?

mostly that against the AI, the player generally has more than enough IC, and is able to use that IC far more effectively, so stuff being expensive isn't really a balancing factor.

 

[vermicious knid]

The AI is never going to pull off feints, tricky encirclements, coordinated invasions, etc... It just isn't going to happen given the limitations of the game.

What I would like to see is the AI using better preset templates. That is possible, Expert AI does it.

Also, it seems like it should be possible to force the AI to group its armor divisions together and drive them at victory points. Nothing fancy, just make a spearhead and thrust with it. That would at least make that piercing/armor arms race matter in single player...which is what the vast majority of us are playing.

 

[Kadanz]

1) The old post/information from podcat about what difficulties players use tells me that if a lot of players are using lower difficulties, then the game is "difficult enough" for a lot of the player base. I know that's not the same as having an AI that can give all players a run for their money, but if I was in the position of seeing that large segments of the player base use lower difficulties, I'd think that these issues with the AI are not the most important thing.

These lower difficulties had to do with severe AI issues in terms of division spam and naval invasions. Going a higher difficulty means more AI resources to make issues for itself worse. The AI would just cripple itself. I think paradox drew the wrong conclusions from this metric imho.

 

[Secret Master]

Going a higher difficulty means more AI resources to make issues for itself worse.

I don't agree.

Under higher difficulties, especially when boosted, the AI can get up to real shenanigans. My person favorite was the time I seized Kyshu from a fully buffed and highest difficult Japan. I'm the US. I have SF awesome divisions sitting across the straits from IJA forces. Even with boosts, I can still beat Japanese divisions in base SA by a large factor, and I have armor advantage due to space marines. I also have the USN in the area and can close the strait at any time.

I decided to leave the strait open just to see how much effort the AI would put into trying to retake Kyushu.

Answer: They were almost able to get through my divisions despite the US having entrenchment, a better general, air superiority, ground attack, armor advantage, the Japanese attacking across a strait without using MAR or amphibious tanks, and with my doctrine tree max SF. Oh, and my tech was better.

The buffs they got with both boosting and higher difficulty were almost enough to dislodge my forces despite inferiority in every measurable metric. It's one of the worst offensive situations the AI could be in, and it was almost able to succeed. I don't feel higher difficulties hamstring the AI. I think they make the AI much more meme-like and able to brute force things when it wouldn't make sense in any other context.

 

[Qswhisper]

What I would like to see is the AI using better preset templates. That is possible, Expert AI does it.

Yes, my proposed auto-generated design template for Paradox is also an attempt to learn Expert AI's approach to problem solving. I don't want to build the same design every time, nor do I want Ai to keep using really stupid designs. How about just giving several options so you can think carefully about the macro strategy. (upgrading scientific research levels , improving military doctrine and having more resources...) Then the system will automatically generate a better design template

 

[GrandVezir]

If it's possible, I think the ideal solution would be twofold:

1. Regular difficulty and below, keep AI mostly the same.
2. Veteran or Elite, script the AI to use (and build equipment for) better templates.

Let newbies learn while players who feel capable enough to choose tougher opponents, actually face tougher opponents.

Just my 2 pfennigs.

 

[Herennius]

If it's possible, I think the ideal solution would be twofold:

1. Regular difficulty and below, keep AI mostly the same.
2. Veteran or Elite, script the AI to use (and build equipment for) better templates.

Let newbies learn while players who feel capable enough to choose tougher opponents, actually face tougher opponents.

Just my 2 pfennigs.

I'm strongly against this approach - the AI should play to its best abilities on all difficulties and the difference between the levels should be solely handled by giving higher level AIs boni (and penalties on lower levels). If AI improvements are made, then the should be applied to all levels and penalties be increased on lower levels, if that is deemed necessary. My reasoning is that the learning effect by this approach is higher - if the AI OTOH is artificially kept limited in its behaviour, there is the danger that new players will look what it does and get teached inefficient startegies and then the higher levels feel even more frustrating. Dimming difficulty solely by numerical boni/mali brings a much smoother scale of challenge.

 

[HugsAndSnuggles]

I'm strongly against this approach - the AI should play to its best abilities on all difficulties and the difference between the levels should be solely handled by giving higher level AIs boni (and penalties on lower levels). If AI improvements are made, then the should be applied to all levels and penalties be increased on lower levels, if that is deemed necessary. My reasoning is that the learning effect by this approach is higher - if the AI OTOH is artificially kept limited in its behaviour, there is the danger that new players will look what it does and get teached inefficient startegies and then the higher levels feel even more frustrating. Dimming difficulty solely by numerical boni/mali brings a much smoother scale of challenge.

And how is, say, limiting production is fundamentally different from altering AI strategies? AI won't have the IC to pull fancier stuff off anyway, so might as well stay focused on WW1-style warfare and be fairly competent at it, rather than be horrible at everything until it suddenly gets enough IC for all those strategies to work.

If teaching efficient strategies is what you're after, SP should just be scraped along with AI. But we all can guess where that one would lead. And you know why? Because majority of players do not want to get better at games. Learning curves lead to niche audience. But if higher difficulty levels are aimed at niche audience anyway, why bother with smoothing out learning curve so much?