Observing the AI... am I the only one cringing, laughing, and crying.

[Mousetick]

If your whole argument is "this is too hard for the devs" and "it's not important enough for the devs", you should provide a source on the claim.

And "take it or leave it" isn't an argument, more so when you're not even on the development team.

I don't have sources on the claims. They're based on my observations that if the developers cared highly about the issue and it were that easy to address, it would have been addressed by now. How long has it been since players started complaining about the stupid frontline management? And based on my educated guesses, having a professional background in technical software engineering, if you must know.

Those are the kind of arguments that are made internally and never communicated to the customers. It all comes down to cost and profit, unsurprisingly. It's a lot easier and more profitable for Paradox to deliver content packs and minor balance patches, than to spend 6 months or more, designing and testing a competent 'AI' by a staff of highly skilled developers, with no clear and immediate impact on sales.

Some here will say "release the improvements as a DLC, I'll gladly pay for it!" That's not possible, nobody except the most hardcore fans would buy that. After the initial flop of Imperator: Rome, did Paradox release the redesigned version as a paid upgrade? No.

"take it or leave it" isn't an argument, it's an advice (that you can take or leave, ) to avoid the frustration and ressentment that shouting to the wind can induce. I suspect, subjectively of course and without any proof, that nobody at Paradox is paying attention to these recurring complaints and discussions.

 

[ThePHD]

I guarantee you this kind of thing would be fixed by some random guy within months if it were moddable. I think a big reason why they don't pay more attention to the AI is because most people playing HOI4 just aren't very good at the game. I'm basing this off of an old dev diary which showed that a shockingly high number of people play on easy. Also basing this on the billion or so threads i've seen on Steam claiming that the game is too hard and overly complex. I think the AI is just fine for the average HOI4 player even if it isn't for the rest of us.

 

[Meglok]

We can only hope the next two patches improve land combat. I know that is one of the goals for the Barbarossa patch, which makes sense as the Russia campaign was the largest land theater by far.

Front mechanics have gotten "better", and incorporating invasions has always been a challenge. Bit the ai can still do things that makes my hair hurt.

As noted above, if you playing SP non-Ironman and see the ai doing something stupid you can save and reload. The ai will recalculate theaters and strategy and usually will fix it's error. The console command theatersrebuild (trebuild) does the same thing iirc. I just save and reload so I can take a quick break.

 

[bitmode]

"take it or leave it" isn't an argument, it's an advice (that you can take or leave, )

golden

 

[SchwarzKatze]

They're based on my observations that if the developers cared highly about the issue and it were that easy to address, it would have been addressed by now.

Now that's objectively untrue:

https://forum.paradoxplaza.com/forum/threads/japanese-unit-and-name-list-suggestions-follow-up-1-9.1373668/

https://forum.paradoxplaza.com/forum/threads/errata-and-suggestion-about-the-japanese-division-name-list-oob.1079514/

https://forum.paradoxplaza.com/forum/threads/problem-with-the-initial-oob-and-the-division-name-list-of-japan.1123017/

HoI 4 - [1.6.1][55d5]Many typos and even Engrish in Japanese division name list and errors in OOB

Description of issue [1.6.1][55d5]Many typos and even Engrish in Japanese division name list and errors in OOB Game Version 1.6.1 Enabled DLC All of the above Do you have mods enabled? No Description I've posted this two...

forum.paradoxplaza.com

HoI 4 - [1.5] Japan's Guard Division appears as "601 Division" due to a typo/inconsistency

Description of issue [1.5] Japan's Guard Division appears as "601 Division" due to a typo/inconsistency Game Version 1.5 Enabled DLC All of the above Do you have mods enabled? No Description In JAP_1936.txt, the division is...

forum.paradoxplaza.com

HoI 4 - [1.5.1] Japanese independent mixed brigades are incorrect in several ways

Description of issue [1.5.1] Japanese independent mixed brigades are incorrect in several ways Game Version 1.5.1, but the issue has been there since release (1.0.0.19987) Enabled DLC All of the above Do you have mods enabled? No Description...

forum.paradoxplaza.com

These are (some of) my threads concerning unit name errors for Japan introduced by the division name list system. Even with clearly elaborated reasons and sources directly provided, it took them 2 years just to fix the names when it's something an intern can do in 5 minutes.

"take it or leave it" isn't an argument, it's an advice (that you can take or leave, ) to avoid the frustration and ressentment that shouting to the wind can induce. I suspect, subjectively of course and without any proof, that nobody at Paradox is paying attention to these recurring complaints and discussions.

There are people trying to help the game get better, and there are people like you shooing away any critics

I guarantee you this kind of thing would be fixed by some random guy within months if it were moddable. I think a big reason why they don't pay more attention to the AI is because most people playing HOI4 just aren't very good at the game. I'm basing this off of an old dev diary which showed that a shockingly high number of people play on easy. Also basing this on the billion or so threads i've seen on Steam claiming that the game is too hard and overly complex. I think the AI is just fine for the average HOI4 player even if it isn't for the rest of us.

I already tried to mod the AI but even with the additions in 1.9, there are still not enough parameters that modders can access. For example, I can force the AI to not invade a certain country with the invade ai_strategy. On the other hand, I can also force the AI to plan for an invasion withinvasion_unit_request, but we modders have no way to tell the AI to assess the defenses in the planned invasion area, or to tell it to pull troops from saturated fronts instead of undermanned fronts.

I can actually set how many divisions the AI will deploy against a certain country with front_unit_request, but I can't get the actual number of division deployed, nor how the front is holding up. There's also front_control, which I can use to make the AI create a specific order, and whether they will execute it or just hold it. However, if I make the AI hold it (execute_order = no), there's no way to tell the AI to execute it afterwards.

What's worse, there's no way to access the AI assessment of enemy strength or modify it, so even if you set execution_type = careful, you still see suicide charges that make even Luigi Cardona pale. I'm talking about 1:10 K/D ratio here that the AI will just keep doing until it runs out of manpower or equipment, and by running out I mean 50% shortage or worse. If the AI can't assess material losses vs gains achieved properly, at least expose it so that we modders can try to tweak things ourselves.

 

[Mousetick]

Now that's objectively untrue:
...
These are (some of) my threads concerning unit name errors for Japan introduced by the division name list system. Even with clearly elaborated reasons and sources directly provided, it took them 2 years just to fix the names when it's something an intern can do in 5 minutes.

Indeed, and this reinforces my arguments: even though the solution was straightforward, your issue was very low on the developers' priority list. Hence why it took so much time to address it. In order for an issue to be resolved in a 'reasonable' amount of time, it has to be both relatively easy, and high on the priority list. 'Reasonable' being subjective and varying depending on the user's expectation, so let's say, from one version of the game to the next.

There are people trying to help the game get better, and there are people like you shooing away any critics

I'm not shooing away the criticism, in fact I agree with your criticism about the issues themselves, and I said so. However I don't think the criticism about the longstanding lack of improvement, helps make the game better or helps in any way whatsoever. It looks like you're trying to mod the game, so that's great. Keep us posted on your progress.

And I'll leave it at that, as I'm afraid we're about to run in circles again, if we're not already doing so.

 

[1 Star General]

I'm also having a love/hate complex in my relationship with the game, so I'll throw in my example.
Playing China, non historical. It's relatively hard at start, stopping Japan not a piece of cake. After a long, multistage marathon (unify, stop Japan, push Jap back, reintegrate all puppeted warlords), really interesting game (kudos to the game for simulating that), comes the time for a grand finale, war with SU, superpower. I've spent hours planning it out and deploying troops on the border accordingly, have no AT capability and know soviets have armor, stressed out. I declare and march in... And keep marching... Bumped onto few divisions in Kazakhstan, literally walked over them. Keep marching, bumped onto another few divisions south of Urals, marched through... I realized it's pretty much over when crossed the Volga and still not see any AI attempt on creating a front line... Surely, SU put up a tough defence, but... in Siberia and in Tajikistan. Meanwhile I kept marching straight line the shortest path to Moscow and apparently it wasn't at all a priority defense axis. That was super anticlimactic, to the otherwise a really fun game. I really had a mixed bag of feelings after this one. SU had solid number of divisions, had lots of armor, more planes, but it chose to defend Baikal Lake and Tashkent, while I just walked into central Russia with no serious opposition. And BTW, SU is huge and marching through Central Steppe (no infrastructure) took me more than a month (I literally had only Infantry). So it wasn't a Blitz, there was time to react, and I kept expecting the AI will reform. But it didn't. And what do you do after that...? ahm....

 

[DaleDVM]

Strangely first day back on the forum in a couple of months and was surprised to see this thread on the front page.

I totally get it. The SU not defending itself against a major power it is a war with is game breaking. It is completely frustrating to invest a lot of time into a campaign only to encounter an AI blunder so bad that it literally destroys the entire play through of the war.

I can honestly say I have never been able to play a war to conclusion without an immersion breaking error by the AI. It is why I haven't played the game since June either. Recently, I have taken to playing the game on 3 computers in multiplayer. This is so I can take control of several countries at the same time. At least I get a decent simulation of the war when I control all sides.

 

[1 Star General]

I have taken to playing the game on 3 computers in multiplayer. This is so I can take control of several countries at the same time. At least I get a decent simulation of the war when I control all sides.

3 computers... So... am I tu understand that you're playing against yourself, who in turn is playing against you're other self??

 

[billcorr]

I want to love it... I repeat, I want to love it so, so badly... However, I could take as much pride in winning against this AI, as beating a squirrel in a chess match.

"...beating a squirrel in a chess match"

Nominated for the best word imagery on Paradox's HoI4 forum.

Here's a visual interpretation of that short squirrel poem:

Oh.

This Internet Thing is just amazing.

Next thing you know, they'll be short films of domesticated cats riding on roombas.

O brave new world.

 

[blue_yonder]

3 computers... So... am I tu understand that you're playing against yourself, who in turn is playing against you're other self??

That's how I played Risk when I was six. Thank goodness I don't have three computers.

 

[TheMeInTeam]

While I wouldn't rate the AI as quite as large of a problem as several broken core gameplay mechanic rules or the developer-intended degradation of the UI, I agree it's still one of the game's major problems. Not just AI's front management, but also it's algorithm for choosing when/how to initiate attack commands and its buildup logic. Maybe equipment build priority is the most egregious since computing allocation of factories for production is the type of thing that should be an AI strong suit.

Or maybe its buildup logic would be fine if it didn't suicide 500k+ all the time or wander away from fronts and get wrecked.

I suspect it's harder to fix AI than something like war score though.

 

[Harin]

Several of the posts mention that if you save, log out and reload, the AI recomputes some stuff and may correct some "squirrel-ish" behavior. (Ok, I could not resist.)

Is that true? If a player were to play in short bursts, then do the save/reload trick, the AI may perform better?

 

[TheMeInTeam]

Several of the posts mention that if you save, log out and reload, the AI recomputes some stuff and may correct some "squirrel-ish" behavior. (Ok, I could not resist.)

Is that true? If a player were to play in short bursts, then do the save/reload trick, the AI may perform better?

In some cases it will. This also clears out a recurring but *since release* where encircled units keep getting supply from other sources than local victory points.

 

[GSP Jr]

Several of the posts mention that if you save, log out and reload, the AI recomputes some stuff and may correct some "squirrel-ish" behavior. (Ok, I could not resist.)

Is that true? If a player were to play in short bursts, then do the save/reload trick, the AI may perform better?

I suggested save/reload to clear memory as a partial (temp) cure to running slow/lag. No idea what happens to AI behavior.

 

[bitmode]

Several of the posts mention that if you save, log out and reload, the AI recomputes some stuff and may correct some "squirrel-ish" behavior. (Ok, I could not resist.)

Is that true? If a player were to play in short bursts, then do the save/reload trick, the AI may perform better?

If you check the contents of a save file, you'll find very little AI data. Most of the "mental" state of the AI gets rebuild from scratch when loading a save and this can lead to reevaluating decisions that it otherwise might not have done (or only much later). But at the level of competence the AI has, it does not necessarily get better results. If it got stuck on something before, it may be helpful. But it may also change its mind on something that it was by chance doing well.

 

[Harin]

In some cases it will. This also clears out a recurring but *since release* where encircled units keep getting supply from other sources than local victory points.

I suggested save/reload to clear memory as a partial (temp) cure to running slow/lag. No idea what happens to AI behavior.

If you check the contents of a save file, you'll find very little AI data. Most of the "mental" state of the AI gets rebuild from scratch when loading a save and this can lead to reevaluating decisions that it otherwise might not have done (or only much later). But at the level of competence the AI has, it does not necessarily get better results. If it got stuck on something before, it may be helpful. But it may also change its mind on something that it was by chance doing well.

Thanks guys. As always, all three of your responses are helpful.

 

[TheMeInTeam]

If you check the contents of a save file, you'll find very little AI data. Most of the "mental" state of the AI gets rebuild from scratch when loading a save and this can lead to reevaluating decisions that it otherwise might not have done (or only much later). But at the level of competence the AI has, it does not necessarily get better results. If it got stuck on something before, it may be helpful. But it may also change its mind on something that it was by chance doing well.

Yeah in both this game and EU 4, the most obvious use case for reloading when it comes to AI is when the player observes it not moving units in a scenario where it should be moving units. Most cases reload helps (not always, EU 4 recent mercenary bug it would still sit there), and in HOI 4 where you can reload without restarting the game there's not much harm in trying if the present situation is already clearly bad.

 

[Kazakk]

Use expert AI 4.0 and buff it to impossible, only way to have a challenge in SP if you're a major.

 

[Ffire]

In another recent thread, bitmode asserted that a good strategy game is "easy to learn, hard to master" (Bushnell's law). I'd argue that HOI 4 is the opposite, it's hard to learn and easy to master

Wow... so true

I don't think there's any chance the AI in HOI4 could be really improved.

in HOI4 the AI should :
-react dynamically (no turn to turn behavior). Something the human brain is way better than any computer.
-manage a very high numbers of parameters. Supply, diplomacy, terrain, weather, fog of war
-not use scripts, but adapt to the ennemy player. expert AI mod use scripts and effectively improved the AI.
-not overload an average computer.
-eat new mechanics from DLCs

That's way too ambitious. Hope they'll think about that for HOI5. But I don't expect the HOI4 AI to be anything but marginally improved. They would need to change the core game so it would be easier for the AI to play with.