Home Trending Latest threads New posts Developer diaries

AI blues

Jump to latest Follow Reply
  • We have updated our Community Code of Conduct. Please read through the new rules for the forum that are an integral part of Paradox Interactive’s User Agreement.
Hearts of Iron IV 45529 Bug Reports 15227 After Action Reports (AAR)
joeenochs said:
Are you sure? Imagine the political situation wouldn't allow you to declare war on the low countries. Wouldn't it then be the right decision to attack the Magino line?
Click to expand...

Generals don't make political decision. Generals make the call and generals do not make decisions to sacrifice hundreds of thousands of men in that way. Breaking a city you need to take is one thing, but a battle that is not necessary is not an acceptable time to sacrifice a massive amount of manpower. The Maginot line was not the only option, the AI ought to know to hold back, do other things, bide its time, research, build ships etc. basic things. Not asking for anything special....

But again, not talking about JUST the Maginot line, talking about consistent decisions of AI to sacrifice too many men. Every WWW I watched, whenever they checked casualties it always seems too high and then I see the AI making bad decisions that do not get made in other PDX games.
 
anam said:
Honestly the AI in HOI3 did exactly the same thing at Maginot in ver 1.0 in fact a lot worse as a lot of time it never went around.

As to other examples of bad AI in HOI4 , please come with some examples. At least then we have something to discuss about :)
Click to expand...

Sorry as reasonable as your request is, I am not sitting here going back over months and months of videos to find the examples. I always brought them up when they happened, the point is the lack of progress on the problem.
 
  • 3
Reactions:
Calahir said:
German AI stupidly charging through Maginot Line is one thing. The other aberration we have seen in the last WWW was UK not only sending expeditionary forces to Poland (ridiculous by itself) but also holding Danzig for months (both pointless and impossible).
Click to expand...

It's easy to conflate AI issues with game design decisions. As SteelVolt said in the stream, his job is not to decide what is and isn't possible - that's the dev's job; the AI simply does its best with the world presented to it. If you think it's implausible for the UK to go to Danzig for example, well fair enough and many would agree, but it would be up to a dev to decide whether or not suicidal lost causes, (or such oddities as Japan sending volunteers to Spain), are good for the game and pass that design choice onto Carl.
 
  • 2
Reactions:
blue_yonder said:
If you think it's implausible for the UK to go to Danzig for example, well fair enough and many would agree, but it would be up to a dev to decide whether or not suicidal lost causes, (or such oddities as Japan sending volunteers to Spain), are good for the game and pass that design choice onto Carl.
Click to expand...

I can't really picture Podcat asking Steelvolt to make the AI more suicidal.
 
  • 1
Reactions:
@blue_yonder
SteelVolt is a dev, a developer.
The British AI going for Danzig is the result of many design decisions working together. How to you know what is intended and what not?
It indeed feels unrealistic but in order to fix it one has to pinpoint what is unrealistic about it and this is not so easy.
In real life I think the Stuka bombers would have bombed the British navy into the ground close to their home basis and the German navy would have done the rest.
Also the British military historically was in a defensive mode and would regard such operation too risky I think.
I wonder whether the British AI followed any kind of higher level plan to hold on to Danzig.
 
  • 1
Reactions:
  • 1
  • 1
Reactions:
If you're seeing a nation uselessly throw away hundreds of thousands of lives, the AI should do a good job simulating the Soviet defense for much of 1941. :rolleyes:
 
  • 5
  • 1
Reactions:
joeenochs said:
Well, I heard it. I suggest you learn the meaning of the word dev.
Click to expand...

eh?? where did I say he wasn't a dev? He's the lead AI in a games company - should I also clarify that he's homo sapiens? If you listened to the stream you'll have heard him say his job is to implement the vision of the design team and work closely with them. I simply don't get what you're trying to say, but nor am I interested in finding out.
 
  • 1
Reactions:
In the battle planner tutorial, the Russian AI does not react when he demonstrates the fallback feature. It just stays there and very, very slowly advances into the unoccupied German territory.

This really gave me doubts when I saw it yesterday....
 
  • 2
  • 2
Reactions:
"Around the Maginot" shouldn't be a NF really, just a common thing or even an event Germany does when at war with France. It's not like it was a revolutionary idea that was just thought of during WWII. It was done in WWI, and pretty much all of the German leadership was in agreement that going through Belgium was the only option to beat France.

Also, you should nerf the Maginot line. It seems that Germany is using its full army to attack it prior to the war with the Netherlands, so they should make some headway when using the full German Army instead of just 2 armies as they did historically and broke through with. It seems like right now the Maginot Line is an impenetrable barrier (at least for players) and it should not be this way. Historically France had a large portion of its army garrisoning it and the Germans still broke through with only a fraction of the troops the Germans have in WWW.
 
  • 2
  • 2
Reactions:
tom_jones said:
Something to keep on mind perhaps; the armchair generals posting here do so with the hindsight of knowing fighting France is just a beginning of a long world war campaign, and they'll need all the manpower, equipment etc for huge war in the east soon enough. So for them the heavy casualties when attacking head on etc, that's inexcusable waste and poor planning.

But the AI doesn't know this -- as far as its concerned, it's simply fighting a country with considerably smaller army and industrial base than its own. From this viewpoint it's not much of a crime to just throw divisions in the meatgrinder, because the enemy is likely to bleed out and run out of the supplies (and the ability to replace them) way sooner than Germany would. It's doing basically what the WWW player did in their war with Spain (and earlier with Switzerland) -- they threw their armies right at the enemy fortified in the mountains, and didn't care how many casualties that inflicted... because given the numeric advantage the outcome was considered sealed from the start.
Click to expand...

No sensible commander EVER thought like you just did: throwing lives away and burning through your divisions just because "my enemy will bleed out faster" is not a viable strategy.
From a historical standpoint, Germany should know that it will need the men for Barbarossa later.
From a gameplay standpoint, the AI should know that it is supposed to fight a long war for about 10 years.
From a strategy standpoint, sacrificing men is not smart, especially when you fight 1 (Germany) vs 2 (France AND UK)
 
  • 1
Reactions:
Lifthrasil said:
No sensible commander EVER thought like you just did: throwing lives away and burning through your divisions just because "my enemy will bleed out faster" is not a viable strategy.
From a historical standpoint, Germany should know that it will need the men for Barbarossa later.
From a gameplay standpoint, the AI should know that it is supposed to fight a long war for about 10 years.
From a strategy standpoint, sacrificing men is not smart, especially when you fight 1 (Germany) vs 2 (France AND UK)
Click to expand...

Um, WWI says hello
 
tom_jones said:
By this metric you can count sensible commanders and wars through history on one hand, if that. Attrition warfare is hardly a new concept, and arguably in use to this day.
Click to expand...


There is no 'arguably' about it. From my personal experience the U.S. military and Colombian military teach it and use it, as does anyone else who trains to fight for keeps. It's how war is. The reason you can't have fraternization between officers and enlisted men isn't just to maintain discipline and cohesion, but also for the same reason you don't name the lobster before you pick him out of the tank and eat him. You must be prepared to send them up and over the hill, end of story. But it's not just about being willing to sacrifice your own. The logic is that if you keep killing off the enemy by the bushel he'll want to quit sooner rather than later. And that's generally true.

Things like general Pershing launching massive assaults even when he knew the armistice was coming in order to "create a legacy" or Stalin pushing to take Berlin by May Day also happened and killed and will kill god knows how many men.

But all that crap aside, we're talking about AI rationale and should focus on that. I think in the WWW the Germans only backed off the Maginot after they lost like 150-300K men dead. That's a sign that you're going to be able to walk on the AI when you shouldn't be able to. I hope it's not as exploitable as it has been in previous Paradox titles, and if any game they make can do it it's HOI.
 
captainobvious2 said:
In the battle planner tutorial, the Russian AI does not react when he demonstrates the fallback feature. It just stays there and very, very slowly advances into the unoccupied German territory.

This really gave me doubts when I saw it yesterday....
Click to expand...

I would have done the same. Its divisions needed to recover their organization, and like this they can have the digging bonus to be prepared for the next attack.
 
You must log in or register to reply here.
Top Bottom
Back