Great answer A. But we do know exactly how many fronts. Get them all and it's an A*
AI incapable of defense - seems to be a very, very big issue
- Thread starter Ricox
- Start date
-
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.
Great on the history. Any more? The question was:
Got to be some history buffs out there who also know a smidgeon about the SCW?
Wow this thread exploded hard. Hopefully the AI has improved to the point where it is able to defend the maginot line, that is at least a must. Overall though i like the way conflicts are fought, so long as the Ai works and i have faith that Paradox will make sure this kinda thing doesnt happen so often before release. If it is still doing that after release, then yeah thats a pretty big problem and i'll be annoyed that Paradox has had the game in development for nearly two years and still, like with HOI3, the AI doesnt work on release. But im sure it'll be fine! No sense stressing about something until you can tell if the problem is still a problem on release! Chill 
8. false The biggest part of the army stayed loyal to trhe Republic. . The were literally disbanded. Almost all the navy and the air forces also stayed with the Republic.
Sometime last year, I had an idea that suggested it be possible for Sanjurjo to survive the plane crash or not get in one at all. Maybe a small percentage. Just curious, can you tell me if they implemented this?
B. Not answered the whole question
It would suggest that the player doesn't have all the facts because he can't see all the French divisions or even most of their territory. So he doesn't have all the the information available. And the viewer has even less.
Good answer. So the Republicans had an actual professional navy with capital ships, not just a bunch of amateur soldiers like George Orwell and Ernest Hemingway. Did any of the Navy join the rebels?
It's a good idea. Someone might be publishing a Spanish Civil War mod next week with your idea in it. Probably. Watch for the modding forum
Yes, that completes that question. I will add your information to my original post.
AI incapable of defense - seems to be a very, very big issue
So things are building up nicely. Still plenty of choices for people to go and have a crack at my questions.
Especially those who were shouting down people in the earlier thread who questioned anything they said that the AI was broken. Perhaps they would like to contribute to some of the discussions? They seemed to know how many fronts the AI had to deal with, how many units it had, where they were, why it wasn't placing units along the German border, and how the civil war mechanics worked. Surely they can answer a couple of the questions?
These for example are still available for someone to try to answer:
I posted on Challenging HOI . From the Streams on Twitch and Youtube I am worried about the defence
(Luxembourg was spelt wrong in the original)
I know AI isn't going to be as good as a player and I was one of the biggest defenders of Stellaris, and still think it is a good game, despite it's drawbacks.
But my problem is that playing as a major should be challenging, this isn't EU where I can go a choose a minor and make it a great nation. I want to play as France and struggle, I want to play as Russia and be defending Moscow, I want to play as UK and see Germany steamroll Europe and be running out of planes trying to defend the UK and finally I want to play as Germany and HAVE to go through Belgium to stretch the French lines.
I shouldn't have to play as Luxembourg to struggle to conquer Moscow....
As great as playing a world domination as Haiti will be... Or conquering Germany as Luxembourg... I also wouldn't mind being able to play a game as Italy where I don't just walk to Suez with no resistance... Is that too much to ask? Even if losing is never a possibility I like to feel like the game knows how to defend the Maginot line, the Suez canal or at least hold a consistent battle line...
I haven't played the game yet so I am not going to be too critical, but from what I have seen I am a little worried, Stellaris has been okay AI wise but it's performance issues and little bugs have annoyed me (they're getting better so I am 100% sure Paradox have it under control).
And I know hard is there but it should never be that the AI requires artificial bonuses to perform simple battleplans and basic tasks, even if they are not challenging to experienced players, they at least defend consistently. (given how vital Suez was to the UK that would be considered more than "oh lets send a division or two")
I haven't played the game yet so I am not going to be too critical, but from what I have seen I am a little worried, Stellaris has been okay AI wise but it's performance issues and little bugs have annoyed me (they're getting better so I am 100% sure Paradox have it under control).
And I know hard is there but it should never be that the AI requires artificial bonuses to perform simple battleplans and basic tasks, even if they are not challenging to experienced players, they at least defend consistently. (given how vital Suez was to the UK that would be considered more than "oh lets send a division or two")
(Luxembourg was spelt wrong in the original)
I know AI isn't going to be as good as a player and I was one of the biggest defenders of Stellaris, and still think it is a good game, despite it's drawbacks.
But my problem is that playing as a major should be challenging, this isn't EU where I can go a choose a minor and make it a great nation. I want to play as France and struggle, I want to play as Russia and be defending Moscow, I want to play as UK and see Germany steamroll Europe and be running out of planes trying to defend the UK and finally I want to play as Germany and HAVE to go through Belgium to stretch the French lines.
I shouldn't have to play as Luxembourg to struggle to conquer Moscow....
- 14
Challenge exactly.
I expect to the AI to provide a challenge. Now if takes giving the computer a significant production boost, or the player a handicap, so be it.
However, I do expect the AI to have a basic understanding of the games system. You take divisions and form defensive or offensive battle plans. Position the air force and navy to assist this plans.
The AI in Stellaris (haven't played the new patch) didn't seem to have a basic understanding of how to develop a sector, which made it very painful to hand over planets to the AI, because it made planets worse not better. Even with a production boost the AI was so bad at developing the economy they didn't have the resource to build large fleets. While the AI was extraordinarily passive in the use of fleets. The combination of not knowing how to develop systems and not understanding warfare, made the middle game, extraordinarily dull. I really hope the HOI IV AI avoids this.
I more forgiving of the not defending the Maginot line than some of the other things, I've observed. Going to war over the Rhineland was a possibility and the thought of another war with Germany good very well trigger a civil war and in the clusterfuck of a civil war I could see how the AI would prioritize putting down the civil war over defending the Maginot line.
On the other hand in Quills game. The UK had plenty of opportunity to reinforce the Suez canal, and lower priority but still important defend Aden. Instead it made a point of building a ton of Naval air. The good news is they did a number on the Italian navy. But not enough to stop troops being sent over to Africa. It appears that AI puts to high a priority on air planes., and doesn't seem to have a concept of critical place that must garrissoned, Maginot line, Gibraltor, Suez, Malta.
- 2
- 1
Nothing in this video though indicates that the AI doesn't understand how to defend the Maginot Line. The original posters who were so adamant that they AI was broken on the basis of an event that literally could occur only one in a 1000 games are still to substantiate their claims. They seem to have disappeared into the ether, rather than answer my questions.
So I'll answer some of them
SPA = { split_troops = 0.55 }
This is why I asked about the SCW, French Mutiny and Russian Revolution
The Armed Forces of the French Commune didn't appear out of nowhere, they are not a rebel pop-up like you could get in the whack-a-mole partisan war in HOI3. They werePART OF the armed forces of France. Whole divisions, ships and air wings.
This split is randomly generated, but the percentage is fixed in the event that generates the civil war. This could be 50/50 or 99/1. In the HOI3 SCW sscript the "0.55" represents a 45% Republican / 55 % Nationalist split IIRC.
Civil Wars are incredibly difficult to handle, because of this random element and the lack of preparation.
In a war the game mechanic that occurs is a wargoal. You have to justify a wargoal. When you do that the other side gets alerted. This includes the AI, if a player starts to justify a wargoal against it.
There is no wargoal in a civil war - it is to the death, winner takes all. So from an AI point of view, this means that a civil war is very confusing - it gets no advance warning that it is going to happen, and suddenly it has significant numbers of it's forces disappeared. Fortunately, this is not usually an issue, because the other side in the civil war suddenly has the same issue. The French Commune is formed, and the AI for that country has to evaluate it's options, decide what fronts it has to defend and redeploy it's forces to do that, and then consider if it might be able to go on the offensive. And it needs to consider completely from scratch what it should do with production, trade, NFs and research. The AI is working hard.
So let's clarify for those who claimed that the country battle plan AI for France removed units from the Maginot Line. No, it did not. The civil war script removed a significant number of forces from France randomly, and gave them to the French Commune.
Howsoever you abstract this it doesn't matter. Some Divisions which were under the control of the French AI sitting in their bunkers in the Maginot Line one day were not there the following day, or they were disbanded, or they were mutinied and not taking orders from France any more, or they just went home.
In the summer of 1917 whole French regiments mutinied, and at least two divisions that had been held in reserve refused to move up to the front to take part in an offensive. In Russia it was even more severe. In the SCW the whole Army of Africa refused to take commands from the government, and followed the orders of the rebelling generals. Significant other parts of the Army, Navy and Air Force in Spain did the same. Some marched out of their barracks and arrested their local mayors, trades union and political leaders etc. But others didn't. Some remained uncommitted either way, some took up with the Republican forces. Most notable was the Navy where on many ships the crews arrested the officers. The only sea-worthy battleship the Spanish possessed Jaime I sent the following signal soon after the attempted coup started:
"Crew of Jaime I to ministry of marine. We have had serious resistance from the commanders and officers on board and have subdued them by force... Urgently request instructions as to bodies."
Such actions were usually spontaneous - they didn't receive orders to do it from the government. The wireless operators on board received the signals from the coup leaders and informed the rest of the men what they their officers were planning to do.
Even among officers there was not universal support for the coup leaders. Some hesitated and waited to see how things turned out, some opposed them directly - there were even secret trades unions of leftists among the officers, as well as Falangist and Carlist groups. The General in command in Grenada opposed the coup, but he was shot by two Colonels below him, who then ordered the garrison on to the streets.
Civil War is a messy business, and it is in the game, for the player or the AI. Armies with a carefully arranged defensive battle plan to defend a border might suddenly have not enough forces to cover all of the provinces, but then also find that there is now a threat to them that is in their rear. The player/AI have to decide how to respond to that.
Normally what you do is abandon the border. Let me repeat that - YOU ABANDON THE BORDER.
Yes, stationing Nationalist troops on the border with Portugal means you will lose the civil war. Stationing Republican troops on the French border means you will lose the civil war. You must commit all of your forces to the civil war, and trust that your neighbours are not going to invade.
But that is not what happens in the video. Because the unthinkable does happen to France - Germany does invade.
So let's look at what the French AI has to deal with. A civil war breaks out and it suddenly loses a significant amount of it's forces randomly, so it has to adjust all of it's forces on it's existing fronts, and now it has TWO ACTIVE fronts. There is an internal border with the French Commune that it must cover. And as rightly pointed out early in the thread, but shouted down, that is close to the capital. The AI must move immediately enough troops to contain that threat.
So let's speculate. Let's say that the French AI has 50 Divisions in the field when these events occur. It might lose 5, 10, 20 even? And remember, every division it loses is an extra division that the Commune gains. If it loses 15 divisions, then the Commune has 15 divisions near Paris. What if it then splits it's remaining forces 50/50?
Can it do that though? There are not two fronts in the civil war. The Commune has naval ships and air wings. What if it conducts a naval invasion or drops paras? Unlikely, I know, since the forces it has have the tech research of France, and if France didn't have paras then the Commune can't possibly. But a naval invasion in Normandy is a possibility. Plus if France had tanks, then the Commune probably has tanks. The French AI must commit forces to at least two fronts, the border with the Commune and the "home" front, ie. defend it's VP cities, and it's ports. Unfortunately, France has quite a few of both, about 20 in the mainland. France also has alot of colonies and the AI (or a player in the same situation) has to decide whether to bring all of their forces their back to the mainland. Can it?
Go to 25:22 in the video. We see France controlling part of Tunisia, and surrounded by the brown of the Commune - the civil war is going in North Africa as well! So that's now three new active fronts that the AI has to deal with.
It is then set in motion to move nearly all of its remaining divisions as it moves some from within France to defend against the Commune and others coming from overseas. They will take weeks to get there.
I'll drop in fog of war here. Probably don't need to say much. But anyone who says they know what the French AI was doing/not doing on the basis of a video playing as Germany needs to consider this part of the game. Germany can't see anything that the French AI is doing beyond their borders. It certainly can't see if divisions are being redeployed from Indochina and North Africa to the mainland. It just doesn't have any way to get that sort of intelligence.
OK, so civil wars are chaotic. But the AI is then also put into a war with Germany. It's difficult to judge here, but we see both things happen around the same time. But does the French AI? Like any computer system microseconds matter here. If the French AI is aware of the civil war first, then it could be in the process of moving most of it's divisions to fight the Commune when it realises that it is also at war with Germany. It won't stop them and reconsider, it can't until they reach their destination, ony then it will reconsider whether to redeploy them again. So it is quite possible that when it gives orders to say 20 divisions to move to the Commune borders at that instance it doesn't know there is a war with Germany. It certainly doesn't know in advance. Germany has no war goal, and there is nothing that would alert the French battle plan AI to prepare for Germany invading within days.
But let's surmise that the civil war and the war with Germany do actually start simultaneously, and that instead of two fronts, the AI has to consider four (the Commune borders, the German border and the "home front"). Germany only has one border with France and it's colonies, it doesn't have anywhere else round the world it can attack overland. But it also has a (strong) navy. Not UK, USA and Japan strong, but the potential equal of France? Especially if France has lost significant naval forces to the Commune. So it's now more complicated for France.
A player might handle this differently, but I'm not so sure they would despite the very strong statements in the thread that the ONLY thing that matters is one province in the Maginot Line. But a human might better predict what COULD happen in the future than the AI. The AI has to make an instant decision on the information available to it. Presumably it can "see" from it's intelligence the strength of the Commune, and how many divisions are close to Paris. But what can it see in Germany?
Fog of war works both ways, and the AI also follows the rules. The people watching the video can see all of the German divisions, but the French AI cannot. It only has limited information.
But what is surprising here is again the force of the argument when everyone watching the video knows that the whole event sequence is around the existence of the DMZ in the Rhineland, and can see that before the war breaks out the German player has a small amount of troops along the Rhine in Wurttemburg but NONE in Mosseland. None because it is a DMZ, and the German player is not allowed to move any divisions there. In the event sequence, civil war and war, the DMZ does get removed. But you can see that it takes a little while that the German player decides to move some of his forces to Mosseland, and longer for them to get there (because he doesn't seem to understand how to use SR ).
So the AI has limited information about Germany because of fog of war, and what information it has tells it that there is no immediate threat to it's border. It doesn't know that troops are on their way to Mosseland, only that there are none there now. Why should it prioritise those provinces above all others, when there is not even any German divisions there to capture them. But that doesn't mean that the AI is ignoring them. It could be that it thinks it has time to get divisions from elsewhere to plug the gaps.
Following in the wake of civil war, war with Germany with no notice, and the confused intelligence because of the DMZ, the French AI starts moving many/most of it's forces to try to counter these dual threats. Then suddenly a week into what ever redeployments it is doing (none of which can be seen in the video), Italy also declares war. And because of the mechanics this also is with no advance warning. Italy doesn't have a wargoal, the French AI can't possibly know that this is going to happen. This is not actually triggered by the events but because Italy (ie the much maligned AI) makes a strategic decision that invading France is a good decision. You have to admit that, from the Italian point of view, this is actually a smart move by the Italy AI. I have no reason to doubt these figures kindly provided by Rommel 459, though I didn't see that myself:
So we see that the Italian AI didn't "attack" France, so didn't need a wargoal. It joined the Axis and seems to have then joined Germany in a "defensive" war. Did you expect that, did the player in the video, did the AI? This is so counter-intuitive because we see that Germany "invades" France, but the mechanics treat France and the UK as the aggressors. You might question that mechanic possibly, though I can see for and against. But it has nothing to do with the AI. The French AI didn't decide to DoW Italy, it just has to deal with the consequences when suddenly and without warning it has another war on it's hands - and what a war!
Unlike Germany, Italy starts in 1936 with plenty of forces on it's border with France. My estimate is that it probably has more forces than Germany has along the Maginot Line, because Germany is still deploying into the former DMZ, and bringing forces from the Czech and Poland borders. So in the middle of the French AIs major redeployment of most of it's forces there is suddenly a fifth front. And it is possible that intelligence available to the AI tells it this front has the most enemy forces. Plus Italy has a strong Navy in the Med and also a strong air force, so the southern coast line is also now strongly threatened by an Italian invasion.
Unlike Germany, Italy has more than one border with France on 1.1.1936. However, we can see in the video it's colony in Libya no longer borders French Tunisia, because of the Commune. But Eritrea (where there are Italian forces involved in the war with Ethiopia) borders with the French colony in Djibouti. So that's a sixth front. And of course Tunisia and core French territory in Sardinia could be invaded from Italy.
Threats there might be low. But the AI is going to start to really regret it's decision to pull units from the Italy border, if that is what is has done. The Italians, unlike the Germans, are already in position. They have probably had an offensive battle plan preparing since 1.1.1936, even if the chances of executing it seemed small. But immediately when they enter the war, they can attack in force.
The AI must respond to all of these things. And it must do it with less forces than it had just a week earlier - three wars, with six fronts. and all started with no advance warning.
It's really beyond me to think how anyone could think that the French AI is "bad" in those circumstances. They should place themselves in that situation as if they were playing as France and these things happened to you. My feeling is that 90% of players would have rage quit and restarted the game, or at least gone back to their last autosave before the war.
Of course, going back a month might not help, a France player could end up in the same situation. But only if they are stupid. Once bitten twice shy and all that...
The German national focus Rhineland kicks off this situation. It is part of the historical AI focuses that Germany under AI control will always take early, and the German player in the video also followed that historical path. Therefore, it is likely that in well over 95% of games this event sequence will start to take place. Yet we have never seen these wars happen before, not just in games where the devs or Youtubers played as France of Germany, but in any of the other countries. We know this because the News Event pops up telling you (whatever country you are playing) that Germany has successfully re-occupied the Rhineland. We can assume it is exceptionally rare.
But a player France? When they get informed that Germany has completed the Rhineland NF they are presented with two options. One to just accept it, the other to decide "This is an Act of War". And the tooltip tells them in red "If war breaks out, we will suffer a civil war due to our Disjointed Government".
You have been warned if you play France. If you want to try to prevent Germany reoccupying the Rhineland you will probably get screwed over in the same way with a three country, six front war, and lose a significant part of your forces.
So I'll answer some of them
He cut the video 0:37 seconds in, when he was talking about save scumming in Ironman mode. We didn't see him hit the button to start the game, or the game load. Fair enough. But did he change the Historical AI mode?
There was a command like:
SPA = { split_troops = 0.55 }
This is why I asked about the SCW, French Mutiny and Russian Revolution
The Armed Forces of the French Commune didn't appear out of nowhere, they are not a rebel pop-up like you could get in the whack-a-mole partisan war in HOI3. They werePART OF the armed forces of France. Whole divisions, ships and air wings.
This split is randomly generated, but the percentage is fixed in the event that generates the civil war. This could be 50/50 or 99/1. In the HOI3 SCW sscript the "0.55" represents a 45% Republican / 55 % Nationalist split IIRC.
Civil Wars are incredibly difficult to handle, because of this random element and the lack of preparation.
In a war the game mechanic that occurs is a wargoal. You have to justify a wargoal. When you do that the other side gets alerted. This includes the AI, if a player starts to justify a wargoal against it.
There is no wargoal in a civil war - it is to the death, winner takes all. So from an AI point of view, this means that a civil war is very confusing - it gets no advance warning that it is going to happen, and suddenly it has significant numbers of it's forces disappeared. Fortunately, this is not usually an issue, because the other side in the civil war suddenly has the same issue. The French Commune is formed, and the AI for that country has to evaluate it's options, decide what fronts it has to defend and redeploy it's forces to do that, and then consider if it might be able to go on the offensive. And it needs to consider completely from scratch what it should do with production, trade, NFs and research. The AI is working hard.
So let's clarify for those who claimed that the country battle plan AI for France removed units from the Maginot Line. No, it did not. The civil war script removed a significant number of forces from France randomly, and gave them to the French Commune.
Howsoever you abstract this it doesn't matter. Some Divisions which were under the control of the French AI sitting in their bunkers in the Maginot Line one day were not there the following day, or they were disbanded, or they were mutinied and not taking orders from France any more, or they just went home.
In the summer of 1917 whole French regiments mutinied, and at least two divisions that had been held in reserve refused to move up to the front to take part in an offensive. In Russia it was even more severe. In the SCW the whole Army of Africa refused to take commands from the government, and followed the orders of the rebelling generals. Significant other parts of the Army, Navy and Air Force in Spain did the same. Some marched out of their barracks and arrested their local mayors, trades union and political leaders etc. But others didn't. Some remained uncommitted either way, some took up with the Republican forces. Most notable was the Navy where on many ships the crews arrested the officers. The only sea-worthy battleship the Spanish possessed Jaime I sent the following signal soon after the attempted coup started:
"Crew of Jaime I to ministry of marine. We have had serious resistance from the commanders and officers on board and have subdued them by force... Urgently request instructions as to bodies."
Such actions were usually spontaneous - they didn't receive orders to do it from the government. The wireless operators on board received the signals from the coup leaders and informed the rest of the men what they their officers were planning to do.
Even among officers there was not universal support for the coup leaders. Some hesitated and waited to see how things turned out, some opposed them directly - there were even secret trades unions of leftists among the officers, as well as Falangist and Carlist groups. The General in command in Grenada opposed the coup, but he was shot by two Colonels below him, who then ordered the garrison on to the streets.
Civil War is a messy business, and it is in the game, for the player or the AI. Armies with a carefully arranged defensive battle plan to defend a border might suddenly have not enough forces to cover all of the provinces, but then also find that there is now a threat to them that is in their rear. The player/AI have to decide how to respond to that.
Normally what you do is abandon the border. Let me repeat that - YOU ABANDON THE BORDER.
Yes, stationing Nationalist troops on the border with Portugal means you will lose the civil war. Stationing Republican troops on the French border means you will lose the civil war. You must commit all of your forces to the civil war, and trust that your neighbours are not going to invade.
But that is not what happens in the video. Because the unthinkable does happen to France - Germany does invade.
So let's look at what the French AI has to deal with. A civil war breaks out and it suddenly loses a significant amount of it's forces randomly, so it has to adjust all of it's forces on it's existing fronts, and now it has TWO ACTIVE fronts. There is an internal border with the French Commune that it must cover. And as rightly pointed out early in the thread, but shouted down, that is close to the capital. The AI must move immediately enough troops to contain that threat.
So let's speculate. Let's say that the French AI has 50 Divisions in the field when these events occur. It might lose 5, 10, 20 even? And remember, every division it loses is an extra division that the Commune gains. If it loses 15 divisions, then the Commune has 15 divisions near Paris. What if it then splits it's remaining forces 50/50?
Can it do that though? There are not two fronts in the civil war. The Commune has naval ships and air wings. What if it conducts a naval invasion or drops paras? Unlikely, I know, since the forces it has have the tech research of France, and if France didn't have paras then the Commune can't possibly. But a naval invasion in Normandy is a possibility. Plus if France had tanks, then the Commune probably has tanks. The French AI must commit forces to at least two fronts, the border with the Commune and the "home" front, ie. defend it's VP cities, and it's ports. Unfortunately, France has quite a few of both, about 20 in the mainland. France also has alot of colonies and the AI (or a player in the same situation) has to decide whether to bring all of their forces their back to the mainland. Can it?
Go to 25:22 in the video. We see France controlling part of Tunisia, and surrounded by the brown of the Commune - the civil war is going in North Africa as well! So that's now three new active fronts that the AI has to deal with.
It is then set in motion to move nearly all of its remaining divisions as it moves some from within France to defend against the Commune and others coming from overseas. They will take weeks to get there.
I'll drop in fog of war here. Probably don't need to say much. But anyone who says they know what the French AI was doing/not doing on the basis of a video playing as Germany needs to consider this part of the game. Germany can't see anything that the French AI is doing beyond their borders. It certainly can't see if divisions are being redeployed from Indochina and North Africa to the mainland. It just doesn't have any way to get that sort of intelligence.
OK, so civil wars are chaotic. But the AI is then also put into a war with Germany. It's difficult to judge here, but we see both things happen around the same time. But does the French AI? Like any computer system microseconds matter here. If the French AI is aware of the civil war first, then it could be in the process of moving most of it's divisions to fight the Commune when it realises that it is also at war with Germany. It won't stop them and reconsider, it can't until they reach their destination, ony then it will reconsider whether to redeploy them again. So it is quite possible that when it gives orders to say 20 divisions to move to the Commune borders at that instance it doesn't know there is a war with Germany. It certainly doesn't know in advance. Germany has no war goal, and there is nothing that would alert the French battle plan AI to prepare for Germany invading within days.
But let's surmise that the civil war and the war with Germany do actually start simultaneously, and that instead of two fronts, the AI has to consider four (the Commune borders, the German border and the "home front"). Germany only has one border with France and it's colonies, it doesn't have anywhere else round the world it can attack overland. But it also has a (strong) navy. Not UK, USA and Japan strong, but the potential equal of France? Especially if France has lost significant naval forces to the Commune. So it's now more complicated for France.
A player might handle this differently, but I'm not so sure they would despite the very strong statements in the thread that the ONLY thing that matters is one province in the Maginot Line. But a human might better predict what COULD happen in the future than the AI. The AI has to make an instant decision on the information available to it. Presumably it can "see" from it's intelligence the strength of the Commune, and how many divisions are close to Paris. But what can it see in Germany?
Fog of war works both ways, and the AI also follows the rules. The people watching the video can see all of the German divisions, but the French AI cannot. It only has limited information.
But what is surprising here is again the force of the argument when everyone watching the video knows that the whole event sequence is around the existence of the DMZ in the Rhineland, and can see that before the war breaks out the German player has a small amount of troops along the Rhine in Wurttemburg but NONE in Mosseland. None because it is a DMZ, and the German player is not allowed to move any divisions there. In the event sequence, civil war and war, the DMZ does get removed. But you can see that it takes a little while that the German player decides to move some of his forces to Mosseland, and longer for them to get there (because he doesn't seem to understand how to use SR
).So the AI has limited information about Germany because of fog of war, and what information it has tells it that there is no immediate threat to it's border. It doesn't know that troops are on their way to Mosseland, only that there are none there now. Why should it prioritise those provinces above all others, when there is not even any German divisions there to capture them. But that doesn't mean that the AI is ignoring them. It could be that it thinks it has time to get divisions from elsewhere to plug the gaps.
Following in the wake of civil war, war with Germany with no notice, and the confused intelligence because of the DMZ, the French AI starts moving many/most of it's forces to try to counter these dual threats. Then suddenly a week into what ever redeployments it is doing (none of which can be seen in the video), Italy also declares war. And because of the mechanics this also is with no advance warning. Italy doesn't have a wargoal, the French AI can't possibly know that this is going to happen. This is not actually triggered by the events but because Italy (ie the much maligned AI) makes a strategic decision that invading France is a good decision. You have to admit that, from the Italian point of view, this is actually a smart move by the Italy AI. I have no reason to doubt these figures kindly provided by Rommel 459, though I didn't see that myself:
So we see that the Italian AI didn't "attack" France, so didn't need a wargoal. It joined the Axis and seems to have then joined Germany in a "defensive" war. Did you expect that, did the player in the video, did the AI? This is so counter-intuitive because we see that Germany "invades" France, but the mechanics treat France and the UK as the aggressors. You might question that mechanic possibly, though I can see for and against. But it has nothing to do with the AI. The French AI didn't decide to DoW Italy, it just has to deal with the consequences when suddenly and without warning it has another war on it's hands - and what a war!
Unlike Germany, Italy starts in 1936 with plenty of forces on it's border with France. My estimate is that it probably has more forces than Germany has along the Maginot Line, because Germany is still deploying into the former DMZ, and bringing forces from the Czech and Poland borders. So in the middle of the French AIs major redeployment of most of it's forces there is suddenly a fifth front. And it is possible that intelligence available to the AI tells it this front has the most enemy forces. Plus Italy has a strong Navy in the Med and also a strong air force, so the southern coast line is also now strongly threatened by an Italian invasion.
Unlike Germany, Italy has more than one border with France on 1.1.1936. However, we can see in the video it's colony in Libya no longer borders French Tunisia, because of the Commune. But Eritrea (where there are Italian forces involved in the war with Ethiopia) borders with the French colony in Djibouti. So that's a sixth front. And of course Tunisia and core French territory in Sardinia could be invaded from Italy.
Threats there might be low. But the AI is going to start to really regret it's decision to pull units from the Italy border, if that is what is has done. The Italians, unlike the Germans, are already in position. They have probably had an offensive battle plan preparing since 1.1.1936, even if the chances of executing it seemed small. But immediately when they enter the war, they can attack in force.
The AI must respond to all of these things. And it must do it with less forces than it had just a week earlier - three wars, with six fronts. and all started with no advance warning.
It's really beyond me to think how anyone could think that the French AI is "bad" in those circumstances. They should place themselves in that situation as if they were playing as France and these things happened to you. My feeling is that 90% of players would have rage quit and restarted the game, or at least gone back to their last autosave before the war.
Of course, going back a month might not help, a France player could end up in the same situation. But only if they are stupid. Once bitten twice shy and all that...
The German national focus Rhineland kicks off this situation. It is part of the historical AI focuses that Germany under AI control will always take early, and the German player in the video also followed that historical path. Therefore, it is likely that in well over 95% of games this event sequence will start to take place. Yet we have never seen these wars happen before, not just in games where the devs or Youtubers played as France of Germany, but in any of the other countries. We know this because the News Event pops up telling you (whatever country you are playing) that Germany has successfully re-occupied the Rhineland. We can assume it is exceptionally rare.
But a player France? When they get informed that Germany has completed the Rhineland NF they are presented with two options. One to just accept it, the other to decide "This is an Act of War". And the tooltip tells them in red "If war breaks out, we will suffer a civil war due to our Disjointed Government".
You have been warned if you play France. If you want to try to prevent Germany reoccupying the Rhineland you will probably get screwed over in the same way with a three country, six front war, and lose a significant part of your forces.
- 3
- 1
Republicans had almost all the ships, but they killed almost all officers. Not very operational until 1938. But the it was too late... nationalist had the Baleares and the Canarias (they were in construction at the beginning) the most modern of the fleet, and the old battleship España, the Cervera and a few little ones). Republicans the rest of the fleet, including all the late generation destroyers)
The entire team doesn't have 10 guys I imagine let alone for just the AI. Also whilst 2 or 3 people might be good if they are coordinated well getting up to 10 people on a project like this seems like it would lead to a lot more bugs and hassle, throwing numbers at programming jobs is generally seen as a not particularly productive thing as far as I am aware.
- 2
- 1
Nothing in this video though indicates that the AI doesn't understand how to defend the Maginot Line. The original posters who were so adamant that they AI was broken on the basis of an event that literally could occur only one in a 1000 games are still to substantiate their claims. They seem to have disappeared into the ether, rather than answer my questions.
So I'll answer some of them
He cut the video 0:37 seconds in, when he was talking about save scumming in Ironman mode. We didn't see him hit the button to start the game, or the game load. Fair enough. But did he change the Historical AI mode?
There was a command like:
SPA = { split_troops = 0.55 }
This is why I asked about the SCW, French Mutiny and Russian Revolution
The Armed Forces of the French Commune didn't appear out of nowhere, they are not a rebel pop-up like you could get in the whack-a-mole partisan war in HOI3. They werePART OF the armed forces of France. Whole divisions, ships and air wings.
This split is randomly generated, but the percentage is fixed in the event that generates the civil war. This could be 50/50 or 99/1. In the HOI3 SCW sscript the "0.55" represents a 45% Republican / 55 % Nationalist split IIRC.
Civil Wars are incredibly difficult to handle, because of this random element and the lack of preparation.
In a war the game mechanic that occurs is a wargoal. You have to justify a wargoal. When you do that the other side gets alerted. This includes the AI, if a player starts to justify a wargoal against it.
There is no wargoal in a civil war - it is to the death, winner takes all. So from an AI point of view, this means that a civil war is very confusing - it gets no advance warning that it is going to happen, and suddenly it has significant numbers of it's forces disappeared. Fortunately, this is not usually an issue, because the other side in the civil war suddenly has the same issue. The French Commune is formed, and the AI for that country has to evaluate it's options, decide what fronts it has to defend and redeploy it's forces to do that, and then consider if it might be able to go on the offensive. And it needs to consider completely from scratch what it should do with production, trade, NFs and research. The AI is working hard.
So let's clarify for those who claimed that the country battle plan AI for France removed units from the Maginot Line. No, it did not. The civil war script removed a significant number of forces from France randomly, and gave them to the French Commune.
Howsoever you abstract this it doesn't matter. Some Divisions which were under the control of the French AI sitting in their bunkers in the Maginot Line one day were not there the following day, or they were disbanded, or they were mutinied and not taking orders from France any more, or they just went home.
In the summer of 1917 whole French regiments mutinied, and at least two divisions that had been held in reserve refused to move up to the front to take part in an offensive. In Russia it was even more severe. In the SCW the whole Army of Africa refused to take commands from the government, and followed the orders of the rebelling generals. Significant other parts of the Army, Navy and Air Force in Spain did the same. Some marched out of their barracks and arrested their local mayors, trades union and political leaders etc. But others didn't. Some remained uncommitted either way, some took up with the Republican forces. Most notable was the Navy where on many ships the crews arrested the officers. The only sea-worthy battleship the Spanish possessed Jaime I sent the following signal soon after the attempted coup started:
"Crew of Jaime I to ministry of marine. We have had serious resistance from the commanders and officers on board and have subdued them by force... Urgently request instructions as to bodies."
Such actions were usually spontaneous - they didn't receive orders to do it from the government. The wireless operators on board received the signals from the coup leaders and informed the rest of the men what they their officers were planning to do.
Even among officers there was not universal support for the coup leaders. Some hesitated and waited to see how things turned out, some opposed them directly - there were even secret trades unions of leftists among the officers, as well as Falangist and Carlist groups. The General in command in Grenada opposed the coup, but he was shot by two Colonels below him, who then ordered the garrison on to the streets.
Civil War is a messy business, and it is in the game, for the player or the AI. Armies with a carefully arranged defensive battle plan to defend a border might suddenly have not enough forces to cover all of the provinces, but then also find that there is now a threat to them that is in their rear. The player/AI have to decide how to respond to that.
Normally what you do is abandon the border. Let me repeat that - YOU ABANDON THE BORDER.
Yes, stationing Nationalist troops on the border with Portugal means you will lose the civil war. Stationing Republican troops on the French border means you will lose the civil war. You must commit all of your forces to the civil war, and trust that your neighbours are not going to invade.
But that is not what happens in the video. Because the unthinkable does happen to France - Germany does invade.
So let's look at what the French AI has to deal with. A civil war breaks out and it suddenly loses a significant amount of it's forces randomly, so it has to adjust all of it's forces on it's existing fronts, and now it has TWO ACTIVE fronts. There is an internal border with the French Commune that it must cover. And as rightly pointed out early in the thread, but shouted down, that is close to the capital. The AI must move immediately enough troops to contain that threat.
So let's speculate. Let's say that the French AI has 50 Divisions in the field when these events occur. It might lose 5, 10, 20 even? And remember, every division it loses is an extra division that the Commune gains. If it loses 15 divisions, then the Commune has 15 divisions near Paris. What if it then splits it's remaining forces 50/50?
Can it do that though? There are not two fronts in the civil war. The Commune has naval ships and air wings. What if it conducts a naval invasion or drops paras? Unlikely, I know, since the forces it has have the tech research of France, and if France didn't have paras then the Commune can't possibly. But a naval invasion in Normandy is a possibility. Plus if France had tanks, then the Commune probably has tanks. The French AI must commit forces to at least two fronts, the border with the Commune and the "home" front, ie. defend it's VP cities, and it's ports. Unfortunately, France has quite a few of both, about 20 in the mainland. France also has alot of colonies and the AI (or a player in the same situation) has to decide whether to bring all of their forces their back to the mainland. Can it?
Go to 25:22 in the video. We see France controlling part of Tunisia, and surrounded by the brown of the Commune - the civil war is going in North Africa as well! So that's now three new active fronts that the AI has to deal with.
It is then set in motion to move nearly all of its remaining divisions as it moves some from within France to defend against the Commune and others coming from overseas. They will take weeks to get there.
I'll drop in fog of war here. Probably don't need to say much. But anyone who says they know what the French AI was doing/not doing on the basis of a video playing as Germany needs to consider this part of the game. Germany can't see anything that the French AI is doing beyond their borders. It certainly can't see if divisions are being redeployed from Indochina and North Africa to the mainland. It just doesn't have any way to get that sort of intelligence.
OK, so civil wars are chaotic. But the AI is then also put into a war with Germany. It's difficult to judge here, but we see both things happen around the same time. But does the French AI? Like any computer system microseconds matter here. If the French AI is aware of the civil war first, then it could be in the process of moving most of it's divisions to fight the Commune when it realises that it is also at war with Germany. It won't stop them and reconsider, it can't until they reach their destination, ony then it will reconsider whether to redeploy them again. So it is quite possible that when it gives orders to say 20 divisions to move to the Commune borders at that instance it doesn't know there is a war with Germany. It certainly doesn't know in advance. Germany has no war goal, and there is nothing that would alert the French battle plan AI to prepare for Germany invading within days.
But let's surmise that the civil war and the war with Germany do actually start simultaneously, and that instead of two fronts, the AI has to consider four (the Commune borders, the German border and the "home front"). Germany only has one border with France and it's colonies, it doesn't have anywhere else round the world it can attack overland. But it also has a (strong) navy. Not UK, USA and Japan strong, but the potential equal of France? Especially if France has lost significant naval forces to the Commune. So it's now more complicated for France.
A player might handle this differently, but I'm not so sure they would despite the very strong statements in the thread that the ONLY thing that matters is one province in the Maginot Line. But a human might better predict what COULD happen in the future than the AI. The AI has to make an instant decision on the information available to it. Presumably it can "see" from it's intelligence the strength of the Commune, and how many divisions are close to Paris. But what can it see in Germany?
Fog of war works both ways, and the AI also follows the rules. The people watching the video can see all of the German divisions, but the French AI cannot. It only has limited information.
But what is surprising here is again the force of the argument when everyone watching the video knows that the whole event sequence is around the existence of the DMZ in the Rhineland, and can see that before the war breaks out the German player has a small amount of troops along the Rhine in Wurttemburg but NONE in Mosseland. None because it is a DMZ, and the German player is not allowed to move any divisions there. In the event sequence, civil war and war, the DMZ does get removed. But you can see that it takes a little while that the German player decides to move some of his forces to Mosseland, and longer for them to get there (because he doesn't seem to understand how to use SR
So the AI has limited information about Germany because of fog of war, and what information it has tells it that there is no immediate threat to it's border. It doesn't know that troops are on their way to Mosseland, only that there are none there now. Why should it prioritise those provinces above all others, when there is not even any German divisions there to capture them. But that doesn't mean that the AI is ignoring them. It could be that it thinks it has time to get divisions from elsewhere to plug the gaps.
Following in the wake of civil war, war with Germany with no notice, and the confused intelligence because of the DMZ, the French AI starts moving many/most of it's forces to try to counter these dual threats. Then suddenly a week into what ever redeployments it is doing (none of which can be seen in the video), Italy also declares war. And because of the mechanics this also is with no advance warning. Italy doesn't have a wargoal, the French AI can't possibly know that this is going to happen. This is not actually triggered by the events but because Italy (ie the much maligned AI) makes a strategic decision that invading France is a good decision. You have to admit that, from the Italian point of view, this is actually a smart move by the Italy AI. I have no reason to doubt these figures kindly provided by Rommel 459, though I didn't see that myself:
So we see that the Italian AI didn't "attack" France, so didn't need a wargoal. It joined the Axis and seems to have then joined Germany in a "defensive" war. Did you expect that, did the player in the video, did the AI? This is so counter-intuitive because we see that Germany "invades" France, but the mechanics treat France and the UK as the aggressors. You might question that mechanic possibly, though I can see for and against. But it has nothing to do with the AI. The French AI didn't decide to DoW Italy, it just has to deal with the consequences when suddenly and without warning it has another war on it's hands - and what a war!
Unlike Germany, Italy starts in 1936 with plenty of forces on it's border with France. My estimate is that it probably has more forces than Germany has along the Maginot Line, because Germany is still deploying into the former DMZ, and bringing forces from the Czech and Poland borders. So in the middle of the French AIs major redeployment of most of it's forces there is suddenly a fifth front. And it is possible that intelligence available to the AI tells it this front has the most enemy forces. Plus Italy has a strong Navy in the Med and also a strong air force, so the southern coast line is also now strongly threatened by an Italian invasion.
Unlike Germany, Italy has more than one border with France on 1.1.1936. However, we can see in the video it's colony in Libya no longer borders French Tunisia, because of the Commune. But Eritrea (where there are Italian forces involved in the war with Ethiopia) borders with the French colony in Djibouti. So that's a sixth front. And of course Tunisia and core French territory in Sardinia could be invaded from Italy.
Threats there might be low. But the AI is going to start to really regret it's decision to pull units from the Italy border, if that is what is has done. The Italians, unlike the Germans, are already in position. They have probably had an offensive battle plan preparing since 1.1.1936, even if the chances of executing it seemed small. But immediately when they enter the war, they can attack in force.
The AI must respond to all of these things. And it must do it with less forces than it had just a week earlier - three wars, with six fronts. and all started with no advance warning.
It's really beyond me to think how anyone could think that the French AI is "bad" in those circumstances. They should place themselves in that situation as if they were playing as France and these things happened to you. My feeling is that 90% of players would have rage quit and restarted the game, or at least gone back to their last autosave before the war.
Of course, going back a month might not help, a France player could end up in the same situation. But only if they are stupid. Once bitten twice shy and all that...
The German national focus Rhineland kicks off this situation. It is part of the historical AI focuses that Germany under AI control will always take early, and the German player in the video also followed that historical path. Therefore, it is likely that in well over 95% of games this event sequence will start to take place. Yet we have never seen these wars happen before, not just in games where the devs or Youtubers played as France of Germany, but in any of the other countries. We know this because the News Event pops up telling you (whatever country you are playing) that Germany has successfully re-occupied the Rhineland. We can assume it is exceptionally rare.
But a player France? When they get informed that Germany has completed the Rhineland NF they are presented with two options. One to just accept it, the other to decide "This is an Act of War". And the tooltip tells them in red "If war breaks out, we will suffer a civil war due to our Disjointed Government".
You have been warned if you play France. If you want to try to prevent Germany reoccupying the Rhineland you will probably get screwed over in the same way with a three country, six front war, and lose a significant part of your forces.
Very nicely thought out and writen post.
- 3
- 2
@potski Thank you for your elaborate analysis, but your assumption seems to be that it's an one-off case and that it was completely justified. That does not seem to be the case when watching many of the other gameplay videos.
I was, though, pleasantly surprised when I saw in one gameplay video that Romania, that joined Axis, was not called to war until Poland (that had defeated Germany) started fighting the Soviet Union. Germany called Romania into the war/or Romania offered to help at that time shortly after the Soviet Union was involved, thus enlarging the front and getting a proper fighting chance, so the call to arms was actually strategic.
I'm not denying the AI looks pretty decent, but the fact that there's often cases of the AI improperly defending multiple fronts is just that - a fact, because it's shown in so many different gameplay videos that I have watched.
You can make any excuses you want for the French AI, but I and many others will never consider abandoning the Maginot Line an acceptable decision. The Germans did not exploit the opening for months, the French had plenty of time to fix it. When the German tanks arrived at Paris, there were tons of spare divisions not being used for the war effort. That is not just an assumptive observation, there were clear facts that pointed towards this enormous AI failure. There was no chance of the communists pushing forward & they didn't even do it due to the minimal odds, yet the French still overprioritized the front.
If I was placed into this situation, I would actually care about the Maginot Line and be much smarter in a multiple front war, because, you know, I'm a human, it's much easier for a human brain, so comparisons to us are a waste of time (maybe some casual would completely fail that scenario despite some superiority). Also, FYI - France declared war on Germany, not the other way around.
You also fail to realize the major point here - I'm not saying the AI is bad, it has a lot of good aspects. I'm saying there is an obvious programming error in how the AI handles multi-front wars and the Maginot Line. It needs to be fixed and then the AI becomes much better because this aspect then is not completely broken.
I was, though, pleasantly surprised when I saw in one gameplay video that Romania, that joined Axis, was not called to war until Poland (that had defeated Germany) started fighting the Soviet Union. Germany called Romania into the war/or Romania offered to help at that time shortly after the Soviet Union was involved, thus enlarging the front and getting a proper fighting chance, so the call to arms was actually strategic.
I'm not denying the AI looks pretty decent, but the fact that there's often cases of the AI improperly defending multiple fronts is just that - a fact, because it's shown in so many different gameplay videos that I have watched.
You can make any excuses you want for the French AI, but I and many others will never consider abandoning the Maginot Line an acceptable decision. The Germans did not exploit the opening for months, the French had plenty of time to fix it. When the German tanks arrived at Paris, there were tons of spare divisions not being used for the war effort. That is not just an assumptive observation, there were clear facts that pointed towards this enormous AI failure. There was no chance of the communists pushing forward & they didn't even do it due to the minimal odds, yet the French still overprioritized the front.
If I was placed into this situation, I would actually care about the Maginot Line and be much smarter in a multiple front war, because, you know, I'm a human, it's much easier for a human brain, so comparisons to us are a waste of time (maybe some casual would completely fail that scenario despite some superiority). Also, FYI - France declared war on Germany, not the other way around.
You also fail to realize the major point here - I'm not saying the AI is bad, it has a lot of good aspects. I'm saying there is an obvious programming error in how the AI handles multi-front wars and the Maginot Line. It needs to be fixed and then the AI becomes much better because this aspect then is not completely broken.
- 1
The reds were no friends of Hitler and would not want to see his troops in France. Rather than disruption from political log jams, strikes, protests, etc., that results in the customary paralysis, an automatic civil war in reaction to a proposed limited military action against Germany's treaty breach seems excessive unless these events are simply a pretext for a red takeover.
If it was an attempted takeover and the Germans went on the offensive with serious forces, then for the Republic defending Paris would be critical and maneuvering to put the Commune between the Republican troops and the anti-communist Germans could be a reasonable step in dealing with both problems.
In reality, France in disorder could not make credible demands and the Germans would probably just march in and treat any French missives as not having been sent.
If it was an attempted takeover and the Germans went on the offensive with serious forces, then for the Republic defending Paris would be critical and maneuvering to put the Commune between the Republican troops and the anti-communist Germans could be a reasonable step in dealing with both problems.
In reality, France in disorder could not make credible demands and the Germans would probably just march in and treat any French missives as not having been sent.
Hopefully some issued can be solved relatively easily by tweaking AI priority, such as always having at least some men guard the border towards a big threatening neighbour even during civil war.