Idea for realism and AI assistance

[bbasgen]

Last changes: Oct. 16, 2008 (vortinex, lazy_boy,wulf145)​

The problem: The AI is incapable of quickly reacting to strategic changes. This creates several AI weaknesses, the most obvious being the overrun strategy and mass encirclement.

A solution: Include a game option (e.g. check box) "Tactical Control: on/off". Tactical control 'on' is normal gameplay. Tactical control off removes a few key controls for the player, including:

• Maneuver Warfare: Provide combat bonuses for long term, in-depth, multi-province maneuvers that are executed. Long term, static orders cannot be less than a X province minimum, which provides the AI opportunities to outflank encircling units, for example.
• Conventional penalty: Increase "re-org" time after a division captures a province in the conventional method (e.g. one province at a time), in order to provide greater appeal/need for multi-province maneuvers, which would have zero re-org time.
• Command Delay: Enforce a X hour period per day for issuing orders. This will help keep player controlled units "locked" in to specific orders for a given period of Y hours per day.(updates from Lazy_boy)
• Doctrine effects: Land doctrines provide different bonus types/amounts of multi-province maneuvers (e.g. Germans excel early on), including division re-org times (e.g. Germans can take a province and move on *much* quicker than the USSR can in 1941), and command delay. (Lazy_boy suggestion)
• Encirclement penalty: (optional and likely controversial) For encircled divisions, have X chance they become "disorganized" and either revert to AI control or are slow to respond to commands, reflecting an inability to communicate with the divisions and issue them orders. Increases in communication technology lowers X chance. (updates from vertinox)
• HQ control: (optional and likely controversial) Limit player control of divisions according to HQ proximity. If an HQ is not within X provinces of a division, that division is under AI control. Alternatively, "out-of-range" divisions are still under player control, but have X chance of acting out on leader initiative (especially for divisions over command limit). (updates from wulf145)

Benefits:

1. This will help the AI when a player "locks-in" to a multi-province maneuver. The AI will have a better chance against the traditional double envelopment, for example, when one of the player's flanks advances rapidly, while the other is bogged down in fighting.
2. This will allow added realism in the form of strategic command: the player simply can't control every last detail of the war effort. Locking players into long term daring maneuvers provides a level of risk commensurate with the level of reward.
3. This makes for quicker and easier games for newbies because pausing and slowing down the game to micro-manage every hour of the battle field isn't necessary or even wise.
4. Multi-province maneuvers better simulates the fog of war element: you shouldn't know the exact status of the battle field immediately after this or that battle is completed
5. Multi-province maneuvers better reflect the advantageous and perils of blitzkreig
6. Allows for improved modeling of morale, in addition to the already modeled attrition
7. Assuming the last option of HQ control was modeled, this would vastly increase the importance of HQ, both to the player and their opponent!
8. The HQ control method could also help the newbie, in that they may choose to have some units under AI control.
9. HQ would need to be behind defensive lines, as well as assisting offenses. This could create an entirely new strategy of destroying/attacking an HQ, causes a collapse of a defensive line.

 

[Ironhead 5]

Some good thoughts, esp regarding controlling encircled units and reduced efficiency for quickly changing orders. However, instead of slowing down friendly units, I would rather see a smarter AI -- ie, when the linear front is broken by fast-moving ARM/MEC/MOTs, it knows to withdraw to avoid encirclement.

 

[Wulf145]

[*] (optional and likely controversial) For encircled divisions, revert them to AI control reflecting an inability to communicate with the divisions and issue them orders.
[*] (optional and likely controversial) Limit player control of divisions according to HQ proximity. If an HQ is not within X provinces of a division, that division is under AI control.
[/list]

[/list]

I find the ideas good as long as they are optional.

Due to the fact that we are talking about a time period in which Radio Comunication was widespread, I do not see the sense in 'removing' encircled divisions from the players controll. In history break out attacks were coordinated between the encircled units and units triing to break them out.

 

[unmerged(52476)]

The problem: The AI is incapable of quickly reacting to strategic changes. This creates several AI weaknesses, the most obvious being the overrun strategy and mass encirclement.

A solution: Include a game option (e.g. check box) "Tactical Control: on/off". Tactical control 'on' is normal gameplay. Tactical control off removes a few key controls for the player, including:

• Enforces a X hour period per day for issuing orders. This will help keep player controlled units "locked" in to specific orders for a given period of Y hours per day.
• Provide combat bonuses for long term, in-depth, multi-province maneuvers that are executed. Long term, static orders cannot be less than a X province minimum, which provides the AI opportunities to outflank encircling units, for example.
• Increase "re-org" time after a division captures a province in the conventional method (e.g. one province at a time), in order to provide greater appeal/need for multi-province maneuvers, which would have zero re-org time.
• (optional and likely controversial) For encircled divisions, revert them to AI control reflecting an inability to communicate with the divisions and issue them orders.
• (optional and likely controversial) Limit player control of divisions according to HQ proximity. If an HQ is not within X provinces of a division, that division is under AI control.

Benefits:

1. This will allow added realism in the form of strategic command: the player simply can't control every last detail of the war effort
2. This makes for quicker, and easier games for newbies (lower learning curve)
3. Allows players to focus on a purely strategic element to gaming
4. Multi-province maneuvers better simulates the fog of war element: you shouldn't know the exact status of the battle field immediately after this or that battle is completed
5. Multi-province maneuvers better reflect the advantageous and perils of blitzkreig
6. Allows for improved modeling of morale, in addition to the already modeled attrition
7. Assuming the last option of HQ control was modeled, this would vastly increase the importance of HQ, both to the player and their opponent!
8. The HQ control method could also help the newbie, in that they may choose to have some units under AI control.
9. HQ would need to be behind defensive lines, as well as assisting offenses. This could create an entirely new strategy of destroying/attacking an HQ, causes a collapse of a defensive line.

...or, alternatively to crippling the players control over the game, you could program a better AI

I have no doubt that your suggestions will make the game more difficult, but I really think that the focus should be on improving the AI and not resign and hinder the player's abilities instead. Some of your suggestions sound like house rules or can even easily be modded. I use, for example, a 48h time lag until a new command can be given. It makes the game a little bit trickier, but after all, the AI will make its dumb moves anyway, you just have more time watching it...
I guess my point is that a dumb AI is no fun, no matter how difficult the game is. The only way to change that is make a better one.

 

[Darth Tracid]

the big problem with every AI is: it has an inherent "way" in which it operates. No matter how good that "way" may be, after X games the player will know what the AI will do and how he can exploit that knowledge.

What we REALLY need is a "forget everything I know about history and the AI in the game" button for your brain you push before each game

 

[Ironhead 5]

Due to the fact that we are talking about a time period in which Radio Comunication was widespread, I do not see the sense in 'removing' encircled divisions from the players controll. In history break out attacks were coordinated between the encircled units and units triing to break them out.

I'm not sure how small the provinces will be in HoI3, but in HoI2 they were a lot larger than the range of terrestrial radio signals. That's not to say they didn't have ways to communicate -- through skywave radio signals, for example -- but I agree with the OP that the player shouldn't have full tactical control over the encircled unit.

 

[Wulf145]

I'm not sure how small the provinces will be in HoI3, but in HoI2 they were a lot larger than the range of terrestrial radio signals. That's not to say they didn't have ways to communicate -- through skywave radio signals, for example -- but I agree with the OP that the player shouldn't have full tactical control over the encircled unit.

SW radio signals travel several thousand kms, and as far as I am aware every divisional HQ had at least one of those sets (major powers). I think it would be extreemly frustrating and a bit 'gamey' not being able to issue general orders if the unit gets encircled. It would render Para Divisions useless for example since once you have droped them beheind enemy lines you cannot do anything with them.
If one would would implement this feature then you would have to have the possibility to enter 'standing Orders' i.e. if encircled move/attack west.
IMHO it will be easier to implement a more flexable Ai than to implement such a feature with all the konsequenses that it entails.

 

[vertinox]

I find the ideas good as long as they are optional.

Due to the fact that we are talking about a time period in which Radio Comunication was widespread, I do not see the sense in 'removing' encircled divisions from the players controll. In history break out attacks were coordinated between the encircled units and units triing to break them out.

Historically, during 1941, most Soviet units encircled lacked effective communication among themselves much less Moscow. (Remember even Soviet tanks lacked radios)

This was not so much and issue towards the end of the war.

That said, I would like realism when it comes to communications and units taking their own initiative.

U-Boats were basically independent while they were at sea and passively took instructions to avoid detection. HQ didn't know where a sub was or if it was still alive until it returned to base.

Not to mention isolated Japanese island garrisons had no communication with HQ either.

Personally, I'd like to see historical leaders react to orders and sometimes react in a different way that they are told, but not in an annoying way.

Lets say you put Rommel in charge of a Division in Africa with the only orders to stay still. If a certain formula is met Romel will auto attack with the unit, but gain an initiative bonus which actually benefits the player instead of attacking an unwiniable battle.

Secondly, certain leaders should refuse to attack if the situation is determined to be suicide (like when Paulus was ordered to break out of Stalingrad and he felt that they simply could not).

 

[bbasgen]

On a "smarter" AI: We all agree we want an improved AI. Tracid is correct to point out, however, that an AI can be learned by the player, especially one that is file-driven rather than heuristic. Thus, the real question is how can a static AI (forgive my ignorance of proper terminology) be improved? It seems to me possible to make gameplay changes that improve realism and reduce micromanagement that also incidentally help the AI.

On "houserules: One of the biggest components to this suggestion is to allow for some amount of long-term strategic planning in order to simulate real-life historical strategies. For example, imagine you have a northern and southern tank army attempting an encirclement, both issue this special 'in-depth maneuver' order, and planned to "meet up" four provinces deep behind the enemy front. This allows for very rapid movement and combat bonuses. Your northern army does exactly as you planned, and is ready for rendezvous in order to collapse the pocket. Your southern army, however, is engaged in fierce fighting and cannot breakthrough!
This is the key. Encirclements are, by definition, quite risky! Yet, AI incompetence combined with the players ability to have unrealistic battlefield command and control make encirclements a no risk, big reward proposition. In this example, your northern tank army is in a compromised position, as well it should be! This doesn't even require a "smart" AI per se, it just doesn't give the player god-like control over warfare.

On encircled divisions: This is a minor idea. The suggestion is to add an layer of difficultly (in addition to being out of supply) in carrying out intelligent orders when in such a compromised position. I think that adds to realism, and it evens the playing field with the AI a bit more. In any event, it need not apply to paratroopers. As vertinox explains, it would allow for historical modeling of leader initiative, which played a very significant role throughout the war.

I think these suggestions can be one of those rare "options" that improves historical realism, immerses your more in the battle, while also creating an easier learning curve.

 

[unmerged(58571)]

The AI, if set to attack desperately when enircled (i.e. set to count encircled units as abnormally strong) makes two hardcoded mistakes that ought to be easy to correct in HoI3 or even in an eventual HoI2 patch:
1. It attacks the weakest surrounding province, disregarding the actual direction to supplied provinces.
2. It leaves divisions on garrison duty in the encircled province during the attack.

 

[vertinox]

On encircled divisions: This is a minor idea. The suggestion is to add an layer of difficultly (in addition to being out of supply) in carrying out intelligent orders when in such a compromised position. I think that adds to realism, and it evens the playing field with the AI a bit more. In any event, it need not apply to paratroopers. As vertinox explains, it would allow for historical modeling of leader initiative, which played a very significant role throughout the war.

Perhaps it should be random and not a 100% type of deal. Lets say an encircled unit has a 1 in 3 chance of becoming "disorganized" which looses its ability to take commands or respond to commands in a timely manner.

If you invest in radio techs, then it decreases the chance to say 1 in 6 or so.

 

[Lazy_Boy]

I like the idea. One thing I've never liked about HOI is that launching attacks is like a spur of the moment thing you can do on a mass scale instantly.

There definitely should be a command delay. And also give you large bonuses to assaults if they are planned routes ahead of time. More time in preparation increases effectiveness. Otherwise the defenders should have a large advantage. Could add all kinds of depth to anticipating your opponent and making intel important.

Doctrines should also make a huge impact on those factors too. Great way to give the Germans early effectiveness that slowly dwindles in the face of reorganized huge armies and supply difficulty.

Also need standing orders for armies. Like retreat if below X org, stand till last man, counterattack in defense support.

Just sort of things that makes you feel like you're the grand strategist of the OKW instead of micromanaging a million units enough to trick the AI.

 

[Hansag]

The problem: The AI is incapable of quickly reacting to strategic changes. This creates several AI weaknesses, the most obvious being the overrun strategy and mass encirclement.

A solution: Include a game option (e.g. check box) "Tactical Control: on/off". Tactical control 'on' is normal gameplay. Tactical control off removes a few key controls for the player, including:

• Enforces a X hour period per day for issuing orders. This will help keep player controlled units "locked" in to specific orders for a given period of Y hours per day.
• Provide combat bonuses for long term, in-depth, multi-province maneuvers that are executed. Long term, static orders cannot be less than a X province minimum, which provides the AI opportunities to outflank encircling units, for example.
• Increase "re-org" time after a division captures a province in the conventional method (e.g. one province at a time), in order to provide greater appeal/need for multi-province maneuvers, which would have zero re-org time.
• (optional and likely controversial) For encircled divisions, revert them to AI control reflecting an inability to communicate with the divisions and issue them orders.
• (optional and likely controversial) Limit player control of divisions according to HQ proximity. If an HQ is not within X provinces of a division, that division is under AI control.

Benefits:

1. This will allow added realism in the form of strategic command: the player simply can't control every last detail of the war effort
2. This makes for quicker, and easier games for newbies (lower learning curve)
3. Allows players to focus on a purely strategic element to gaming
4. Multi-province maneuvers better simulates the fog of war element: you shouldn't know the exact status of the battle field immediately after this or that battle is completed
5. Multi-province maneuvers better reflect the advantageous and perils of blitzkreig
6. Allows for improved modeling of morale, in addition to the already modeled attrition
7. Assuming the last option of HQ control was modeled, this would vastly increase the importance of HQ, both to the player and their opponent!
8. The HQ control method could also help the newbie, in that they may choose to have some units under AI control.
9. HQ would need to be behind defensive lines, as well as assisting offenses. This could create an entirely new strategy of destroying/attacking an HQ, causes a collapse of a defensive line.

I like this, being restricted from tactical control (as an option) as well as having HQ's being more important.

An option would also be to have at least two types of HQ's
HQ - Division sized unit, basically the same as in HOI2.
FHQ - Field HQ, an attachment to a division (which may enable other missions for units, such as generating logistical "supply dump" missions etc.).

 

[Ironhead 5]

An option would also be to have at least two types of HQ's
HQ - Division sized unit, basically the same as in HOI2.
FHQ - Field HQ, an attachment to a division (which may enable other missions for units, such as generating logistical "supply dump" missions etc.).

I recall posting in "What would you like to see in a hypothetical HoI3?" type threads a couple of years ago. The conventional wisdom back then, and I hope it still holds true, is that HQ, as a type of unit, has got to go. I mean, EVERY division had a HQ, so why make it a brigade that had to be attached? And EVERY corps, army, army group, etc, had a HQ, so why make it a division? I think I have heard that they are replacing it with a "chain of command" system. That, I think, is the best thing they could have possibly done with it.

 

[Wulf145]

Historically, during 1941, most Soviet units encircled lacked effective communication among themselves much less Moscow. (Remember even Soviet tanks lacked radios)

This was not so much and issue towards the end of the war.

That said, I would like realism when it comes to communications and units taking their own initiative.

U-Boats were basically independent while they were at sea and passively took instructions to avoid detection. HQ didn't know where a sub was or if it was still alive until it returned to base.

Not to mention isolated Japanese island garrisons had no communication with HQ either.

Personally, I'd like to see historical leaders react to orders and sometimes react in a different way that they are told, but not in an annoying way.

Lets say you put Rommel in charge of a Division in Africa with the only orders to stay still. If a certain formula is met Romel will auto attack with the unit, but gain an initiative bonus which actually benefits the player instead of attacking an unwiniable battle.

Secondly, certain leaders should refuse to attack if the situation is determined to be suicide (like when Paulus was ordered to break out of Stalingrad and he felt that they simply could not).

Being no expert on the Soviet Army, I must say that I am surprised that Divisional HQs were not equiped with SW radio, always happy to learn something new .

In not being able to manage divisions directly one would loose a lot of gameplay, IMO.

I like the idea of Commander initiative but it would be tricky to implement as one would have to take into account the radical differences in how Armies were commanded and led during WW2. The Soviet Army was notorious for comanders not even going to the toilet without a Battleplan sanctioned from the Kremlin. The Wehrmacht used the method of "Auftragstaktik" (basicaly everyone was told what was ment to be achieved and one could do it how one thought it was best) and encouraged Soldiers to use their initiative. These are two diametricaly oposed command methods - each would have to be modeled and implemented in the programm with the 'friendly AI' calculating how the divisions are to be used and what they should and can do. For example:
The Germany Player would give an HQ the goal of taking Province X to capture industry, the AI would the try to do so using all rescources attached to this HQ, if something unforseen happens (overlooked Guard Unit) the AI would try to capture the adjasent province since it has almost as much IC. So far so good. The Soviet Player would on the other hand have to order his HQ to take Province X, and plott the paths of each division. Due to comunication restrictions the Soviet divisions would not be allowed to diverge from the plan if something unforseen happened, i.e. a Panzer division was overlooked on one of the flanks. If one looks at how the Armies of WW2 were led and organised many such examples can be found.

Though the ideas from the OP are good I feel that, if followed through they would take more effort to implement than would be required to beef up the AI.

 

[unmerged(58571)]

There are two things I would like to point out:
1. Lack of proper equipment, for communication or anything else, on the tactical level is best illustrated by unit stats and combat modifiers.
2. Not giving an attack order until the time the attack is to begin does not mean it was not planned several game-days, or even weeks, ahead. The plan might be something like "corps 1 is to capture provinces a, b and c, then hold until corps 2 signals the rear is secured, upon which point corps 1 attacks either province d or e, depending on which one seems weaker defended."

 

[Ironhead 5]

There are two things I would like to point out:
1. Lack of proper equipment, for communication or anything else, on the tactical level is best illustrated by unit stats and combat modifiers.
2. Not giving an attack order until the time the attack is to begin does not mean it was not planned several game-days, or even weeks, ahead. The plan might be something like "corps 1 is to capture provinces a, b and c, then hold until corps 2 signals the rear is secured, upon which point corps 1 attacks either province d or e, depending on which one seems weaker defended."

Agreed -- with one caveat. What if something drastic happens suddenly that changes the battlescape, and orders are given to react to it? For example, the Soviet player is surprised by a German counterattack and wants to rush divisions to a province to shore up the defense. Those divisions weren't planning their movement days or weeks in advance; they were notified suddenly, and told to move immediately and with great speed to a certain area. Should there by a greater loss of org associated with this as opposed to an attack that has been planned for weeks or months?

 

[unmerged(58571)]

What if those divisions are a strategic reserve whose commanders have been told to be ready for a march order?
Like others have said, I think spending money and effort on improving the AI is a far better investment than making unit control as complicated as possible for players.

 

[Ironhead 5]

What if those divisions are a strategic reserve whose commanders have been told to be ready for a march order?

Weren't US 101st and 10th Armored strategic reserve in December 1944? Yet they suffered some problems with a rapid deployment order.

I concede, that in game terms, there may be no value added to making control of the units more difficult. But the idea of receiving a movement or attack order and executing it immediately, is impossible. It would take a couple of hours just to disseminate that order down to regiments, battalions, and company level commands. And that totally disregards any time to plan the attack/movement.

 

[bbasgen]

Great discussion everyone. I've made updates to the top post based on input/suggestions from vertinox and Lazy_boy.

On difficult unit control: I'm not convinced these suggestions increase difficulty. No one would be required to use the multi-province feature -- the combat bonuses could be small enough (like combined-arms, for example) that HOI junkies and geeks alike would call it essential, but noobs could get away with not using it without any problem.

Everything else being suggested is not a change to the end-user experience so much as a change to game mechanics (e.g. command delay, improved fog of war, etc).