SC2 Suggestion - Probing AI

[Edwin P.]

Often in a game against a human player I will land a corps to take an unguarded city behind enemy lines, once I even managed to take Rome.

Of course the secret to doing this is to patiently wait until you believe that the enemies forces are occupied elsewhere, and you should not always use the same strategy otherwise your opponent will be waiting for you.

If the AI could adopt a variant of this strategy it would make things really interesting. Currently the AI seems only to follow this strategy towards the Norwegian city of Bergan.

Some tactics for the AI to consider:

1. Transport a russian corps to the shores of Romania and land it to take Bucharest in one turn.

2. Transport a UK corps to the shores of Northern africa and land it to seize an Italian city in one turn.

3. Transport a UK corps near the capital of Italy and land it to sieze the Rome in one turn.

4. Land a Corps on northeastern germany to sieze one of two German cities.

Now the hard part for the AI will be timing and gathering intelligence to minimize the chance of sending a corps on a suicide mission. Before I land a corps near Bucharest I move my Russian warship near the coast to see if the city is empty and ungarded. Then I move in my corps. In Noth africa I will use a bomber or air fleet to see if an Italian city is ripe for plunder. The only city I can't do this for is Rome, but sometimes this gamble can turn a losing game around rather quickly.

[ May 10, 2004, 02:16 AM: Message edited by: Edwin P. ]

[James B.]

This isn't even realistic. In real life, would the Allies (or Axis) have landed a single corps in a city just to have it smashed a few days later, no! We need to maintain historical realism, not tear it up!

[norvandave]

I agree with James B. A corp is thousands of soldiers who wouldn't be sacrificed so easily. I don't think there were any such "invasions" of that size unless they were part of a larger attack.

[Saukopf]

Approximately 5000 Canadians had landed at Dieppe; 100 were wounded, 1,900 were evacuated , 900 were killed and 1,874 were POWs . Only 2200 made it back to England.

[Edwin P.]

Not Realistic? You threaten to land units behind enemy lines to force the enemy to guard his rear, cut his supply and/or weaken his front lines.

If you can't threaten the enemies rear then your your opponent would not station troops there, I believe that is more un- realistic.

You may seize a city in the rear in the hopes of holding it until reinforcements can arrive - Operation Market Garden. Of course if they don't arrive .....

During the Korean War the US landed troops behind enemy lines at Inchon. This caused the North Korean front lines to collapse.

During WWII the Allies landed troops under the command of General Lucas behind enemy lines at Anzio in Italy. Eisenhower removed him for not advancing off the beach and seizing Rome, which was undefended at that time.

[ May 11, 2004, 02:41 PM: Message edited by: Edwin P. ]

[kurt88]

We need to maintain historical realism, not tear it up!

Hi! Nice to have anew poster! Welcome to the forum . Good to see you appreciate Edwin's suggestions as much as we do.

[SeaMonkey]

I'm in agreement with you Edwin, not that it carries any clout. My abstractly constructed mind says this is as good as a simulation of commando raids we'll get out of any SC version. Maybe not detailed realism, but in the spirit of SC, its good enough for me, how about for YOU! :eek:

[Edwin P.]

Good enough for me. I play SC for a fun game, and if its too predicatable - ie the AI will never make a surpise behind the lines landing - it loses some of its fun factor.

Besides, I find my best games against human players are when they do something unpredictable - ie landing 5 corps behind my lines on the northern coast of Germany while at the same time landing in France.

Something the AI never does, it will never invade North Western Germany via Sea, while landing at Brest. The narrow focus of its landings in Western Europe always allows the Human player to concentrate his forces. (Hint for HC). It would be nice if the AI had several invasion strategies to choose from.

Also the AI needs to time its invasions better - ie wait for an American HQ to be built and it needs to withdraw its invasion forces if German resistance is overwhelming.

[ May 12, 2004, 12:38 AM: Message edited by: Edwin P. ]

[Night]

The A.I in SC1 was not to good at all. I have played it many many times and always get the same "D-day" landings. Almost always the Allies don't land an HQ on the first turn of the invasion, and often wait until they take Brest to do so, meaning many of their forces are choped up peice meal with no supply or readiness.

Repeling an A.I. launched D-day assult is a joke, they never use tanks, air power is mis-allocated, HQs are under-used, additional landings behind the lines (if the allies are traped in Brest with a shrinking pocket the A.I. never seems to releave this preasure with a landing elsewhere in France)

Also, how come the Allied A.I. never lands in Italy? It would be so much more interesting. All these things should be fixed with SC2

[James B.]

I understand that landing behind enemy lines was an important tactic in world war ii, but the fact remains that a single corps would not have been able to to just land and take a city. Every city was defended. Now, I understand that human players will always be able to try this tactic, but I think the computer should be barred from this attempt, as it was (or just didn't try) in SC1.

[Edwin P.]

Anzio

The Anzio invasion began at 0200 on 22 January 1944 and achieved, General Lucas recalled, one of the most complete surprises in history. The Germans had already sent their regional reserves south to counter the Allied attacks on the Garigliano on 18 January, leaving one nine-mile stretch of beach at Anzio defended by a single company. The first Allied waves landed unopposed and moved rapidly inland. On the southern flank of the beachhead the 3d Division quickly seized its initial objectives, brushing aside a few dazed patrols, while unopposed British units achieved equal success in the center and north. Simultaneously, Rangers occupied Anzio, and the 509th Parachute Infantry Battalion seized Nettuno. All VI Corps objectives were taken by noon as the Allied air forces completed 1,200 sorties against targets in and around the beachhead. On the beach itself, the U.S. 36th Engineer Combat Regiment bulldozed exits, laid corduroy roads, cleared mines, and readied the port of Anzio to receive its first landing ship, tank (LST), an amphibious assault and supply ship, by the afternoon of D-day. By midnight over 36,000 men and 3,200 vehicles, 90 percent of the invasion force, were ashore with casualties of 13 killed, 97 wounded, and 44 missing. During D-day Allied troops captured 227 German defenders........

Yet the campaign did accomplish several goals. The presence of a significant Allied force behind the German main line of resistance, uncomfortably close to Rome, represented a constant threat. The Germans could not ignore Anzio and were forced into a response, thereby surrendering the initiative in Italy to the Allies. The 135,000 troops of the Fourteenth Army surrounding Anzio could not be moved elsewhere, nor could they be used to make the already formidable Gustav Line virtually impregnable. The Anzio beachhead thus guaranteed that the already steady drain of scarce German troop reserves, equipment, and materiel would continue unabated, ultimately enabling the 15th Army Group to break through in the south. But the success was costly.

I think the key is that the threat of being able to land divisions/corps behind enemy lines forces the enemy to divert resources from his front lines. If the AI is not allowed to threaten the human player's rear line then the human player is free to concentrate all of his forces at the front lines, something that was not done in real life.

In SC1 Human players would never station any troops in Italy or Beglium as they new the AI would never invade there, a most unrealistic situation.

Just a threat of the AI being bold and audacious means that the human player must hold reserves in his rear, much as happened historically. This in turn weakens the human players front line and makes for a more interesting game.

[ May 14, 2004, 09:51 AM: Message edited by: Edwin P. ]

[Edwin P.]

Originally posted by Night:

The A.I in SC1 was not to good at all. I have played it many many times and always get the same "D-day" landings. Almost always the Allies don't land an HQ on the first turn of the invasion, and often wait until they take Brest to do so, meaning many of their forces are choped up peice meal with no supply or readiness.

Repeling an A.I. launched D-day assult is a joke, they never use tanks, air power is mis-allocated, HQs are under-used, additional landings behind the lines (if the allies are traped in Brest with a shrinking pocket the A.I. never seems to releave this preasure with a landing elsewhere in France)

Also, how come the Allied A.I. never lands in Italy? It would be so much more interesting. All these things should be fixed with SC2

I strongly agree with everything that you said.

The AI's D-Day is very weak. Part of this is because if the AIis landing in France the Axis can strip all other areas, including Italy, of units as it knows that the AI will never attack there.

Another issue, the AI does not know how to bluff. It will never move units adjacent to the coast and then withdraw them without landing.

Just the action of threating a landing weakens that Axis as it is forced to proceed on the assumption that an invasion is coming.

Additionally, the AI never launches a serious effort to liberate the Nordic Countries or even gain experience by attacking any unit in Bergan. This failure on the part of the AI allows the Axis to protect Norway with a minimum number of units.

It should be noted that in SC2 the AI will have access to amphibious units that can land on the same turn that they move adjacent to the coast.

------------------------------------

Norwegian Liberation Strategic Options for the Allied AI

1. Bombard Bergan with Ships and Carrier Air if no Axis Air in Area.

2. If have LR2 & Russia at War then move Air Units to Northern UK and Attack Bergan with Land Based Air and Carrier Air and Land Corps to take Bergan.

----------------------------------------

Another Issue is that the AI does not know how to use the Intelligence Reports. If it knows that the Axis has six air units and all six have been spotted on the eastern front then it can confidentialy move its carriers to attack in the west. Nor will it use its bomber unit to spot the location of enemy units prior to invading France. Nor will it use land based air fleets to draw off interceptors before making carrier based attacks.

[ May 14, 2004, 11:40 AM: Message edited by: Edwin P. ]

[Retributar]

I really like your ideas and suggestions 'Edwin P.'... idea-wise it's a great-idea...but, it might be very difficult to come up with an AI programme that can do all that!.

Maybey it can be done!. Now to continue on your line of thought...'Saukopf' mentioned the Canadian Raid at Dieppe...and it was just that...a RAID, not an INVASION.

What im thinking of is that the AI could have choices...to use various sized smaller raiding parties like the 'British Commando's' & the 'Canadian' invasion for harrasment tactics to de-stabalize the Germans (by forcing Force-Buildup in the Area) and also use Larger Force Sizes when necessary for actual-invasions...like that at 'ANZIO'.

The Allied-Invaders should also have a work-in to with-draw if resistance is too stiff!.

---------------

As for 'INTELLIGENCE' regarding troop-deployments..., yes...the Allied-AI could use Probing air-reconnaissance efforts as well as using the 'Free-French' resistance fighters (Invisible to the Germans) as a resource to help determine where the German Forces are stationed!.

Free-French Units would be 'INVISIBLE' and would not be restricted by the zone-of-control from enemy units. They could become visible once they have detected the actual composition of an enemy unit.

To Detect the 'ACTUAL' make-up of the enemy...the resistance unit would bump up against a grey-colored unit...but, would not know what it is...unless you opt to do a close-in 'RECON' fact-finding mission.

How that concept could really work... or should it be at all in the game is just a 'Who-Knows'...would have to play-test it to see if that idea is worth anything!.

You would only have a very few of these units in the game ...and so would need to use them 'ONLY' when absolutely necessary...like for the Invasion Of Europe!.