Optimal Western Front Invasion Tactcs for AI

[Edwin P.]

What changes in the AI would be needed to make it more likely to launch a successful invasion of France?

Santabear stated in another post that

The AI ... was very persistent about trying to invade France when the Russian battle was at its height--three attempts; on the third he kept sending in ships and transports as fast as I could air them into oblivion. And lots of Allied air support, too.

The overall strategy would have defeated the Axis if it had been timed a little better--the invasion of France actually started after the Russian counter attack had been got under control.

I would consider having the AI send tansports to the borders of France and then withdraw them if the Germans reinforced with troops, instead of landing anyway and being destroyed.

I would have the AI hold transports ships at sea if there are no German subs in the Atlantic, and then landing along a wide front at the same time, instead of making small isolated landings.

The AI should make diversionary invasions along the coast of northern Germany / Belgium/ Denmark when garrison troops are not in the area.

I would have it consider making a serious invasion attempt at Norway after the US enters the war. If the Allies control Norway then I would give the Russians an extra 20MPP per turn, reflecting increased merchant ship convoys traveling to Russia.

The timing of the AI invasions also needs to be improved. To often it attacks at the wrong time. Too early or too late. it needs to take into account the forces it has available and the situation on the Russian front and the forces stationed in France and Italy.

Perhaps, it should consider invading Italy if the Italian navy has been destroyed or making a three front attack on France - Norther France, Western France, and Southern France.

[ April 15, 2003, 07:32 PM: Message edited by: Edwin P. ]

[santabear]

Edwin--all good points and ideas; especially about the "Dieppe raids." If the British are too passive during the campaign in Russia they'll lose every time. I'm sure human players for the Allies are much more aggressive during Barbarossa.

My guess is that there are three basic modes (passive, middlin', and aggressive) for AI that influence all of his decisions, and that one of these modes is selected early in the game (or perhaps before play starts) randomly.

The aggressive AI will always attack by air when it gets the chance, and will try mulitple Overlords. There is also a "sneaky aggressive" AI that will build air units in northern Britain, then try to swoop transports, ships and air in quickly. The aggressive AI is not deterred easily by losses; it will persist in operations and repeat them until very substantial losses have been inflicted.

The problem that I've seen is that the aggressive AI is often too aggressive for its own good, especially with British naval units. It's also too easy to gang up several German air units to take out the British/American air in southern Britain.

The most enjoyable games, though, are the ones against the aggressive AI, because they offer the most challenge to strategic decision making--setting priorities, allocating resources, temporizing, etc.

In the latest patch, Hubert said the tweaked the AI to attack less when surrounded in cities--which is a good start. Perhaps later AI's could have another semester at staff college to become more sophisticated about unsustainable loss rates. If there is a way to improve communication between Stalin and Churchill that would help, too.

Then an aggressive AI would become a very formidable opponent.

AI should also "look before it leaps" more carefully in the "opportunity" phase of its turn. Too often, it will zip an army onto a mine hex, placing it in a position to be wiped out the next turn. The mine hex in the middle of Russia is nice for baiting the AI into depleting his forces piecemeal.

[ April 16, 2003, 05:38 AM: Message edited by: santabear ]

[Edwin P.]

Santabear - thanks for the background on the SC AI. It was most interesting.

I agree that the AI should look before it leaps more often and that it is too aggresive with its naval units.

One strategy I have used is to base six air fleets in Denmark and then send a lowly corps transport or Cruiser into the north sea. The British fleet will move in to attack it and be destroyed by air units and waiting German naval units. After tempting the British fleet into range several times the entire British fleet is destroyed.

Once that occurs Germany is free to field a submarine force that has a reasonable chance of stopping any invasion of France, especially since the Allies will not produce any more naval units to sink them.

It would be nice if the UK Naval AI in the North Atlantic could be tweaked to avoid sailing into range of an overwhelming air fleet attack force that is certain to destroy them and is safe from attack. A better strategy would be to wait in the North Atlantic for the German fleet to come to you.

Another strategy that I have used against the AI is to hold Normandy with a single Canadian army supported with air and carrier units and battleships (and transporting in fresh army units). The AI will continue to launch attack after attack on a one hex front even through it loses a unit every time. If only it would move in all 3 of its air units to support the attack it could take the city. The current AI only allocates 1 or 2 air units to support the attack which leaves the defender with a strength of 2 or 3. Just enough to hold the city.

A similar situation holds true if the Allies seize Sicily and defend that one hex front that links it to the Italian mainland.

[ April 16, 2003, 07:45 AM: Message edited by: Edwin P. ]

[santabear]

I like your Normandy "rope a dope" move.

Just to be clear about my contribution: That is just my guess from having played a lot against AI. I don't have any real 'background' information--the game designers may be laughing hysterically at that post (well, perhaps giggling a bit). It does fit my experiences, though.