SC Patch Request - German Sub AI Mod

[Edwin P.]

The two German subs in the North Atlantic execute the same moves in every game and are usually destroyed by the British Navy at the end of Turn 2.

I would like to see these two subs select a random stategy each game;

A. 20% - Current Routine

B. 20% - Move to interdict sea lanes

C. 20% - Move to north west Atlantic and then do Strategy A or B or D or E.

D. 20% - Move to South Atlantic to intercept transports heading towards the Mediteranean Ocean via the transit hexes or Gibraltar

E. 20% - Move to South Atlantic and then execute Strategy B or C or D

This would make the UK hunt for these subs a real mouse hunt and the game more interesting from a naval point of view. The chance that the German subs might move to the South Atlantic also means that the UK fleet can't limit its anti-sub search to the north Atlantic.

[ May 10, 2003, 02:04 PM: Message edited by: Edwin P. ]

[pzgndr]

Edwin, all interesting ideas as always. Most of the current sub strategy problems in SC result from the too-small map. Subs can't break out of the North Sea into the Atlantic, and can't maneuver much when they get there. A larger map in SC2 will permit more options.

At the end of the Designer Notes in the Strategy Guide, Hubert describes in general terms how the AI works. What's interesting is that it thinks on 3 different levels, so it's deciding on its own what to do and how to do it. Free will. If we start adding canned strategies, we infringe upon its free will. Is that a good thing? I'm not sure.

The AI needs some enhancement and that's going to take a lot of effort even with minimal changes to the game. HOW the AI thinks deserves more development work than WHAT canned strategies should be programmed into the code and their chances for execution. And in the long run that's probably easier to program and better for us the gamers. We're already seeing quite a bit of variation in AI strategies; we just need a bit more, yes?

[Night]

The sub problem has alway's bothered me, another problem is that they are much to expensive to be practical in reducing enemy MPPs. I think thier price should be lowered, and also thier effectivness when attacking enemy warships to prevent people from just buying mass subs to destroy enemy fleets.

[Edwin P.]

Bill

I like the concept of focusing on how the AI thinks and then how to execute that strategy. At the same time I realize, that developing a more robust AI takes a lot of work. As Hubert stated in your strategy guide writing the AI took about 50% of the time allocated to the game.

In the short term spicing up the SC AI with a few relatively simple canned strategies is simpler and faster than developing the long term AI improvements that I hope to see in SC2. This is especially true in the opening moves of a game where there are only a limited number of options and countermoves.

On example is the movement of the French Corps in Beruit. Currently the AI does not move this unit. Why? I do not know. Most human players move this unit to Egypt after Italy enters the war.

To resolve this issue you could 1) update the AI to account for the use of this unit or 2) you could add a simple subroutine (Ie move Unit XX to Hex XX,YY) which would execute after Italy enters the war (Idea: Perhaps - 10% at Green AI, 50% at beginner AI, 70% at intermediate AI, 90% at expert AI). The second option would be easier to program.

Writing a routine that tells the AI to look ahead 1 to 4 turns and see if it can save a unit (ie French) by moving from French to UK controlled territory would be a better solution as the routine could be applied to units in Algiers, France, or Syria. Such a forward thinking routine would take more work to develop and test and would be too time consuming to qualify for a patch to SC.

Another example is what to do with the UK Egyptian fleet during the opening turns of the game. Essentially there are only 3 options: move to the Atlantic, defend Egpyt or mass the fleets to destroy the Italian Navy. It seems to me that programming in these canned stratgies would also be a simpler job that updating the top level AI.

Adding 4 or 5 opening game movement options to the next SC patch would add variety and spice to SC while we await SC2 with a much improved AI.

[ May 10, 2003, 10:57 PM: Message edited by: Edwin P. ]

[Edwin P.]

Night

I have found that subs, as is, can be quite useful against human players and the AI, if you can destroy the bulk of the british fleet before fielding such a force to the Atlantic. Normally this occurs if the Brits target the German fleet in the Baltic or move in close to shore to give the French naval support and attack German units along the French coast (Hint: Bait the Brits in and use airpower to sink them)

How so? If the Germans can deploy 5 or so subs in the Atlantic they can damage the UK merchant ship convoys while also preventing US transports from reaching the UK and allied landings along the west coast of France. In fact, I have played a few games where the Allied AI never launch a D-Day invasion. Of course, if you do decide to use the Sub stratgy its useful to invest 1 tech chit in submarines for the longer range and firepower that it gives them, along with the increased chance of diving.

[ May 10, 2003, 02:01 PM: Message edited by: Edwin P. ]

[pzgndr]

In the short term spicing up the SC AI with a few relatively simple canned strategies is simpler and faster than developing the long term AI improvements that I hope to see in SC2.

Edwin, personally I don't believe this is as simple as you may imagine it to be. Since we're not likely to see another patch in the short term, we'll both have to wait and see what happens with SC2.