AI Consideration

[OTGamer]

Here's a map at conclusion of IJN (Human) victory over Allies (CPU). Sea lanes win.



I'll give the AI credit for at least recognizing the danger and making concerted attempts to "open" the lanes. Allies successfully took Samoa, and then threw a good bit of effort into trying to take Fiji until time ran out.

Would it really be that hard to program an additional layer of AI "thinking". The AI knows it must take Fiji (or at least destroy the base). A simple trace program between the desired target island and major IJN ports would identify Tarawa as the key link. A successful Allied invasion of Tarawa, would isolate not only Fiji, but the entire sea lines island chain.

[Skanvak]

Congratulation,

You have set a new record to beat.

I have experienced the same problem. The AI is not aggresive enough about the protection of its sea lane. Though in one early game it did try to take tarrawa to isolate Espiritu Santo, but it never tried to take Espiritu Santo which result in a a victory as even an isolated base will count.

I don't know how to realy improve the AI. The game is quite complexe to analyze in term of game theory.

[ScottWAR]

Not really that complex.

If Japan doesnt have enough oil the allies should be focusing on not letting them get oil.

If japan has oil but doesnt have any of the sea lane islands, the allies should be focusing on taking away Japans oil or taking islands to get bomb points.

If Japan has enough oil AND owns a sea lanes island the allies should be doing EVERYTING in their power to take back the sea lane island.

If Japan owns a sea lane island and Japan has less months of oil than it will take to get a sea lane cut victory, the Allies should be concerned with not letting Japan get oil. Since it will lose from not having enough oil before the allies lose becasue of the sea lanes being cut.

Of course these are for later in the game,.....early in the game the allies mainly need to slow Japan down as much as possible without taking many losses.

[Skanvak]

That is overall strategy (and if you follow that every time I defeat you neat, except for point three that is obvious and lacking), that does not tell you how to send your fleet, or how to slow Japan and prevent them from getting close to your sea lanes.

The game is complexe because have you several possibilities each turn (what to defend, what to attack) without knowing what your opponnent will do. In game theory the solution is a strategy mix, that is setting a probability that each strategy is good then randomizing the decision (computer don't know psychology) because if you are predictable, you will be easily defeated. There are no perfect strategy in simultaneous game, you must surprise the opponnent and out guess him.

This process which is quite natural for a human brain (we have short cut that the computer don't have, but those can lead to logical error the computer will not do) is quite complexe to write in mathematical term (those needed for the computer to decide).