Creating decent games v. the AI.

[daystar777]

Has anyone tried playing the AI by deliberately positioning yourself from the outset in such a way that the AI gains a severe advantage, and THEN contesting the game?
Two examples: giving up India to Japan, or allowing Germany to take control of the Atlantic, or Africa.
I've found that you can play a decent game this way, avoiding the AI's usual UK Cruiser/Destroyer build in the English Channel, or that weird U.S. Carrier/Sub move which then stays down in the South Pacific for the whole game.
As Axis, gifting the Allies one of the big islands can make for an interesting permutation.

[simon_hibbs]

Sounds like that could be fun, sure. The only problem I have with play vs humans is the long waits between turns, but if i get bored and want a game in an evening I'll give this a try..

[daystar777]

Originally posted by simon_hibbs:

Sounds like that could be fun, sure. The only problem I have with play vs humans is the long waits between turns, but if i get bored and want a game in an evening I'll give this a try..

If you want a real challenge try gifting the Germans Moscow (but keep all your Russians in one piece back behind the Urals, don't just have them all wiped out!)

[Nihonsean]

Allied AI is pretty stupid so this shouldnt be that hard. The Soviet AI, for whatever reason, likes to turtle in Moscow with lots of infantry it doesnt attack with. UK does the same on Britain, just builds a lot of ships, and funnels troops from India to Russia. If Africa falls they don't try and get it back. Both are sort of weird.

[simon_hibbs]

Originally posted by Nihonsean:

The Soviet AI, for whatever reason, likes to turtle in Moscow with lots of infantry it doesnt attack with. UK does the same on Britain, just builds a lot of ships, and funnels troops from India to Russia. If Africa falls they don't try and get it back. Both are sort of weird.

I think they make sense in terms of going for the simplest behaviours to program. Stacking defensively and funnelling troops in a corridor is pretty simple behaviour to program and still provides something of a challenge for a newbie to tackle. Programming a german attack on Africa is straightforward, just transport troops to Egypt and head south. Allies re-taking Africa on the other hand is quite complex and would conflict with other behaviours.

[jlklopfenstein]

My attempt at this was playing as the Allies and throwing the game so badly I was down to just the continental United States. I had lost the UK and Russia. (Note I started actually TRYING to win the round before UK fell). What resulted was a long game where I painfully fought back and ended up clean-sweeping the board (I had set it to complete victory). It was an adventure.

[Abscondrel]

working on the AI would be a very nice thing for the Devs to do. Difficulty settings and strategic personalities.

maybe if we give them candy...

[daystar777]

Originally posted by jlklopfenstein:

My attempt at this was playing as the Allies and throwing the game so badly I was down to just the continental United States. I had lost the UK and Russia. (Note I started actually TRYING to win the round before UK fell). What resulted was a long game where I painfully fought back and ended up clean-sweeping the board (I had set it to complete victory). It was an adventure.

Nice. I've had similar game.

[BakedAlaska]

It would sure be nice to have better AI from the outset. It has been discussed previously with the reasoning of new players encountering anything that would set them off from using the game. When playing as a single nation I can sometimes get interesting breakthroughs from the other nations but they are few and far between. It is just something I do to pass time in hopes that one day some proper AI comes to the game. The hope of competitive efficient AI opponents keeps us all wanting...

[MenjoBleeko]

I've had fun editing the local save game to make myself the only one on my "team" in a "hold every victory city" victory condition.

The only thing I do to mitigate it being 4 on 1 is give my capital territory an extra 10 RU in value (Take that amount away from other territories so it doesn't unbalance your income for an extra challenge).

Russia is nearly unwinnable 4 on 1 unless you get very lucky. USA is simple, especially if you can build an Atlantic fleet on Turn 1 (and if you've retained any of your Pacific fleet after Japan's turn). Germany is great fun, Japan or the UK vs the world is also quite a challenge, given the limited industrial base and island starts.

[seldon]

The AI could be made so much better by reviewing the underlying rules which seem off in places. I'd be happy to advise- Beamdog !

This should be the AI decision tree:


1/ Can AI push into any enemy territories strongly so they can't counter. If so, do it. Prioritize VP cities if there are options.

2/ Can AI push into any enemy territories weakly to deny / steal VP? If so, do it. Prioritize VP cities if there are options.

3/ Can AI hold each territory I currently have, with current forces or by moving into it? If so, do it.

4/ Can any of the AI territories not be held? If so, retreat.

5/ As Russ, build to maintain 2 inf to 1 art ratio.

6/ As UK and US build enough ground forces for transports - hard code the transport builds / movement shuck to Finland with a check to confirm the transports are safe. If not safe, spawn fighting ships instead.

7/ Germany 2 rounds of inf and art marching to Russia, followed by 2x rounds of mostly tanks, 1x round of bombers for 1-2-3 punch hitting at same time.

8/ Japan just needs to build mix of inf and tanks, ignore India, and mass where safe off moscow with an alternative strategy of subs if the allies are judged to be doing KJF.

Not much creativity involved and would lack any sort of flair but would be a reasonably solid game.