CEAW 3.0 Grand Strategy AI
Posted: Fri May 08, 2015 1:34 am
I recently upgraded to CEAW 3.0 GS. Overall I find V3.0 to be a much improved version than what I was using (1.0). However I did note that somewhere in the huge (and somewhat repetitive) users guide, that AI has not been updated. I have completed several games playing as the Axis player vs AI. I have found that as in earlier versions of the game, AI does some not too intelligent things. I'd like to pass on some observations that significantly impact the game and would not realistically happen. I understand from conversations with Slitherine support that an AI upgrade is in the works. Can these be passed on to your designers for possible fixes in future updates?
1. The US leaves NY and Washington without a garrison late in the war. That made it easy to send a 'suicide' invasion of several units protected by subs to the US and seize both cities forcing a US surrender. In one game, France also left Paris without a garrison, allowing a break thru unit to seize Paris early in 1940. I suggest your AI code needs to change to always put a garrison in major power capitals.
2. Despite US, UK, and Russian PP levels being quite large, AI does not build the units it needs as the war goes on. In one game I played, Russia had over 2000 pp stored buy mid 1943, yet did not replace lost units and allowed the Germans to eventually take both Russian capitals. I observed the same issue with the US and UK (which contributed to #1 above). When it became obvious that the Russians would not spend pps, it no longer was strategically necessary to try to stop conveys to Russia. Axis subs could then be used to stop allied troop transports, which were usually unescorted. Also, subs could be put in French ports and snipe at the invasion units stacked up there (see #3 below). Since the subs could only be attacked once per turn while in port, they got off with little damage turn after turn.
3. The US/UK launchs an invasion fleet in mid 1942 which sits off the French coast literally for years, allowing German subs and air units to attack it constantly. In one game, these units tried successive invasions which racked up higher and higher amphibious overuse penalties to the tune of hundreds of pps each game turn. Eventually the pp balance in both the US and UK went negative, preventing building new units or repairs to existing units. Interesting though that a negative pp situation did not stop invasion attempts and that drove the pp value even further negative. By the time the US fell (see #1) the negative pp value was minus 4000.
So in summary, AI fixes are needed to:
1. Garrison capital cities at all times
2.
Stop creating invasion fleets that just sit there
3. Stop trying successive invasions once pp values hit zero.
4. Build units when pp values allow it.
1. The US leaves NY and Washington without a garrison late in the war. That made it easy to send a 'suicide' invasion of several units protected by subs to the US and seize both cities forcing a US surrender. In one game, France also left Paris without a garrison, allowing a break thru unit to seize Paris early in 1940. I suggest your AI code needs to change to always put a garrison in major power capitals.
2. Despite US, UK, and Russian PP levels being quite large, AI does not build the units it needs as the war goes on. In one game I played, Russia had over 2000 pp stored buy mid 1943, yet did not replace lost units and allowed the Germans to eventually take both Russian capitals. I observed the same issue with the US and UK (which contributed to #1 above). When it became obvious that the Russians would not spend pps, it no longer was strategically necessary to try to stop conveys to Russia. Axis subs could then be used to stop allied troop transports, which were usually unescorted. Also, subs could be put in French ports and snipe at the invasion units stacked up there (see #3 below). Since the subs could only be attacked once per turn while in port, they got off with little damage turn after turn.
3. The US/UK launchs an invasion fleet in mid 1942 which sits off the French coast literally for years, allowing German subs and air units to attack it constantly. In one game, these units tried successive invasions which racked up higher and higher amphibious overuse penalties to the tune of hundreds of pps each game turn. Eventually the pp balance in both the US and UK went negative, preventing building new units or repairs to existing units. Interesting though that a negative pp situation did not stop invasion attempts and that drove the pp value even further negative. By the time the US fell (see #1) the negative pp value was minus 4000.
So in summary, AI fixes are needed to:
1. Garrison capital cities at all times
2.
Stop creating invasion fleets that just sit there
3. Stop trying successive invasions once pp values hit zero.
4. Build units when pp values allow it.