I have a potentially controversial change I intend to make. This post sets out the background and I'd appreciate anyone that feels strongly (either agreeing OR disagreeing) posting please.
The past couple of days I have been working on a save game from Faustmouse that illustrates some issues the AI had with attacking past blockading points. I'm not going to go into the specifics of the issues I found here (I'll detail it in the SVN thread post when I push the fixes), but the interesting aspect from the perspective of this thread is that once I had resolved the problems the end result was actually WORSE than before!
The situation in Faustmouse's game was this. On the GEM map Faustmouse owns the area around the Middle east up to it's border with Africa (at Egypt). This connection is a 2-tile-wide region through which anything wishing to cross between Africa and Eurasia must pass (if by land anyway). Faustmouse has wisely built a fort on one of these tiles (a hill) and populated it with a decent garrison (more on that later). Because it's a fort it exerts a ZOC, so the ONLY way through is a direct attack that takes the fort.
Slightly west of this choke point, in Egypt is an AI city, in which an AI attack stack of OVER 250 units has gathered.
The originally reported problem was that it just sat there and never did anything. This was because of a number of detailed bugs/issues that prevented it seeing that it had to take the fort in order to get anywhere.
Once I resolved the problems so that the AI realized what needed doing I was amazed to find that when it attacked the fort it lost all 250+ units WITHOUT killing a SINGLE defender!! Now, the garrison is not a bad one, and in particular it includes a Warlord Commander and a hero (Sargon). Both of these units are approximately twice the strength of most the AI's attacking units, and obviously they benefit a lot from being in a hill-fort.
Regardless of this it seems crazy that (even after I modified the AI to do maximum archery bombardment first, so they were not at full strength), they can defeat a horde of 250 units (horsemen and similar with a smattering of crossbowmen and light swordsmen). In fact they came out of it at nearly full health!
The reason for this is battlefield promotions. Basically the defenders get promoted often enough to get healed back up faster than they are beaten down. To me this seems quite wrong - battlefield promotions are a reward for surviving tough battles, no some miraculous healing you get in the heat of battle.
I therefore propose to change things so that any earnt battlefield promotions are awarded at the END of ALL the combat, not between attacks by individual units. The net effect is that they will kill a LOT of the enemy but ultimately won't survive to see their promotions (in this case).
Opinions please...
The past couple of days I have been working on a save game from Faustmouse that illustrates some issues the AI had with attacking past blockading points. I'm not going to go into the specifics of the issues I found here (I'll detail it in the SVN thread post when I push the fixes), but the interesting aspect from the perspective of this thread is that once I had resolved the problems the end result was actually WORSE than before!
The situation in Faustmouse's game was this. On the GEM map Faustmouse owns the area around the Middle east up to it's border with Africa (at Egypt). This connection is a 2-tile-wide region through which anything wishing to cross between Africa and Eurasia must pass (if by land anyway). Faustmouse has wisely built a fort on one of these tiles (a hill) and populated it with a decent garrison (more on that later). Because it's a fort it exerts a ZOC, so the ONLY way through is a direct attack that takes the fort.
Slightly west of this choke point, in Egypt is an AI city, in which an AI attack stack of OVER 250 units has gathered.
The originally reported problem was that it just sat there and never did anything. This was because of a number of detailed bugs/issues that prevented it seeing that it had to take the fort in order to get anywhere.
Once I resolved the problems so that the AI realized what needed doing I was amazed to find that when it attacked the fort it lost all 250+ units WITHOUT killing a SINGLE defender!! Now, the garrison is not a bad one, and in particular it includes a Warlord Commander and a hero (Sargon). Both of these units are approximately twice the strength of most the AI's attacking units, and obviously they benefit a lot from being in a hill-fort.
Regardless of this it seems crazy that (even after I modified the AI to do maximum archery bombardment first, so they were not at full strength), they can defeat a horde of 250 units (horsemen and similar with a smattering of crossbowmen and light swordsmen). In fact they came out of it at nearly full health!
The reason for this is battlefield promotions. Basically the defenders get promoted often enough to get healed back up faster than they are beaten down. To me this seems quite wrong - battlefield promotions are a reward for surviving tough battles, no some miraculous healing you get in the heat of battle.
I therefore propose to change things so that any earnt battlefield promotions are awarded at the END of ALL the combat, not between attacks by individual units. The net effect is that they will kill a LOT of the enemy but ultimately won't survive to see their promotions (in this case).
Opinions please...
