Since I discovered this bug today : http://www.realmsbeyond.net/forums/showt...p?tid=8238 , I think it's a good time to rethink what type of resource advantage the AI needs, and how much of it.
Population growth
As of now AI receives some of this advantage but not too much, they grow 1.5 times faster on impossible than default. I believe this is good as is, and they definitely need this, since they waste quite some population on settlers and ofc the human player is more creative at killing population in cities than the AI.
Outpost growth
This was removed from the advantage tables and hardcoded to be +25% per level of difficulty. Not much to say here, the AI cities are extremely vulnerable while outposts, since the AI generally does not garrison them. The advantage is needed both to help the AI spread and to reduce the window when they are vulnerable.
Overland Casting cost discount
This one was added for the mod and works really well. It effectively allows the AI to cast more overland spells, since they lack the tactical ability to focus on one goal and use appropriate spells, instead do a little bit of everything with some weighting, an advantage is definitely needed here to be able to build scary fantastic unit armies, or use enough curses to actually hurt another player, etc. The AI casts twice as much spells on extreme than on normal, and 2.5 times on impossible.
Production multiplier
The AI builds buildings and produces units faster. Not much to say about this,it's needed, this is the main source of game difficulty afterall. It scales from 75% on easy to 250% on impossible currently.
Gold multiplier
How much gold the AI gets from their cities through taxes and whatever else. Has no effect on Noble heroes or leftover food converted into gold, but affects everything else. Uses the same values as production at the moment. If the AI's maintenance discount is fixed, this might need a major reduction.
Power multiplier
Multiplies the power from all cities and nodes. Has no effect on volcanoes (making Armageddon a pretty crappy spell for an AI wizard, except for wrecking the economy of others). Has no effect on Ritual Master hero ability because it would be difficult to squeeze that into the code and it's insignificant.
This advantage scales from 75% on easy to 300% on impossible currently.
Reseach multiplier
Multiplies all research obtained from cities. Has no effect on any other research income. Since research is a very important, game changing effect, this has been toned down to a maximal value of 130% at impossible.
The AI has more than enough research advantage from having a larger power base, this part only really matters if the AI spends all their power on producing mana due to constant fighting, in that case they have to rely on research they produce from buildings.
Food multiplier
How much extra food AI cities produce. This directly reduces the effectiveness of employing city curses against AI players, since cities will need fewer farmers to avoid starving.
At the same time it doesn't do anything meaningful aside from that. AI receives a huge discount on food maintenance, so the amount they produce is mostly irrelevant, and excess food converts poorly to gold as well.
I think this modifier should not exist - it should be set to 100% on all levels of difficulty.
Maintenance multiplier
See the link above, this one is buggy.
It reduces the maintenance cost for checks on disbanding units, so the AI can keep larger armies and more spells, however it does not reduce the actual cost subtracted from their gold and mana reserves, so even if they can keep their units, they'll be left with zero gold or mana each turn. While the latter is rare as they can raise their mana production to compensate, the former is frequent. It also reduces food maintenance but since food only matters for the disbanding of units, and does not produce an actual income (well aside from that few pieces of gold from the excess), the bug didn't affect the food part of the multiplier.
This multiplier is definitely needed, since we want the AI to be able to afford and maintain large armies - as AI players are way less efficient at moving them around and using them for any purpose than a human player is. They also need it to be able to utilize the advantage of higher production, without this they couldn't afford those extra units.
However, if I fix the bug, that'll have a major impact. Gold not lost on the maintenance, and will effectively raise the AI's income per turn, allowing them to rush buy more buildings and units. It also makes it harder to tribute gold to them, since they have more of it. But this is the small part of it. Mana not lost of maintenance not just means more mana crystals piling up, no it means the AI can reduce mana production and push their research or skill higher, both of which can have a major impact on the game. Imagine if the AI researched Armageddon 40% faster than usual due to not wasting that power on maintenance.
So if the bug is fixed, most likely all gold and power income numbers need to be lowered...and I have no idea by how much. It is already surprising the AI can operate well with only 200% gold and mana when vanilla game used 400% for the highest difficulty, and not just well but far more effectively, as current extreme - heck even hard - is more difficult than old impossible ever was.
Population growth
As of now AI receives some of this advantage but not too much, they grow 1.5 times faster on impossible than default. I believe this is good as is, and they definitely need this, since they waste quite some population on settlers and ofc the human player is more creative at killing population in cities than the AI.
Outpost growth
This was removed from the advantage tables and hardcoded to be +25% per level of difficulty. Not much to say here, the AI cities are extremely vulnerable while outposts, since the AI generally does not garrison them. The advantage is needed both to help the AI spread and to reduce the window when they are vulnerable.
Overland Casting cost discount
This one was added for the mod and works really well. It effectively allows the AI to cast more overland spells, since they lack the tactical ability to focus on one goal and use appropriate spells, instead do a little bit of everything with some weighting, an advantage is definitely needed here to be able to build scary fantastic unit armies, or use enough curses to actually hurt another player, etc. The AI casts twice as much spells on extreme than on normal, and 2.5 times on impossible.
Production multiplier
The AI builds buildings and produces units faster. Not much to say about this,it's needed, this is the main source of game difficulty afterall. It scales from 75% on easy to 250% on impossible currently.
Gold multiplier
How much gold the AI gets from their cities through taxes and whatever else. Has no effect on Noble heroes or leftover food converted into gold, but affects everything else. Uses the same values as production at the moment. If the AI's maintenance discount is fixed, this might need a major reduction.
Power multiplier
Multiplies the power from all cities and nodes. Has no effect on volcanoes (making Armageddon a pretty crappy spell for an AI wizard, except for wrecking the economy of others). Has no effect on Ritual Master hero ability because it would be difficult to squeeze that into the code and it's insignificant.
This advantage scales from 75% on easy to 300% on impossible currently.
Reseach multiplier
Multiplies all research obtained from cities. Has no effect on any other research income. Since research is a very important, game changing effect, this has been toned down to a maximal value of 130% at impossible.
The AI has more than enough research advantage from having a larger power base, this part only really matters if the AI spends all their power on producing mana due to constant fighting, in that case they have to rely on research they produce from buildings.
Food multiplier
How much extra food AI cities produce. This directly reduces the effectiveness of employing city curses against AI players, since cities will need fewer farmers to avoid starving.
At the same time it doesn't do anything meaningful aside from that. AI receives a huge discount on food maintenance, so the amount they produce is mostly irrelevant, and excess food converts poorly to gold as well.
I think this modifier should not exist - it should be set to 100% on all levels of difficulty.
Maintenance multiplier
See the link above, this one is buggy.
It reduces the maintenance cost for checks on disbanding units, so the AI can keep larger armies and more spells, however it does not reduce the actual cost subtracted from their gold and mana reserves, so even if they can keep their units, they'll be left with zero gold or mana each turn. While the latter is rare as they can raise their mana production to compensate, the former is frequent. It also reduces food maintenance but since food only matters for the disbanding of units, and does not produce an actual income (well aside from that few pieces of gold from the excess), the bug didn't affect the food part of the multiplier.
This multiplier is definitely needed, since we want the AI to be able to afford and maintain large armies - as AI players are way less efficient at moving them around and using them for any purpose than a human player is. They also need it to be able to utilize the advantage of higher production, without this they couldn't afford those extra units.
However, if I fix the bug, that'll have a major impact. Gold not lost on the maintenance, and will effectively raise the AI's income per turn, allowing them to rush buy more buildings and units. It also makes it harder to tribute gold to them, since they have more of it. But this is the small part of it. Mana not lost of maintenance not just means more mana crystals piling up, no it means the AI can reduce mana production and push their research or skill higher, both of which can have a major impact on the game. Imagine if the AI researched Armageddon 40% faster than usual due to not wasting that power on maintenance.
So if the bug is fixed, most likely all gold and power income numbers need to be lowered...and I have no idea by how much. It is already surprising the AI can operate well with only 200% gold and mana when vanilla game used 400% for the highest difficulty, and not just well but far more effectively, as current extreme - heck even hard - is more difficult than old impossible ever was.
