MoLAoS
Guest
TL;DR: Amusing gimmick where the AI competes with the player in the diplomacy metagame instead of in the gameplay itself. Probably more fun as an optional game mode.
I'll use EU4 from Paradox as my direct example but this could apply to even games like Galactic Civilizations 3 or Distant Worlds.
Players hate when the AI cheats, and when the AI backstabs them when they are down. Makes em rage. But imagine a new kind of AI that learns along with the player BUT IT ONLY LEARNS ONE THING!
Who kicked the most ass.
Imagine if instead of the shitty AE/Coalition system keeping track of expansion in a single game the AIs each tracked the overall winner of the game, who actually did the most damage to other AI or players, and who personally took down them.
The AI would assign threat bonuses based on each of these stats for determining what to do in future games. So in the first few games France, Austria, Ottomans, Russia, Spain, Poland etc. would be kicking ass and doing really well. In later games whichever of them won would be dogpiled early. Whether this meant total elimination or just royally screwing them up and then hitting them harder next time if they still did good is up in the air right now.
Aside from this AI would look at who historically, hehe, ate them or smashed them enough to be eaten in previous games. AI would then assign bonuses to alliances with states that also got killed by that other AI. So Ryazan, Novgorod, Kazan, etc would ally up and try to smash Muscovy early.
As games kept being played the AI would adapt their strategy game theory style based on who won the next game.
Players already do this, especially in EU4. Depending on where you play you focus on taking down certain nations that start super strong. Furthermore, this is how players play multiplayer games like the famous Diplomacy. Players who kick too much ass get double teamed in the next game.
Which brings me to the part of this that affects the player. The player should ideally be trying not to be the finisher of nations, even small ones, and should attempt to win by as little as possible. If the player slams through the game slaughtering everyone with his unfair computational resources, ability to read guides on the internet, and knowledge of unintended exploits, this will incentivize the AI nations to gang up on him in later games. Currently the ideal exploit for the AI is to do what humans do and focus the most dangerous player first, which is always the human cause that's the point of strategy games. The AI can't do the stuff the player can and it can't adapt and evolve gameplay the same way. But this AI strategy would merely require a few bits per country file to record a nation ID and a value.
This is slightly less viable in more standard 4x games with random starts and more symmetry between entities but still works from a balance perspective. If the Drengin in GalCiv3 have a better AI for w/e reason the other races can counter that after losing the first time.
I'll use EU4 from Paradox as my direct example but this could apply to even games like Galactic Civilizations 3 or Distant Worlds.
Players hate when the AI cheats, and when the AI backstabs them when they are down. Makes em rage. But imagine a new kind of AI that learns along with the player BUT IT ONLY LEARNS ONE THING!
Who kicked the most ass.
Imagine if instead of the shitty AE/Coalition system keeping track of expansion in a single game the AIs each tracked the overall winner of the game, who actually did the most damage to other AI or players, and who personally took down them.
The AI would assign threat bonuses based on each of these stats for determining what to do in future games. So in the first few games France, Austria, Ottomans, Russia, Spain, Poland etc. would be kicking ass and doing really well. In later games whichever of them won would be dogpiled early. Whether this meant total elimination or just royally screwing them up and then hitting them harder next time if they still did good is up in the air right now.
Aside from this AI would look at who historically, hehe, ate them or smashed them enough to be eaten in previous games. AI would then assign bonuses to alliances with states that also got killed by that other AI. So Ryazan, Novgorod, Kazan, etc would ally up and try to smash Muscovy early.
As games kept being played the AI would adapt their strategy game theory style based on who won the next game.
Players already do this, especially in EU4. Depending on where you play you focus on taking down certain nations that start super strong. Furthermore, this is how players play multiplayer games like the famous Diplomacy. Players who kick too much ass get double teamed in the next game.
Which brings me to the part of this that affects the player. The player should ideally be trying not to be the finisher of nations, even small ones, and should attempt to win by as little as possible. If the player slams through the game slaughtering everyone with his unfair computational resources, ability to read guides on the internet, and knowledge of unintended exploits, this will incentivize the AI nations to gang up on him in later games. Currently the ideal exploit for the AI is to do what humans do and focus the most dangerous player first, which is always the human cause that's the point of strategy games. The AI can't do the stuff the player can and it can't adapt and evolve gameplay the same way. But this AI strategy would merely require a few bits per country file to record a nation ID and a value.
This is slightly less viable in more standard 4x games with random starts and more symmetry between entities but still works from a balance perspective. If the Drengin in GalCiv3 have a better AI for w/e reason the other races can counter that after losing the first time.
