3 ms·

> You can also mitigate the problem by playing on the hardest levels, which lets the enemies cheat shamelessly. I honestly despise games that do this. I unders

> You can also mitigate the problem by playing on the hardest levels, which lets the enemies cheat shamelessly. I honestly despise games that do this. I understand why they do though: writing a difficult AI can be quite hard to do.

criley2 7y ago
Age of Empires 2 is an interesting case where the original versions had cheater AI, and the subsequent re-releases have taken the time to examine competitive meta and create new difficult AI that plays fair based on actually competitive tactics.
JohnFen 7y ago
I actually agree with you, and tend not to play on the hardest levels because it irritates me so. But it is a legitimate option. My love for MOO and MOO2 is great, but doesn't blind me to the fact that the AI is dismal.
setr 7y ago
the problem I think is that the cheater AI is unstoppable cheating -- his cheating becomes almost independent of his state within the game (true independence is the behavior of actual cheating), such that your path to victory really ends up being complete and total domination. You can't really apply normal strategies like crippling the opponent by poking the edges, or baiting a fight here and having a small force go around to there, or focusing down the infrastructure over the army itself; there is only the main body, and that's all that matters.. I think cheater AI is probably a perfectly good strategy; its just the implementation is never done that well, and it ends up cheating in a fashion that isn't actually interesting. Alternatively, the cheater AI could simply be given an aesthetic reason to validate cheating, instead of trying to pretend its a normal player -- eg AI war simply defines the AI as an overwhelming force; so overwhelming that it doesn't actually pay much attention to your activity, and that's the only reason you're not dead (your goal is to continue growing without letting the AI noticing). Total War and other 4X games generally have you playing where the AI is also generally unintelligent, but feels more fair, because everyone is in various starting states of strength. The cheating AI gets a set of arbitrarily strong units, justified because its simply a strong country 10x the size of yours. And then the amount of blatant, uninteresting, cheating that has to occur is much more limited -- if the AI is not sufficiently challenging, then it's generally simpler to increase his relative size, or resources, or production facilities, or whatever. What the AI does after that, after the game has started, becomes less important, so what cheating occurs becomes negligible from the perspective of the player. A 50% cheating bonus to production time doesn't mean much when he already started with a 500% bonus in resources and production. Games like MoO and Civ make the mistake of having everyone start at the same state, even when its just you vs the AI(s), because they don't treat the two differently -- there is only one setup, one goal, one victory condition, regardless of the number of human players. That is, they don't admit to the weakness of their AI, and try instead to make up for it "under the hood"; and when you as the player catch this cheating, and it's never that hard to, you feel inexplicably stupid to have ever thought you were playing a fair match, and losing/winning to fair rules. The same as when you catch a human cheating. Most other 4X games are defined as you vs the AI, and there's little to no multiplayer in the first place, so the idea doesn't trivially translate, but you vs the AI should probably be treated differently than a legitimate multiplayer map anyways -- as a campaign setup or a skirmish map.
- imtringued 7y ago
  I have played a tactical strategy game called tanks of freedom where the enemy was blatantly cheating. Even though the enemy has the same amount of capture points as you do he seemingly has an endless army. What happens is that the fog of war is filled with pre-spawned units which means you are stuck at a choke point feeling helpless but if you play long enough the AI will run out of his starting advantage and suddenly you overwhelm him. The game became more and more grindy the closer you were to the end.
smacktoward 7y ago
Especially in the days when these games came out -- the machines they were built to run on were absurdly under-powered by today's standards. When MOO1 came out in 1993, a high-end desktop PC would have had a single-core 80386 CPU running at a blazing 25MHz, and a whopping 4 megabytes of RAM. DOS was a single-tasking operating system, so at least your program didn't have to share those limited resources[1], but still your entire program had to fit within those constraints. That didn't leave a lot of room for AI. [1] Well, technically it was a single-tasking operating system, but there were hacks like TSR programs (see https://en.wikipedia.org/wiki/Terminate_and_stay_resident_program https://en.wikipedia.org/wiki/Terminate_and_stay_resident_pr...) that people used to run multiple programs at once. So that was an additional wrinkle you had to deal with.
Buttons840 7y ago
Despite all our AI advances recently, with AlphaGo and AlphaStar, nobody has yet created an AI that can defeat the best humans in 4X games without cheating, right? One game I respect is AI War, which fully embraced that AI must cheat and didn't even pretend that the AI played by the same rules. Instead, they gave insight into the rules the AI did play by, and made manipulation of the AI's constraints part of the game.

criley2 7y ago

Age of Empires 2 is an interesting case where the original versions had cheater AI, and the subsequent re-releases have taken the time to examine competitive meta and create new difficult AI that plays fair based on actually competitive tactics.

JohnFen 7y ago

I actually agree with you, and tend not to play on the hardest levels because it irritates me so. But it is a legitimate option. My love for MOO and MOO2 is great, but doesn't blind me to the fact that the AI is dismal.

setr 7y ago

the problem I think is that the cheater AI is unstoppable cheating -- his cheating becomes almost independent of his state within the game (true independence is the behavior of actual cheating), such that your path to victory really ends up being complete and total domination. You can't really apply normal strategies like crippling the opponent by poking the edges, or baiting a fight here and having a small force go around to there, or focusing down the infrastructure over the army itself; there is only the main body, and that's all that matters.. I think cheater AI is probably a perfectly good strategy; its just the implementation is never done that well, and it ends up cheating in a fashion that isn't actually interesting. Alternatively, the cheater AI could simply be given an aesthetic reason to validate cheating, instead of trying to pretend its a normal player -- eg AI war simply defines the AI as an overwhelming force; so overwhelming that it doesn't actually pay much attention to your activity, and that's the only reason you're not dead (your goal is to continue growing without letting the AI noticing). Total War and other 4X games generally have you playing where the AI is also generally unintelligent, but feels more fair, because everyone is in various starting states of strength. The cheating AI gets a set of arbitrarily strong units, justified because its simply a strong country 10x the size of yours. And then the amount of blatant, uninteresting, cheating that has to occur is much more limited -- if the AI is not sufficiently challenging, then it's generally simpler to increase his relative size, or resources, or production facilities, or whatever. What the AI does after that, after the game has started, becomes less important, so what cheating occurs becomes negligible from the perspective of the player. A 50% cheating bonus to production time doesn't mean much when he already started with a 500% bonus in resources and production. Games like MoO and Civ make the mistake of having everyone start at the same state, even when its just you vs the AI(s), because they don't treat the two differently -- there is only one setup, one goal, one victory condition, regardless of the number of human players. That is, they don't admit to the weakness of their AI, and try instead to make up for it "under the hood"; and when you as the player catch this cheating, and it's never that hard to, you feel inexplicably stupid to have ever thought you were playing a fair match, and losing/winning to fair rules. The same as when you catch a human cheating. Most other 4X games are defined as you vs the AI, and there's little to no multiplayer in the first place, so the idea doesn't trivially translate, but you vs the AI should probably be treated differently than a legitimate multiplayer map anyways -- as a campaign setup or a skirmish map.

imtringued 7y ago
I have played a tactical strategy game called tanks of freedom where the enemy was blatantly cheating. Even though the enemy has the same amount of capture points as you do he seemingly has an endless army. What happens is that the fog of war is filled with pre-spawned units which means you are stuck at a choke point feeling helpless but if you play long enough the AI will run out of his starting advantage and suddenly you overwhelm him. The game became more and more grindy the closer you were to the end.

smacktoward 7y ago

Especially in the days when these games came out -- the machines they were built to run on were absurdly under-powered by today's standards. When MOO1 came out in 1993, a high-end desktop PC would have had a single-core 80386 CPU running at a blazing 25MHz, and a whopping 4 megabytes of RAM. DOS was a single-tasking operating system, so at least your program didn't have to share those limited resources[1], but still your entire program had to fit within those constraints. That didn't leave a lot of room for AI. [1] Well, technically it was a single-tasking operating system, but there were hacks like TSR programs (see https://en.wikipedia.org/wiki/Terminate_and_stay_resident_program https://en.wikipedia.org/wiki/Terminate_and_stay_resident_pr...) that people used to run multiple programs at once. So that was an additional wrinkle you had to deal with.

Buttons840 7y ago

Despite all our AI advances recently, with AlphaGo and AlphaStar, nobody has yet created an AI that can defeat the best humans in 4X games without cheating, right? One game I respect is AI War, which fully embraced that AI must cheat and didn't even pretend that the AI played by the same rules. Instead, they gave insight into the rules the AI did play by, and made manipulation of the AI's constraints part of the game.