Rushing AIs... Are you turning on Aggressive AI?

[bestbrian]

Actually, if the AI would rush more often and more efficiently, it would make an early rush the only viable starting strategy every time you have a neighbour other than Gandhi. There's no point to start a builder game if there's a significant risk that the AI will (c)rush you.

I was just thinking the same thing.

 

[Tyrant Roger]

This has been a fascinating thread to read. I think in large part it has been a debate about what would make BTS a "better" game. To answer that question,of course, requires us to define what the goal of this game is. I would contend it is to make as "enjoyable" a game experience as possible for the human player. Not to make a game that is "fair" to the AI in some sense. Not to make a game that is as difficult to win as possible. Not to make a game where the AI's behavior is more like human behavior or where the production bonuses to the AI are reduced, etc.

Now what makes a game like BTS "enjoyable" to the human player. It seems to me it must have significant intellectual challenge, it must have variability in setting and in optimum strategy, and it must leave the human player with the impression that his opponents [the AI's] are playing rationally. In my view BTS succeeds remarkably in the first two criteria, but plainly falls far short in the third. The remedy is better algorithms for the AI, but that is a major programming challenge.

Under this view whether regular AI or aggressive AI is better depends largely on which produces the most rational - the optimum - level of play by the AI. That, of course, depends on the map, the other AI's, and a host of other factors. At this point, however, under either version AI play is so sub-optimal that the difference in play between the two versions is not very consequential.

Whither Civ5?

 

[Perplexity]

This has been mentioned in the thread but not often -- wouldn't a mix of standard and aggressive AIs in the same game be very clearly a good thing?

 

[Tyrant Roger]

For my part I think a mix of AGG AI and regular AI's as an option would be an improvement. I wonder whether it would be feasible to give the human player a broader array of AI style choices than just two. Wouldn't it be better if we could choose from say 5 different AI styles of play just as we now can change the terrain maps on which we play?

Some AI style choices might allow earlier AI rushes, better AI defenses to early human rushes, and so on.

 

[aelf]

For my part I think a mix of AGG AI and regular AI's as an option would be an improvement. I wonder whether it would be feasible to give the human player a broader array of AI style choices than just two. Wouldn't it be better if we could choose from say 5 different AI styles of play just as we now can change the terrain maps on which we play?

Some AI style choices might allow earlier AI rushes, better AI defenses to early human rushes, and so on.

But isn't this already somewhat the case in normal Civ? We get different personalities as it is. Aggressive AI is for people who want even Gandhi to be actively hostile.

 

[DanF5771]

For everybody who wants to experiment with AggAI on and off during different phases of an ongoing game (like AggAI on till 1000BC and then off), I just wanted to share that this can be achieved without the WB-detour and its inherent amnesia problems by using the python console in debug mode and entering:
gc.getGame().setOption(6,1) for AggAI on and
gc.getGame().setOption(6,0) for AggAI off.

 

[Andvare]

For everybody who wants to experiment with AggAI on and off during different phases of an ongoing game (like AggAI on till 1000BC and then off), I just wanted to share that this can be achieved without the WB-detour and its inherent amnesia problems by using the python console in debug mode and entering:
gc.getGame().setOption(6,1) for AggAI on and
gc.getGame().setOption(6,0) for AggAI off.

Ooh, nice. Thanks.

 

[Airefuego]

Actually, if the AI would rush more often and more efficiently, it would make an early rush the only viable starting strategy every time you have a neighbour other than Gandhi. There's no point to start a builder game if there's a significant risk that the AI will (c)rush you.

I do agree that you should not start a builder game if you are about to be rushed

I don't agree at all that the early rush would become the only viable strategy.

A rush is a good idea when the following conditions are met:

-neighbour starts close to you
-neighbour has good capital site / holy shrine / early wonder worth having
-you have strategic resources early (or resourceless UU)
-neighbour is underdefended.

That's not every game. Not even close. But the AI should have a CHANCE of rushing when the above conditions are fulfilled. After all, a human would.

Also, I advocate AI rushes on HIGH LEVELS. Not on noble or something.

So - look at the above list from point of view of the targetof the rush. If you are the target of the rush - underdefended, with a prime capital, close to a neighbour with metals, and playing high levels - I do not think it is very wise to 'start a builder game'. If you do, you deserve to be rushed.

 

[aelf]

Since when does the AI rush under these conditions? When they want to rush, they will rush no matter what. And usually the human player is the target, since you have fewer troops than the AI in the early game.

 

[MkLh]

I do agree that you should not start a builder game if you are about to be rushed

So no builder games anymore on higher levels. I don't think this would lead to a more varied gameplay.

I don't agree at all that the early rush would become the only viable strategy.

A rush is a good idea when the following conditions are met:

-neighbour starts close to you
-neighbour has good capital site / holy shrine / early wonder worth having
-you have strategic resources early (or resourceless UU)
-neighbour is underdefended.

That's not every game. Not even close. But the AI should have a CHANCE of rushing when the above conditions are fulfilled. After all, a human would.

[/quote]

The AI is not supposed to act like human, its purpose is to entertain human after all, not to aim for win at any cost. If AI would more likely DoW early eg if you have no metals, it would just make a hard starting position even harder.

Also, I advocate AI rushes on HIGH LEVELS. Not on noble or something.

So - look at the above list from point of view of the targetof the rush. If you are the target of the rush - underdefended, with a prime capital, close to a neighbour with metals, and playing high levels - I do not think it is very wise to 'start a builder game'. If you do, you deserve to be rushed.

On higher levels you either build infra (workers, settlers, granaries, libraries) or then you build a lot of troops. There's really no way to do both before 1000 BC. And if you are going to build a lot of troops, attacking your neighbor is the only viable strategy, troops sitting in undeveloped cities just cost a lot.

 

[PieceOfMind]

The AI is not supposed to act like human, its purpose is to entertain human after all, not to aim for win at any cost. If AI would more likely DoW early eg if you have no metals, it would just make a hard starting position even harder.

Forgive me if I'm pulling this out of context but who decides this? An AI that would genuinely entertain me is one that tries to win. If it means attacking me when I have no copper then so be it. If it did not, then I would prefer not to do it to them when they don't have copper.

 

[Belisar]

For everybody who wants to experiment with AggAI on and off during different phases of an ongoing game (like AggAI on till 1000BC and then off), I just wanted to share that this can be achieved without the WB-detour and its inherent amnesia problems by using the python console in debug mode and entering:
gc.getGame().setOption(6,1) for AggAI on and
gc.getGame().setOption(6,0) for AggAI off.

Now that is something interesting,
If you have a similar trick for changing game speed during a game (for example starting epic and thhen change to marathon in 1AD you would be my hero...

 

[DanF5771]

While gc.getGame().getGameSpeedType() works to retrieve the value, the corresponding function to set this value is not exposed to Python - messing around with this part of the game's InitCore does not seem to be allowed (but I'm far from being an expert, just like to snoop around ). I could also imagine that changing the speed in an ongoing game might cause some rather serious side effects (too much depends on it, just think about production/resource costs, cultural levels of cities, whip/draft anger, cottage growth, workers building improvements, ........).

You CAN still achieve/try it via the WB-detour, where all (most?) of this critical stuff might simply be dropped.

 

[Airefuego]

Since when does the AI rush under these conditions? When they want to rush, they will rush no matter what. And usually the human player is the target, since you have fewer troops than the AI in the early game.

I'm not talking about what the AI does. I'm talking about what the AI should do.

 

[Airefuego]

I can see that my suggestion that the AI should be taught how to rush properly has aroused some opposition.

Perhaps another way of putting it is this. Should there be an "unequal playing field" option that you can tick at the start of the game? It could have the effect of stopping the AI from using strategies that are open to the human.

I would not have such a box. I would just have difficulty levels, with a more sophisticated AI in the higher levels.

So no builder games anymore on higher levels. I don't think this would lead to a more varied gameplay.

No. The New More Human AI should not rush EVERY game.

But it should have a well-executed rush among its options. Currently the AI does not rush. It is hard-programmed not to DOW before 1000BC (I believe - pls feel free to correct me on this Dan!) and when it does, the war is poorly executed with no clear objective.

The AI is not supposed to act like human, its purpose is to entertain human after all, not to aim for win at any cost.

That's your opinion but (I agree with PieceofMind on this) who decides this?

Most people complain that the current AI is 'dumb' and unstrategic. I'm not sure a limited and predictable AI is the most 'entertaining' option.

Repeating: I'm not suggesting the AI should rush all the time. I'm suggesting that on high levels, it should have the option of rushing in a situation where a human would.

 

[aelf]

I'm not talking about what the AI does. I'm talking about what the AI should do.

I don't disagree, but I think we're referring to the current situation. I interpreted "more often and more efficiently" as more likely to and with a better concentration of troops that are built for the specific purpose of rushing in the early game, but still operating within the same mindset of the AI, which I take as a given.

 

[sylvanllewelyn]

Alright, let's join the mud bath!!!

Yes, this happens. And even though you kill wave after wave of their units, they still refuse peace and you're forced to mount a costly counter-offensive to take a few of their cities that might be very far away. The 2 of you get left behind eventually and the game is unplayable.

Does this make the game more fun? Count me out.

That's very historical actually. It's why tribal wars are so frequent amongst nomadic pastoralists, because they lack good grassland and water. It's also very common that two groups fight, exhaust themselves and a third party takes them both out. It truly adds to the flavour.

*******

There is an option in the game that says "random personalities". I wonder if it is possible to do it even better - randomize the personalities, then randomly assign them "normal AI" or "aggressive AI". Now you don't even know whether you should rush or not. And if the aggressive AI's all win or all lose, that's even better. You conquer 2 neighbours, you have someone that conquered 3 on the other continent.

*******

Why do so many people assume that the decision to rush is so heavily based upon AI settings? I think it's got a lot more to do with your start and your game in general. For example, if you found out after animal husbandary that you settled your capital right on top of horses, I'm rushing the hell out of my neighbour, aggressive AI or not.

 

[aelf]

That's very historical actually. It's why tribal wars are so frequent amongst nomadic pastoralists, because they lack good grassland and water. It's also very common that two groups fight, exhaust themselves and a third party takes them both out. It truly adds to the flavour.

Except the flavour is the kind that leaves a bad aftertaste in your mouth...

I'm not going to argue with the generalization you make, but I must say that there a lot of things that are historically accurate that would make the game suck if added. For one, which civilization really chooses what to research like that? If we're to be historically accurate, techs should come at random and at different intervals, especially in the early game. Some will never get certain vital techs at all. How's it like to be stuck with no BW or AH until someone comes and takes you out?

Why do so many people assume that the decision to rush is so heavily based upon AI settings? I think it's got a lot more to do with your start and your game in general. For example, if you found out after animal husbandary that you settled your capital right on top of horses, I'm rushing the hell out of my neighbour, aggressive AI or not.

I thought the AI gets strategic resources by default in the higher levels? And, heck, if I knew I started with a couple of archers and my neighbour with only 1 warrior or none at all, I'd rush the hell out of him.

 

[Andvare]

Actually, to be able not only to set aggAI for individual AIs, but also individual difficulty would further the game quite a bit. A diety AI next to a monarch AI would pose a bigger problem than two deity AIs, as the deity AI would quickly get the upper hand.

 

[vvcv]

I always play with AGG AI, and yesterday i tried to play with the normal AI, and the difference was too big. In normal AI the player can be relaxed because the AI probably doesn't declare war many times, however, in AGG AI any AI can attack you with many troops.

Yesterday, France attack me in normal AI, and he has a very poor army, while AGG AI has a huge army.

I think that the difference in the difficulty is too evident. I play in noble.