Brutal AI

[Funak]

I think I've found a slight problem with the way the AI tech-discount works. Per definition of how research-agreements work (both sides always earning the same amount of beakers) it is close to impossible to get a fair exchange when making one. I mean if we are both getting X beakers from it and their techs cost a fraction of what my techs cost it is absolutely never worth it for a player to sign a research-agreement, which means there is absolutely no benefit to keeping the research-agreement option activated.

 

[Txurce]

There seems to be no question that the AI is having a late-game tech snowball effect. Any specific explanation doesn't matter as much as the fact that it's noticeable, and feels unfair.

 

[Forsti]

There seems to be no question that the AI is having a late-game tech snowball effect. Any specific explanation doesn't matter as much as the fact that it's noticeable, and feels unfair.

I disagree. The problem cannot be fixed unless the cause is found. To me it seems like the AI might get some massive instant yields compared to the tech costs, since there are several AIs gaining a tech on many consecutive turns.

 

[Txurce]

I disagree. The problem cannot be fixed unless the cause is found. To me it seems like the AI might get some massive instant yields compared to the tech costs, since there are several AIs gaining a tech on many consecutive turns.

No one's going to fix anything until it's agreed that there is something to fix.

Presently on this thread, no consensus has been reached.

 

[Forsti]

No one's going to fix anything until it's agreed that there is something to fix.

Presently on this thread, no consensus has been reached.

That's a good point, although it should be obvious to anyone that AI shouldn't be getting techs on that kind of pace on any difficulty.

 

[Adam_S]

That's a good point, although it should be obvious to anyone that AI shouldn't be getting techs on that kind of pace on any difficulty.

It dosent seem to be one off bonuses. In a Prince game Im 2-3rd world ranked, With 447 Science and 382 Culture per-tern. excluding one offs from GP. The 2nd Ranked, Germany, has 1494 Science and 1354 Culture per turn, with less cities and less population over me. IGE shows me that 90% Of those yields come from cities. While Germany does have 5-6 Academys down and the full rationalism tree, the actual yield for science from terrain dosent seem to explain those numbers, especialy true for culture.

I for one really dont like this end game rush. I have really enjoyed this mod and Communitas mod before it. But there just dosent seem to be a middle ground now, Warlord is easy, Prince now to difficult.

Its really killed my enjoyment of the mod and Civ 5.

 

[Revolutionist_8]

I think I've found a slight problem with the way the AI tech-discount works. Per definition of how research-agreements work (both sides always earning the same amount of beakers) it is close to impossible to get a fair exchange when making one. I mean if we are both getting X beakers from it and their techs cost a fraction of what my techs cost it is absolutely never worth it for a player to sign a research-agreement, which means there is absolutely no benefit to keeping the research-agreement option activated.

That's a nice catch, good point. Maybe Gazebo could give his opinion on this?

 

[Gazebo]

It dosent seem to be one off bonuses. In a Prince game Im 2-3rd world ranked, With 447 Science and 382 Culture per-tern. excluding one offs from GP. The 2nd Ranked, Germany, has 1494 Science and 1354 Culture per turn, with less cities and less population over me. IGE shows me that 90% Of those yields come from cities. While Germany does have 5-6 Academys down and the full rationalism tree, the actual yield for science from terrain dosent seem to explain those numbers, especialy true for culture.

I for one really dont like this end game rush. I have really enjoyed this mod and Communitas mod before it. But there just dosent seem to be a middle ground now, Warlord is easy, Prince now to difficult.

Its really killed my enjoyment of the mod and Civ 5.

The AI does not get a secret 'per turn' bonus via the CBO. It gets chunks of yields every once in a while for:

Entering a new era.
Founding a new city.
Winning a war.
Entering a Golden Age.

That's it. These bonuses scale based on difficulty, but they're actually somewhat lower than they were a few months ago.

I'm sorry that the AI's difficulty is ruining the game for you, however that...that was kind of the point, wasn't it? To make a stronger AI that needs fewer bonuses?
G

 

[wobuffet]

I for one really dont like this end game rush. I have really enjoyed this mod and Communitas mod before it. But there just dosent seem to be a middle ground now, Warlord is easy, Prince now to difficult.

Can I ask how many games you've played on each difficulty? Geography, opponents, etc. make a big difference, and I can't help but think you might be running into random variance that would smooth out if averaged over a larger sample size.

 

[Txurce]

The AI does not get a secret 'per turn' bonus via the CBO. It gets chunks of yields every once in a while for:

Entering a new era.
Founding a new city.
Winning a war.
Entering a Golden Age.

That's it. These bonuses scale based on difficulty, but they're actually somewhat lower than they were a few months ago.

I'm sorry that the AI's difficulty is ruining the game for you, however that...that was kind of the point, wasn't it? To make a stronger AI that needs fewer bonuses?
G

I think his main point is not having a level between Warlord and Prince, given present difficulty. Playing some more, as Wobuffet suggests, may address this.

Personally, I'd prefer to not notice more of a jump late game than I do at other points (when the jumps are more seamless). The very fact that people are noticing it shows the balance isn't quite right (no matter how much sense it might make in theory).

 

[Adam_S]

Can I ask how many games you've played on each difficulty? Geography, opponents, etc. make a big difference, and I can't help but think you might be running into random variance that would smooth out if averaged over a larger sample size.

On 7-7 three or four games, over King, Prince and Warlord.
On 7-15 three games, Prince.

Some Epic and Huge, some, and my last, on Standerd and Huge.

I won my Warlord game as a cultural victory before we got to the 'late game', so avoided the issue.

I think his main point is not having a level between Warlord and Prince, given present difficulty.

My point If i didnt clearly explain it. Im not "Your mod is rubish", Im "Your mod is one of the best ive ever seen, but this change is to much for my tastes." as an 'average' player.

 

[Funak]

The AI does not get a secret 'per turn' bonus via the CBO. It gets chunks of yields every once in a while for:

Entering a new era.
Founding a new city.
Winning a war.
Entering a Golden Age.

That's it. These bonuses scale based on difficulty, but they're actually somewhat lower than they were a few months ago.

I'm sorry that the AI's difficulty is ruining the game for you, however that...that was kind of the point, wasn't it? To make a stronger AI that needs fewer bonuses?
G

Didn't you mention earlier that AI techs are cheaper?


Btw, does the AI get some kind of bonus whenever it adopts a new policy? Because the first AI to get an ideology always instantly jumps up to 5 tenets in that ideology (instead of the normal 2 early adopter tenets)

 

[Gazebo]

Didn't you mention earlier that AI techs are cheaper?


Btw, does the AI get some kind of bonus whenever it adopts a new policy? Because the first AI to get an ideology always instantly jumps up to 5 tenets in that ideology (instead of the normal 2 early adopter tenets)

The AI gets a per era modifier that reduces tech costs, unit/building costs, etc. That's all vanilla difficulty stuff, though.

G

 

[Adam_S]

The AI gets a per era modifier that reduces tech costs, unit/building costs, etc. That's all vanilla difficulty stuff, though.

G

Perhaps the issue is that you/the team have done so well getting the AI to play, the bonuses are now an advantage rather than a crutch?

 

[Txurce]

Perhaps the issue is that you/the team have done so well getting the AI to play, the bonuses are now an advantage rather than a crutch?

Maybe, but even if they're an advantage, the solution would then be for the player to just drop down a level. I get the sense that I would be seeing what looks like an anomaly at a certain point, even if I can shrug it off because I'm way ahead.

 

[Funak]

The AI gets a per era modifier that reduces tech costs, unit/building costs, etc. That's all vanilla difficulty stuff, though.

If the AI always have cheaper techs, then signing research-agreements is never worth it. I'm not bringing this up as a joke, this is absolutely serious, RA's are capped so that both sides always gets the exact same amount of science no matter how high their output is at, but if the AI have cheaper techs they are always going to go winning out of that exchange.

 

[Gazebo]

If the AI always have cheaper techs, then signing research-agreements is never worth it. I'm not bringing this up as a joke, this is absolutely serious, RA's are capped so that both sides always gets the exact same amount of science no matter how high their output is at, but if the AI have cheaper techs they are always going to go winning out of that exchange.

That's a vanilla thing, don't look at me.

G

 

[Funak]

That's a vanilla thing, don't look at me.

And you haven't changed the numbers at all? Then where does all this AI tech come from?

 

[Fakoleon]

To be clear about tech rush, we are talking about almost 1 tech per turn(or 2-3 max) for AI in late game, when it takes you 8-10-12 for you, even civ that where late the whole game, when entering late eras rush tech and overtake you, it's ok to have difficulty and not be the first, but at that rate it's clearly unfair and kind of frustrating.
Does bonus yield apply to science yield AND science bonus from great scientist? it would make a massive boost as the scientist get advantage of the boosted science+ the bonus is boosted itself.

I noticed that when Haile the Ethiopian is in a game, he is always far away in every field, someone noticed this too? he can be like two or 3 era away from latest civ

 

[Morgawr]

As long as the AI bonuses are intentional per difficulty level and the math is working as intended , doesn't bother me. I agree with "if AI gets too hard, drop your difficulty next time." But if there is indeed some calculation bug that is giving AI unintended extra science yields or techs, we have a problem..