v137.1 on Emperor

[Txurce]

As of t160, Greece (with an average score and pop) controls all 7 CS with totals of 334, 291, 168, 164, 158, 133 and 89. They are near the north pole, and don't seem to have any desert tiles. This seems to break the CS system for me.

Arabia has over 4000g - over the limit?

 

[Seek]

perhaps I am oversimplifying the issue as my basis of comparison is vanilla games with AI set to play on Prince (which somehow mimics your HandicapInfo values), and in those games I see a smooth flow of growth without such huge runaways...

Probably because there are mechanics in VEM which increase science dramatically (if the AI plays a certain way) - making these mechanics less powerful would mitigate the VEM runaways. How about this:

• We set a middle-ground between the old cheaper/longer GAs and lower the science from happiness to .5%.
• Set RAs to grant 2% combined research (4% with Alliance) since we're getting the vanilla boost (50% of median tech) anyway.

I'd bet $100 that these changes would mean a shift away from "runaway" civs to simply "leading" civs.

Several buffs were made to food income in vem:

• Better granaries.
• Aqueduct added.
• Different citystate yield distribution.
• The + surplus food policy in the Tradition tree.

Until now, there were no corresponding adjustments to food costs, which means growth rate is much faster than originally. The adjustments in that chart above should compensate for this so growth rates in vem are closer to vanilla. Tall empires still have an advantage because they have more happiness.

Also, on a personal note it felt weird to have much bigger jumps in food costs for small cities than big ones (the 'diff' column). Logically, it should feel like city growth slows down in big cities, not speeds up. Bigger cities still have smaller cost jumps than small cities (6% vs 18%) but not as much as vanilla (4% vs 60%).

The buffs to food were incorporated from TBC into vanilla* because larger cities are more fun for the player; like Txurce and Zaldron, I don't think we need these adjustments either (I can only imagine what Ahriman would say about 2 yield bonus resources!). pthmix's idea of removing the worker is much more elegant in my eyes, and alternatively we could just delay the worker (until the AI researches two techs, say) - if possible.

*It doesn't seem fair that you're comparing VEM to vanilla-beta (Civ5 when it was released) and not with vanilla as it is now, when the latter is what everyone else refers to when they make the comparison. As of now, there are comparable amounts of food between VEM and vanilla.

 

[Zaldron]

Several buffs were made to food income in vem:

• Better granaries.
• Aqueduct added.
• Different citystate yield distribution.
• The + surplus food policy in the Tradition tree.

Until now, there were no corresponding adjustments to food costs, which means growth rate is much faster than originally. The adjustments in that chart above should compensate for this so growth rates in vem are closer to vanilla. Tall empires still have an advantage because they have more happiness.

Also, on a personal note it felt weird to have much bigger jumps in food costs for small cities than big ones (the 'diff' column). Logically, it should feel like city growth slows down in big cities, not speeds up. Bigger cities still have smaller cost jumps than small cities (6% vs 18%) but not as much as vanilla (4% vs 60%).

You can't count the aqueduct since that's a vanilla building now. The tradition and granary points seem quite valid and I'm willing to give a try to the latest beta and see how it plays out.

 

[Txurce]

Probably because there are mechanics in VEM which increase science dramatically (if the AI plays a certain way) - making these mechanics less powerful would mitigate the VEM runaways. How about this:

• We set a middle-ground between the old cheaper/longer GAs and lower the science from happiness to .5%.
• Set RAs to grant 2% combined research (4% with Alliance) since we're getting the vanilla boost (50% of median tech) anyway.

I'd bet $100 that these changes would mean a shift away from "runaway" civs to simply "leading" civs.

Why are you so sure, if the changes affect the human player as well (just like the v140 food nerf)?

Why not lower the AI growth %? Does that work differently than it sounds?

 

[Thalassicus]

@Txurce
At least citystate influence isn't in the 1000+ range anymore. I think I'm getting it towards a good range now, just a little lower... perhaps 100 or so.

You can watch Arabia's spending habits in the tuner. AIs with high stored gold could be due to several different reasons:

1. The AI might have nothing to buy.
2. Income rate could be higher than the one purchase per turn limit.
3. Could be a bug.

Since #1 and #2 are opposites of one another, the basic quantity of stored gold is not information we can use.


@Seek
Good idea with golden ages, the happiness incentive, and research agreements. I'll do that once I have the bugs and oddities with the AI purchase system worked out.

I agree that larger cities are more fun to play, but cities that get too large are less fun. I read feedback recently that some people's cities were getting into the 30 range where there's no tiles to work, and no specialists slots available... they just had a lot of unemployed citizens sitting around. That feedback is why I raised the later costs. The change to earlier costs was to balance the recent Granary and Palace buffs.

 

[Txurce]

No problem with Arabia - just reporting!

The CS issue does seem to be receding, although it's still becomes essentially unusable for happiness by the human player about a third of the way into the game.

I don't know who would come out ahead on a poll on nerfing 30+ city growth, but I think what we've been reading today is a preference that small cities grow the way they once did.

But the overall issue of runaways still exists. It's obviously a waterfall of the increased AI gold spending. As we've seen with various attempts to address the issue, only AI-specific changes help noticeably. Dropping the extra worker makes sense from this perspective. Like I asked earlier, what would happen if you lowered the AI growth % by 10 or 15?

 

[Thalassicus]

I do not believe growth is slower for small cities (1-5 pop). We've got a new 15 per turn to spend on 33% cheaper Granaries, which gives us much higher food potential in the early game.

I feel it's more logical to reduce AI gold/science/production bonuses than add an AI food penalty, because food affects gold/science/production. Extra workers are focused on militaristic AIs, who traditionally do very poorly, so I think it would be counterproductive to reduce their strength.

 

[Txurce]

I feel it's more logical to remove AI bonuses than add an AI food penalty. Only militaristic AIs get extra workers, and militaristic AIs traditionally do very poorly, so I think it would be counterproductive to reduce their strength.

Okay. I just don't know what bonuses you're going to cut at this point that would make the sort of difference we're seeking.

 

[Thalassicus]

I suspect the main thing is citystates are currently unattainable for the human player. I'm working on fixing that, and I removed the AI production and science bonuses in v140. I want to see how these changes play out before making further changes. At this point the only AI bonuses are:

Lower unit maintenance and upgrade costs
Higher unit experience
Bonus sight range
Extra starting units
Faster worker rate​

If further changes are needed I'll alter the unit maintenance/upgrade bonuses.

 

[Txurce]

Definitely agreed on the CS issue being paramount. I've already submitted partial results regarding pop/sci, but look forward to hearing results from the less dedicated players (!).

 

[pthmix]

...

I agree that larger cities are more fun to play, but cities that get too large are less fun. I read feedback recently that some people's cities were getting into the 30 range where there's no tiles to work, and no specialists slots available... they just had a lot of unemployed citizens sitting around. That feedback is why I raised the later costs. The change to earlier costs was to balance the recent Granary and Palace buffs.

One thing to note: when you play a tall empire you need cities this big to compete. It is a issue of city placement. If you place them far enough from each other you will have enough good tiles to work, especially since there are at least 7 specialist slots of each type available.

Regarding the resources: the changed wheat really hurts city locations in plains. And I no longer have to think about whether I want super growth (cattle) or good production (stone) which is a bad thing imo. Why do you not remove the on the palace if you want to limit early growth?

Edit:

...

@Seek
Good idea with golden ages, the happiness incentive, and research agreements. I'll do that once I have the bugs and oddities with the AI purchase system worked out.

...

This could be a chance to change the + while the Empire is policy in rationalism to something like double/triple the bonus from happiness. It would make both the policy and the happiness science boost more interesting.

 

[albie_123]

Seconding that the wheat change is a bit odd.

It takes away a lot of the variety of early-game terrain.

 

[Seek]

One thing to note: when you play a tall empire you need cities this big to compete. It is a issue of city placement. If you place them far enough from each other you will have enough good tiles to work, especially since there are at least 7 specialist slots of each type available.

Regarding the resources: the changed wheat really hurts city locations in plains. And I no longer have to think about whether I want super growth (cattle) or good production (stone) which is a bad thing imo. Why do you not remove the on the palace if you want to limit early growth?

Agree with everything here.

 

[Txurce]

Why are you so sure, if the changes affect the human player as well (just like the v140 food nerf)?

@Seek
Good idea with golden ages, the happiness incentive, and research agreements. I'll do that once I have the bugs and oddities with the AI purchase system worked out.

Why reduce RA's? That hurts the human about as much, in terms of staying even. Even GA's are a mixed bag, in that it will help the situation, but make the game less fun.

I don't understand why making small-bore, mixed-bag changes like these are a good idea, but simply reducing AI pop growth is not, when that is seemingly at the crux of the major problems.

 

[Seek]

Reducing RAs would make peaceful AIs (or humans for that matter) that are doing well in science not end up as runaways. But reducing the science from happiness is more important to this end (because the AI hordes happiness at a level most players won't) - I won't mind if we decide not to nerf RAs.

I really don't think that AI growth is the crux, and I thought we were on the same page with this: I argued earlier in this thread that there is no direct causation of population leading to runaway status based on your reports - and you agreed with me! So you'll forgive my confusion here I hope. We've seen massive AI populations since the game was released, and runaways have just recently cropped up. There is something else going on here besides population, and giving the AI a penalty for growth doesn't seem like the answer to me.

Also, how are the changes in the quote "mixed-bag"? They're all related to science and reducing runaways. In the past we had longer/cheaper GAs to incentivize accruing happiness, then we changed the incentive to a science bonus which seems like an obvious candidate for the rise of runaway AIs (sorry if I'm repeating myself, but I want to be clear). Turning back a bit to in this case could work out well, I think: 1% per is *very* strong, and halving it while making GAs slightly more available would still have me, at least, trying to keep double-digit happiness throughout a game.

Edit - I have not played a game with some of the recent changes (like no AI bonus for production, etc.) - are we still seeing the runaways like we were? In my most recent game I played (Persia wide-ish culture game with v137) I was first or second in science throughout and I didn't even take Rationalism!

 

[albie_123]

I can agree with Seek on three major points:

1% for 1 is very strong, that large pops are not the cause of the problem and are often entirely unrelated to runaway civs, and that science runaways don't seem to be as big an issue anymore - or at least, not in my last game. (Admittedly, one game is maybe too little experience to base an opinion.)

In my opinion the one big issue is City State friendship, at the moment. Once that is sorted, I think runaways will be much less pronounced.

 

[Txurce]

I really don't think that AI growth is the crux, and I thought we were on the same page with this: I argued earlier in this thread that there is no direct causation of population leading to runaway status based on your reports - and you agreed with me! So you'll forgive my confusion here I hope. We've seen massive AI populations since the game was released, and runaways have just recently cropped up. There is something else going on here besides population, and giving the AI a penalty for growth doesn't seem like the answer to me.

The misunderstanding is that I also don't see a 100% correlation between pop and science, but do see a very strong correlation. (I've posted a lot of stats that I think show this.)

Still, your point about massive AI populations having already existed is a good one. It seems obvious to me that the increased AI gold spending is at the heart of VEM's major problems. But gold and science production increases with large population, and therefore make the problem worse. We agree that we want to have the AI spend its gold is an ideal adjustment, so changes have to be made elsewhere. AI population is unrealistically too high, so why not lower it? Aesthetics aside, wouldn't it automatically lower their gold and science? With all of the recent adjustments, why is testing this verboten?

Beyond all this, I have maybe a bigger question. Why does the AI grow the way it does, with its attendant science boost, if it's lost seemingly all its applicable bonuses? Where is the gold coming from? Maybe it's a bug, but its source should be easy for someone knowledgeable to identify with FireTuner.

Also, how are the changes in the quote "mixed-bag"? They're all related to science and reducing runaways. In the past we had longer/cheaper GAs to incentivize accruing happiness, then we changed the incentive to a science bonus which seems like an obvious candidate for the rise of runaway AIs (sorry if I'm repeating myself, but I want to be clear). Turning back a bit to in this case could work out well, I think: 1% per is *very* strong, and halving it while making GAs slightly more available would still have me, at least, trying to keep double-digit happiness throughout a game.

They're a "mixed bag" in that they hurt the human player as well (weaker RA's), are less fun (shorter GA's). I did agree with your proposal to lower the happiness science bonus. I could see shortening GA's if it comes to it. But I would rather try approaches that solely penalize the ultra-successful AI's.

Edit - I have not played a game with some of the recent changes (like no AI bonus for production, etc.) - are we still seeing the runaways like we were? In my most recent game I played (Persia wide-ish culture game with v137) I was first or second in science throughout and I didn't even take Rationalism!

Thanks to whatever produces lots of beakers for the AI (happiness? pop?) the big AI finish strongly. I have been able to be competitive (but not seize control) despite playing with Korea (and occasionally Babylon) and generating large amounts of science (i.e., I'm not playing poorly). As of the genesis of the CS issue, I had also begun to wonder if the science issue was becoming manageable, meaning relatively minor adjustments could be all that was needed. (When the Inca won via Culture in t251, they were ahead by far in science). As I mentioned earlier, I am okay with waiting on further changes until the CS issue is resolved. But I objected to certain approaches being tried, and others not.

I can agree with Seek on three major points:

1% for 1 is very strong, that large pops are not the cause of the problem and are often entirely unrelated to runaway civs, and that science runaways don't seem to be as big an issue anymore - or at least, not in my last game. (Admittedly, one game is maybe too little experience to base an opinion.)

In my opinion the one big issue is City State friendship, at the moment. Once that is sorted, I think runaways will be much less pronounced.

As per above, I agree that CS is by far the bigger issue today. But I have posted stats of multiple games showing a strong correlation between pop and science on Emperor. Again, it may not prove anything - I just think it's worth testing.

 

[Seek]

The misunderstanding is that I also don't see a 100% correlation between pop and science, but do see a very strong correlation. (I've posted a lot of stats that I think show this.)

Still, your point about massive AI populations having already existed is a good one. It seems obvious to me that the increased AI gold spending is at the heart of VEM's major problems. But gold and science production increases with large population, and therefore make the problem worse. We agree that we want to have the AI spend its gold is an ideal adjustment, so changes have to be made elsewhere. AI population is unrealistically too high, so why not lower it? Aesthetics aside, wouldn't it automatically lower their gold and science? With all of the recent adjustments, why is testing this verboten?

I think the reaction of many people (myself included) to giving the AI a penalty at the higher difficulty levels just seems wrong. That's not to say that it is wrong - it simply goes against how we see the AI, and we, perhaps subconsciously, think to ourselves "the last thing we want to do is nerf the already poor AI". Looking at it from your point of view as you describe above, I concede that I can't think of any truly valid reason not to try it out - however, I would prefer it if we could find the root causes of the problems and eliminate them.

Beyond all this, I have maybe a bigger question. Why does the AI grow the way it does, with its attendant science boost, if it's lost seemingly all its applicable bonuses?

I am pretty sure the AI never goes off of the default city focus, which, when I press that button, in almost every circumstance yields me more food than I normally would work. I have no idea how long it's been since I went an entire game without changing the city focus (if ever) but it wouldn't surprise me if the populations in my cities were significantly higher if I did so.

They're a "mixed bag" in that they hurt the human player as well (weaker RA's), are less fun (shorter GA's). I did agree with your proposal to lower the happiness science bonus. I could see shortening GA's if it comes to it. But I would rather try approaches that solely penalize the ultra-successful AI's.

In no way did I propose shorter GAs. I think you misread what I wrote: I proposed lowering the science from happiness and buffing GAs by making them more available (cheaper to reach and/or perhaps longer duration) - just not back to the levels we were seeing before the science mechanic was added. A thread posted by Thal earlier today asking if anyone knew how to adjust the GA production modifier interests me - perhaps he's considering making GAs more powerful to boot!

 

[Bernd-das-Brot]

I played a game on v140.1 (emperor/standard/continents) yesterday/today. I experienced no greater bugs, just some smaller weird things. I played till around t190 and quit then because it was obvious I could win in any way I wanted. I was leading in science and culture production and was near the leading AIs regarding everything else.

I observed the gold prod of the AI and could not discover anything strange. Competition for the CS was as it should be. And the AI definitely had not too much food! My capitol was the biggest city on my continent until I stopped growth because of happyness reasons. There was no city beyond pop 20 the turn I quit on the world as far as I could see. At least from my experience of this game I see absolutely no reason too nerf AIs food production.

On the contrary main problem was that the game was a bit too easy. Even with a mediocre starting position (coast, small river, silver and pearls, 2 sheep, no real food ressource like wheat or fish) and one of the weakest civs (England) there was no real challenge.
Perhaps 3% per era science and a bigger prod-modifier for (at least) unit building on emperor would be okay.

Some things I observed which could perhaps need a bit tweaking:
- Buying tiles with cultute points seems at a very good level in this version. But buying tiles with gold was way too cheap. 20-60 Gold even in the 4th row? And the prices did not really seem too rise. Intended or a bug?
-Great Scientists give too much beakers. I had to save them for the 3rd level renaissance tech if I do not wanted to waste a lot of beakers. And with the change to national college (25 % instead of 50%) it is obvious not to settle them. Lightbulbing now gets more.
- +1 on farms is still very good and gets better with -1 food consumation for specialists. I think it should not be the opener of freedom.
- RA are still always useful so I would support to nerf them.

Some weird things I observed:
- One out of ~10 AI units had a lots of promotions. 14(!) was most I saw.
- I conquered an iron mine with 10 prod (opportunty?).
- Mecca (conquered) had +3 or 4 culture and science on its city field.

All in all a very fun game and thanks again Thal for all your hard work for this game!

I hope to play another game tonight with the actual version!

PS: for giving AI +50 % city attack.

 

[Txurce]

Looking at it from your point of view as you describe above, I concede that I can't think of any truly valid reason not to try it out - however, I would prefer it if we could find the root causes of the problems and eliminate them.

I am pretty sure the AI never goes off of the default city focus, which, when I press that button, in almost every circumstance yields me more food than I normally would work. I have no idea how long it's been since I went an entire game without changing the city focus (if ever) but it wouldn't surprise me if the populations in my cities were significantly higher if I did so.

In no way did I propose shorter GAs. I think you misread what I wrote: I proposed lowering the science from happiness and buffing GAs by making them more available (cheaper to reach and/or perhaps longer duration) - just not back to the levels we were seeing before the science mechanic was added. A thread posted by Thal earlier today asking if anyone knew how to adjust the GA production modifier interests me - perhaps he's considering making GAs more powerful to boot!

1. I'm with you on this.

2. I don't think the "default" setting is the reason the AI has 2-5X my poplation!

3. Thanks for the clarification. It sounds good to me.

I played a game on v140.1 (emperor/standard/continents) yesterday/today.

Competition for the CS was as it should be.

There was no city beyond pop 20 the turn I quit on the world as far as I could see. At least from my experience of this game I see absolutely no reason too nerf AIs food production.

This game sounds like it should be put under the microscope! It's the only example I've seen of the AI not having an absurdly high lock on CS. With regard to population, I've posted enough examples of relatively runaway populations.

Some things I observed which could perhaps need a bit tweaking:
- Buying tiles with cultute points seems at a very good level in this version. But buying tiles with gold was way too cheap. 20-60 Gold even in the 4th row? And the prices did not really seem too rise. Intended or a bug?
-Great Scientists give too much beakers. I had to save them for the 3rd level renaissance tech if I do not wanted to waste a lot of beakers. And with the change to national college (25 % instead of 50%) it is obvious not to settle them. Lightbulbing now gets more.

I was going to post the same note about tile buying. I am back to settling near the AI, then buying every single worthwhile tile on that turn.

I've also been saying for a while that GS give too many beakers in certain eras (mainly Medieval, Renaissance). I'm in favor of restoring the old NC boost , which would benefit tall civs anyway, and that would at least address the bulbing issue.