How does AI advance so fast in science?

[Seek]

Ouch. Thing is, the Pioneer Fort doesn't really look all that powerful on paper - perhaps we should change it to a flat bonus though.

What's the situation diplomatically in that game? Specifically Washington's?

 

[wobuffet]

I'm actually the only Research Agreement that Washington's involved in at the moment (this is several turns after the screenshot in the last post), although there's also one Civ I haven't met yet:

Spoiler :

I think I was mistaken about InfoAddict not counting RAs, though, because its reports on my Science output (but not Washington's) changed when I declared war on Washington to cancel the RA and see if it'd change the reported Science numbers:

Spoiler :

before:


after:

The math doesn't work out though... if 246 and 211 are my Science outputs before and after cancelling the RA, Washington's before-RA output x must be given by
(246-211)=0.03(211+x),
which yields x = 955.67... far below 1317 - 35 = 1282. That gap of 326 is definitely too large to be coming from an RA with the one civ I haven't met yet, so tl;dr: I have no idea what's going on with the math here.

Any chance there's a math error in the RA coding, Thal?


In any case, I'm pretty sure the main issue at hand is runaway population:

Spoiler :

 

[Txurce]

There is definitely a consensus that the AI has a population-based runaway science problem on Emperor that makes staying ahead in the Modern Era (if not sooner) very difficult. The current difficulty level is probably equivalent to what I think Immortal should be.

As I've already said, my view is that the least unsatisfying way to adjust it is by nerfing the AI's base era bonuses.

Nerfing public schools, pioneer forts, RA's - specifics - hurts the human player as well, makes certain civs less fun to play and, most importantly, makes the game less fun. I love feeling the effect of each one of those treaties and improvements, and I like seeing America and Siam (and others) flex their specific muscles almost as much.

 

[Dunkah]

I don't think it's fair for you "emperor" players to say that the Tech levels at King are not runaway, or specifically that Tech levels are runaway at higher levels.
I have never played higher than King and feel very strongly that they in fact are runaway at that level.
I've got to the point where I am not playing anymore. I play for a couple hours look at the tech scores sigh and sign off. King is too difficult and Prince is too easy.

 

[Txurce]

I don't think it's fair for you "emperor" players to say that the Tech levels at King are not runaway, or specifically that Tech levels are runaway at higher levels.

I don't think anyone said anything specific about King. And no one said that tech levels are runaway at higher levels only. Emperor players are quite reasonably talking about the level they play on.

I have never played higher than King and feel very strongly that they in fact are runaway at that level.

You're probably right. If they're runaway on Emperor, then it's likely that they are runaway on King.

I've got to the point where I am not playing anymore. I play for a couple hours look at the tech scores sigh and sign off. King is too difficult and Prince is too easy.

That's a different point. Thal recently smoothed the differences between Prince, King and Emperor. If you could specify the current too-large spread between Prince and KIng, I'm sure it would be helpful.

 

[Dunkah]

Agreed that needing to grab the Great Library just to keep pace with the leading AI civs in Science is no fun at all.

...

In any case, I think by this point (extensive discussion in two 3-page threads) it's safe to say there's a consensus that VEM has a problem with runaway AI Science leaders, at least at higher difficulty levels. Washington, in particular, seems to always have an incredible population and beakers per turn in my games.

I don't mind having one or two leader Civs that I have to watch out for if not deal with directly (i.e., cripple or eliminate) each game, but when I consistently have no chance at all at nabbing the late Medieval and Renaissance World Wonders (e.g., Notre Dame and The Forbidden Palace), the game's simply less fun to play.

Over the past two weeks I have played a King and a Prince game.

King game (Elizabeth), I was 10 -15 techs behind the majority of the field. I was 5 - 7 Policies behind the majority of the field. I had happines and gold producing problems. was ranked dead last to all but the lowliest players that had been defeated miliatarily, and was in the process of being over run by supperior numbers of Artillery, Light Infantry and Infantry to my Trebs, Muskets and Scouts. I had 2 or 3 CS's. Got one or two Wonders.

In my Prince game (Neb) I have been at the top of the heap in Science since the very begining. Only Korea has one more Tech than I. every other Civ has at least 10 less than us. I am 6 or 7 policies away from the best cultured Civ, closer 15 or 20 to the rest of the pack. I outnumber just about everyones miliatary, and just before the game crashed, I had taken 4 of egypts cities in one turn after declaring war. I have 4 or 5 loyal CS's. After Capturing 4 wonders from egypt I have almost every wonder in the game. 10 or 12 I think. (I lost count).

If I am playing peacefull, I feel like I can close my eyes and win at Prince, and I feel like I must perform a certain sequence of moves in order to even come close to performing well at King, but still end up chasing everyone from the begining.

I feel like if I play Conquest that I am simply trying to manage happiness. The conquering part is almost simplistic.

Niether way is very fun to play.

 

[Txurce]

Over the past two weeks I have played a King and a Prince game.

King game (Elizabeth), I was 10 -15 techs behind the majority of the field.

In my Prince game (Neb) I have been at the top of the heap in Science since the very begining.

If I am playing peacefull, I feel like I can close my eyes and win at Prince, and I feel like I must perform a certain sequence of moves in order to even come close to performing well at King, but still end up chasing everyone from the begining.

1. Along with what I said earlier, Wobuffet said there's a runaway science problem at least on the higher difficulty levels (which is all he can comment on) - not only on those levels.

2. Babylon is going to give you better science results than England, but obviously that doesn't account for the disparity. Did you have at least as many good cities with England?

I played 100 turns of Prince yesterday and found it easier, but not that much easier than Emperor in that time frame. That's because the AI era bonus multipliers hadn't kicked in yet (or aren't there at all) compared to the higher levels. Maybe Thal can comment on the literal differences between Prince, King and Emperor, so that all of us can be on the same page. I know I found the stats for Emperor very interesting.

 

[Dunkah]

1. Yes I see that he did not specifically leave that out, however I want to make sure King difficulty doesn't get lost in the fray. I read just about everything you guys write on these forums and can't usually comment because I don't try to have the understanding and overall picture of the game mechanics that you have, so when I do comment it's because I feel strongly about the subject matter and want to make sure the little guys get represented as well.

I teach software applications. I feel that once most people hit a certain level they can't lower themselves back to the level of a beginner and see the software from their point of view. They assume too many things that a beginner wouldn't know. I feel this is similiar. You guys play at such a high level and have such an intimate knowledge of the process that you can't possibly put yourselves in the shoes of a player like me.

2. Yes I would expect a discrepancy between those two leaders.


I play all of my games similar, Get a good base of Science and Culture early, worry about troops later on. I typically build three cities until I get a feel for the surrounding area, and what kind of VC I should go for. I usually have a VC in mind but circumstances change.

Recently with my King game playing Elizabeth I wound up trying to conquer because I was so far behind by turn 100. Ended up with 5 decent cities pops in the high teens, 2 of those converted capitals. A dozen puppeted cities.

With my Prince game playing Neb I had three cities and Thebes all with pops over 27. Decided to stay peaceful throughout due the the huge lead I had in Science and Culture. Egypt declared on me.

With King I am behind from turn 1 until the end of the game. I sometimes joke because after I place my first city I am ahead until I hit "Next Turn". Then I know for the rest of the game I will be looking up to most of the other Civs.

With Prince by turn 10 I am usually near the top of the pack and stay in the top 4 or 5 for the whole game. Usually number 1.

With Prince I feel like I can do anything. Like I have a choice of what VC I wish to go for and that no matter which Civ I am playing with I can accomplish that VC fairly easily, though lots of time by the time I get to the point where that VC is a reality I have one or more of the other VC's as viable options. Sometimes I have to put one of those off just to get the VC I was going for.

With King I feel like I am constantly behind and that I have to keep on adjusting majorly to the game. That I have very few choices as far as grand strategy is concerned. If I don't go for a very specific strategy and stay strictly on that path that I am not going to compete with the top Ai's. And that by doing so I am going to be seriously lacking in one or more of the other areas.

Prince = Active Style (Choose my destiny)
King = Reactive Style (Chase other players)

 

[Txurce]

Thanks for the thorough explanation. I see your point and of course want the game to be satisfying on all levels. The details you provided certainly indicate a huge disparity between Prince and King. Once again my initial reaction is to wait for Thal's comment, so we can see just how these levels differ. Maybe the solution is as simple as reducing the gap, but it's possible that a player's skill level would exacerbate the differences beyond something someone playing on a higher level can see. (I say this knowing that old Deity may not be that much harder than old Immortal, but I know I would get my head handed to me in the former, whereas I could compete with the latter.)

 

[Zeppelin4]

I play on King level and find the challenge just about right. It depends on what civ I'm playing on how I place in the scoring. I have come to notice playing Korea now that I'm doing much better. It shows me that I have been ignoring science to much in my other games. Korea has taught me to be a better player.

If I play on prince its way to easy. I find King to be perfect for it is making me a better player without getting my butt kicked. One thing I have done lately is to try out more civs to learn the game better. I was only playing about 4 different civs and found that I became a better player by trying others. my 2 cents.

 

[Txurce]

That's good to hear, this dialogue aside. I'm curious: did you have a hard time adjusting to AI gold-spending (and all the other changes between say, v131 and v137)? Was it tougher for a while before you improved your skills.

 

[Jaybe]

I believe the catch-up mechanic you describe, however, is already in the game: see here for details.

The relevant value is TECH_COST_TOTAL_KNOWN_TEAM_MODIFIER in
...Steam\steamapps\common\sid meier's civilization v\assets\Gameplay\XML\GlobalDefines.xml , set to 30 by default. I think corresponds to a beaker discount of up to 1 – 1/(1 + [7/8]*30/100) = 20.8% on a standard size (8 player) map if every civ other than you has already researched a technology.

I'd be more than okay with experimenting with bumping this value up, although I doubt this alone will satisfactorily solve the AI runaway Science problem.

This is a holdover from Civ4's permanent alliance (research from alliance members A and B were not simply added together), not a discount for techs known by x number of civs in Civ5.

 

[Dunkah]

Keep in mind Korea and Babylon are two of the more Scienc-ey types so thus the desparity between Science levels would be less with those Civs.

Science is key to just about every VC. If you are doing well in science you are typically doing well in most of the VC's.

High Science effects...
Culture - Get to culture buildings faster, have a better chance to get later Wonders.
Conquest - Get better units, get better production buildings, get better buldings for gold support and more happiness buildings.
Diplo- Get to UN before anyone else.

I'm sure there are many more but Science seems to me to be one of the biggest factors in victories. Thus turning up science gives all AI's a better chance of winning simply because they are AI's.

 

[Txurce]

Keep in mind Korea and Babylon are two of the more Scienc-ey types so thus the desparity between Science levels would be less with those Civs.

Science is key to just about every VC. If you are doing well in science you are typically doing well in most of the VC's.

I'm sure there are many more but Science seems to me to be one of the biggest factors in victories. Thus turning up science gives all AI's a better chance of winning simply because they are AI's.

Agreed. But turning up science helps the AI more only because the AI has bonuses that give it an edge there as well. It's those bonuses I'm hoping we lower.

 

[Zeppelin4]

That's good to hear, this dialogue aside. I'm curious: did you have a hard time adjusting to AI gold-spending (and all the other changes between say, v131 and v137)? Was it tougher for a while before you improved your skills.

First to say I'm not a great player. I have a tendency to start over a lot. I have yet to finish a game for I like the early part the best. My scores are based off when I start over not a complete game.

I have found it harder to get higher scores lately so not sure if that is what your asking. Like I said Korea which I had never played until the recent changes showed me I was ignoring science to much in my other games.

My next game I'm going to try another civ and apply what I have learned playing Korea and see how I fair.

 

[Patum333]

Science is key to just about every VC. If you are doing well in science you are typically doing well in most of the VC's.

I'm sure there are many more but Science seems to me to be one of the biggest factors in victories. Thus turning up science gives all AI's a better chance of winning simply because they are AI's.

This is absolutley right!
You can't win on deity without well in science and military power.

 

[Thalassicus]

Thanks for all the helpful feedback! The tech runaway problem is something I haven't experienced... probably because I kill everyone on the map.

RAs will Allies jumped from being 50% to 200% better than normal RAs in v131

Did you mean 150% to 200%? Alliances used to make RAs give 4.5% of combined science; now it's 6%. I want to keep this because alliances mainly help peaceful empires.

 

[Txurce]

Did you mean 150% to 200%? Alliances used to make RAs give 4.5% of combined science; now it's 6%. I want to keep this because alliances mainly help peaceful empires.

Agreed. Have you been able to play with your notion of making the AI bonuses 100% gold based?

 

[Seek]

If RAs give 200% as advertised the it is a jump from 4.5 to 9%. Was it an error in the patch notes? As I said, there is some confusion here because there is no documentation regarding an interim jump (if any) from 50% to 150%.

I also just realized that the science-from-happiness mechanic is also certainly a factor in tech runaways (I can't believe I didn't think of this earlier!) - perhaps lowering this bonus would even things out a bit. Since peaceful AIs frequently have 30+ happiness that would account for some of the disparities here.

 

[pthmix]

I think Thal is talking about the base value (150% of 3 = 4.5 and 200% of 3 = 6 ) and you are talking about the increase (3 + 50% of 3 = 4.5 and 3 + 100% of 3 = 6 ).