New Beta Version - August 16th (8/16)

[Zanteogo]

Imperfect does not have to mean shooting yourself in the foot. There are other ways to balance puppets which are less irrational.

They are balanced though. Having a downside that's different than other options does not means it's not balanced.

I truly think people just want and option for a "free city". Sorry, against that.

If we mess with the current balance between all three options we are making a go to choice. There has to be a downside of wide conquest.

 

[Stalker0]

1. Yields from Puppets

Let us assume for a second that a City only produces Gold, Science, and Culture and that each of these Yields has the same utility value of 1 (Gold, Science, and Culture output of a City are all equally useful).
Let us further assume that the net Gold utility of Gold is the combination of -2 utility from Building maintenance and +3 utility from Gold production (Building maintenance is equal to 2/3 of raw Gold production).
For a regular City we thus get a net utility of 1 + 1 + 3 - 2 = 3.
For a Puppet without Martial Law all positive utility is cut by 80%, resulting in the following net utility value: 0.2 + 0.2 + 0.6 - 2 = -1.
For a Puppet with Martial Law (60% cut) we get: 0.4 + 0.4 + 1.2 = 0.
Under these assumptions Puppets are actually draining your resources without Martial Law and even with Martial Law they do not give you a net benefit.

Some things that were not considered for this analysis:

1. Food/Production: these only have localized effects and do not contribute to your empire as a whole. I therefore think they're irrelevant.
2. Faith: If you have a low number of Cities Puppets are not worthwhile in the first place. If you have a large number of Cities (which you do later in the game) the utility of Faith is low when compared to other Yields. I thus think it can be neglected.
3. Tourism: If you have a low number of Cities Puppets are not worthwhile in the first place. If you have a large number of Cities you get more Tourism from Annexation than from Puppets. You still get some Tourism from a Puppet though.
   
4. Build order: Puppets can have really bad build orders which results in overall lower Yields compared to regular Cities.
5. Military Utility: Both regular Cities and Puppets provide some military utility but this is hard to quantify. I would argue that regular Cities are a little better though.

3. My Strategy for going Ultra-Wide

Each non-puppeted City that you own increases the Needs modifier in your Cities by 9% (independently of map size).
If no countermeasures are taken a large empire will therefore become increasingly unhappy as it gains Cities.
Public Works can reduce Unhappiness to some extent but I think it doesn't fulfill the purpose it was added for.
If I remember correctly its purpose was to be a last resort for unhappy Cities.
However, if you only start building it once Unhappiness starts to become a problem it is basically useless.
Cities that are unhappy typically have very large Need deficits so if you build Public Works in such a City the Happiness and the Needs reduction has literally no effect.
You need to build a lot of Public Works to get any effect, but once you do you can quickly eliminate all Unhappiness from Needs.
Conversely, if you build Public Works in a City that is already pretty happy you get immediate benefits without the need for a large up-front investment.
This has led me to develop a barbell-like strategy for dealing with Unhappiness: instead of trying to make all of my Cities kind of happy I only try to keep some percentage of my Cities very happy while completely neglecting Happiness in all other Cities; as long as I have more total Happiness than I have total Unhappiness I do not suffer any penalties.
In practice this means delegating some Cities to build Buildings and Public Works while delegating other Cities to build Buildings and Units.
(In case it isn't obvious: you also need to build as many Factories as possible to get enough Production.)
With the above strategy I was able to control ~40 non-Puppet Cities while staying at ~60% Happiness.
Keep in mind that the above strategy needs a lot of long-term planning and experience to pull off well; when in doubt designate too many Cities to building Public Works rather than too few.
Alternatively just found TwoKay Foods to completely eliminate Unhappiness (but I don't know if that is intended).

Alright I'm digging in to this one, as I have some comments:

Gold, Science, and Culture output of a City are all equally useful) - I am going to challenge that assumption, as gold is very commonly seen as less important than the other yields. For common "napkin math" that the community has done in the past, Production = Science/Culture, and gold is generally half as valuable as prod. So I think a better assumption is 1 Culture = 1 Science = 2 Gold.

Building maintenance is equal to 2/3 of raw Gold production - So I'm not sure where the basis for this assumption lies, have you studied building maintenance compared to its average yield generation to make this conclusion?

For a regular City we thus get a net utility of 1 + 1 + 3 - 2 = 3 - So the way I read this, you are suggesting that every building generates (effectively) 1 culture, 1 science, and 3 gold, and costs 2 building maintenance....or at least in that ratio. Is that correct?


On to your strategy for going Ultra-Wide, I will say just playing my last game, I agree that building a few public works in every city does not have much happiness impact, and that was a "mere" 11 cities for me. I'll give your barbell idea a shot, as I can't imagine supporting 40 non-puppets on standard at the moment doing what I've been doing.

 

[JamesNinelives]

There has to be a downside of wide conquest.

You're clearly ignoring that puppets already have other downsides. Puppets working specialists isn't the worst part of them, it's just the one that makes the least sense.

I truly think people just want and option for a "free city".

They how do you explain my arguments? I'm a peaceful player, I almost never go for domination victory. You need to move past this assumption.

 

[tothePAIN]

I'm playing with these new resource distributions. My initial reaction is I don't like them, but I want to be fair.

Could anyone point me towards the discussion that lead to this? I can't find it and I'd like to read the reasons for the change. Thanks.

Its a different map script - communitas. I did not see any discussion leading towards the script for the new resource distribution.

The resource distribution is definitely different. I enjoy communitas more because it makes naval combat a bit more bearable, because there are more bays allowing for protected cities.

 

[ElliotS]

What's the % increase to social policy and technology costs on large? I think ElliotS might have the right idea for the wrong reason. Going wide with each city increasing by 1 more % would be a huge deal.

My point is that if the tall/wide balance is balanced at 7% on standard, what would make 5% on large a reasonable change? That just tips the scales in favor of wide for no reason. People always sight having more cities as the reason, but If having more cities is bad at 7% and good at 5% then it should be 5% for all maps. If having more cities is good at 7% but more balanced, then it should be 7% for all maps.

Not to mention that having cities add 10% on small maps is absurd.

 

[InkAxis]

They are balanced though. Having a downside that's different than other options does not means it's not balanced.

I truly think people just want and option for a "free city". Sorry, against that.

If we mess with the current balance between all three options we are making a go to choice. There has to be a downside of wide conquest.

Of you truly think that people want a "free city" then you are misrepresenting what we are saying. There is a downside of conquest, making the governor not a complete idiot won't somehow make puppets OP. Just because we want balance doesn't mean we want it to be easy.

I still don't see a reason why a governor would work an objectively bad tile. You say it's "random" but it's not. The governor selects the best tiles in their "opinion" - I don't see why it should be different in this case. Puppets are weak as is.

 

[Stalker0]

I just thought this one was funny. Behold the almighty Air God! It is truly invincible! (I do 1 hp a hit to it).

EDIT: Darn, I guess my picture didn't upload in time. Well here's a different picture, but it doesn't look nearly as impressive

Spoiler :

 

[Stalker0]

I thought this was worth highlighting. Definite improvements to the amphibious assault portion of tactical AI.

Spoiler :

1) Good use of the island up north to weaken any naval forces I bring to bear.
2) Nice use of melee on the west wing to protect the land cannon.
3) That knight on the right is a bad move, I have too much ranged firepower and I'll mow it down quickly. If he had brought his brother knight from the north at the same time it might have had some effectiveness, but realistically there should not be any incursions around the city until he's softened me up more, that trio of xbows + city will basically take down anything that lands in the immediate vicinity.
4) I do think this invasion needs a bit more troops to start. I think that's the biggest aspect of amphibious that took me a while to learn, and the AI needs to learn it to. Unless you have a beachhead to work from (like a friendly CS)...if your going to land, you have to land BIG. As soon as you land several of your units are going up in smoke, so you have to land on enough beach to spread out the defender's damage, and ensure you are going to kill units on your first strike. If you can't do that, the defender will just chew you up and spit you out.

Edit: And now a few turns later

Spoiler :

The AI did land his second knight on the east side, and as expected I blew those 2 away. However, it does seem to be focusing forces on the left side where I am weakest, a good sign. He had a trio of galleassas protecting this fleet but I took one out, if he doesn't bring some more naval support soon I may take out his remaining fleet, so we will see.

However, he is definitely bringing a bit more power to bear which is great to see.

Edit: And the next turn

Spoiler :

Excellent formation here. The AI did push its troops to the southwest. The trio of cannons in the back are a nice fortified line, with knights to provide frontline meat.

The danger here are my caravels, ideally the AI once to take this out to control the sea and allow for continuing reinforcements. It should be able to take out my northmost caravel with cannon fire pretty easily (that one killed a key caravel so I am willing to trade it). Then we will see what it does with its Galleasass, it should pull them back away from my caravels to allow his land forces to deal with my ships...otherwise I may be able to snipe them, and retake control of the sea.

I am doing a strong buildup of musketmen now, so we will see what the AI does...

EDIT: Next Turn!

Spoiler :

Ok, here the AI falters a bit. First, its pushing its knights out aggressively instead of keeping them close to the cannons. Effectively the AI needs to change its mentality to a "siege posture". Its got 2 cannons on a weak coastal city, with a 3rd ready to take over if one gets wounded, this is a solid position, it now needs to hold this. Instead its squadering its melee protection with pushes. Those knights are dead, and its going to cost the AI.

On the naval side, the AI snagged 2 of my caravels, which at first seems good. But its left that Galleass completely exposed. Not only will I be able to hit it from land, but my finishing caravel won't take any retaliation damage. This is dangerous, the galleases offered a lot of flexibility to support the land, and without the threat of it, my remaining caravels are going to dominate that open ocean.

Spoiler :

As expected, the AI is crushed, the meat line was removed, and with that my musketmen have no fear of the cannons, so they begin the assault. It also brought another galleases near the cannons, which was an immediate snack for me.

The mistake was both in pushing with what melee it had, and not bringing additional tercio support. Skirmishers would be great once the beachhead is established, but they aren't going to tank well for your cannons.

So decent try AI, better luck next time.

 

[BiteInTheMark]

You can always annex and starve. Buildings already in place still give all their yields and at 1 pop you only get 1 local unhappiness, as opposed to a useless growing puppet that gives lots of unhappiness, almost no yields, sub-optimal build orders, terrible tile management and requires courthouse investment.

I disagree. Annexing the city and starving it down to 1 might cause even more unhappiness than staying with it as puppet. Every additional city rise the empire modificator for ALL other cities, and unless you have only few cities, this annexiation will cause much more unhappiness in your other cities than the new city is causing.
Some versions ago, it was nearly impossible to go above 20 controlled/annexed cities on a standard map cause the empire modificator was causing exponential unhappiness in all you cities...

There has to be a reason not to puppet, or else why have any other option?

If you pass a certain point of empire size, you have no choice, either you raze the city or stay with it as puppet. Anexxing would kill otherwise your empire.

 

[Voremonger]

So the way I read this, you are suggesting that every building generates (effectively) 1 culture, 1 science, and 3 gold, and costs 2 building maintenance....or at least in that ratio. Is that correct?

To quantify how good or bad something is I'm abstracting the benefits you get from a City as "utility".
A positive utility value is beneficial, a negative utility value is harmful.
The larger the utility value is, the better something is considered, the lower the utility value is the worse something is considered.
It's a metric to compare different situations and to determine which would give you the "best" results.

For example, if we were talking about military tactics:
We could give each enemy unit killed some amount of positive utility while giving each of our own units lost some negative utility.
By comparing the total utility value that results from a tactic we could then sort the tactics and determine how "good" they are.
By picking the tactic with the highest utility (the "best" tactic) we would pick a tactic where few of our own units die while a lot of the enemy units die.
Typically this would be an actually good tactic that would get us closer to our ultimate goal of winning the game.

Gold, Science, and Culture output of a City are all equally useful) - I am going to challenge that assumption, as gold is very commonly seen as less important than the other yields. For common "napkin math" that the community has done in the past, Production = Science/Culture, and gold is generally half as valuable as prod. So I think a better assumption is 1 Culture = 1 Science = 2 Gold.

There seems to be a misunderstanding.
I am not saying that 1 Gold has the same utility as 1 Science or 1 Culture.
What I am saying is that for my rough calculation I am assuming that the total production of Gold, Science, and Culture in a City are about equally useful to a player (thus providing the same utility).

For example:
If a specific City were to produce 40 Gold, 20 Science, and 10 Culture per turn it would still be assigned only a utility of 1 for each of these totals.
This means that each Gold would have 1/40 = 0.025 utility, each Science would have 1/20 = 0.05 utility, and each Culture would have 1/10=0.1 utility.

Think of it this way: if the total production of Gold, Science, and Culture of a City each have a utility of 1 then losing half of your Gold/Science/Culture production would each have a utility of -0.5.
This means that in the framework I described each would be considered equally "bad".

Building maintenance is equal to 2/3 of raw Gold production - So I'm not sure where the basis for this assumption lies, have you studied building maintenance compared to its average yield generation to make this conclusion?

It's a rough estimate based on my personal experience.
Of course this parameter is not constant and depends on tech level, Policy choices, Civilization, etc.
I might be biased in selecting this parameter because I typically go Authority -> Fealty -> Imperialism -> Autocracy where the Gold production per City is relatively low.

 

[BiteInTheMark]

Because it's supposed to imperfect. It's a puppet. If it made perfect choices then the whole "lack of control" aspect is pointless.

Nobody wants an "always best choice" city mechanic. But a puppet working a specialist, which is generating 1 or 2 science but eats 6-8 food and create 1 unhappiness is irrational, not imperfect.
By the way, if I remember correctly, the puppet governor building logic is the same as the normal AI logic. Would you want, that the AI would do such bad decisions?

I am for a very simple solution.
BAN SPECIALISTS FOR PUPPETS.
It is simple, it is effective, its understandable.

 

[BiteInTheMark]

Simple question, when was the jungle/forest pantheon changed? Maybe I have missed it but I noticed it now, that you are now able to get 3 yields, very valuable yields, simply for working 2 forest/jungle tiles, which seems a bit overpowered to me.
Comparable pantheons like God-King and especially Ancestor look in my eyes extremly weak against it.

 

[Stalker0]

To quantify how good or bad something is I'm abstracting the benefits you get from a City as "utility".
A positive utility value is beneficial, a negative utility value is harmful.
The larger the utility value is, the better something is considered, the lower the utility value is the worse something is considered.
It's a metric to compare different situations and to determine which would give you the "best" results.

Ok, I am following your metric now. So at the end of the day, if we boil your math down, you are saying that a normal city generates:
5 units of "Benefit" for every
2 units of "Cost".

Ok I understand now. However, I don't feel that you have "validated" this ratio at any point of your analysis. This is critical, because the entire conclusion rests on the accuracy of this ratio.

For example, if I assume a 5 to 1 ratio. That means a regular puppet is straight up 0, and then a martial puppet would have additional utility, a very different conclusion than the one you are making.

So I feel that in order to utilize this analysis for decision making, you would need to prove more evidence that the 5/2 ratio is accurate.

 

[Stalker0]

I am for a very simple solution.
BAN SPECIALISTS FOR PUPPETS.
It is simple, it is effective, its understandable.

I might argue that banning specialists may in fact making Puppets even weaker.

The theoretical benefit of Puppets is that any Culture, Science, and Tourism they make is "free". You aren't paying the extra X% for the new city, so any of the prementioned yields are "pure gravy". You also have the secondary benefit of holding land that the enemy can't take (this is my main issue with razing, the AI can replant and regrow a city so quickly that it often feels like razing doesn't do much, but I digress).

Specialists are the main way you generate culture and science. Take those away, and what benefit is the puppet actually providing?

 

[BiteInTheMark]

Specialists are the main way you generate culture and science.

Thats actually not true. Especially not in any discussion about puppets, cause specialists lose their GPP generation (in puppets), which is the only part which might rescue your statement.
Ive done the math in 3 of my games from late midgame to endgame, and any "population based" yields (citizen working tiles, specialists yields, yields from buildings based on city size) diminish more and more. They always ended at around 40% of my total empire yield generation. Food and hammers percentage was higher, culture and science a little bit lower. Another 35+% was coming from instant yields (from policies, buildings, GP,....), the rest from buildings, trade routes, UA, religioin, etc....

The only part where flat yields from specialists might have a weight is the early mid game, when writers, scientists and artists come online. But other yields are overshadowing those the more the game progresses.
Maybe you are able to do some math in one of your games too. The sheer amount of instant yields from all corners are simply overwhelming the global yield generation.
Its a bit sad, atleast for me, cause this creates some distance from the core of the game, city management and working tiles.

 

[InkAxis]

I might argue that banning specialists may in fact making Puppets even weaker.

The theoretical benefit of Puppets is that any Culture, Science, and Tourism they make is "free". You aren't paying the extra X% for the new city, so any of the prementioned yields are "pure gravy". You also have the secondary benefit of holding land that the enemy can't take (this is my main issue with razing, the AI can replant and regrow a city so quickly that it often feels like razing doesn't do much, but I digress).

Specialists are the main way you generate culture and science. Take those away, and what benefit is the puppet actually providing?

I've never had a game where specialists were the majority of my culture. And again, that culture/science is extremely small, and at the cost of happiness and growth. I don't care if they generate 1000 culture if it costs 1000 happiness and 2000 food.

 

[Voremonger]

Ok, I am following your metric now. So at the end of the day, if we boil your math down, you are saying that a normal city generates:
5 units of "Benefit" for every
2 units of "Cost".

Ok I understand now. However, I don't feel that you have "validated" this ratio at any point of your analysis. This is critical, because the entire conclusion rests on the accuracy of this ratio.

For example, if I assume a 5 to 1 ratio. That means a regular puppet is straight up 0, and then a martial puppet would have additional utility, a very different conclusion than the one you are making.

So I feel that in order to utilize this analysis for decision making, you would need to prove more evidence that the 5/2 ratio is accurate.

Despite how it may seem I am not dogmatic about my views; I more or less spontaneously wrote down what I thought without spending too much time on the details (I had already made the annexation vs puppeting figure for myself).
In my original post I probably overvalued Gold.
Looking back giving Science and Culture a utility of 3 would probably be more accurate.
If nothing else is changed that would mean that regular Cities are valued 3 + 3 + 3 - 2 = 6, Puppets without Martial Law are valued 0.6 + 0.6 + 0.6 - 2 = -0.2, and Puppets with Martial Law are valued 1.2 + 1.2 + 1.2 - 2 = 1.6.
If you want that can be interpreted as Puppets having basically no utility in terms of Yields without Martial Law and some utility with Marital Law.
However, the ultimate factor in deciding whether a City should be puppeted or annexed is still opportunity cost: what really matters is which option is better than the other.
Especially with how Factories currently work the build order makes a huge difference to Yields and I would argue that for that reason alone Puppets kind of blow.

In any case, it is important to note that the model I proposed assumes that the utility of any Yield scales linearly; 2 Gold is twice as good as 1 Gold, 6 Science is 3 times as good as 2 Science, and so on.
In general, the utility from Gold in particular does not scale linearly; most simple models for complex systems have this problem.

Concerning Gold:
The most important part about Gold is that you are net positive; once you have reached that amount getting any more Gold is less useful.
In other words: losing Gold is potentially much more dangerous than gaining Gold.
Notable even with Martial Law Puppets are a drain on GPT.
This means that having a large percentage of Puppets will make your GPT negative which I would argue hurts you overall.
There is definitely some headroom where you can trade some GPT for some Culture/Science but I think this headroom is relatively small
Especially in the context of wide, militaristic play you have large expenses for Units and tactical Roads while your income from Trade Routes stays constant or even declines due to a shortage in trading partners.
I would also argue that low GPT exposes you to catastrophic risks: if you have low GPT and things take a turn for the worse you are more likely to become GPT negative; I personally would not be comfortable with more than ~15% Puppets.

 

[tu_79]

Why? Shouldn't I be able to play the kind of game I enjoy?

Sure. If you enjoy playing a tundra map by all means do it, but don't expect it to be balanced. All efforts go into balancing the standard map, with such player density. If you play on lesser density, then you are giving an edge to civs that are good at expanding, making them better than they usually are. Maybe the difference is not big enough to be an issue, but it's there. You don't need to play a completely balanced game, as long as you know and accept the terms.

 

[Stalker0]

Despite how it may seem I am not dogmatic about my views; I more or less spontaneously wrote down what I thought without spending too much time on the details (I had already made the annexation vs puppeting figure for myself).
In my original post I probably overvalued Gold.
Looking back giving Science and Culture a utility of 3 would probably be more accurate.
If nothing else is changed that would mean that regular Cities are valued 3 + 3 + 3 - 2 = 6, Puppets without Martial Law are valued 0.6 + 0.6 + 0.6 - 2 = -0.2, and Puppets with Martial Law are valued 1.2 + 1.2 + 1.2 - 2 = 1.6.

I did not take your analysis for "dogmatic", I'm just attempting a peer review. In order for number based decision making to be useful, it has to stand up to scrutiny...and I would expect no less from the various analysis I have posted over the years.

Ultimately, I give the same critique to your new value as I did to the old one....it at the moment seems completely arbitrary. 3, 5, 10...I don't see any backing from where these numbers are coming from other than "feel". And its okay to say that you "feel" that puppets are weak, a lot of our changes are made that way if its a common consensus. But if your trying to use math to back up an answer, it needs to come from somewhere concrete, and I don't see where the utility number's value is really coming from.

 

[arthurF]

I thought this was worth highlighting. Definite improvements to the amphibious assault portion of tactical AI.

Spoiler :

View attachment 567923

1) Good use of the island up north to weaken any naval forces I bring to bear.
2) Nice use of melee on the west wing to protect the land cannon.
3) That knight on the right is a bad move, I have too much ranged firepower and I'll mow it down quickly. If he had brought his brother knight from the north at the same time it might have had some effectiveness, but realistically there should not be any incursions around the city until he's softened me up more, that trio of xbows + city will basically take down anything that lands in the immediate vicinity.
4) I do think this invasion needs a bit more troops to start. I think that's the biggest aspect of amphibious that took me a while to learn, and the AI needs to learn it to. Unless you have a beachhead to work from (like a friendly CS)...if your going to land, you have to land BIG. As soon as you land several of your units are going up in smoke, so you have to land on enough beach to spread out the defender's damage, and ensure you are going to kill units on your first strike. If you can't do that, the defender will just chew you up and spit you out.

Edit: And now a few turns later

Spoiler :

View attachment 567925

The AI did land his second knight on the east side, and as expected I blew those 2 away. However, it does seem to be focusing forces on the left side where I am weakest, a good sign. He had a trio of galleassas protecting this fleet but I took one out, if he doesn't bring some more naval support soon I may take out his remaining fleet, so we will see.

However, he is definitely bringing a bit more power to bear which is great to see.

Edit: And the next turn

Spoiler :

View attachment 567926

Excellent formation here. The AI did push its troops to the southwest. The trio of cannons in the back are a nice fortified line, with knights to provide frontline meat.

The danger here are my caravels, ideally the AI once to take this out to control the sea and allow for continuing reinforcements. It should be able to take out my northmost caravel with cannon fire pretty easily (that one killed a key caravel so I am willing to trade it). Then we will see what it does with its Galleasass, it should pull them back away from my caravels to allow his land forces to deal with my ships...otherwise I may be able to snipe them, and retake control of the sea.

I am doing a strong buildup of musketmen now, so we will see what the AI does...

EDIT: Next Turn!

Spoiler :

View attachment 567928

Ok, here the AI falters a bit. First, its pushing its knights out aggressively instead of keeping them close to the cannons. Effectively the AI needs to change its mentality to a "siege posture". Its got 2 cannons on a weak coastal city, with a 3rd ready to take over if one gets wounded, this is a solid position, it now needs to hold this. Instead its squadering its melee protection with pushes. Those knights are dead, and its going to cost the AI.

On the naval side, the AI snagged 2 of my caravels, which at first seems good. But its left that Galleass completely exposed. Not only will I be able to hit it from land, but my finishing caravel won't take any retaliation damage. This is dangerous, the galleases offered a lot of flexibility to support the land, and without the threat of it, my remaining caravels are going to dominate that open ocean.

Spoiler :

View attachment 567929

As expected, the AI is crushed, the meat line was removed, and with that my musketmen have no fear of the cannons, so they begin the assault. It also brought another galleases near the cannons, which was an immediate snack for me.

The mistake was both in pushing with what melee it had, and not bringing additional tercio support. Skirmishers would be great once the beachhead is established, but they aren't going to tank well for your cannons.

So decent try AI, better luck next time.

Your tech advantage, i.e. muskets and tercio vs knights, is not conducive to a good landing. Additionally the AI did not come with overwhelming force. I'm not seeing what you thought was good about this assault.