Armies

[Txurce]

I don't think I quite understand this sorry. They can always construct more happiness buildings, or research the tech for the next tier happiness building, or trade for more luxuries through diplomacy or via city state (unless they have them all), or stop growing their cities so much, or raze some cities, or sell them, or focus their social policies on happiness ones.

Yes, eventually at the top of the tech-tree and once you have stadiums in every city there is a cap, but I still don't see this as a problem.

Let me try to be more specific, and correct me if I'm wrong.

If you can't order a happiness build in a puppet, but have lots of puppets, then a serious or increasing happiness hit could limit growth even after trading for luxuries. This would be my theoretical concern in a fix focused solely on happiness. Obviously this won't happen if a more moderate happiness hit is combined with a general puppet yield nerf. (And I could be wrong about my basic premise. Keep in mind that I am speaking as someone who has probably never had more than 4 puppets in Civ 5!)

 

[Ahriman]

If you can't order a happiness build in a puppet, but have lots of puppets, then a serious or increasing happiness hit could limit growth even after trading for luxuries. This would be my theoretical concern in a fix focused solely on happiness.

To me, this is a feature, not a bug.
If you're on a conquering rampage and you puppet everything, then you *should* hit a point where you can't keep expanding without driving yourself into unhappiness.

Slow down your conquests, make peace, let the puppets build some happiness structures by themselves (which they'll tend to prioritize if you're unhappy), or start annexing, courthousing and building/buying colosseums and such.

To me, this is the whole point of using happiness as an expansion limiter. It is exactly by design that you can't keep conquering without driving your economy and growth into the ground thanks to negative happiness.
[I'd also consider having very negative unhappiness add penalties to culture and science, if needed (not penalties to production or gold, or it could be too hard to dig yourself out).]

If this *didn't* happen, then there would be no check on expansion.

I don't think we need a specific happiness nerf for puppets, and I'm not proposing one. The only *extra* happiness hits I'm proposing are for while razing, and maybe for the few turns during disorder.
The vanilla game already uses happiness to check expansion, since when you puppet a new city you get
Puppets are slightly worse for for happiness in my proposal because each citizen still gives a full unhappiness point, but each puppet citizen is less productive, due to yield nerfs. So puppets are less efficient in happiness terms, but its an indirect effect, through the yield nerfs, not a direct extra unhappiness.

 

[alpaca]

Ahriman makes a good point (though it wasn't exactly his point): I prefer a science hit to a production hit for unhappiness. -50% production is very severe and directly influences your ability to get out of the unhappiness trap, making for a high frustration if you get there. What I would like even better is the ability to create some happiness for a while by spending money on festivals and such to absorb some happiness hits until you can build up the necessary infrastructure, which should obviously be much more costly than buildings and capped. For Gandhi, the extreme example, an annexed city yields 10 unhappiness which makes it almost impossible to annex instead of puppet (well you can buy the courthouses with the last patch but that wasn't possible before)

 

[Thalassicus]

I have two aims: The player should only keep veteran units around, and the extremely powerful "upgrade all your units the turn you research a new techs" strategy should be nerfed.

I agree, I've also been experimenting with ways to balance upgrading vs training, and looking at your mod was a help in this regard.

It was difficult to follow the conversation about puppet states because it came up simultaneously on several different threads and bounced around a bit. Scripting buildings to add/remove to puppets for modifying them is something I've been exploring though, and just finished today, so we're thinking along the same lines.

Culture generation and border expansion of puppets is something I haven't tested yet because I don't know how it'll interact with the policy, wonder, and UB that also affect border expansion.

My goal with conquest is balancing things so a player can go for technological quantity or quality, but not easily both. This is how past versions of civilization worked and I think it made a lot of sense... if you can get a big and highly advanced army, what's the downside? CiV currently lets us get both 1) a super-big empire with a large army, and 2) huge tech lead.


@antdog
I've avoided buffing muskets mostly because of their unique status among leaders, actually. More UUs are based off the Musketman than any other unit, so buffing muskets can cause problems elsewhere.

If you (general, untargeted "you")

I often use "we" in this situation... shows comments aren't personal.


@alpaca
I agree... I find it odd the 'very unhappy' state lets us still research like crazy yet not build anything to get out of that state. It makes more sense to have the opposite... an empire can more easily force people to build than to force people to think (well unless you're in an Orwellian dystopia ).

 

[Ahriman]

hat I would like even better is the ability to create some happiness for a while by spending money on festivals and such to absorb some happiness hits until you can build up the necessary infrastructure

Gold-buying the buildings is good enough I think, or gold-buying a luxury through diplomacy or city state.

Border expansion is something I haven't tested yet because I don't know how it'll interact with things like the policy and wonder. (Like you said, the solution is to have something like -50% border expansion costs with -50% culture generation... but I don't know how this'll combine with similar modifiers.)

I don't think its a problem if puppets have reduced border expansion, as long as they still have some.
-25% science, gold, -50% culture (as Alpaca suggested) would be fine, even if this meant the puppets expanded borders half rate.
My one concern would be; would this then prioritize the puppet AI towards production boosters (since these aren't nerfed, so are relatively more powerful) or towards culture boosters (since culture is very low, so the AI needs to focus on that more to bring up cultural generation to an "acceptable" level)?

 

[Thalassicus]

AI priorities are called "flavor" settings by Firaxis in the files, and unless we manually change them the AI doesn't alter its behavior. Currently puppets have a +50 gold flavor and -100 military flavor.

 

[alpaca]

Culture generation and border expansion of puppets is something I haven't tested yet because I don't know how it'll interact with the policy, wonder, and UB that also affect border expansion.

The border expansion modifier currently only works on a global level so don't bother using the xml value, you're like to break things (and yes, this means the Krepost is bugged).

What you can do is use Lua. The SetJONSCultureStored value does not affect social policies. So you can add a building that provides -100% culture to remove everything, then loop through the city's buildings and add the culture via Lua. Alternatively, you could use the City:SetBuildingYieldChange function to change the yield of the building class to 0 locally. Both have advantages and disadvantages: The -100% modifier can be overridden by the broadcast tower and other modifiers, while the yield change would leave specialists out of the equation.

 

[Ahriman]

AI priorities are called "flavor" settings by Firaxis in the files, and unless we manually change them the AI doesn't alter its behavior. Currently puppets have a +50 gold flavor and -100 military flavor.

I think its more complex than that.
I think the AI is still doing cost-benefit analysis.
It will build structures that it calculates as having high net benefits, either because the structure has high yields (eg: hypothetical buildnig that gives +20 culture), low costs (structure that only costs 2 hammers and has no maintenance) or fulfill a particular need (happiness when the civ is in negative happiness, food when city is starving or barely growing, etc).

The flavors are *adjustments* to the cost-benefit analysis, they don't replace it entirely.

Even if you set all the flavors to zero, the AI wouldn't just build things randomly.

I'm just not sure exactly what it counts as "fulfilling a particular need", and whether it calculates high yields in a gross sense (ie building gives +X hammers) or a net sense (ie building gives +X hammers combined with all the other modifiers currently affecting this city).

 

[alpaca]

I think its more complex than that.
I think the AI is still doing cost-benefit analysis.
It will build structures that it calculates as having high net benefits, either because the structure has high yields (eg: hypothetical buildnig that gives +20 culture), low costs (structure that only costs 2 hammers and has no maintenance) or fulfill a particular need (happiness when the civ is in negative happiness, food when city is starving or barely growing, etc).

The flavors are *adjustments* to the cost-benefit analysis, they don't replace it entirely.

Even if you set all the flavors to zero, the AI wouldn't just build things randomly.

I'm just not sure exactly what it counts as "fulfilling a particular need", and whether it calculates high yields in a gross sense (ie building gives +X hammers) or a net sense (ie building gives +X hammers combined with all the other modifiers currently affecting this city).

I think so, too. Puppets definitely build more colosseums if you have bad happiness.

How the algorithm works is probably impossible to tell without massive testing or DLL access

 

[Thalassicus]

That's why I just guesstimate for the most part.

When fiddling around with this stuff I looked at how they set up other values, and put a -10 for some values so they don't try and overcompensate. I'm not really sure if -5 or -10 will be better, needs testing.

 

[Ahriman]

When fiddling around with this stuff I looked at how they set up other flavor values, and put a -10 for culture and science so they don't try and overcompensate.

But then we risk encountering your sausage-tube problem: if they won't build science or culture buildings, they'll build production buildings and gold buildings. Do we want to risk them building a factory before a library or monument?

I've just leave it with gold flavor +50, military flavor -100, seems safest when we don't know how the AI works. Not that big a deal though, -10 probably doesn't have that big an influence.
It might be hypthotically: science flavor value -10 reduces the "value" it places on a science building by 10%; not that big a deal.

 

[Thalassicus]

Oh I don't want to prevent it entirely, just somehow balance out the reduced yields by also lowering their priorities to get up to the old values. Like say... if they used to want to have X culture and I cut culture in half, their new goal should be X/2 culture. As alpaca pointed out there's no way to know exactly how to do that... so I'm taking a guess.

Based on what I've seen of their lua code, and some comments from a developer in a press release a year ago, I think it's weights.

Options
A - 5 weight
B - 10 weight
C - 10 weight

==

becomes
A - 20%
B - 40%
C - 40%

Now, we have no way of knowing for sure if the weights convert into % odds or something else, but it's an educated guess. The advantage of a weights system is even if an item has an extraordinarily low value (like the -100 for military) it still becomes 100% if it's the only option remaining, so there's no need for special code in that circumstance. This would be important for puppets before the wealth/science processes are researched for example, the only remaining option could theoretically become unit production.



-----------------

One thing I really like in Alpaca's mod is more transparent unit maintenance. Up-front costs specific to each unit type makes more sense to me than vanilla's opaque method, and is easier to balance. Anyone have thoughts about this?

• Vanilla: All units cost the same, unit maintenance scales upwards over time from an unknown & hidden formula.
• Option: Each unit type has a fixed maintenance cost, possibly equal to /100.

 

[Polycrates]

Since this seems to be the puppet-talk thread now

Just wondering; what do people think the main problem with puppets is? Is it that each one is too powerful individually, or that you can get too many of them and their power is from the proportion of your empire they can represent?
I’m not sure what I think, but I’m drifting towards thinking that if you (for example) only puppet one or two cities, the balance of puppet vs annex maybe isn’t too bad. As a reward for your successful warmongering labours, you’ve got a choice of a fully productive city that you haven’t had to develop yourself, with all the positives and negatives that entails; compared to a rather crippled city that will never develop nearly as well as a full city (and mostly just be useful for gold) but fewer of the expansion penalties.
If you’ve only got a few puppets, I think this is fine. They stay small and don’t focus science, so their science output is going to be relatively minor compared to big, focused science supercities (which can have such huge stacked bonuses if developed well that they’re easily worth 10 or more other cities for science output). The main thing they produce well is gold, and even that they’ll never quite do as well as a full city. The science output, to my mind, really only becomes problematic when you start getting lots of them. I’m interested to try it and I wonder if nerfing puppet outputs will actually reduce the number of cities people puppet (vs either annex or just not bother taking at all); or if it will just make those same number of puppets just less effective overall. I’m not even sure which would be the more desirable outcome, to be honest.
What I do like is that by reducing the gold and science output of puppets, you help mitigate the current bias towards puppeting good gold cities and annexing good production cities (rather than annexing the best cities or the ones more contiguous with your country proper).

But I don’t know. My inclination is that the EU3 (and particularly the ruthless Magna Mundi mod) model had it right with the idea that a big empire ought to be hard to manage, and that there ought to be a limit both to how fast you can expand by conquest, but also a meaningful limit on how much you can conquer and hold before the empire starts falling apart under its own weight. So the value of each additional conquest decreases when balanced against its increasing negatives, to the point where conquering beyond a certain point just isn’t worth it. But this is a bit of a departure from the general vanilla civ design, and I suspect would need a proper concept of explicit vassals (a la Civ IV, for example) to really play smoothly (and not just result in mass razings, for example). So hence why I suggested the spiralling unhappiness costs for additional puppets; I do like it as a mechanic but I suspect maybe that’s something that belongs more in a mod to change things to personal taste rather than an overall balance mod.

Anyway that’s just some rambly thoughts, the proof will be in the testing of the system you’ve implemented, which I suspect will be a massive improvement. Thanks again for all your excellent work on this, Thalassicus!

EDIT: Regarding unit maintenance, I really like the idea in Alpaca's mod of explicit maintenance for each unit type (the vanilla inflationary model really doesn't work so well, and it balances terribly); there's just a couple of things I'd flag as potential hazards - the first being that it may promote having more units around the place as obsolete units are kept rather than scrapped, which might cause some congestion issues with 1UPT; the second being the rather cheesy tactic of keeping your army outdated to save on maintenance and mass-upgrading at wartime. The second is probably dealt with just fine by the increased upgrade costs you've implemented; the first I'm not so sure about.
I'd love to give it a try though!

 

[Thalassicus]

Those are some good questions.

My first thought about the maintenance concerns is units mostly cost 1/turn until about halfway through the game, so it'll only really have an impact on the late game, which matters less anyway.

Part of the reason alpaca's maintenance method appeals to me is I think individual control over unit maintenance costs might make it easier to manage balance of warfare economies. For example, the cost increases with maintenance based on /100 take effect at the start of the industrial revolution, when most trading posts get a buff - this is why I decided to try it out. It might also be a way to control the value of individual units... could make all ships cost less maintenance for example.

 

[Polycrates]

So you're suggesting 1 gold per hundred hammers (presumably based on costs at normal speed)? That certainly seems like a good starting point! I like that it might also promote rush-buying/"drafting" rather than always keeping a huge standing army. I'm all for trying this change!

Something that would be impacted, and could use a fix anyway, would be nukes - they can currently be rush-bought (for a couple of thousand gold - less with the right policies; which is not that huge in the late game) which somewhat makes a mockery of their resource requirement if you can e.g use 4 uranium to make 4 nukes, then use them and then (if you've stockpiled the gold) rush-buy another 4 the next turn then another 4 the turn after that. They're also powerful enough that they feel like they ought to be earned the honest way, so I wonder if they could/should have rush-buy disabled like wonders?

It would also stop late-game workers being such a drain on the economy, which is great (though I also wonder whether it wouldn't also be worth giving them an improvement speed boost at some point anyway to cut down on the late-game glut).

 

[Thalassicus]

It's very easy to disable rush-buying of something like nukes, I just set the modifier to -1 (courthouses used to have this too). It seems reasonable to do so for nukes and I'll add that.

And yes... by the very late game, military units like Modern Armor cost 7 times as much in maintenance as a worker, which should hopefully reward people who like the building aspect of the game. Even in the renaissance era workers cost half that of military units.

Edit: upon investigating this I discovered nukes actually have no rush buy penalty at all in vanilla. Even warriors cost more per than a nuke.

 

[Ahriman]

I think the problem with puppets is that they are superior to annex and courthouse, and that they allow you to expand through conquest without suffering any super-normal expansion check, either through occupied city penalty and courthouse maintenance, or efficiency cost. The problem at the moment is that a puppet *isn't* a fairly crippled city; if its in a good location, then its still a powerful, valuable city.

I like explicit unit maintenance costs as an ideal (and this lets you boost some units even further but give them high maintenance costs), but I worry about whether the AI would understand it. Maybe tat doesn't matter much though.

I think any maintenance cost high enough that it encourages rush-buy over maintaining an army is problematic, because the AI won't be able to exploit that very effectively. The AI *does* maintain a large standing army, so you're giving the human player a big advantage if the gold maintenance cost is high enough to make a difference.
But 1 gold per 100 hammer cost doesn't seem unreasonable as a starting point.

 

[Thalassicus]

My goal is to avoid significant maintenance increases and keep it in the +25%-50% range or so, similar to increases in upgrade costs, purchase costs, and so on. Basically, make all aspects of warfare a little more expensive, but not in a way that has too much of an impact on gameplay.

Since maintenance is roughly equal among units of an equal tech level (until the Modern period or so), it should hopefully have minimal impact on the AI. The exception is workers, which will cost less in proportion to other units, dunno how that will work out.

 

[Txurce]

I just started a game with dev v.18, researched IW, have no iron, but cannot build a warrior anymore. Could you adjust this?

 

[Thalassicus]

I can if there's a strong feeling about that, though we were discussing it a few pages back I think.

Warrior - 6 for 40 = 0.15 /
Spearman - 8 for 50 = 0.16 /, and +100% vs mounted units

The question is, with spears buffed to be more cost-effective than warriors all-around, is there really a reason to build warriors once we're in the classical era? That +1 buff for spears gives them significantly better combat odds than before against ancient-era units.