ICS best of the best!

[alpaca]

I disagree, I feel like it's already a big problem with the AI on immortal/deity. One time, I fought a war where it took me more than 100 turns to grind through Monty's gigantic army, so that I could finally start taking his cities. Anyway I think everyone agrees that scientists are by far the most powerful specialist- not only from their high base yield, but also because the great scientists are so powerful.

But that has nothing to do with research speed. It will take a hundred turns no matter the tech level, it's just a function of the number of units. If the game takes longer, the number of units doesn't significantly increase anymore for the aggro AIs because they will always have tons of them.

 

[Valkrionn]

my 2 cents.

To underpowered ICS (*) and make viable other strategy easily, i really think that the mathematical function used to link science and population should be a powerfunction (^Y).
The value for each technology in the tech tree should be scaled in consequence (the hard part of the thing -i think this scale should be somewhere link to the number of turns...-).

For exemple :
5 pop city gives 25 sciences.
10 pop city gives 100 sciences.
25 pop city gives 625 sciences.

I mean if ICS use only maritime states to boost the population, strategies which use farms and maritimes states will have more population... then more science. It will create a balanced choice between go for total money/production or go for total science/culture. I guess ICS is overpowered because it's the best way the have money/production... AND science. It's too much. Only culture is crushed in ICS. Science should, too.

In my example, two 25 pop's cities will produce a bit more sciences than twelve 10 pop's cities. (**). But, the factor ^Y can be chosen correctly if it seems to be too much... or not enough.
The power function here can be a good way to powered the verticality.

(*) : I just want to underpowered ICS (or make it less automatic).
(**) : We can notice than this behaviour happens in reality : if scientist A produces a total of research equals to W1 and a scientists B a total equals to W2, then these two scientists, by working in team, will produce a total of research greater than W1+W2 (in reality it's true not only for sciences, but for workers and marchands, too)

Note : In this design super-scientist should give a X% boost and not an absolute value.

This is something I plan to work on, actually.

 

[LordTC]

The big issue with a pop-size science model, is that farmers generate more science than science specialists over the long run if you use an N^2 pop model.

I think what you actually want is a floored linear model with a base penalty and a high constant. At a guess the first thing I'd try would be max(0, 2n - 5) where n is the city population. Size 1 cities make no science, size 2 cities make no science, size 3 cities make 1 science, etc..

This way every time you decide to expand early you do so at the expense of technological growth, and cities don't support gaining technology until they reach a sufficient size (in the example 3).

This enables you to nerf growth in a very real way, while still keeping value to science specialists. If growing from 10 to 11 results in going from 100 to 121 science per turn from population you just never make science specialists at all, but would instead go crazy with farms for science growth. Also the libraries and such giving linear bonuses would just never get built at all (your +50% bonus to base science instead becomes increasing science from 100 to 105 in a size 10 city, which is such a trickle its not worth the upkeep). I mean if you want to make a mod where the population based science amount is based on the square of the population I can pretty much guarantee the optimal strategy will be to expand your first city as large as possible and only expand outward when it doesn't stall your science growth (i.e. when your number of turns for growth is low enough and your luxury resources are high enough to settle more). Pretty much you'd want to always avoid hitting the growth penalties and grow horizontally as much as possible. OCC might even be easier in such a mod because it would prevent someone from expanding when they shouldn't, and India's civ bonus would be completely and totally unfair, they'd easily be the best civ in the game with even quadratic pop to science ratio.

 

[Valkrionn]

The big issue with a pop-size science model, is that farmers generate more science than science specialists over the long run if you use an N^2 pop model.

I think what you actually want is a floored linear model with a base penalty and a high constant. At a guess the first thing I'd try would be max(0, 2n - 5) where n is the city population. Size 1 cities make no science, size 2 cities make no science, size 3 cities make 1 science, etc..

This way every time you decide to expand early you do so at the expense of technological growth, and cities don't support gaining technology until they reach a sufficient size (in the example 3).

This enables you to nerf growth in a very real way, while still keeping value to science specialists. If growing from 10 to 11 results in going from 100 to 121 science per turn from population you just never make science specialists at all, but would instead go crazy with farms for science growth. Also the libraries and such giving linear bonuses would just never get built at all (your +50% bonus to base science instead becomes increasing science from 100 to 105 in a size 10 city, which is such a trickle its not worth the upkeep). I mean if you want to make a mod where the population based science amount is based on the square of the population I can pretty much guarantee the optimal strategy will be to expand your first city as large as possible and only expand outward when it doesn't stall your science growth (i.e. when your number of turns for growth is low enough and your luxury resources are high enough to settle more). Pretty much you'd want to always avoid hitting the growth penalties and grow horizontally as much as possible. OCC might even be easier in such a mod because it would prevent someone from expanding when they shouldn't, and India's civ bonus would be completely and totally unfair, they'd easily be the best civ in the game with even quadratic pop to science ratio.

The formula I intend to use at least as a base is (1+Pop/15)^(1+Pop/20). It tracks fairly close to the current model (only double science at pop 10), but increases quickly after that, resulting in 100science at pop 20; That level of population is difficult to reach, however.

Basically, my intention is to find a formula that would track closely to the current setup, but increase quickly above a certain pop level... Benefiting large cities disproportionately to small ones.

An excel sheet showing science for the first 20 pop: https://spreadsheets.google.com/ccc...WN0tKX1B2WHFSRUE&authkey=CN-_s4wE&hl=en#gid=0

 

[Paeanblack]

The formula I intend to use at least as a base is (1+Pop/15)^(1+Pop/20). It tracks fairly close to the current model (only double science at pop 10), but increases quickly after that, resulting in 100science at pop 20; That level of population is difficult to reach, however

Be careful with exponential rewards. According to that spreadsheet, this city would be earning about 9,000,000 beakers per turn.

http://forums.civfanatics.com/showpost.php?p=9926039&postcount=27

 

[Valkrionn]

Hah. I wonder how much of that was from maritime city states?

There are things I'd like to do to limit that, but I can't touch city state effects, not really. Unfortunately.

 

[Dale]

Hah. I wonder how much of that was from maritime city states?

There are things I'd like to do to limit that, but I can't touch city state effects, not really. Unfortunately.

You can to some extent, but not as much as we'd like to.

 

[blind biker]

Be careful with exponential rewards. According to that spreadsheet, this city would be earning about 9,000,000 beakers per turn.

http://forums.civfanatics.com/showpost.php?p=9926039&postcount=27

Dude... WTH is that? I haven't played Civ V in more than 3 weeks, now (not anything serious, just checking if it crashes after the patches, that's all), so I am not sure anymore: does Civ V allow for such monstrous megalopolis?

 

[LordTC]

The big issue with a pop-size science model, is that farmers generate more science than science specialists over the long run if you use an N^2 pop model.

I think what you actually want is a floored linear model with a base penalty and a high constant. At a guess the first thing I'd try would be max(0, 2n - 5) where n is the city population. Size 1 cities make no science, size 2 cities make no science, size 3 cities make 1 science, etc..

This way every time you decide to expand early you do so at the expense of technological growth, and cities don't support gaining technology until they reach a sufficient size (in the example 3).

This enables you to nerf growth in a very real way, while still keeping value to science specialists. If growing from 10 to 11 results in going from 100 to 121 science per turn from population you just never make science specialists at all, but would instead go crazy with farms for science growth. Also the libraries and such giving linear bonuses would just never get built at all (your +50% bonus to base science instead becomes increasing science from 100 to 105 in a size 10 city, which is such a trickle its not worth the upkeep). I mean if you want to make a mod where the population based science amount is based on the square of the population I can pretty much guarantee the optimal strategy will be to expand your first city as large as possible and only expand outward when it doesn't stall your science growth (i.e. when your number of turns for growth is low enough and your luxury resources are high enough to settle more). Pretty much you'd want to always avoid hitting the growth penalties and grow horizontally as much as possible. OCC might even be easier in such a mod because it would prevent someone from expanding when they shouldn't, and India's civ bonus would be completely and totally unfair, they'd easily be the best civ in the game with even quadratic pop to science ratio.

The suggestion I made has issues in that it makes early warrior rush far more favorable compared to any sort of settler building because there is a lot of difficulty in stopping food in your biggest city to expand. Basically blitzing a neighbor and using their city as a settler farm while you grow your main city becomes a clear favorite as a strategy. Also the slowed down early tech rate makes it harder to defend against warrior rushes because spears, archers, horsemen and swordsmen will all come way too late even in situations where the warriors have to travel longer than typical. I'm really not sure if this problem is solvable though, it might need a lot of tweaking dealing with some low-order polynomial and being very careful about constants.

Basically here are the cautions noticed so far:

1. Anything exponential goes crazy with very large cities. This also applies to even quadratics like n^2 if not very careful about small constants. (Even then, 90^2= 8100 points and unless you're dividing by a large constant that's out of control, and if you need to divide by say 100 than this component isn't significant as it only produces 4 science at size 20, so most of the early science will still be in the linear component).

2. Anything with a negative constant can makes early rush strategies, particularly warrior rush too strong.

3. Anything with a strong emphasis on the linear component or with positive constants rewards ICS.

 

[Valkrionn]

You can to some extent, but not as much as we'd like to.

Well, you can tweak the amounts. You cannot, however, make maritime effects only apply to a handful of cities.

 

[alpaca]

Well, you can tweak the amounts. You cannot, however, make maritime effects only apply to a handful of cities.

Actually you could code that in Lua, I think. You can modify city base yields and you can check city state alliances. Alternatively, you could add a pop point to a random city every n turns in analogy to how military CS work.

 

[Valkrionn]

You'd have to completely remove the maritime trait in the xml, and then do the entire thing via lua.

So yes, you could, but not by playing with the existing system.

 

[alpaca]

You'd have to completely remove the maritime trait in the xml, and then do the entire thing via lua.

So yes, you could, but not by playing with the existing system.

Sure but you said you "can't touch them", which isn't true

It wouldn't be a vastly complex script, either. Depends on how complex you want to make the system, really.

 

[Dale]

I'm with alpaca on this one. You're wrong this time valk.

You'd have to hack the system like we are doing with happiness. But it's certainly doable.

Now to implement Bureaucracy unhappiness.

 

[alpaca]

I'm with alpaca on this one. You're wrong this time valk.

You'd have to hack the system like we are doing with happiness. But it's certainly doable.

Now to implement Bureaucracy unhappiness.

By the way, how's that happiness mod coming along, I'd like to "borrow" your code so I don't have to write it myself

More seriously, how did you code it? Just used Player.ChangeHappinessFromBuildings?

 

[Valkrionn]

I'm with alpaca on this one. You're wrong this time valk.

You'd have to hack the system like we are doing with happiness. But it's certainly doable.

Now to implement Bureaucracy unhappiness.

Hey hey, I admitted you could.

I just meant that you cannot do anything without discarding the existing mechanic. There are other things I personally would rather focus on (Dark Ages, atm ) than rewriting the Maritime bonus in lua; In any case, the issues with that system are VERY well known, there's no need (IMO) to test any changes there. Just argue for their implementation.

Also: Once Dale (finally) releases his ICS mod, I'll make my own happiness tweaks available as well. Changes the amount of happiness you receive from resources; +3 from each type of resource, and +0.5 from any extra instances of that resource. This means with 1 Gold, you gain 3... With 2 Gold, you gain 3.5 (3)... But with 3 Gold, you're up to 4. The exact numbers can all be tweaked with xml/sql.

Happiness apparently IS tracked decimally, and is rounded, but .5 is rounded down; .75 is rounded up.

 

[Dale]

By the way, how's that happiness mod coming along, I'd like to "borrow" your code so I don't have to write it myself

More seriously, how did you code it? Just used Player.ChangeHappinessFromBuildings?

The mod's coming along, just one last thing to implement and it'll be ready to test. So hopefully tonight (my time) I'll try to get v1 on this thread.

As to how, I tapped into TopPanel:OnUpdate() and recalculate happiness and then just use the Player:SetHappiness() dll link.

 

[boehj]

I'm looking forward to giving this mod a go. Without delving into fancy stuff like mathematics or logic to support my feeling, nonetheless I do have the feeling I'll still be able ICS in a disgusting way.

The proof will be in the pudding as they say.

Thanks for putting the time into this Dale.

 

[Dale]

No probs.

 

[Dale]

Okay, the mod is available here. Note it's alpha, and the demographics don't reflect the new happiness system. Rely on the tooltip instead for guidance.

Please let me know your results so I can tinker with this mod to get it right.