AU mod: Balancing the Governments

[ducki]

Quoting alexman from the MilAcad thread - warning, this is out of context, go read the original for all tidbits, but I felt responding to it belonged here.

The proposal in the government thread makes the AI unlikely to choose Fascism, so they will usually not get the Army bonus.

And now noone will get the SPHQ - this seems a big move for something we are actively trying to prevent the AI from using.

Fascism will become a non-factor in the AU mod for the AI, as it is for the human.

If so, why try to fix it? Why not really make it a non-factor? If it's so bad that the AI should never use it and we're tweaking things to try to prevent just that, why not get rid of the Government altogether?

But even if the AI does choose Fascism (because they don't know Communism yet), double Armies are a good thing! Since the AI doesn't use leaders properly, it needs as many Armies as it can get, especially if it's unfortunate enough to choose a terrible government like Fascism.

This seems like an odd way to address it as noted above. If Fascism is so bad that the AI should - really - never use it, then let's make it so that the AI never can use it instead of simultaneously giving Fascism a boost and ensuring that the AI prefers Communism.
It seems an odd way to go about it.

[alexman]

The change is so that Fascism is not entirely worthless. It can actually be worth it for small empires, or for someone who doesn't know Communism.

If you remove Fascism, it's a highly visible change from stock (tech tree graphics downloads again anyone?) that I'd rather avoid.

What you see as "odd" was my best attempt at a compromise between a) helping the AI, b) keeping the strategic choice of Fascism under certain special cases, while c) changing as little as possible. I think it is a good solution.

Nevertheless, I suppose we could split the proposal in two and see what happens:

1. Yes/No: Increase Fascism corruption level to problematic.
   Restore Feudalism unit support cost to stock level.
   All AIs favor Republic and shun Fascism.
   Remove SPHQ from Communism.
2. Yes/No: Only if 1 results in a Yes:
   Add SPHQ to Fascism.
   SPHQ spawns an Army every 20 turns.

[ducki]

Ah, I see...
I do like the split proposal better.
Thanks for clarifying.

[Cryptor]

It seems to me that there is constant debate over whether fuedalisms unit suport should be one or three? Why not just make it two? This preserves the fact that fuedalisms should not have extremely large armies, whilst allowing them to support a few more units over their support limit. I don't see the problem with unit support 2, just like republic, but for some reason this proposal is usually ignored.

[alexman]

I think the reason that proposal is not popular is because it doesn't have enough effect to justify a change from stock rules. Somewhere in this thread there is a graphical comparison for Feudalism at 1gpt, 2gpt, and 3gpt unit support.

Anyway, it's time to put the following proposal under consideration:

1. Yes/No: Increase Fascism corruption level to problematic.
   Restore Feudalism unit support cost to stock level.
   All AIs favor Republic and shun Fascism.
   Remove SPHQ from Communism.
2. Yes/No: Only if 1 results in a Yes:
   Add SPHQ to Fascism.
   SPHQ spawns an Army every 20 turns.

Voting in 1 week.

[nbarclay]

Alexman, have you thought about how changing the favored and shunned governments would affect AI attitudes? I'm starting to think that if we can't make Feudalism useful to human players without either having AIs use it when they shouldn't or changing other aspects of the game to prevent AIs from doing so, we'd be better off going back to the stock rules for it.

I think we really ought to have separate votes on whether to add the SPHQ to Fascism (assuming we remove it from Communism) and on whether, if we do that, we should have it spawn armies. If I recall correctly, the idea of moving the SPHQ to Fascism predates the idea of having it spawn armies and was originally proposed for separate reasons, so I think combining the two issues into a single vote is inappropriate.

My current views on these issues are:

- I definitely like the idea of removing the SPHQ from Communism.
- The rest of "1" seems better than what we have in the last version of the Mod, but I'm not convinced it's preferable to the stock rules.
- I'm undecided on both aspects of "2".

[alexman]

It's true that an AI in Republic will have a minus 5 attitude adjustment towards you when you are in Republic, whereas normally the AI adjustment is minus 1 when you are in the same government as the AI.

A difference of 4 attitude points, or 40% of the range of "Polite", is considerable, but then again who here knows the current favorite/shunned governments of the AI civs, and adjusts their strategy accordingly? I would guess nobody.

Anyway, here is the new proposal, even more split-up:

1. Yes/No:
   - Increase Fascism corruption level to problematic.
   - All AI civs shun Fascism.
   (Both are required for the AI to choose Communism over Fascism).
2. Rank the choices, to make the AI choose Republic over Feudalism more often:
   A: No change
   B: Restore Feudalism unit support cost to stock level, all AIs favor Republic.
   C: Change corruption and unit support cost of Feudalism back to stock levels.
3. Yes/No: Remove SPHQ from Communism.
4. Yes/No: If 3 results in a 'Yes', add SPHQ to Fascism.
5. Yes/No: If 4 results in a 'Yes', SPHQ spawns an Army every 20 turns.

[Theseus]

1. Y
2. ABC
3. Y
4. Y
5. N... too far afield, wait for the patch.

[MqsTout]

Isn't there a problem, if you want the AI to prefer Communism, and yet you remove the SPHQ from it?

[alexman]

Not really. Communism is better than Fascism even without the SPHQ.

[nbarclay]

The main reason for removing the SPHQ from Communism is that Communist warmonger strategies are currently overpowered in human hands. With a relatively large empire, a human player can go into Communism for the rest of the game and conquer AIs right and left while still maintaining about as fast a research pace as he could have had under Democracy. Thus, there is very little tradeoff involved in the choice to be a communist warmonger. With the SPHQ removed from Communism, the tradeoff would be at least somewhat more serious.

Also, if I understand how things work correctly, the impact of the SPHQ is much greater for large empires than for smaller ones. If that is correct, removing the SPHQ from Communism would have a vastly greater impact on the typical human warmonger than on the typical AI.

Nathan

[alexman]

Originally posted by alexman

1. Yes/No:
   - Increase Fascism corruption level to problematic.
   - All AI civs shun Fascism.
   (Both are required for the AI to choose Communism over Fascism).
2. Rank the choices, to make the AI choose Republic over Feudalism more often:
   A: No change
   B: Restore Feudalism unit support cost to stock level, all AIs favor Republic.
   C: Change corruption and unit support cost of Feudalism back to stock levels.
3. Yes/No: Remove SPHQ from Communism.
4. Yes/No: If 3 results in a 'Yes', add SPHQ to Fascism.
5. Yes/No: If 4 results in a 'Yes', SPHQ spawns an Army every 20 turns.

Time to vote. AU mod panel, you have 48 hours.

My vote:
1. Yes
2. BAC
3. Yes
4. Yes
5. Yes

[lockstep]

1: YES
2: BAC
3: YES
4: YES
5: NO

[Stuie]

1. Yes
2. BAC
3. Yes
4. Yes
5. Yes

[Nor Me]

1. Yes
2. BCA
3,4,5: Yes

[ZargonX]

1. Yes
2. BAC
3. Yes
4. Yes
5. No

[nbarclay]

1. Abstain. I'm really not sure whether pushing AIs to choose Communism over Fascism would be a good thing or not. AIs tend to have relatively small empires, which suggests Fascism, but also often relatively widely spread ones, which suggests Communism. There are also some other complications that give me additional mixed feelings.
2. BCA (mostly because I'm nervous about how our reduced unit support for Republic might interact with the standard rules for Feudalism in regard to AI government choices).
3. Yes
4. Yes
5. No, at least for the moment. If we find the AIs still making significant use of Fascism, I'll reconsider, but if AIs almost always choose Communism instead (which, if I understand correctly, is Alexman's goal), free armies from a Fascist SPHQ would not help with the problem of AIs' not getting armies.

[Theseus]

Sorry for jumping the gun on voting, alex.

I'll stand by my previous vote, but, again, would like a couple days leeway in re Armies, to try to get some more meaningful feedback from Firaxis.

[Strollen]

I realize this subject has been discussed a great deal, but....


After playing a handful of games with the AU mod 1.04/1.05 (and starting dozens more) I think we still have some work to do in balancing the governments.

First the good news, I think the changes to Republic and Democracy have significantly improved the game. From a human perspective I no longer research Republic ASAP, switch to it and stay there. Sometimes I am Monarchy sometimes Republic, the benefits of Democracy are sufficient that I almost always try to research it and switch to it. But the timing of the switch depends on many factors.

I more of an builder than a war monger, so I don't use Communism. If the consensus is Communism is too good with SPHQ, than probably removing it makes Communism a good, but not always better, alternative to Democray.

Unfortunately, we are still left with 2 goverment Feudalisms, and Fascism which good humans players don't use but hurt the AI if the switch to them and in the case of Fascism by researching them.

I'd like to see the next AU mod address the government rebalancing. I don't think it is necessary to have a new course but rather just make a mod that we can play with a give feedback about how the new goverment work out. Of course if somebody would like to do the AI research that would be invaluable.

It seems to me we have several choices.
1. Do nothing
2. Make some minor tweaks.
3. Remove Feudalism and Fascism a suggestion made by ducki a while ago.

Originally posted by ducki
r

Otherwise, really, honestly, no tongue-in-cheekiness, it would be better(help the AI) if they [inferior goverments] were removed.

4. Make some major changes to the governments.

Along the lines of #4. I advocated switching the effects of Monarchy and Fuedalism.
So Monarchy support is now 5/2/1, corruption Problematic, war weariness=low and Rush building requires population.

Fuedalism is 2/4/8 corruption problematic (or perhaps minimal) war weariness=none and rush build require cashs.

It is still worth while to research Monarchy because it allows the building of a very good wonder, and switching from Despotism to the new Monarchy makes sense if it is done by either
A. A religious goverment
B. Very early
C. by the AI under higher playing level.

The new Feudalism is good solid middle of the road Government which is superior to Depotism and almost always better than Monarchy. Researching Republic is slightly more expensive (34 vs 32) than Feudalism but can generally be acquired faster than Feudalism.

My suggestion for Fascism is to combine it with Communism and call it totalitarism. Perhaps we can figure out some way of making Fascism a good goverment for either Humans or the AI to play but it in the meantime, the AI won't waste research points going after it.


Any thoughts?

[alexman]

I don't agree that Feudalism needs any more changes. It doesn't fit all styles of play, but pop-rushing and the huge unit support for towns are unique elements that are quite useful in certain situations. With minimal corruption in the AU mod, Feudalism gives you more shield production than any government before Democracy, and you don't have to research any optional technologies to get it.

Fascism is not as bad as you think either. In stock rules it suffers because you have Communism as the alternative war government. But even in stock rules, Fascism is more efficient than Communism for small empires. With the SPHQ moved to Fascism in the AU mod, the gap has been narrowed. Also, the AI does not research or choose Fascism nearly as often as in stock rules.

[Strollen]

If Feudalism or Fascism aren't so bad, than how come we haven't seen KAI win the game using those govt types?

In fact, my DAR is one of the few (only?) I've read where a human player used Feudalism. In hindsight I would have been far better off in either Republic or Monarchy. The only reasons I didn't switch was because I wanted to see how Feudalism fared,and secondly I didn't have access to Republic for a long time, and Monarchy while better than Feudalism isn't sufficiently superior to endure 6 or so turns of Anarchy.

Not having played Facism I can't really comment about how suitable it is, but I would be curious to hear of players who have played at Emporer and above level who used Fascism extensively in their games.

However, Feudalism is a losing govt. because the longer you stay in it the worse you get. Pop rushing + war weariness leads to a very unhappy people, requiring a 20-30% luxury setting, and/or a large number of MPs.
Plus each person you pop rush is one less person producing for your empire. The 20 turns of unhappiness means you typically have to assign an entertainer for every 2 pop rushes you do.
Even for maximum corrupt cities, you are generally better off assigning tax collectors or scientists and paying for rush build than using pop rushing.

[ducki]

Originally posted by Strollen
If Feudalism or Fascism aren't so bad, than how come we haven't seen KAI win the game using those govt types?

I don't think we see KAI win very often at all, let alone with certain governments, but I know I'm playing below my ability for the sake of a higher fun factor.

The only reasons I didn't switch was because I wanted to see how Feudalism fared,

To get the most out of Feudalism, you have to make a serious effort to "fit the mold" - scads and scads of small towns, lots of units, and lots of very short wars against different opponents. It's really well suited to cramped, dry environments, and Firaxis have stated that the only real reason for a human to use it is if they are playing catchup.

However, Feudalism is a losing govt. because the longer you stay in it the worse you get. Pop rushing + war weariness leads to a very unhappy people, requiring a 20-30% luxury setting, and/or a large number of MPs.

Keeping your wars short, your pop-rushes strategic, and your towns small can offset a lot of this. Maybe not all, but with the huge support of towns size 6 and under, you can have a lot of MPs and still have enough of a force to fight short, decisive, oscillating wars, assuming your units are not woefully obsolete.

I do wish the F governments were more compelling for the human, but they have their niches for now, and are not necessarily the kiss of death for the AI that stock rules were.

[alexman]

Originally posted by Strollen
If Feudalism or Fascism aren't so bad, than how come we haven't seen KAI win the game using those govt types?

First of all, I don't think it's wise to evaluate the strategic value of elements of the game based on how they are used by the AI. If we did that, we would have to make artillery dirt-cheap, for example (since they are not used by any killer AI they must be useless! ).

Killer AIs are by definition large, well-developed empires. You can't make all governments good for such empires.

Feudalism and Fascism are good for other types of empires, which might not be the most powerful kind of empires, but by choosing Fascism/Feudalism, at least they are be better off than under any other government.

[nbarclay]

I'd like to propose increasing the flat free unit support for Republic in the AU Mod from 12 to 18. That way, Republic won't be quite so painful when civs first transition to it, but building enough units for large-scale fighting under Republic will still eat heavily into Republic's commerce bonus.

As a background note, I originally picked the number 12 almost out of thin air when I proposed it, using my REX in AU 501 where we started confined to a relatively small island as a data point. But in AU 502, 503, and 504, I averaged about 18 cities per game that I built myself before I started conquering additional territory, so replacing one unit support per city with 18 flat support seems reasonable. For me, anything less than 18 flat support would average making me worse off than 1/2/2 even before I start conquering other civs, and that goes farther in nerfing Republic than I'd intended to go when I suggested the 12+0/1/1 support model.

For players who don't claim as much territory and build as many cities REXing as I do, 18+0/1/1 would provide a small advantage before they start conquering additional territory, but that advantage would quickly transition to a disadvantage as they claim additional territory. Also, the times when 18+0/1/1 would work most in players' favor in the early game compared with 1/2/2 wold be the times when players need the help most.

[alexman]

I'd have no problem increasing the flat unit support for Republic/Democracy to 18. Of course, keep in mind that the greater that number, the more complaints we'll hear about the effect of map size on strategy.

Assuming that a larger fraction of units in a peaceful Republic or Democracy are Workers than in a wartime government, would removing support cost for Workers be worth considering?

Zero-support Workers combined with a minimum unit support might ensure that representative governments are not abused for warmongering. War-time governments would need to be balanced by removing an estimate of average Workers per city. 1 fewer free units per town/city/metro is a good one.

On the downside, building as many Workers as possible would become somewhat of a no-brainer, and that's not what we want here. It's also a big change.

[Aqualung71]

Disagree with free unit support for workers. Too exploitative....as Alexman says, it makes building lots of workers a no-brainer for a good human player.

[nbarclay]

I agree that removing support costs for workers would make building massive numbers of workers too easy a choice. The need to factor in support costs when deciding the right number of workers, and when deciding whether to keep workers around or to add at least some of the initial workers to cities and replace them with new ones when the time for railroads approaches, is an element of strategic challenge, and removing that element would go against the AU Mod's goals.

[pvzh]

I would suggest another change to republic/democracy

1/1/1 + 12.

As it now I am somewhat dissatisfied with AU republic because it forces Monarchy or extends desptism for very long even if you are plan to be peaceful, but need a lot of workers due to jungle/marshes/rough terrain.

This 1 free unit per any type of settlement would not benefit warmonger too much, but will still let use republic when getting cities to size 7 is a problem. Plus, AI does not understand concept of delayed switch and will switch ASAP even if it means big hit in unit support cost.

[alexman]

Originally posted by pvzh
AI does not understand concept of delayed switch and will switch ASAP even if it means big hit in unit support cost.

That's not true. A big factor in deciding whether to switch governments for the AI is unit support cost that it would pay.

As for your 1/1/1 +12 proposal compared to the 0/1/1 +18 proposal from Nathan, I could go either way. But I think we all agree that we can give a small initial benefit to Republic/Democracy compared to the current version.

[nbarclay]

I don't like 1/1/1 + 12 for three reasons.

1) It takes away the tradeoffs inherent in getting extra unit support when cities grow to size seven or larger. I think that would make the strategic choices a bit less interesting.

2) It actually makes Republic better than even under the standard rules (which have Republic 1/3/4) until a civ grows six cities to size seven or higher, and better than 1/2/2 until a civ grows twelve cities to size seven or higher.

3) It gives Republic back a large percentage of its value to warmongers at least until the late medieval era. During that period, a significant number of AI cities haven't grown past size six and even those that have tend to starve back down below size six rather quickly. With 0/1/1 + 18, captured cities don't start contributing to unit support until they grow back to size seven, probably after the war in which they were captured is over. (Those size eight or higher when captured - seven after the pop point lost during capture - do contribute until they starve down to size six.) In contrast, with 1/1/1 + 12, captured towns would start contributing to unit support immediately regardless of their size. That makes the 1/1/1 + 12 configuration, in my view, a stronger one for using Republic in a heavily warmongering style of play.

By the way, if it looks like 0/1/1 + 18 is stronger than 1/1/1 + 12 in the long term, that appearance is a bit misleading. The 0/1/1 + 18 configuration is stronger only when a civ reaches a point where no more than five of its cities are size six or below. Reaching that point isn't too hard for builders, or for players in the build stages of a "build first, then fight" strategy, although it will generally take a while even for them. But for civs that have engaged in significant conquest, getting distant, highly corrupt cities to size seven or higher can be a lot more difficult - and, for genuine warmongers, perhaps impossible because wars keep adding new small towns faster than old ones grow.

In regard to support for workers, remember that prior to C3C, Republic had no free unit support but had a unit support cost of only one gold per unit. So the break-even point between C3C's version of Republic and the previous version comes when a civ's free unit support covers [b]half[/i] its units. Also keep in mind that in the original and PtW versions of Civ 3, we tended to regard Republic as a bit overpowered compared with Monarchy.

With the current 0/1/1 + 12 configuration, that break-even point can be as low as 24 units for a civ with no cities size 7 yet, which I agree can pose a huge problem for a civ that needs a lot of workers. But with 0/1/1 + 18, break-even compared with earlier versions of the game would climb to a minimum of 36 units. A civ with ten cities size seven or higher would have a break-even point of 56 units, while one with fifteen would have a break-even point (compared with earlier versions of the game) of 66 units.

So while paying 2 gpt per unit for units not covered by free support might seem painful, the reality is that a civ can pay support for quite a few units and still have Republic work out very nicely.