AU mod: How to help the AI with happiness

alexman · December 13, 2003, 01:52

The Problem

The AI doesn't use the luxury slider. As a result, the AI assigns an inordinate number of entertainers, which cripple its growth, economy, and production. The fact that the AI pop-rushes and drafts more often than most humans, makes the problem worse. Also, the AI's combat tactics often result in war weariness, which forces a switch to inefficient wartime governments and a serious research setback late in the game.

Possible Solutions:

Double the effect of entertainers. This change was implemented in the PTW version of the AU mod, with success. The added value of entertainers is usually not significant enough to affect human strategy, except in totally corrupt cities. On the downside, it is a change visible to the human player, made just to help the AI.
Increase the citizens born content for the AI. This will allow AI cities to grow larger and more productive before needing entertainers. The AI currently has 2 content citizens. A possible solution could be to double that number to 4. However, this change would make it impossible for the human player to play at Chieftain difficulty using the mod.

Please share your comments on the above, as well as any other ideas to entertain the AI.

Dominae · December 13, 2003, 02:27

With the scarcity of Luxury resources in Conquests, I believe double-strength Entertainers will be a lot more important for human strategy. This is yet another change (like the Military Academy) that we should not just copy/paste without some thought.

I'm fine with the AU mod being incompatible with Chieftain difficulty; if you're playing the mod, you should be beyond Chieftain already (or will be after about a day on this forum!). One question though: why does doubling the number of Content citizens for the AI make it impossible to play Chieftain?

Dominae

alexman · December 13, 2003, 02:36

Holy cow, it doesn't!
In C3C you can actually add difficulty levels!

The AI plays at Regent in the stock rules, so it gets 2 content citizens. In order to increase the number of content citizens for the AI, we need to modify the difficulty level where the AI plays. Instead of modifying Regent, I thought we would have to modify Chieftain to be identical to Regent, except that it would allow 4 content citizens, and have the AI play at Chieftain

But from what I see in the editor, we can just add a new difficulty level. Edit: It appears on the setup screen as the hardest level though (beyond Sid), so that's kind of awkward.

nbarclay · December 13, 2003, 03:16

Has anyone verified that the effects of going from two citizens born content to four for the AI aren't too strong? The advantage of the AI's being able to get for free what human players can get only through significant (and significantly expensive) use of the luxury slider seems likely to have a major impact on balance in the ancient era.

Nathan

Harovan · December 13, 2003, 06:41

I would advise small steps. First try three, test it thoroughly (may be in an AU course) and if it is a good change, doesn't hurt the human too much (beware the fun killers!), but does still not suffice, increase it to four.

lockstep · December 13, 2003, 07:59

Quote:

Originally posted by Dominae
I'm fine with the AU mod being incompatible with Chieftain difficulty; if you're playing the mod, you should be beyond Chieftain already (or will be after about a day on this forum!).

At the first moment, I was uneasy about erasing a difficulty level because I view the AU mod as 'smoothing out the basic game', and the basic game includes Chieftain after all. But in the end, I think Dominae is right.

However, increasing the citizens born content for the AI raises another, more serious problem IMO: Isn't Regent difficulty supposed to be the level that offers 'playing by the same rules', i.e. bonuses neither for the human player nor for the AI? With the suggested change, this won't be true anymore. (Okay, the AI trades more freely with other AI's at the moment, but a happiness bonus is beyond that).

In other words, I'd rather tinker with entertainers to solve the AI's happiness problems.

ducki · December 13, 2003, 11:03

Hmm... has anyone ever played an experiment game where they disallow themselves the use of the lux slider?

This might be a good way to gain insight into possible solutions. Maybe I'll abandon my current epic and see if anything jumps out at me, unless someone's already done this, in which case, do share!

geniemalin · December 13, 2003, 12:43

Put me down for 'double value entertainers' - giving the AI more free happiness just allows more early AI dominance without changing the fact that an AI without enough luxuries or at war in a WW government will still have cities full of entertainers later in the game.

Jaybe · December 13, 2003, 16:05

I have seen the AI use the luxury slider in C3C.
10% Entertainment, but it WAS just before slipping into anarchy (war weariness).

Re 2xEntertainers: The human is much more likely to use other specialists (as available) rather than entertainers. Have you seen the AI use police or civil engineers yet? (I haven't)

Dominae · December 13, 2003, 19:47

Quote:

Originally posted by Jaybe
Re 2xEntertainers: The human is much more likely to use other specialists (as available) rather than entertainers.

Not if you're having happiness problems, which you always do in Conquets.

Dominae

Stuie · December 14, 2003, 08:36

I like the idea of increasing the citizens born content for the AI. This will give no advantage for the human, whereas the 2x Entertainers in C3C will (as pointed out by others).

Quote:

But from what I see in the editor, we can just add a new difficulty level. Edit: It appears on the setup screen as the hardest level though (beyond Sid), so that's kind of awkward.

You could get around this by adding a new difficulty, then changing Chieftain to something else ("AI Only" or somesuch) and performing the change suggested, then change Warlord to Chieftain (and inputting the Chieftain difficulty data), Regent to Warlord, Monarch to Regent..... etc. etc. so that the Sid level ends up being the one we added. A lot of work, but then the "AI Only" level shows up before Chieftain, not after Sid. I think that would look better. Does that make sense?

lockstep · December 14, 2003, 16:34

How about adding a new 'AI only' difficulty level similar to regent, but leaving the no. of citizens born content alone and instead changing the no. of citizens quelled by military from 1 to 2? This change would help the AI in the later game at least as much as in the early game, given that the AI usually has a bunch of defensive units in its cities and uses governments with military police at least as often as the human player.

alexman · December 14, 2003, 16:55

I think the number of units quelled by military refers to resisting citizens, not unhappy citizens.

If that's not the case, and if the government MP limits are in terms of units, not citizens, it's a great idea!

lockstep · December 14, 2003, 17:02

From the editor help file:

Number of Citizens Quelled by Military

Determines how many rioting-citizens are quelled with the presence of military units.

'Rioting' could mean 'resisting' as well as 'unhappy'. This needs to be tested.

lockstep · December 14, 2003, 17:07

I just checked the 'General settings' editor page: Under 'citizens mood', there's an entry 'Chance of Rioting', and according to the help file this is clearly connected to cities in civil disorder.

lockstep · December 14, 2003, 17:07

EDIT: From the editor help file

Military Police Limit (Government Properties)

Determines how many military units (units with attack rating greater than zero) can be garrisoned in a city to reduce unhappiness.

Nor Me · December 14, 2003, 20:17

Quote:

Originally posted by lockstep
I just checked the 'General settings' editor page: Under 'citizens mood', there's an entry 'Chance of Rioting', and according to the help file this is clearly connected to cities in civil disorder.

That's when improvements get destroyed in civil disorder. Individual citizens don't riot so I doubt that's relevant.

On adding another difficulty level, not everything about the difficulty levels can be modded. The multiplier to the score depends on how far down the list the human's difficulty level is (so deity now gets a higher scores because demigod has been added before it). I'm fairly certain hut results depend on difficulty level and I don't know how. I'd rather have the AI get chieftain hut results than beyond Sid. So adding a difficulty level might have side-effects and it might be better to lose chieftain.

alexman · December 15, 2003, 13:30

Good point. We don't want different scores from stock, and we certainly don't want the AI popping barbs every time if it's playing beyond Sid.

So moving the AI to a modified Chieftain might actually improve its chances to get good things out of huts... I hope not, but I don't think it's a huge deal if that's true.

By the way, Ision had a good suggestion in another thread, that might solve some of the AI happiness problems as an indirect effect: Put Workers in the AI build-often category.

Nor Me · December 15, 2003, 15:05

Come to think of it, If any AU Mod Chieftain players object to having only a 200% combat bonus against barbarians, they should speak up now.

With 4 citizens born content and keeping the 20 cost factor, the new chieftain won't be much different from the old. I doubt many Chieftain players will notice the 90% OCN.

Ision · December 15, 2003, 15:25

I consider the '4 born content' idea to be excessive. I believe that the effects would be drastic in terms of the game dynamics at emperor and especially diety level.

The doubling effect for entertainers sounds much more reasonable. While it is true that from time to time this will be a greater advantage to the human than the AI - this will be far more of an exception than the rule. On average it will benefit the AI far more often.

I agree with a previous posters commentthat stated,

"Put me down for 'double value entertainers' - giving the AI more free happiness just allows more early AI dominance without changing the fact that an AI without enough luxuries or at war in a WW government will still have cities full of entertainers later in the game."

Ision

sabrewolf · December 16, 2003, 11:25

this is definetly something firaxis has to solve. it can't be too difficult to calculate the scenarios each turn for the ideal combination of the luxury slider and entertainers.

in fact, i'm sure this would make the AI a whole lot more potent. in the early game and the higher difficulty levels wouldn't need to have that many bonuses.

for what we can do: i prefer seeing the entertainers effect doubled and the number of content born increased by 1.

the entertainer makes it more worth it in cities larger that 5, the born-content helps the sizes 3 and 4. sizes 1 and 2 don't suffer any problems anyway

vulture · December 17, 2003, 08:38

Ideally we would like the use of entertainers to more or less match the use of the luxury slider for effectiveness (too much power in entertainers, and we use them instead of the lux slider, too little and the AI is handicapped by its non-use of the slider).

In despotism or monarchy, each tile 'typically' produces 1 or 2 trade, 1 shield and 2 food. In republic (or democracy) that's 2 or 3 trade, 1 shields, 2 food. Using the luxury slider eats 1 trade per population point, but each trade used for luxuries allows you to produce 1 shield and say 1.5 trade instead, a profit of 1 shield and 0.5 trade. Plus food, so you keep on growing.

Using an entertainer that gives 2 happy faces (plus an unhappy citizen becoming the entertainer in the first place) takes 1 worker off the fields - costing you 1.5 trade, 1 shield, 2 food, but lets you produce work from an extra 2 tiles. Investing 1.5 trade, 1 shield gains you 3 trade, 2 shields. But in order for you to keep growing, you want one or both of those guys to be working irrigated tiles, so you are producing fewer shields, more food, so more likely your two workers will be producing 3 trade, 1 shield (and 3 food, slow growth). Next benefit 1.5 trade, 0 shields.

So the luxury slider allows normal growth, and in despotism / monarchy gives you benefits primarily in terms of shields, with some benefit to trade. 2-happy-face entertainers benefit primarily in terms of trade, with slower growth and little or no effect of shield production. This benefits the human player - we use the luxury slider, and get the benefit in terms of shields, which are harder to come by than cash in the ancient era. The AI irrigates liberally, so is better placed to use the entertainers, but only increases its cash flow by doing so, which in the ancient era means it all goes to research - already a strong suit of the AI at higher levels.

With 3 happy face entertainers, an investment of 1.5 trade and 1 shield gives you a return (eventually) of 4.5 trade and 2 shields (for the sake of argument), a profit of 1 shield and 3 trade, again at the expense of slower growth (again offset by the AI's tendency to irrigate anything that stands still long enough). I think this is a decent level - the AI's bonus is still in the direction of trade (it always will be), but it doesn't hamper production unduly.

Consider a size 5-6 city, 2 naturally content citizens. Each tile, for the sake of argument, is 2/1/2 food/shields/trade (mined) or 3/0/2 f/s/t (irrigated)

Size 5:
With luxury slider. Everything mined. Gross output is 6 shields, 12 trade. Luxury slider at 30% gives 4 happy faces (1 is wasted (too much drink

)). Result: 6 shields, 8 trade.
With 2HF-entertainers. 1 square irrigated. 1 entertainer needed. Gross output is 5 shields, 10 trade.
With 3HF-entertainers - same as 2HF entertainers in this case.

Size 6:
Luxury slider at 30%. Gross output is 7 shields, 14 trade. 4 happy faces from luxuries. Net output 7 shields, 10 trade.
2HF-ents: 2 irrigated tiles, 2 entertainers needed, gross output is 5 shields and 10 trade, with no further growth possible without more irrigation.
3HF-ents: 1 irrigated tile, 1 entertainers, gross output 6 shields, 12 trade.

Comparing the entertainers with the luxury slider, you are basically giving up growth and shields in return for trade - science, essentially. Going to more happy faces per entertainer means less of a penalty in terms of growth and shields, and more of a boost to trade (relative to luxury slider use).

Ideally, it'd be nice to see an a worker on a 2/1/1 tile become an 'entertainer' that produces 2/1/0 (and a worker on a 2/1/2 tile become an 'entertainer' that produces 2/1/1 - mirroring the effect of the luxury slider exactly, but fine-tunable on a city by city basis...).

Is it possible to create a new type of specialist (the extra-happy entertainers we are discussing here)? And set it up so that only AI civs can use it (attach it to a tech that all AI start with, and is far too expensive to trade for, steal or beat out of the AI). Is it possible to assign a tech to all AI while still allowing the player to pick any civ?

ducki · December 17, 2003, 11:51

What if we made the AI prioritize the happiness wonders the way humans do?

With Conquests, any time I can, I'm building the Mausoleum - it's relatively cheap and those extra contents in (usually) my capital give me a fairly well developed wonder-building city. If I happen to be on a river, look out, especially when the Govt switch comes.

Anyway, I agree we need to help the AI until Firaxis teaches it to use the Lux slider, especially since the Despotic Irrigation that Soren fixed in PtW seems to be back.

Personally, I second the idea for +1 born content as a first test. I know for a fact that I'm a bit more likely to use entertainers in size 6+ cities if they are doubled, as I leaned that way last time we did this. It sure makes WLT*D easier than the slider, IMO.

(You know what C3 needs? Two rulesets - one for Human, one for AI. That way we could tweak for the AI's weakness without strengthening the player.)

Dominae · December 17, 2003, 12:16

Quote:

Originally posted by ducki
What if we made the AI prioritize the happiness wonders the way humans do?

We do?

nbarclay · December 17, 2003, 12:56

Not all human players make happiness wonders a priority. I don't mind getting them, but I'm usually too busy REXing to build ancient wonders and they don't tend to be my top priority in other phases of the game either.

I agree that the biggest problem with boosting the number of happy faces for entertainers is that it makes WLT?D easier for human players in distant cities. But the the increased power of specialists of other types mitigates that in C3C, especially in the later game once civil engineers become available. Do you use entertainers to put a city in WLT?D to boost production, or do you use civil engineers (or possibly policemen) to boost the city's production in a more direct way? Or do you forget about production, irrigate as much as possible, and have a city of scientists/tax collectors? Considering how rarely I've been inclined to give up tax collectors, scientists, civil engineers, or police in favor of entertainers with the default rules, I think double-value entertainers might actually enhance strategic choices for the player at that stage of the game.

If there's a problem, it's more earlier in the game when WLT?D is the only way to boost a distant city's production to a "reasonable" level. But personally, I think being able to do that makes the game more enjoyable. So as long as AIs benefit at least as much from double-value entertainers as I do, I'im inclined to view them as an improvement.

Question: are we sure that AIs still use mostly or at least largely entertainers rather than other types of specialists when they run into happiness problems? If so, double-value entertainers woud tend to favor them. But if AIs tend to use mostly other types of specialists now and don't know how to adjust for double-value entertainers, there could be a problem. I haven't investigated the issue, so I thought I'd bring it up just in case it might be a reason for concern.

Nathan

ducki · December 17, 2003, 14:18

I think perhaps my word choice confused you, so let me try to clarify.

Quote:

What if we made the AI prioritize the happiness wonders the way humans do?

I added emphasis where I did, because that is the important part of the sentence.

I'm not saying players make happiness wonders a priority. I'm saying players are able to prioritize happiness(or any other type) wonders where it makes sense.

When you get a SGL and don't need a research boost and have several "good" wonders available and coming soon, what do you do?
You prioritize them.
If you have little or no luxury, happiness probably gets prioritized at the top of the list. Probably.
If you are stuck on a small island alone, seafaring wonders get prioritized at the top and continent-wide happiness _probably_ gets prioritized at the bottom.

Anyway, my point was, humans are able to see when they have happiness problems and if they have a good opportunity to fix it with a wonder, they can consider that and determine how valuable it would be versus other solutions. They prioritize their options.
It would be nice if the AI did that.

nbarclay · December 17, 2003, 14:59

Changing the AI priority for happiness wonders would have only a limited impact on the overall picture because each great wonder can be obtained by only one civ. The AI that gets a happiness wonder might be more competitive, but what about the others that don't? And what if the human player ends up with most of the happiness wonders in spite of the AIs' best efforts? If we're going to make happiness less of a problem for AIs in general (rather than maybe for one or two in each game), it needs to be done in some deeper, more fundamental way than adjusting wonder-building priorities.

Nathan

ducki · December 17, 2003, 16:04

Ahh! I just re-re-re-read my post and I see where the confusion is coming from.

Instead of "...if we made..." I meant to say "...if we could make..."

Basically, the first two paragraphs of that post were a C3C version of the Beach Boy's "Wouldn't It Be Nice". There's nothing we can do to make the AI analyze his empire. The rest of the post was agreement about +1 Born Content and concern about abusing Improved Entertainers(I am guilty).

In an effort to pull things back on track:
Pretend I didn't post those last two items. It was desire for smarter AI(from Firaxis) that got me excited enough to speak slow and garbled.

On topic:
I'd prefer to give the AI more Born Contents or Tell them to quit irrigating so much than improve Entertainers. That's just personal preference not supported by any tests.

lockstep · December 20, 2003, 13:59

Quote:

Originally posted by geniemalin
Put me down for 'double value entertainers' - giving the AI more free happiness just allows more early AI dominance without changing the fact that an AI without enough luxuries or at war in a WW government will still have cities full of entertainers later in the game.

I have to agree with geniemalin here - a double-value entertainer may affect human strategy somewhat, but this tweak not only assists the AI, but assists it in the late game at least as much as in the early game.

BTW, entertainers that yield two happy faces may be more powerful than the other 'old' specialists (taxmen and scientists - even in their C3C version), but IMO they aren't more powerful than the new civil engineers.

Jeem · December 20, 2003, 22:25

Quote:

Originally posted by ducki
Hmm... has anyone ever played an experiment game where they disallow themselves the use of the lux slider?

That sounds like me, just about every game.

I only ever use luxuries when I'm in a close race to build a wonder and need everybody working at my capital (or wonder building city) in order to leap ahead of the AI.

I think the way the AI is able to wage war with relative impunity towards war-weariness (and governments) helps to offset it's use of entertainers. By modding them to have happier citizens than what they get at regent, I think there could be a danger of AI democracies running riot.

I'd love to try this mod out though, just to see how much harder the AI would be. Any Religious AI that got the Chapel and Bach's would surely be able to wage war as a democracy for the rest of the game if it chose to?

There is also the issue of WLTK day. Increasing happiness levels might allow the trading AI to have WLTK days far too easily. I think the religious civs would get a massive bonus from this.