Pick Your Leader Challenge 1

[davidlallen]

3. Suggest replace Tleilaxu palace with a UB that gives +1 priest GPP. That way the shrine will get founded much earlier. Take the shrine from 2 free priests down to 1 free priest (or 0 free priests and just ~+3 priest GPPs), and maybe have it add a priest specialist slot or two.

Your wish is my command. Please find attached a modified buildinginfo which has a tleilaxu palace with +1 priest GPP and a zensufi shrine with one free priest instead of two. Also I changed the civilizationinfo file so that the tleilaxu palace is a UB for tleilaxu. Please unzip into your mods directory and start a new game, to see if it makes enough difference. You can probably tweak the XML for the two buildings if you want a small change from there.

 

[Ahriman]

Thanks!

Testing now on the same map.
First, something to note; score under-estimates the true faction power for Tleilaxu, because it doesn't value settled specialists. So actual hammer and gold output is much more than that indicated jsut by their population adn tech.

Game 1(old version) scores:
Turn Score
50 173
100 519
150 947
200 1302
250 1499

Game 2 (with patch)
Turn Score
50 143
100 473
150 1022
200 1350
250 1540

Ai does a different early game; in game 1 it built settler then worker. In game 2 it built soldier, scout thopter, solider, mushtamal, worker, settler. So was much slower to expand.
Ai still doesn't have Faith tech by turn 150. Unfortunate.
Ai still won't found more than 3 cities!!
Ai doesn't build a Zensufi temple in its capital until turn 227. so isn't running a priest economy until late despite the shrine.

Recommendation: have the shrine provide 1 slot to turn a citizen into a priest.
Maybe look at AI flavor Weights? Put another unit at Faith, or increase its weighting, or change the flavor value of the AI leaders?

Another thing I notice:
The AI builds a *lot* of infantry and missile troopers in the early game, and then in the midgame it takes it tech slider way down and starts building gold in all its cities in order to accumulate enou
This chokes off its tech progress.

The AI isn't building turbines or solar farms anymore; it sees cottages as superior (presumably becuase of BBAI which recognizes that cottages grow).

Game3 (with patch, identical to game 2)
Turn Score
50 166
100 495
150 1138
200 1934
250 2299 (3rd place)

Again, shrine build on turn 51. I think this is a good change.
Ai builds a settler earlier.
Ai has Faith tech by turn 150.
Ai builds 4 cities this time (though the 4th is turn ~170, and not in a good place), but still ignores the excellent city site (2x water + mesa + cereal) ~10 tiles west of the capital.
Ai gets a good 5th city because it captures a nearby barbarian city.
This propels it into first place in the rankings (briefly, it then falls to second Atreides who terraform to get big population).
Big war with Harkonnen, Tleilaxu capture a Harkonnen city on turn ~244.

Tleilaxu did very well this game.

It seems that sometimes the Tleilaxu player does well relative to others until they get religion and then declines.
To test if this is because of techwhoring, I test a game4 which is identical to 2 and 3 *except* that it has tech-trading option off.

Game 4
Turn Score
50 143
100 483
150 1085
200 1751
250 1961

Again, AI has weird build. It builds soldier, worker, scout thopter, mushtamal, soldier, thopter.
I wonder; does the AI's build get messed up when its scout thopters are eaten by worms? It feels the need to replace them, which slows its expansion?
Is there a way to give the AI player only free victory against animals for the first 1-2 fights, to stop this from happening?

Ai builds 3 good cities.

Ai is 3rd place in score for most of the game.
Still no Faith tech by turn 200, so its not running a priest economy.
~Turn 210 Ai invades Ecaz, and starts giving them a severe beating, but takes no cities.
Harkonnen invades us turn 230, but is driven off.
Ai falls to 4th place, partly because pillaging starves some cities.

Conclusion; no tech trading does weaken other factions much more than it weakens Tleilaxu, but as shown by game3 they can still perform well even without it.
The main effect of no tech trading is to increase the score variance; weak factions who are behind in tech are the main beneficaries of tech trading, it doesn't really help the leaders to push ahead.

 

[Ahriman]

From the above: the palace change is a very good one.
I think that will be enough, provided we can also make espionage more valuable.

AI settling pattern is still weird, they ignore good city spots while building in some weaker ones further away.

We still might want to look at flavor issues for Faith tech though, or add something else to make it look more valuable. Ai isn't getting it early enough.
How about a generic monastary building (available to all factions, no particular religious affiliation) that gives a priest slot, +1 culture and +1 health? Or something.

 

[davidlallen]

The AI isn't building turbines or solar farms anymore; it sees cottages as superior (presumably becuase of BBAI which recognizes that cottages grow).

One of many things I had meant to test, but never did. Time to wake back up the thread about balancing improvements from this post and this post.

Game3 (with patch, identical to game 2)

Your description of the AI performance between game 2 and game 3 seems very different. If the start position is the same, I would think a new play would give the exact same result. Is there any difference in the setup?

We still might want to look at flavor issues for Faith tech though, or add something else to make it look more valuable. Ai isn't getting it early enough.

Might we address this by bumping up the FLAVOR_RELIGION for the BT LH? It is 2 now, maybe 8 or 10?

AI settling pattern is still weird, they ignore good city spots while building in some weaker ones further away.

There has been no (published) change to colony site selection, it is just as stupid as it has always been.

 

[Ahriman]

Is there any difference in the setup?

No, no differences. They were the same map, same settings, same start positions. The AI doesn't always do things identically (there's a lot of random), and I wonder if scout thopters dying to sandworms makes the AI try to replace them.
In game 2, Fremen were very weak and Corrino conquered most of them. In game 3, just the reverse. Where Imperium spreads also makes a big difference for the diplomatic alliances.

Might we address this by bumping up the FLAVOR_RELIGION for the BT LH? It is 2 now, maybe 8 or 10?

Yeah, this is probably the best way to go. Maybe 5 or 7? Most of the other factions are 5/2 for major/minor (this is what they were before, I didn't change them).

There has been no (published) change to colony site selection, it is just as stupid as it has always been.

Oh, agreed, I didn't mean to imply this was anything new. Just frustrating.

A great city site just gets totally ignored.... even when its so very close.
And the AI keeps settling on top of resources.

 

[davidlallen]

Yeah, this is probably the best way to go. Maybe 5 or 7? Most of the other factions are 5/2 for major/minor (this is what they were before, I didn't change them).

This is the FLAVOR_RELIGION setting in xml/civilizations/civ4leaderheadinfos.xml, for LEADER_GOYA and LEADER_SCYTALE. Care to give it a quick try? The only difference should be researching faith earlier.

 

[Ahriman]

Care to give it a quick try? The only difference should be researching faith earlier.

Will do. Obviously this won't be the only affect, it will also divert them towards other religion techs (fanaticism, jihad, religious mandate, academies, etc.)

 

[Ahriman]

Game5
Same as game 2-3 except flavors changed from science/religion 5/2 to 2/5.

Build has soldier, scout thopter, worker, water cache, mushtamal, soldier, barracks, refinery. All before getting a settler.

By turn 100, no Faith tech, only one other city founded.

I think the AI is being weakened by building the House Shield very early each game; I think it needs to be made cheaper. The benefits aren't very high in the super-early game (when there is no war), but the cost can slow economic development. The human player never really wants to build it super-early.

Atlternatively; if we're renaming suspensor device to Holtzmann field, maybe we should move the House Shield Generator there?

The AI finally builds a city to the west in this game (as its 3rd city), but its in a dumb position (right on top of the nitrate), with only 2 water sources and one of those is already in the city BFC.

Faith AND fanaticism researched by turn 150, and working on Divine Mandate, and 3 settled great prophets in the capital, and priest economy going strong.

Tleilax is 3rd place at turn 100, falls to 5th place by turn 150, and stays there all game.
Still only 4 cities (including capital).

Scores:
Turn Score
50 136
100 467
150 941
200 1317

So this flavor change seems to work.

I think that once we can get espionage to actually be effective and powerful, AI-controlled Tleilaxu will be as powerful as anyone else.

 

[davidlallen]

OK, I'll keep the change of flavors and the Tlei UB and shrine change. Which civ is up next? Ix, to try vehicles? Ordos, to see if their spies are just as lame?

 

[Ahriman]

Ok, can do a bit more.

Ordos
Game1

Build is soldier, scout thopter, worker, water cache, settler, barracks.
There is a non-trivial diference in AI performance I think depending on whether the 3-mesa chain at the capital gets 2 windtraps or 1 (ie do they build windtrap on the center tile first, or on an end one).

They build a 2nd settler for a 3rd city pretty soon too.
By turn 100 they are in first place, and have a 3rd settler built.

I have no idea why Roma is expanding so much more and faster. She doesn't have an Expansive trait or flavor.

Ai found Imperium religion. This will be a big help, diplomacy-wise.

By turn 150, down to 4th/5th place. Partly because the 4th city is built 1 SE of the nitrates, which removes the windtrap from the capital's BFC, so it stops growing and starves down to size 5.

1 Saboteur by turn 150 (just discovered law of arrakis).

Fallen to 6th place by turn 175 (but 2nd through 8th place are all score 1200-1400).
Again because of poor city placement, lack of expansion. One of the places (island to the east) where the AI normally expanded had a barbarian city spawn there, that was captured by Corrino.

3 saboteurs, but very low EP income.

6th place by turn 200.
4 saboteurs, but barely enough EPs to use any missions. Prison camps just being constructed.
Saboteurs don't really seem to be going anywhere.

Again observe lots of cities buildign wealth, getting up gold to upgrade units.
This is quite inefficient. IIRC, building wealth = 2 hammers per gold. Upgrading units = 2 gold per hammer.
So, building Wealth to upgrade units costs 4x as much as it would be to just disband the units and build new ones.
I'm not sure how we should fix this.

Saboteurs *are* traveling around in Corrino territory.

Fremen invade, and destroy a lot of units and pillage some stuff, but don't take any cities.

[Aside: Fremen Raider should be renamed Raider; otherwise the game refers to them as a Fremen Fremen Raider. Also, I'd make them strength 10, 30% withdraw chance, melee unit. And maybe move them to a different tech.]

Turn 250
Down to 7th place, thanks to the Fremen war.

Saboteurs aren't in Fremen territory, they're still generally wandering the deserts.
I wonder if they're constantly changing their mind about what mission to carry out, and changing direction?
250-1000 EPs accumulated against each other player.

6 Saboteurs, all generally sitting around. All fully promoted, but do the promotions do anything much?

500 EPs against Rabban.


So, espionage is pretty useless. We need to drastically improve the AI somehow, and maybe make some of the minor missions cheaper (eg improvement destruction) or more effective (eg poison water for Tleilaxu)

 

[AnotherPacifist]

I think the AI hoards cash because it wants to be able to upgrade units as soon as the tech is available and buy units offworld. But the main problem is still, as Ahriman points out, that they don't expand either by building more cities or conquest.

 

[davidlallen]

I think the AI hoards cash because it wants to be able to upgrade units as soon as the tech is available and buy units offworld.

I am not certain, but I don't think that is the reason for "build wealth". The AI for this, written by koma13, says that the AI will spend money as long as it is not in financial trouble. But it does not intentionally save up money in order to buy units. To test whether this is related, one could easily modify the code to prevent the AI from buying units. In assets/python/MercenaryUtils.py, find "def computerPlayerThink" and add the line in red:

	def computerPlayerThink(self, iPlayer, mercenaries):
		[COLOR=red]return[/COLOR]
		player = gc.getPlayer(iPlayer)
		currentGold = player.getGold()
		
		if player.isHuman():
			return

What could be another reason for "build wealth"?

 

[Ahriman]

But it does not intentionally save up money in order to buy units.

I wouldn't quite interpret it like this.

I would say:
The AI build lots of cheap tier1 units (infantry and rocket troopers).
The AI spends gold on homeworld units and on upgrading this large army (mostly upgrading, not purchasing). This depletes its gold supply.
So the AI uses "build wealth" in order to replenish its gold stocks.
I noticed sometimes as well that the AI would take its science slider down to ~10% or so to have a gold income of ~+110 or so per turn, after it researched tech for heavy troopers (to upgrade its large army of infantry/maula guardsmen).

It doesn't need to be "saving" pre-emptively in order to exhibit a similar behavior.

The alternative hypothesis is that the AI has built all available buildings and wonders, and feels it has enough units already relative to its population size (even if they're obsolete units of low-strength), and so it reverts to building wealth because it doesn't see anything else worth spending on (like... a 4th settler).

This is partly I think because of normal game speed; on normal game speed the AI tends to finish all the availible buildings and have a large army sooner than it does on Epic.
I think we could consider adding some more buildings, so that there are more hammers to spend on buildings. For example, adding a second health provision building (after removing the extra bonus health from greenhouse, water cache and the grocer building whose name I forget (at Protective trade), and converting the Spice refinery into a gold building and adding another Hammer booster.

 

[negyvenketto]

I have played as Bene Tleilax, had Axolotl Tanks and got a GG. I used it to create a super unit. I could clone that unit with all it's promotions, I know that's intended. But it's not fair to have unlimited GG units with 0 upgrade cost. Is it possible to make the new units have all but GG specific promotions?

 

[davidlallen]

I have played as Bene Tleilax, had Axolotl Tanks and got a GG. I used it to create a super unit. I could clone that unit with all it's promotions, I know that's intended. But it's not fair to have unlimited GG units with 0 upgrade cost. Is it possible to make the new units have all but GG specific promotions?

Good feedback! I had not thought of that exploit. There are only a couple of promotions which are GG specific. What should happen if a unit gets a GG and then picks all standard promotions with the experience?

It has been suggested a few times that instead, we should make an "immortal" promotion. If a unit has this promotion and it gets killed, it re-appears at your capital city alive again, with high damage, and without the immortal promotion. Do you think that would be better?

 

[negyvenketto]

Good feedback! I had not thought of that exploit. There are only a couple of promotions which are GG specific. What should happen if a unit gets a GG and then picks all standard promotions with the experience?

I gave only standard promotions, but there is one which guarantees the no cost upgrade, added automatically. It saved me a great bunch of gold.

It has been suggested a few times that instead, we should make an "immortal" promotion. If a unit has this promotion and it gets killed, it re-appears at your capital city alive again, with high damage, and without the immortal promotion. Do you think that would be better?

Actually I don't like if someone can get a unit for free. It should cost gold or hammer.
To nerf the current system I could think about limiting the number of gholas of a single unit. Let's say one could build 5 gholas of the most experienced unit, after then 5 gholas of the 2nd most experienced and so on.

 

[AnotherPacifist]

To nerf the current system I could think about limiting the number of gholas of a single unit

But Leto II had many gholas of Duncan Idaho, so that's not really true to the story.

 

[davidlallen]

I gave only standard promotions, but there is one which guarantees the no cost upgrade, added automatically. It saved me a great bunch of gold ... Actually I don't like if someone can get a unit for free. It should cost gold or hammer.

I can exclude certain promotions from the ghola. The free upgrade one is a good example. Perhaps the Ix mechanical ones should be excluded as well. When the ghola is created, there is a good chance that one promotion will be removed or decreased by one level (combat 3 on the original, decreased to combat 2 on the copy). Perhaps each subsequent copy should be reduced by more. This represents the fact that it isn't a "perfect" copy. Somebody also suggested an increasing chance that the ghola will go insane, which could have some bad effect.

BTW, since you have played at least one game using these files, do you find any funny results in founding religions? AnotherPacifist has found weird things reported earlier in this thread, such as when he researches a tech to found a religion, some other civ gets the religion when they should not. I haven't been able to reproduce this, and I don't see how it could be specific to these game files. Still, if anybody besides AP can reproduce it, please let me know.

 

[negyvenketto]

When the ghola is created, there is a good chance that one promotion will be removed or decreased by one level (combat 3 on the original, decreased to combat 2 on the copy). Perhaps each subsequent copy should be reduced by more. This represents the fact that it isn't a "perfect" copy.

Perfect copy is quite rare. I guess around 10%. It seems there is at least 50% chance to loose more than 1 promotions.

BTW, since you have played at least one game using these files, do you find any funny results in founding religions? AnotherPacifist has found weird things reported earlier in this thread, such as when he researches a tech to found a religion, some other civ gets the religion when they should not. I haven't been able to reproduce this, and I don't see how it could be specific to these game files. Still, if anybody besides AP can reproduce it, please let me know.

Yes, I wanted to report that as well.
I think it's not very hard to reproduce, but for the early religions it's hard to believe nobody could have that tech on that turn in a clean game. But if you use WorldBuilder to give Tleilax all prereq techs for Academies or Landsraat and accelerate research with some settled GS ... You will see.

 

[davidlallen]

AnotherPacifist has found weird things reported earlier in this thread, such as when he researches a tech to found a religion, some other civ gets the religion when they should not.

I think it's not very hard to reproduce, but for the early religions it's hard to believe nobody could have that tech on that turn in a clean game. But if you use WorldBuilder to give Tleilax all prereq techs for Academies or Landsraat and accelerate research with some settled GS ... You will see.

Thank you for the suggestion. I can reproduce this particular case but I do not understand why it should happen. In this particular case, Tleilaxu is prohibited from founding the religion, so nobody should found it at that time. In some other thread AP mentioned he had the "Tech Diffusion Mod" turned on, which is apparently turned on by default in RevDCM 2.6. I can reproduce this problem *only* when I have this option on. When I have this option off, I do not reproduce the problem.

Could you try a quick test to ensure that when you turn off Tech Diffusion in the custom game screen, this problem no longer happens?

Although I understand the idea of tech diffusion, I do not see how it could cause this problem. Tech diffusion should grant other civs some beakers towards certain techs. When I grant Benevolence to BT and then research Academies using 15 settled GS and an academy, I get Academies around turn 10. With tech diffusion *on*, some other civ founds Qizarate. But, using WB and the tech chooser screen, I can see that no other civ has learned Academies. With tech diffusion off, nobody founds Qizarate which is what should happen.

Somehow tech diffusion is causing another civ to found the religion without having the tech. This is a bug. I will look over on the RevDCM subforum to see if there is a solution. For now, let us turn off Tech Diffusion. I will remove it as a game option and make sure the option is false. We have had similar problems with other RevDCM features.