Call me crazy, but I think I managed to fix the AI farm/morale problem for good this time.

Okay, maybe not totally, but it has been definitely severely reduced. The biggest issue has always been the amount of farms compared to the amount of morale improvements the AI builds on it's worlds. Too many of the former and too few of the latter. In my previous attempts to fix this, I standardised the base amount of food the Civ Capital and the Initial Colony provide, and changed the food bonus of farms from an integer to a percentage value so the bonus tiles are no longer an issue. Now, I also standardised the amount of food the farms themselves provide. I originally wanted to keep the standard farms unique, by providing more food than the Robotic Farms and Charging Stalks, just like it is in the vanilla game. As it turned out, that was the wrong way to go. It also blinded me to the obvious solution. So what, if the AI builds four farms on it's worlds? As long as the food cap doesn't go above 20b, it's fine.

What this means is, that there won't be any 1pp limitations for farms anymore. Except for those, that already had them, and the new Nourishment Facility of the Thalan, of course. Yes, I gave the Thalan a new farm improvement. It has the single biggest bonus (+50%; i.e., 4mt food), but also the highest maintenance cost (3bc). All other farms have a max bonus of +40% (i.e., 3mt food), a standard maintenance cost of 1bc (2bc for the Robotic Farms and the Harvester), and can be build as often as you want.

The other part of the issue, is that the AI needs to build enough morale improvements. My previous attempt was a combination of a high AI value and a 1pp limitation for the starting improvement, to force the AI to build them, while preventing it from spaming them. Once I noticed how AIP 7 handles morale improvements, it should have been obvious, that the fear of spaming was unfounded. The AI is much smarter than I give it credit for sometimes. All I needed to do was to remove the 1pp limitation, set all morale improvements to a high AI value (currently 200), and let the AI handle it by itself. I still need to iron out some kinks, but the results look very good so far.

Is my new solution perfect? No. There are still occasions where the AI isn't building enough morale improvements on some worlds, while running very high tax rates (sometimes beyond what should be possible), but that was also happening in all of my previous attempts and the vanilla game. In the end, I think it's good enough. The only way to totally solve this problem is to rewrite the AI, and that is beyond my capabilities.


As for the Yor, I made some progress with them. I still haven't managed to get them to research Efficiency Studies early on, but their morale issues have been reduced. Well, mostly because I finally added some Morale ability bonuses to their tech tree. I also added a pop-growth bonus to the Charging Stalks. Yes, it's more of a band-aid, treating the symptoms, instead of a cure for the actual problem, but it works for now.

Speaking of pop-growth bonuses, I disabled the Fertility Clinic. The AI just couldn't handle it correctly, and none of the additional bonuses I added to it worked out. The econ bonus made it too similar to the Recruiting Center and the Orbital Hospital, while the the morale bonus made it compete too much with the other morale improvements for the bonus tiles. There is also the issue of there being too many must-have improvements. By turning the Fertility Clinic into another one, it would have made it even more difficult for the AI to use it right. A planet has only so much space, after all. So, I opted to remove the building and increased the Pop-growth ability bonus of the Fertility Acceleration tech instead.

I haven't been able to come with a replacement description for the tech yet, so if anyone got some ideas, please let me know.

Beta 7 is out now.

I consider this to be my release candidate, and updated the OP to reflect this. Unless I get reports of major issues, I'm not going to make further changes. Well, except for the AI, of course. I haven't yet been able to get it back to how it was in my previous versions, so expect it to be a little weaker in some places for now.

Okay, what is new in this version?

For one thing, I added the fixed campaigns. I would greatly appreciate it, if somebody could test them. While I did so myself, there is always the possibility of a bug slipping through. Not to mention typos or grammar mistakes. I also want to make sure, that the installation instructions, I provided in the OP, are detailed enough.

I went through the planetary improvements again, and cleaned up some of the descriptions. If it is still unclear what a improvement does, drop me a line, and I'll see what I can do.

I also had another look at some of the SPs and GAs. For example, the Omega Research Center saw both a buff and a nerf. The research bonus has been reduced from +200% to +100%, but it now also provides a +20% bonus to the research ability. I cut the cost for the Brain Washing Machine and the Aul Power Center in half. The previous costs were just too high for both of them. The economics bonus from the Aul Power Center has also been reduced from +25% to +15%. This should make it less similar to the Galactic Stock Exchange. The Festival of Capitalism is now a GA instead of a SP. The cost and economy bonus have also been slightly reduced (100 -> 70; +25% -> +10%) to bring it more in line with the other starting GAs. I removed the speed bonus from the Innovation Complex, because it just made no sense, and replaced it with a +10% research bonus. I also increased the base-research from 5rp to 6rp, and added a small (+5%) bonus to creativity.

The Yor saw some more changes. I replaced the influence and PQ bonus from Biological Studies with a +10% bonus to morale. This means, that the Yor don't have a bonus to their influence ability at all. The PQ bonus wasn't really necessary, because the Yor don't need that much space to work. The bonuses from Computational Matrix, Advanced Computing, and Synthetic Ingenuity have been cut in half. In exchange, the Research Matrix now has a +10% research bonus (but also a maintenance of 2bc). I also added a new tech, Auto-Repair System v1.52, which grants a +20% repair bonus. Last but not least, I re-added the Warp Field Disruptor module. I'm not entirely sure about this change, but the Yor needed something to make them stand out more. Compared to the vanilla version, the WFD is now weaker (-1 to speed, instead of -2) and more expensive (400bc, instead of 0bc). This should make it a bit less OP. Tell me what you think about it.

As usual, for the rest of the changes, please take a look at the changelog.

Speaking of changelog, I still need to finish writing it before the 1.0 release. This will take at least another week, if not more, so you should have enough time to play-test and provide some feedback.

Thanks, great work !

Yesterday, I started to play the latest version of your mod. It seems the Korath did not colonise much, a least after ca. three years game time, they still kept only two or three planets. Maybe just chance / bad luck ?

I started a new game (this time without the Korath), so I can't give more details.

- farms now provide a percentage-bonus instead of a fixed value

The disadvantage is that this renders the bonus soil tiles useless, as they give a percentage bonus to a fixed farm output value only.

The AI is insane about using fixed-output farms on bonus tiles without regard to the impact on morale.  This makes the AI play a weaker game.  It's a bummer, but there's nothing we can do about the AI directly.

That is the point of the change. The AI doesn't know how to handle those bonus tiles properly. It uses them with no regard to the pop-limit or approval. While the problems, that arise from this behaviour, are still happening in my mod (especially in case of the Iconians), they are much rarer and easier to address.

The AI builds a maximum of four farms per planet. Only the Yor build that many. The other races usually build between one and three. With my current values, I know, that this will lead a max pop-limit of 20b and can balance around that. With a fixed value, even if it provides the same amount (3mt), the max limit can be anywhere between 20b and 30b+, due to the bonus tiles. There is no way, that I can balance around that many people, or prevent the AI from using the bonus tiles.

I wish, that I could use fixed values for the farms (at least for the biological lifeforms), but, due to the way to AI works, it just isn't possible.

Edit: Ninja'd. I should have checked, if somebody already replied, before posting. Well, that happens when you take an hour to write your reply.

I understand. For my personal pleasure, so that the farming bonus tiles will not be completly wasted, I have modified the farm improvements so that they give, in addition to the percentage bonus, a food production of "1" at all levels. I will see what happens ...

I've been busy so I haven't had too much time to play.

 

Anyways, it looks like the AI still isn't building the amount of morale improvements needed. I think that the bonus tiles - no matter what we do are not going to be improved at this point. We would need to go in and modify the AI itself for anything to happen.

 

Perhaps the best solution for the morale is to make the morale structures more potent altogether. Another solution is to put morale onto the farm buildings to compensate. However for very high yield farm bonuses, unless you want to make morale structures OP, it really is not going to solve much.

 

The AI is building enough of them. Just not on every world. That, combined with running very high tax rates, is the main problem.

Increasing the power of morale improvements is not going to solve this. In fact, it might make it even worse, because it allows the AI to run even higher taxes. The planets with enough morale improvements won't have any problems with this, while the ones without are in even more trouble than before. I could increase the AI value of those buildings again, but that isn't going to help either. If there is one thing I learned in the five years trying to fix this, then it is that the AI builds whatever it wants. You can try to guide it, but if it doesn't want to build something, then it won't. Take the Recruiting Center, for example. I'm using an AI value of 1000 for it, which is the highest value I'm using (only the Orbital Command Center has it too). Yet, the AI still isn't building it on all it's planets.

In short, only an AI re-write can truly fix this problem.


The AI already makes poor choices when it comes to the morale bonus tiles (or bonus tiles in general), so I'd rather not add another pitfall. It's bad enough that I re-added the morale bonuses to the Graft of Ages and the Space Cannon. Besides, a small bonus won't help much. In the best case, it doesn't change anything. In the worst case, it allows the AI to run higher taxes. Then we have the same problem as with more powerful morale improvements. Using a bigger bonus will only make this more apparent. Not to mention that it would make farms too similar to the Charging Stalks, and I'd rather not infringe on one of the Yor trademarks.

EDIT:  LOL I just saw that the post right in front of this replies to the morale+food idea.  It still seems like it might be worth pursuing considering how much trouble population growth and morale causes the AI.

 

I'm late to this conversation but I've read the thread.  I've played and "lurked" on and off for years so I'm not new to the game.  I'm just about to try version 7 and thank you for making it.  Before I go play I've got a couple ideas to throw out there.

--Could farms be made to add morale as well as increase population?  I think the Yor already do this in vanilla or some mods.

--Perhaps the morale increase could be related to the population increase?...perhaps the morale could receive a multiplier based on whether the space has a fertile bonus on it the same way the population limit does?  Some combination of these ideas could solve the AI problem with growing population but not keeping them happy.  Pure morale buildings would still be useful but it wouldn't kill the AI so much if they didn't get around to them.  It would also allow usage of the fertile bonus spaces on planets as originally intended.  The storyline "rationale" would be that fed people are happier people.

 

--Thalan Industry and Research. I know their buildings are supposed to have the exact same output, only changing the cost to maintain them but would it help the AI into thinking the upgrades were attractive if the buildings "slightly" improved?  Perhaps the first building adds 15, the second 16, third 17?  (or first building 14, second 15, third 16).  The gains would be small enough to not alter output greatly but would be worth it if the AI started using the upgrades.  Maybe the Thalan wouldn't fall behind late game.  The storyline "rationale" for the slight change could just be that the upgraded maintenance efficiency also causes slightly more output efficiency.

 

--It's been mentioned that the AI economy gets hurt by building upgrade costs while players just skip upgrades until practical or until they reach the top level building, then do the upgrade just once.  We can't stop the AI from upgrading each time even if their economy can't handle it but could the costs be additive?...in other words not possible to skip the upgrade cost at each upgrade?  Then players wouldn't have the advantage of skipping these costs either.  In other words, either a building couldn't be upgraded to level three until it first upgraded to level two OR the cost to upgrade two levels at one time be made twice as much?

 

Just some thoughts.  They may already be obsolete or not attainable but better to throw them out there and see what shakes.

Hey Gaunathor, why do you have to do the quicksave fix for the minors on this and prettymuch every other smart minors mod?

 

I never understood it...for instance, what does it change???

EDIT:

Also, have you adressed the Market Center becoming Terran only by either removing the requirements or cloning it?

An interesting idea, but, sadly, not possible. All game mechanics are hardcoded. I cannot change them.


The answer is apparently "no". I already tried this, despite not being very fond of the idea, but the AI kept behaving like it always did. Maybe I should have tested it a bit more, but the apparent lack of any change in behaviour at all made me abandon this approach pretty quickly.


The main problem is, that the cost of the buildings are extremely high, the further up the tech tree you go. I fixed this by significantly reducing the cost of the improvements, and using a fixed value for the cost-increase (instead of big jumps, like it is in the vanilla game). This had also the effect of making it easier to build up new planets, later in the game.
As for skipping upgrade-costs, this is already addressed by the game itself. The value of the upgrade-cost is the difference between the cost of the original building and the new one. At least as far as I know. The time for each upgrade sometimes seems to take more (or less) time than it should, but I haven't yet figured out why. Plus, it is again something I cannot change, even if I knew why it is happening.


The minors still use their old AIP at the start of the game, despite it being set to a different one. The quicksave fix forces the game to update the AIP to the correct one. If you don't do this, the minors will not be able to colonise planets, and pretty much behave like they always do. Well, until the first time you load the game, that is. All the other changes to make the minors smarter work even without the quicksave fix.


I changed the requirement from NeutralHistory (the InternalName for A History of Extremes and A History of Perseverance in the vanilla game) to my new Capitalism tech, which I added to all appropriate races. The minor races have their own version of it, because they could loose access to it, depending on what major races your are playing against. I also had to make a copy of the Market Center for the minors, due to this, because the game doesn't like it when an improvement has tech requirements from two (or more) different tech trees.

Thanks for the quick response Gaunathor. I know you've tried all sorts of work a rounds to get past hard code oversights.

Thanks for the reminder about saving and loading a game to get minors working as intended.

I was looking through some tech trees and noticed Stellar Cartography is located past an expensive tech in the Yor tech tree. It only costs 20 but can't be researched until a 187 tech is researched first. This effectively puts it out of reach for quite some time for any custom race. Is this as intended? The base Yor race starts with it and based on it's cost as well as how many other races have it to start it seems like it was intended as a possible starting tech.  It's a relatively minor thing.  Just wondered if it was intentional.

Do you have any idea what the Yor techs "Anti Virus Software" and "Firewalls" were supposed to do? I'm assuming they played some part in the campaign (which I've never touched) but they've always been empty on info and never done anything when playing a custom race. They just drain 20 points from choosing other starting techs.  I still take them out of principle since they seem like they'd fit when playing a Yor based race but essentially they are just a point drain. Cybernetic History and Expert Robotics on the other hand must be taken or the Yor will start with no buildings available to build and never get an option to research factories of any kind.  I'll look closer at Thalan after the Yor.

 

Btw, I really like your balances to the starting ability choices. Some of the vanilla choices were not balanced well.  The color changes to the races are also good improvements.

 

Time to start playing version 7 for me!  Will check in after a while and thanks again. 

Yes, it is intentional. In the GalCiv lore, some of the major races had already mapped out the galaxy, which grants them a headstart in the colony rush.

The real reason, though, is a bit different. You see, in GalCiv 1, the AI actually knew where all the good planets were, and that part of the backstory was created to justify this knowledge. Once you researched Stellar Cartography yourself, you could see how many planets a star had, and if one of those planets had a high PQ. In GalCiv 2, however, things have changed. The AI no longer knows where everything is, and needs to scout the map, just like you. Stellar Cartography still tells you how many planets a star has, but no longer if a planet has a high PQ. It is still useful, but not as good as before. I would put it into the "nice to have, but not essential" category.


They are the cures for the Plague and Disease Mega Events respectively. If you have Mega Events enabled, there is a chance for either one to appear. The Plague will reduce your abilities to zero, while the Plague will slowly kill your population. The same is also happening to all the other races. To solve this, you have to research the cure, which will appear in your tech tree. However, both cures have Xeno Medicine as a hardcoded requirement, which the Yor don't have. This means, that the Yor are effectively blocked from ever researching the cures themselves. To fix this, the devs decided to make the Yor start with the cures already researched. If they had just put the cures into the Yor tech tree, the AI would immediately try to research them, which would be crippling in the early game, due to the cost (1000) of the techs.


Thanks, that is good to hear.

Well, here it is: version 1.0.

If you already have beta 7, then there is no need to download v1.0, as I made only some small tweaks since then. (Xeno Mysticism now uses Pure Research as Category instead of Culture, Divergent Evolution requires Precursor History again instead of Xeno Biology, the Altarians now start with Stellar Cartography (while the Korx and Thalan no longer do so), and I made some changes to the Espionage entries to make them easier to read)

From now on, improving the AI is the main focus. However, I'm not ruling out any further changes to the tech trees and other areas, because I realised how futile that is. Back when I released the first beta version, I thought that I was already done, except for the AI. Maybe a few tweaks to the balance, but that would be it. Going through this thread, you can see how true that turned out. I'm not going to make the same mistake yet again.

Realistically at this point, I think you've done all you can given the limitations.

 

What we really need is a code re-write of several portions of the AI. That some of the AI's have to be fixed altogether.

I'm not so sure. The AI can still be improved, within certain limits, and I'm also considering to make some changes to the Korath-dialogues. They are currently using the generic entries, which feels a bit weird.

However, before I do that, I'm going to take a break. I'm quite exhausted at the moment.

Gratz on research V1!

Long time lurker, first time poster. Just wanted to drop in and say thank you for all the amazing work that's gone into this mod. I've been tinkering with Galciv2 mods for a few years, but the depth of the changes and testing that has gone in Autumn Twilight is really impressive.

This weekend I started playing V1.0, along with some of my standards tweaks (e.g. increasing colony module cost significantly to slow down the initial rate of colonization). Having a great time so far, and the behavioral changes of the various AIs is quite interesting. I'll need to get some more game play in before I have a good sense of how they've changed, but so far they definitely seem quite a bit more confident in early game play.

Thank you Gaunathor, and everyone else involved. You're why I'm still so excited for playing Galciv2 after all these years!

Maybe this is already known but I ran into some weirdness with upgrades. Upon gaining Collective 1 technology my Basic Collectives began upgrading as expected.  However their cost for the upgrade was over triple the cost of just building a new collective.

--The new Collective 1's can be built for the same industrial output as the Basic Collectives before them.

--The current Basic collectives with the high upgrade price can also be demolished manually and Collective 1's can be built in that spot for the original cheap basic price too.

 

I don't know if the building price of the Collective 1's (industrial output 7) is supposed to be higher than the Basic Collectives (industrial output 5) but currently it isn't, except when leaving already built structures to auto upgrade themselves.  Either the auto upgrade price is messed up or the cost of the Collective 1's is messed up.  I've not checked whether the same problem is present for Collective 2 and beyond yet.

There is a problem with auto upgrading. The upgrade sometimes costs more (or less) than it should. However, that shouldn't be happening with the first upgrade.

In fact, I just tested this and it worked fine. I kept the production sliders how they are at the start of the game, built a Basic Collective on Iconia, and gave myself Collective Manufacturing I. The upgrade takes two weeks, while building a new Collective takes eight weeks.

As for the build cost, the Basic Collective has one of 45, and the Collective has one of 65. Both are correct in the PlanetImprovements.xml. If your production output is high enough (quite easy with the Yor), the time it takes to build either one is the same (1 week), so that may be why it seems like the Collective is cheaper than the Basic Collective.

Still, I'm not sure what is going on with the upgrade cost. It doesn't make sense. Is it happening on all planets, or just some of them? Also, did you make any changes to the files?

 

 

I'm trying to determine what triggers the bug.  I ran into the bug after waiting six months into the game but have not been able to recreate it at the start of a game so it's difficult to quickly recreate. I'm using the Yor tech with Super Hive ability so the costs are different but appear to be in line (Collective 1 is about 50% more expensive than Basic Collective).  I know the exact cost of each structure and the auto upgrade was way out of line as I mentioned before.  You said auto upgrade can have weird cost fluctuations in some instances. I will update if I can find a consistent trigger.

Out of curiosity, is there anything that can be done about certain random events like the Peacekeepers? I ended up in a game where the Peacekeepers were very potent towards the end of the game and it brought a partial stalemate.

The Peacekeepers, like any other Mega Event, are hardcoded. We can't do anything about them, except for changing parts of the text. That is the only thing we have access to.

Out of curiosity, what races are you finding are underperforming right now?

 

I've had issues with the Korath, sometimes the Yor, the Arceans, and the Thalans chronically underperforming in some of my games.