Dune Wars

[Ahriman]

What do you think should happen?

(Religion + inquisition)

I have no strong feelings, I can see a good case either way.

Under the status quo (buildings stay), we have the problem that you can get a religion a city, construct the buildings for it, then lose the religion (and lose the unhappiness penalty from nonstate religion) and so be better off.
Also, there is a basic logic problem; if you don't have religion X in your city, why should you still be able to get benefits from it?

Under the alternative (buildings go), inquisition is potentially too powerful, particularly for wiping out cathedral-type buildings. It also makes Quizarate too strong, and might make Imperium too weak. It does increase the value of spiritual and political straits though.

A possible compromise might be some mix of making buildings destroyed if the religion is lost, while:
1) Increase the cost of inquisitors (and quizarate missionaries)
2) Increase the value of the Imperial temple (its already pretty weak, maybe bump up its culture yield?).
3) Decrease the cost of the cathedral buildings (too expensive to bother building if they are easily destroyed by inquisitors).

It is possible that either of those, or both together, has caused the AI to lose interest in spice. I will look into that, thanks for pointing it out.

Interesting. I have not tested yet, but that would be an important find.
Its possible that the corporation mechanic directly encouraged the AI to try to acquire multiple copies of the same resource, whereas this is missing once we just give the bonus from a building in a way the AI can't see?
Does the diplomacy value on the resource increase the AI's tendency to try to build improvements to get it?
Or it could be the worker AI, which is probably more likely.

On slaves: I haven't tested, interesting feedback.
Allowing slaves to have a ~10 hammer mini-great engineer would help reduce any excess slave problem, if we could somehow get an AI to work for it. What happens if the unit gets both great engineer AI and worker AI?
Part of the problem seems to be that you are in the very late-game, with no more improvements to build. This happens in vanilla too; workers are pretty useless in the very late-game.

The city with 20+ slaves I captured from Fremen had maybe a dozen defenders but there was also a lot of thopters and such flying around

Odd.... I thought the slave capture chance was ~50%, so it should require 40 units to get 20 slaves.
I would have no objections to removing thopters and suspensors from slave generation (these get shot down; no prisoners).

 

[davidlallen]

The city with 20+ slaves I captured from Fremen had maybe a dozen defenders but there was also a lot of thopters and such flying around. In fact, there still are, and provide for more slaves - screenshot attached.

Sorry, my question was not very clear. A Harkonnen unit has a 50% chance to generate a slave on a combat win. So, you have 20 slaves there. When you first approached the city, I assume there were no slaves there since only Harkonnen can create slaves. So, where did the 20 slaves come from? It must have been from 40 combats on average.

 

[Ahriman]

Another thought on religions; another possibility (to prevent temple-whoring) is to make the various temples require *state* religion, not just religious presence.
Or maybe do this for the various cathedrals and/or missionaries (while leaving the temples alone).

I dislike the strategy favored by the current (and by the vanilla) religion system where you want to get as many religions as possible in your cities, so that you can just build all the different temples and cathedrals.

[Problem with doing it with missionaries; no-one will adopt a newly founded religion when it only has one city, so late-game religions would never spread.]

 

[embryodead]

Sorry, my question was not very clear. A Harkonnen unit has a 50% chance to generate a slave on a combat win. So, you have 20 slaves there. When you first approached the city, I assume there were no slaves there since only Harkonnen can create slaves. So, where did the 20 slaves come from? It must have been from 40 combats on average.

Those slaves don't come just off city defenders but all the units around that I killed just before and after the capture, e.g. this endless swarm of thopters. That screenshot is from several turns after the actual capture. All in all it was much more than 40 actually (as I said I also lost many slaves of those created 1-2 tiles from the city).

 

[davidlallen]

OK, thanks. By coincidence one of Ahriman's requests would reduce the problem; if thopters and carryalls and hornets did not generate slaves. This is certainly reasonable. But, the main part of the problem seems to be that the city had 40 + units defending it and you were able to conquer it. Given 40 + defenders, I cannot see any good reason *not* to generate 20 slaves.

What difficulty level are you playing, and how far along in the game is this happening? In games I play or autoplay (on Noble difficulty), I rarely see as many as 10 defenders. So 40 seems like an awful lot. What are the sizes of your attacking stacks?

 

[ThaDoews]

Really great mod! I enjoy it alot THX very much

 

[davidlallen]

Thanks for the feedback! Please let us know any suggestions you have to improve the mod.

 

[embryodead]

What difficulty level are you playing, and how far along in the game is this happening? In games I play or autoplay (on Noble difficulty), I rarely see as many as 10 defenders. So 40 seems like an awful lot. What are the sizes of your attacking stacks?

This is Emperor difficulty, Large Arrakis map w/ default map settings + NoVassals & TechDiffusion. The screenshot shows turn 311, the only techs I have left are the nuke ones, golden path etc. I attacked the Fremen to stop them from winning by Terraforming and had no problem with conquering their cities (4 so far) since my attacking force is also around 40 top-tech units and while Fremen may be numerous, they are in their middle ages. Hard to lose when you have those top sword dudes, devastators etc. each w/level 5+ and Fremen defend with heavy troopers and naib's chosen, at best.

It does seem too easy, however, I'm pretty sure the AI's lack of interest in spice is at fault, because I was behind in both score and techs for most of the game, struggling to defend from Tleilaxu who almost thrashed me with superior units and plague. That was until spice production difference became too big. For instance I've built Guild Research Facility + Academy and my research in Carthag went up from something like 150 to 400. In general, commerce-related wonder stacking in the capital and 50+ spice quickly put me over everyone else.

On a related note, what would you suggest as optimal game settings for DW? I disabled vassals because I didn't want to see silly combinations like Fremen being vassals of Ix or w/e and figured TechDiffusion helps the AI mostly (but it helped me early).

 

[Ahriman]

Tech diffusion is not intended for the DuneWars mod. It makes tech progress run far too fast, particularly because the tech-tree is much "wider" than the fairly linear vanilla tech-tree, and so even tech leaders will benefit from diffusion on lots of techs that they haven't researched yet. It also has the net effect of increasing average effective research rates globally.

It sounds like you were sufficiently ahead of everyone else that you had already won the game long ago. Balance isn't really designed to be dealing with the after-you've-already-won type situations - you could have won at that point even without the slaves. If you have a huge army of 40 units, you shouldnt' really be worried about getting a bunch of extra slaves. The game is always going to be a bit lame at that point.
Also, the very late-game is currently weak, and we are working on a redesign.

If the AI isn't building harvesters, then yes that is a big penalty, and we need to figure out what is going on.

I recommend:
Arrakis map, 24% land area, standard size, 8 players, Emperor or Immortal difficulty, no-tech brokering, Epic gamespeed.
(I play standard speed for most mods, but I find Epic works better for Dunewars - you need units moving faster relative to tech speed and such because of the large distances and lack of roads/railroads).

This used to be fairly challenging, at least when the AI used spice correctly.

Its intereseting that you're reporting that spice is too strong. In some sense this is an issue of map size. The larger the mapsize, the more powerful spice will be. I think we should aim at balancing for standard size.

*edit*
Another thoughth; maybe the problem is the Guild Research Facility?
Maybe we should nerf this - 2 beakers per spice resource, or less?

 

[embryodead]

Thanks for the suggestions regarding game settings.

As for the spice, many values in CIV are modified based on map size. If map size throws the balance of spice yield off, maybe it should be treated similarly? The current settings: +3 Commerce / +3 Beakers / +3 Gold per spice from Palace / Wonders and 0.15 per spice from Silo would be the standard map setting. Large map could be +2 / +0.10 and Huge map +1 / +0.5 etc.

 

[davidlallen]

Its intereseting that you're reporting that spice is too strong. In some sense this is an issue of map size. The larger the mapsize, the more powerful spice will be. I think we should aim at balancing for standard size.

Why should this be related to map size? If the map size and the number of players scale together, then the area per player is the same, and it seems to me the strength of spice should be the same. Of course if you fix the map size and vary the number of players, the area per player goes up or down, and I suppose a lot of things change then.

 

[Ahriman]

Of course if you fix the map size and vary the number of players, the area per player goes up or down

I find that larger maps tend to end up with more area per player.
Map area (total tiles) increases faster than number of players (normally).

I'm not sure that map size is important enough to worry about though.
If you get big enough on any map that you control 60 spice, then you are clearly the dominant player and winnning will be trivial.

The main issues are probably:
a) AI not building harvesters (I still haven't had a chance to test this properly, some auto-runs are probably necessary)
b) guild research facility was too powerful, and maybe a few of the other wonders?
Prescience chamber should probably also be +2 espionage per spice.

 

[davidlallen]

a) AI not building harvesters (I still haven't had a chance to test this properly, some auto-runs are probably necessary)

Auto-runs by themselves will not tell us the answer. The AI is building "some" harvesters. The question is whether it is building "enough", that is, are there insufficient workers or is the AI choosing to build something else instead which is less useful? What we need is a side by side comparison of a human player vs the AI.

 

[embryodead]

If you get big enough on any map that you control 60 spice, then you are clearly the dominant player and winnning will be trivial.

That wasn't true in my game - before I got to the top, my territory was roughly the same as other major players. Some AI players could easily harvest 40+ spice like me but they harvested 5 - like I said, I would peek at enemy territory and see clusters of 15 or so spice with just one harvester active. I can provide saves if needed, I'm not sure if I have ones corresponding to that breaking point in particular, but I have 10 different saves from various points of the game.

 

[Ahriman]

AHR95
Spice silo is massively over-priced. The benefit is small; it should be very cheap.
Maybe 50 hammers?

 

[Ahriman]

AHR96
Something *is* bugged with worker AI I think.
Spice workers on automate do not reliably build harvesters.

Automating a spice worker in a city had it just sit there, even when there were spice tiles available in the cities BFC without harvestesrs.

Automating a spice worker on a spice tile (without harvester) had it move to the nearest city and sit there.

* * *
I'll also confirm that the AI isn't building enough harvesters. Unclear whether or not this is entirely from worker AI, or whether it also isn't building enough workers.

* * *
Spotter control is overpriced/underpowered I think.

 

[davidlallen]

Automating a spice worker in a city had it just sit there, even when there were spice tiles available in the cities BFC without harvestesrs.

That is weird in two ways. After thinking about it more, the AI does not realize it will get any income from harvesters outside the BFC; the AI accounts for corporation income, so it used to work. I did not change the AI to look at the new system of buildings directly generating yield, so that is probably the reason.

It is a little more weird that the AI won't build a harvester inside the BFC. In this case, it still has to see the regular commerce income from working the tile. Maybe it just decides that the regular commerce income alone, is exceeded by some other use for that population point, so there is no point in building the improvement.

I will study how corporation yields affect the decision about building improvements, and copy that for building yields.

 

[Ahriman]

the AI accounts for corporation income, so it used to work. I did not change the AI to look at the new system of buildings directly generating yield, so that is probably the reason.

I agree that this is a very likely cause.
Even if it is insufficient workers, these could be related; the AI might build more workers if it sees lots of unconnected bonus resources that it puts a value on.

Maybe it just decides that the regular commerce income alone, is exceeded by some other use for that population point, so there is no point in building the improvement.

It could make sense for a worker to think that some other tile was worth improving over a harvester tile.

It does not make sense for a worker (or spice worker) to think that not building any improvement at all is better than building a harvetser. And a spice worker *can't* build anything except harvesters.

The automated worker was (sometimes) going back into the city and staying there, in the same way that an automated worker does when there are no more valid targets. (Eg: in vanilla if you set a workers to "improve trade network", and every single tile in your empire has railroads and bonus improvements (pasture/mine/farm/plantation/etc.) built, then that automated worker will return to a city and stay there until a new mission becomes available.)

So the "return to city" effect means the AI doesn't see any more worker missions worth building.

 

[davidlallen]

So the "return to city" effect means the AI doesn't see any more worker missions worth building.

I agree. Please keep in mind that in general, an improvement in the BFC which is not worked, does not contribute anything. (Exceptions such as linking a resource.) So my point was only that the AI is probably thinking the population point to work that resource is probably better spent somewhere else. Rather than tying up the worker on an improvement it can't use right now, it leaves the worker idle.

It probably isn't worthwhile to dig into that any further; let me make the AI place the same value on spice from a building, as it used to place on spice for a corporation. Then all the problems may go away.

 

[Ahriman]

Rather than tying up the worker on an improvement it can't use right now, it leaves the worker idle.

I would be surprised if that were what was going on; that would be very bad worker AI.
And I have not observed that in vanilla or any mods for normal on-land terrain. I have only ever observed the idle worker when there were *no* valid improvements that could be built.

The AI always keeps building the "best" improvements possible on all of its BFC tiles, whether or not the city is big enough to actually work the tile or not.

The vanilla worker AI will prioritize tiles that need to be worked, or improvements that the nearby city is short of, but once its done with those, the AI will build improvements on other BFC tiles. It will never decide that it is better to be idle than to be building improvements - even redundant improvements.

I think that there could be some separate issue for ocean tile worker AI going on here.

let me make the AI place the same value on spice from a building, as it used to place on spice for a corporation. Then all the problems may go away.

I completely agree that this is the first thing that should be tried. Hopefully this will be sufficient.