C3X: EXE Mod including Bug Fixes, Stack Bombard, and Much More Release 24

[Flintlock]

If a caravan costs 100 shields and a city is only 99 (or 24 at a transfer ratio of 75%) shields short of completing a building, does that mean that the AI will not decide to use the caravan to speed up production at that moment and will instead send the caravan elsewhere? When is the decision made on how to use a caravan? I am referring to cases where, before the caravan arrives at its destination to speed up production, the building may already be nearing completion and will therefore not meet this condition ("it tries to get the full shield value from disbanding,"). If the condition is not met, will the caravan wait in the city or leave for another location, if one exists?

The caravan AI knows that shields that overflow the box are wasted. It won't necessarily avoid any particular city because of overflow, rather it deducts the overflowed shields from the estimated value of disbanding there. It tries to get the most value it can by disbanding, which naturally leads it to avoid wasting shields on overflow. However, it will waste shields if there is no better option.

The AI does not save caravans, it tries to spend them immediately. I think it would save them, though, if it's impossible to get any benefit from them but I haven't tested that. It searches for the best city each turn, so it can waste time by running to one city then seeing that there's a better option on the other side of the empire, and then running there to see yet another better option elsewhere. I've seen caravans waste 10+ turns doing that in my test, but eventually they find somewhere to disband themselves.

If they use a terraform AI strategy, does that mean they will respond to threats by hiding in the city like ordinary workers? Are they able to assign a defensive escort to themselves like artillery?

Caravans don't really use the terraform AI strategy. They just have to have it set as part of how the mod identifies caravans. The caravan AI does avoid threats. The logic for that is based on the base game's threat avoidance logic for the leader AI. It works but there's a lot of room for improvement. Caravans can't assign themselves escorts and the AI sends them out without any.

On one side, the capital city with no corruption, and on the other side, the city with the highest level of corruption. All cities are connected by railroad. Then I would create 200 caravans. Let's assume that all cities are building something that can be accelerated by a caravan. What behavior can we expect from the caravans?

The caravans would contribute to the capital until doing so would waste too many shields, then they would start contributing to the other cities in order of increasing corruption.

If the optimal number of cities is 10, to how many cities will the AI never send a caravan? Or rather, what percentage of corruption is classified as highly corrupt cities? In my opinion, the AI should strive to transform peripheral cities, which are only a burden on thriving cities. And the sooner it manages to eradicate corruption, the better and more effective the AI will be.

The AI counts a city's corruption rate against the value of disbanding there if that city is constructing a building. In other words, disbanding in an 80% corrupt city is estimated to be worth 80% less than disbanding in a completely non-corrupt one or one that's building a unit. The idea there is relatively simple, if a city has 80% corruption, building for example a library there is worth 80% less because the library's effect only applies to the city's non-corrupt commerce. Peripheral cities would be an even larger burden if they were sucking up caravans that would otherwise go to more effective infrastructure in core cities.

If I place all the caravans in one place, will they all go to speed up construction in one city, and once the building is finished, will they go to speed up construction in another place, or will each caravan go to a different destination to speed up construction?

Caravans do not coordinate with each other, so a large stack of them would all go to the same place. The AI can't transport caravans between continents. It can't even transport workers between continents, as far as I know.

Are caravans immediately assigned tasks, or can they be kept in reserve (caravan escrow amount = 1 per 5 cities)?
These caravans in reserve should be specifically focused on accelerating the construction of buildings that reduce corruption in cities with high levels of corruption. Is it so difficult for AI to recognize that a city has high levels of corruption and therefore designate it as a city that needs caravan assistance when it builds a building that reduces corruption?

There is no reserve for caravans, the AI looks to spend them immediately. It is not difficult to program the AI to prioritize its caravans toward corrupt cities building corruption reducing improvements, but I doubt that's the best way for it to use them. Infrastructure in core cities is worth so much more than in the periphery. If I were to try to make caravans focus on city builds that are especially urgent or valuable, that opens a whole can of worms that ends with second guessing all of the AI's production choices. It makes more sense for caravans to assume that shields are shields and that all city builds are approximately the same value. Improving the AI's production choices is another project. It's true that means the AI won't take advantage of what makes caravans strategically interesting for the human player, but the Civ 3 AI fails in so many ways and suboptimal caravan use will be one of its most minor issues.

Yes, an elaborate caravan AI is something that should wait for Lua.

I consider this whole point to be highly challenging and therefore more like something from the world of dreams, but I can't help but ask about its feasibility?

It would be a relatively big job but I don't see any reason this wouldn't be feasible.

Is it true that the number of shields also tells the AI how much to value a given unit? If so, the loss of a caravan worth 400 shields during the war would be far more painful in the eyes of the AI than its actual value as aid.

I don't know if the AI considers the shield value of units in losses during war. I'm pretty sure war weariness depends only on the number of units lost and whether they were lost in foreign territory. The only place I know of that the AI considers the shield costs of units is when deciding how many escort units to assign to naval transports. Transports with more expensive passengers get a larger escort. (However, that almost never matters in game because of how it was balanced. The AI almost always ends up assigning three escorts to each transport.)

I am playing a test game with the next version of RARR and that version modified by starting with 31 civs and the C3X option "Barbarians can capture cities" enabled. To my surprise the Barbarians were able to capture big parts of a continent and are the "player" who eliminated the most civs in the game. It is a pitty, that I cannot spy into those Barbarian towns.

Wow! Can you use a save editor to turn on debug mode and see what the barbs are up to? I think I could do that through C3X by inserting & running a bit of logic but I've never actually tried.

 

[Civinator]

Wow! Can you use a save editor to turn on debug mode and see what the barbs are up to? I think I could do that through C3X by inserting & running a bit of logic but I've never actually tried.

Unfortunately I have absolutely no experience with save game editors. I attache the current save file of that test game and the scenario.c3x_config file (version R20) to this post. If the modified 1.9 RARR biq should be needed, I will send it to you by a pm (I don´t want to post it here as I want to do several changes and additions and this would trigger confusion for some civers who would download that work in progress and later mix it up with the released new version).

In that test game I had a terrible start haunted by early waves of disease in floodplain terrain, especially when my capital reached population size three. In one case my capital in two following turns was reduced from size 3 to size 1 if remembering well. Flintlock, as you found out that tech number 8 (in RARR Mathematics) in reality cures the disease of some terrains, have you found out how the scalable strength of disease is working, too ? Does it make sense to scale disease down to only 20 instead of the current 50 in the game or is it only working in steps of fifties?