Discussion in 'Civ4 - General Discussions' started by Bhruic, Oct 5, 2007.
As you should know by now, I can't fix it without a sample
My remark was about how railroads work: they use 1/10 th of the total movement points of a unit for each step a unit takes from a railroaded tile to a railroaded tile and that has worked this way since vanilla civ 4 1.00. The civilopedia description has always been wrong.
They made a change for gunships and other fastmoving units (more than 4 moves, maybe modded) so that they can take the best movement rate of roads or railroads because it was weird that in vanilla civ4 1.00, the gunship moved slower over a railroaded tile than over a roaded tile.
I vaguely recall that originally, the gunship didn't even use road movement bonuses, but I could be wrong about that. It's quite some time ago.
I read a post that you've posted in this thread a while ago which suggested that the AI couldn't spread a corporation among its cities very well because it would have spend all of its money before it came to the phase of the turn where it would do such moves. (at least that is what I understood) Would the AI be much better at spreading corporations when the gold cost of spreading a corporation was removed? I'm considering doing that as a list of my personal modifications to the game (in this case to improve the AI at this part of the game). I just need to know if it might be effective.
(I consider the costs of spreading a corporation very minor compared to the income of a late game empire and compared to the building costs of corporate executives. It would barely change the game for me, but it might improve the AI.)
Absolutely. That's the reason that it ends up having so many executives sitting around idle, it can't afford to use them. But because it has cities without the corporation, it still thinks it needs more. Remove the cost to use the executives, it'll be able to use them, and that will stop it from building so many.
Do you work for Firaxis? I heavily disagree with your "worker AI" points. In the game without Bhruic's patch, the worker "defense AI" isn't just bad - it's nearly non-existent. The opponent AI literally cripples itself by not taking care of his workers and letting them fall into the hands of the enemy. It's not just a 1 in a hundred thing either. I could grab AI workers 50% of the time. Not only that - my workers on auto-improve were too dumb to run for cover about 1/2 the time. My conclusion is you haven't played the game enough if you haven't seen this problem in all its glaring inadequacies.
Yes. AI should never be perfect and there's a delicate balance a programmer must do to prevent a Chess-like AI strategy. But worker AI is NOT the place to introduce these random variables. Brad Wardell at Stardock has spoken about the challenges of making AI excellent, not perfect, and not stupid. He does a great job. There are places you can introduce random variables for unpredictability and failure. And like I said, in Civ... it's not Worker AI.
Interesting. What cost are you referring to? The cost to build them? I haven't used them enough to konw all the applicable costs. What file & line are you editing?
He's referring to the gold cost of establishing the corp. in the city.
There's a gold cost for every city? I never paid attention since I put them on "auto".
Instead of cutting the price to zero, would it be better to try it at a discounted rate (50%? 75%?) first, and seeing how that works? Does the AI always spend ALL of its available cash?
Just wondering if zeroing out the price might give too much of an edge to the AI.....kind of goes against the idea of minimizing AI cheats unless absolutely necessary.
A discounted rate isn't likely to help. The sample games I was using generally involved the AI spending all of its gold either upgrading units, or rush buying. If the costs worked out, they occassionally had less than 10 gold at the end of their turn.
you could just cut out the cost full stop and increase the cost of the executives to build?
The reason why I was considering such a change (removing gold cost from spreading a corporation) for my personal mod was because it would improve the AI. But the change would be for both the human player and the AI, so it's not a handicap bonus for the AI.
At present, spreading a domestic corporation costs (at epic speed) 150 hammers (for the executive) and 50 gold (for the actual spreading). Since gold is much easier to come by at this stage of the game, the gold cost is almost inconsequential. You might not even notice it. At the same time, as Bhruic has told us, the AI uses all of its gold for other things and regularly doesn't have the measly 50 gold left to spread a corporation (probably less of a problem when the AI is not using the universal suffrage civic). So while it is an inconsequential cost for the human player, it is a major obstacle to the AI.
Since the gold cost of spreading a corporation isn't a major game defining aspect, I think I will remove it. I for instance don't think anyone would be complaining about the implementation of corporations if Firaxis had initially designed them without such a spreading cost.
Oh, of course it would be better if the AI could be improved. No need to discuss that.
The cost can be found in the file ...\Civilization 4\Beyond the Sword\Assets\XML\GameInfo\CIV4CorporationInfo.xml and the value controlling it is named <iSpreadCost>50</iSpreadCost>. This cost is set for each corporation independently.
Disclaimer: This is a change for myself, so it won't affect Bhruic's patch or any of your games.
Ah, that makes sense. Can the AI do purchasing of hammers for partial construction of buildilngs and units -- as it could do in Civ II and SMAC? (As opposed to being limited to making a full purchase of the building/unit.)
@Roland: I've noticed that the costs for spreading a corporation for me are usually around $100 -- with the costs of spreading to foreign cities hovering between $150 and $200. Being able to spread corporate branches without costs does represent a significant change from the intended rules.... What would happen, for instance, if a "buy out" was attempted -- like if a Cereal Mills executive tried to replace a Sid's Sushi franchise? (I had this situation occur in a recent game....)
The cost is 100 for spreading a corporation in your own lands? Hmm, are you playing at marathon speed? Then the building cost of the corporate executive is also a lot higher (300 hammers I guess) and thus the founding cost is still relatively low.
You are right that spreading a corporation in foreign lands is a bit more costly. It's a multiplier to the normal cost (set in globaldefines.xml), so if you set the domestic spreading cost at 0, then the foreign spreading cost will also be 0. I think that buying out another corporation is also just a multiplier, but I don't know where it is stored. I agree that these multipliers are a good game mechanic to make switching between corporations expensive and to make spreading corporation to foreign lands expensive. I still think that even for these more expensive missions, the hammer cost to build the executive is the main cost factor.
I personally don't think that removing the gold cost of spreading a corporation will change gameplay a lot. I think that the fact that the AI can't spread corporations effectively is hurting gameplay a lot. So I choose the lesser of two evils.
Or reinstate the limit on the number of Corporate Executives. I still don't see why the limit was apparently such an onerous burden. I never found it to be a burden at all, because I would set up a production queue such as:
If building a Mech Inf doesn't provide enough time to allow those Corporate Execs to travel to the nearest destination city, then add a little more to the queue.
This will stop the AI from building a massive number of unused executives. But it doesn't allow the AI to actually spread the corporation. The massive number of unused executives isn't the core of the problem, it's just the result of the AI being unable to spread the corporation. You can take away the symptoms of the problem (massive number of unused executives), but that doesn't really solve the core problem (AI unable to spread the corporation).
While we're on the topic of movement/pathing bugs, I observed one a while ago. Unfortunately I don't have a savegame file, but maybe if I describe it accurately you can reproduce it.
Here are the tiles in question:
x B O
A C x
where A = tile with Road
B = tile with Forest but no Road
C = tile with Road
O, x = any tile
Group a Woodsman 3 Warrior and a plain Warrior, both with full movement points, at A. Send them to O. The movement indicator correctly says "1", but it takes 2 moves when they are grouped. It's because they're forced to stick together. The Woodsman guy is first in the stack and therefore heads straight to B, dragging the normal guy with him. Then the normal guy is stuck, and since they're grouped, nobody else moves, and thus they're stuck at B.
Is there a sort of priorities list for what the AI spends its money on every turn? You could simply move up building executives to make it more important. It won't interfere with the other priorities that much (e.g. buying units, buildings, deficit research etc) since as you said founding corporations is not that expensive anyway. If there was a way to move it up in the code I bet that would fix it.
a simple question (as I don't wanna read the 41 pages of this topic)
i've installed the official patch 3.13, do I need to install this unofficial BTS 3.13 patch? the two patchs work well with eachothers?
This unofficial patch patches the official patch. So yes, if you want Bhruic's awesome fixes, then you need to install the 3.13 official patch and then this unofficial patch.
Separate names with a comma.