Railroads

[Koshling]

I just researched Railroads and a few thing struck me:

1. My workers still cannot build railroads (the route). I can't see why not. Anyone know?
2. ALl the buildings associated with railroads require steel wares, which are not available until a number of technologies LATER then railroads. This makes me think that railroads is a bit misplaced in the tech tree
3. One exception to the previus point - I CAN build a national raliroad museum, despite not being able to buiod the route or stations, or anythign else. Guess it's gonna be a kinda empty museum!!

 

[Dancing Hoskuld]

I just researched Railroads and a few thing struck me:

1. My workers still cannot build railroads (the route). I can't see why not. Anyone know?
2. ALl the buildings associated with railroads require steel wares, which are not available until a number of technologies LATER then railroads. This makes me think that railroads is a bit misplaced in the tech tree
3. One exception to the previus point - I CAN build a national raliroad museum, despite not being able to buiod the route or stations, or anythign else. Guess it's gonna be a kinda empty museum!!

Railroads used to require coal and steel. I am not sure if this is still the case.

 

[Koshling]

Railroads used to require coal and steel. I am not sure if this is still the case.

Has steel moved later in the tech tree or smething? You get railroads a long time before steel, but it basically does nothing due to lack of steel.

 

[Thunderbrd]

It'd probably be where the steel is converted to Wares that's too late (or that railroads is too early - one of the two)

 

[Dancing Hoskuld]

Originally it was coal and iron but Zappara changed it to steel when he introduced the steel resource. Perhaps it should "go back" to coal and iron wares to build railways.

 

[Hydromancerx]

Well when Vokarya redid the tech tree he insisted that Railroads be put BEFORE Steel. The problem now I think is tags and requirements. Railroad paths have ...

			<BonusType>BONUS_WARE_STEEL</BonusType>
			<PrereqOrBonuses>
				<PrereqOrBonus>
					<BonusType>BONUS_COAL</BonusType>
				</PrereqOrBonus>
				<PrereqOrBonus>
					<BonusType>BONUS_OIL</BonusType>
				</PrereqOrBonus>
				<PrereqOrBonus>
					<BonusType>BONUS_DIESEL</BonusType>
				</PrereqOrBonus>
			</PrereqOrBonuses>

But I do not think you can set it for (Iron Wares OR Steel Wares) AND (Coal OR Oil OR Oil Products)

Or can the property system work on routes?

 

[Raven Destroyer]

why not create 3 types of railroad, Early Rails with Iron, and Railways with Steel, and then later with Electricity, Electrified Rails (note i have not build rails in months )

that way it also signifies how the railway evolved over the years and you could also add a wonder or a national wonder for HSL Rails (if it is possible to keep them as straight as possible )

 

[Koshling]

Well when Vokarya redid the tech tree he insisted that Railroads be put BEFORE Steel. The problem now I think is tags and requirements. Railroad paths have ...

			<BonusType>BONUS_WARE_STEEL</BonusType>
			<PrereqOrBonuses>
				<PrereqOrBonus>
					<BonusType>BONUS_COAL</BonusType>
				</PrereqOrBonus>
				<PrereqOrBonus>
					<BonusType>BONUS_OIL</BonusType>
				</PrereqOrBonus>
				<PrereqOrBonus>
					<BonusType>BONUS_DIESEL</BonusType>
				</PrereqOrBonus>
			</PrereqOrBonuses>

But I do not think you can set it for (Iron Wares OR Steel Wares) AND (Coal OR Oil OR Oil Products)

Or can the property system work on routes?

You are correct that it cannot be expressed (except possibly with the generic expression system, though I don;t think that tag implements it, and the AI would not understand it). However, IMO you should just change it to iron wares. In practice you won;t have steel wares without iron wares anyway (or only a TINY fraction of the time would that be plausible).

 

[LumenAngel]

But I do not think you can set it for (Iron Wares OR Steel Wares) AND (Coal OR Oil OR Oil Products)

If you want to be perfet it must be AND (Wood OR Prime Timber OR Concrete) (between track, to keep them equidistant)

 

[n47]

But I do not think you can set it for (Iron Wares OR Steel Wares) AND (Coal OR Oil OR Oil Products)

I do not know, does Railroad paths schema matches against <TrainCondition>, but if so wouldn't something like this work?

			<TrainCondition>	
                                <And>
                                <Or>
					<Has>
						<GOMType>GOM_BONUS</GOMType>
 						<ID>BONUS_COAL</ID>
					</Has>
					<Has>
						<GOMType>GOM_BONUS</GOMType>
						<ID>BONUS_OIL</ID>
					</Has>
                                </Or>
                                <Or>
					<Has>
						<GOMType>GOM_BONUS</GOMType>
						<ID>BONUS_WARE_STEEL</ID>
					</Has>
					<Has>
						<GOMType>GOM_BONUS</GOMType>
						<ID>BONUS_WARE_IRON</ID>
					</Has>
				</Or>
                                </And>
			</TrainCondition>

Sorry for spaces, but I'm too lazy to remove them.

 

[Koshling]

I do not know, does Railroad paths schema matches against <TrainCondition>, but if so wouldn't something like this work?

			<TrainCondition>	
                                <And>
                                <Or>
					<Has>
						<GOMType>GOM_BONUS</GOMType>
 						<ID>BONUS_COAL</ID>
					</Has>
					<Has>
						<GOMType>GOM_BONUS</GOMType>
						<ID>BONUS_OIL</ID>
					</Has>
                                </Or>
                                <Or>
					<Has>
						<GOMType>GOM_BONUS</GOMType>
						<ID>BONUS_WARE_STEEL</ID>
					</Has>
					<Has>
						<GOMType>GOM_BONUS</GOMType>
						<ID>BONUS_WARE_IRON</ID>
					</Has>
				</Or>
                                </And>
			</TrainCondition>

Sorry for spaces, but I'm too lazy to remove them.

I don't think he expression system is enabled for the tag it would need to be (its a new system that is currently only enabled on a few tags). Also, the AI cannot interpret it, so it would screw the AI up if used on a tag that needed to be evaluated for the AI to make decisions (which this one does)

 

[n47]

Ok. But AI is screwed anyway. I play third game as a monarch and in all 3, if I would like to I could conquer any nation in any point of the game except the very beginning. Also was technologically something like half an age before CPU players and the distance was growing. It is actually more like SimCity then 4X. -- Still I like SimCity.

AI based just on points calculated from all currently possible choices + 2 steps forward clairvoyance and taking the best one, was probably good for for a less complex game like Vanilla or BtS, but for C2C without long term planning I do not see a chance to make it a challenge. -- Yes yes, I can take deity difficulty, but it's not like CPU is going more cleaver by this.

I am not an expert on AI, but have a feeling the complexity of C2C could overgrew possibilities of most of game AI developers of this world.

 

[Nimek]

We dont want to do things that AI dont understand. Avoid it as often as possible.

Please try to play on Immortal and Agresive AI. Game is really chalange in this dificulty.

WE made big progress to make AI competetive. It is a lot to do I know but now AI is more "inteligent " than it was before v30 when even new players won game at deity.

 

[n47]

@Nimek,

Yes yes, I can take deity difficulty, but it's not like CPU is going more cleaver by this.

Ok, I put it wrong. It is easy to make each game challenging this way. Make all units of cpu almost invincible, the game become also "challenging" then. I was saying, about cpu which tactics would be challenging, not forces. Which would enforce me to reflect on my tactics and change it. Which can surprise me with something.

I'm a scientist, not specialized at AI, but I know some basics. And I am really afraid, there is a chance, that there can be no chances for a good AI for C2C. At least, if there is no of some creative AI scientists wanting to take care about it. I know few scientists of AI, but believe even they would be hopeless.

What I am thinking of is, a CPU player would need to to approximate far future to make strategic decisions, but there are too many states to do it effectively. Maybe there is a chance to somehow "blur" the future, so reasoning about it would be computable and still it would be precise enough to guide the CPU. But I would probably learn few months, before I would be able to say something serious about it.

 

[JosEPh_II]

Ok. But AI is screwed anyway. I play third game as a monarch and in all 3, if I would like to I could conquer any nation in any point of the game except the very beginning. Also was technologically something like half an age before CPU players and the distance was growing. It is actually more like SimCity then 4X. -- Still I like SimCity.

AI based just on points calculated from all currently possible choices + 2 steps forward clairvoyance and taking the best one, was probably good for for a less complex game like Vanilla or BtS, but for C2C without long term planning I do not see a chance to make it a challenge. -- Yes yes, I can take deity difficulty, but it's not like CPU is going more cleaver by this.

I am not an expert on AI, but have a feeling the complexity of C2C could overgrew possibilities of most of game AI developers of this world.

Your assessment is really quite "on" the money.

What I've bolded in Red, is actually what has been happening for a long time to C2C.

The major content modder loves Sim City and has shaped many of the buildings/building paths to emulate that game. Also throw in major reworkings to combat and all the added content both in objects and coding to handle the objects and other similar projects, the Mod content is outstripping the AI guys abilities to catch up on a consistent and orderly basis. koshling, AIAndy, and to a lesser degree Thunderbrd do a great job of squashing the game killers. But, as the last 2 eras are still under construction, content is continually increasing even in the Eras that are supposed to be finished.

Long term planning has been inconsistent and mostly sporadic as the modders generally do their own thing then add it in. This is not always true, but was mostly the way it has been. Recent additions to the modders with apprentice modders has allowed some focus. But over all co-ordination is loose.

As I once posted this Mod is still 3 Mods. Prehistoric to get you up to the original Civ IV BtS Ancient era (very good mod section). Then from Ancient thru Modern Eras you have a buffed up and Fat AND (which needs to go on a diet and where most games come to an End), with finally the 3rd Mod the "space age" Transhuman and Galactic eras still under construction.

Just to clarify I've been a player/tester of this Mod since it's inception and it's 2 predecessors. I've spent 6 years of leisure time playing, commenting, criticizing, making suggestions, irritating the team and fellow players and complimenting the same. (Some hate me others love me, oh well can't win them all they say) All because I love playing this RoM/AND/Caveman2Cosmos 6 year experiment in pushing the Civ IV BtS game to new limits.

But a Project Director IS really needed. Strategyonly is the Mod owner/progenitor/maim man, but there is no "firm hand" when needed. Why? Because it's all volunteer work from players that love this experiment and no one wants to have to be the one to say No, not now put it on the back burner till this (current problem whatever it may be) is fixed and finalized.

JosEPh

 

[n47]

The major content modder loves Sim City and has shaped many of the buildings/building paths to emulate that game.

It is some curse. I can't even joke, to not put some truth in it.

the Mod content is outstripping the AI guys abilities to catch up on a consistent and orderly basis

I believe, that the point in this project is, that it constantly grow. AI should deal with it. But it would require some automated learning mechanism, which I can't even imagine in this moment. It is hard to learn AI to play chess, so learning it to play properly C2C may be technically impossible. -- Currently, I believe there is no learning mechanism for AI in the mod at all.

Because it's all volunteer work from players that love this experiment and no one wants to have to be the one to say No, not now put it on the back burner till this (current problem whatever it may be) is fixed and finalized.

Johnny is thinking: I would fix those high voltage cables in the wall, they're in a really bad condition.
Jimmy is thinking: I would fix this power generator outside, it sounds like an old HMG.
... 15 minutes later ...
Jimmy starts to smell something really well roasted ... ...
Not quite sure, does such cooperation follows the best patterns. Yes, I believe the project needs some long term coordination, but am not sure how to do this.

 

[Thunderbrd]

I believe, that the point in this project is, that it constantly grow. AI should deal with it. But it would require some automated learning mechanism, which I can't even imagine in this moment. It is hard to learn AI to play chess, so learning it to play properly C2C may be technically impossible. -- Currently, I believe there is no learning mechanism for AI in the mod at all.

What there is is an evaluation system. At some point it will need fine tuning. Koshling is really quite good at developing AI and its improved immensely since we began this mod. It makes some very odd decisions strategically but he's fully aware of and has plans on how to get past the hurdles that the combat mechanisms present. Given time and focus on that section I'm sure he'll have a strategic AI that will best even some of the most experienced players because all he does asks the computer to utilize its incredible powers of evaluation to determine both short and long term 'best course' determinations. It's very clever. At the moment we only see the full expression of how clever it really is in the earliest portion of the game where its priorities seem to be spot on - later in the game it needs to learn to shift its priorities a bit I think but that's not going to be impossible to manage to teach it.

Anyhow, I have a LOT of confidence in Koshling here... what now may look to be some really stupid AI determinations is merely due to a need for a little more structure and calibration of the genius system design base that's in place.

 

[Hydromancerx]

But a Project Director IS really needed. Strategyonly is the Mod owner/progenitor/maim man, but there is no "firm hand" when needed. Why? Because it's all volunteer work from players that love this experiment and no one wants to have to be the one to say No, not now put it on the back burner till this (current problem whatever it may be) is fixed and finalized.

The reason I personally think this works so well is because SO is a benevolent leader who allow the modders of C2C to have creative freedom in what they do. Rather than being controlling and restrictive he encourages us to "Just have fun!"

 

[Il Principe]

I must say: Although the content development sometimes outpaces AI development, the AI work so far is really extensive on its own.
It was even made aware of completely new concepts like crime, and handles revolutions far better then it did in RoM days.

And like Thunderbird already said: It already works pretty well in the earlier game. In my current game for example I got 2 big AI competitors so far, one outracing me to most important techs, because instead of piling up even more military when already second best with no border with the first, it concentrates on research, by "building" it in cities that have nothing better to do. Vanilla AI would have made those cities spill out countless units it doesn´t really need.

 

[n47]

The reason I personally think this works so well is because SO is a benevolent leader who allow the modders of C2C to have creative freedom in what they do. Rather than being controlling and restrictive he encourages us to "Just have fun!"

I'm not a fan of directors either. But are you aware of the situation I wrote about? No being aware what the others do, makes possible your work will collide with their work. And not being aware of capabilities of the engine, makes possible your work will collide with everything. I'm afraid, it can end up in the Never Ending Bug Story.

@Thunderbrd, the abilities of CPU are not so great as you can think. Do you know the Go game? It is no such a complex game. But did you know, the greatest specialists can't build AI which would properly play it? And there are problems way much more complex then that. I can't be sure about it, as I am not in C2C AI, but I guess the best what can be actually done, is giving AI some hard coded patterns and a short range forecasting. But the case is, a human being can learn those patterns and learn how to fight them. And the current AI can't probably learn the patters of the human, so it can only choose an option hard-coded by the programer. There is no creativity. And a not creative opponent is a boring opponent in a long-term game play. -- Or at least it is for me.

For example, I attack a CPU player. I have overwhelming forces and I've put them next to its cities. But I still haven't bring artillery. The CPU have railway tracks. So in his place I would: put some espionage at his territory to find out were is the artillery and how well it is guarded. Try to concentrate my fast units but of weaker experience, as near the border as I can, but out of the sight of the enemy. Promote them in mobility and some in fighting the type of units guarding the artillery. Attack when the artillery would be near enough and then I use the more mobile units to destroy as much as I can of his roads and railway paths, by the way killing nearby workers. During that time I would produce some units for crashing at least the part of enemy army which is most near the border, try to sabotage production and happiness in the enemy cities, and count, I will withstand up to the time when his cities will start to rebel due to prolonging war time. Then I could probably overgrow the enemy forces.

You can of course try to encode such tactic in the AI. But there are lots of others which can be invented and they form depends on the situation, and moreover they need to adapt on opponents' moves. Seriously, I'm afraid it is impossible to hard-code things like that.

@Il Principe, this rather not a big deal. When you start all players are in quite similar situation, so there are already relatively few strategic decisions to be done -- in compare to all decisions done up to for example medieval. So even, if CPU is not too inventive, the result of his choices still had no time to accumulate and the do not impact the odds in game so greatly. But if there is no long term planning there may be no chances to extend balanced game-play for the later ages.

Uh, enough. I'm getting confused because all of this writing. I need a break.
Can't you talk a little bit about make up or cookies?