AU mod: AI Naval exploration

[donZappo]

After my initial tests it appears that the AI will not use curraghs to transport settlers unless it can take an escort with them. That means that we'd either have to let curraghs carry 2 units, or fundamentally change the settler so that it could defend itself similar to Civ 1. I don't know if it would be possible to get the AI to use the curraghs for settling without a major change to the fundamentals of the game. If you make curraghs able to carry 2 units, then they're just cheap galleys. If you gimp their movement to not be as effective, you remove the curragh-as-scout function and make the unit profoundly different. If you change the settler to be 0/1/1 so that it could defend itself (and maybe entice the AI to let them go unescorted), this strikes me as being outside the spirit of the AU. Plus, I don't like the idea of using it to only transport "foot units" as I don't think that military angle is the intent of the seafaring trait or this early unit. I guess something like this could work for Naval Exploration:

Give Curraghs the load flag (this makes the AI build them and use them)
Don't let the Curraghs traverse ocean squares
Make the curragh seafaring only

This would make it a true seafaring-exploration unit similar to the land-based scout for expansionist. Not as useful as it is in its current form, but something has to be done to not give the human player a major exploit.

[Fosse]

I'd like to see the AI build and use curraghs without having to add unit transport ability (what about missle transport, or an air mission range, or some other flag that the unit can't actually use?).

Humans should NOT be able to enter the oceans with curraghs, and oceans SHOULD cost a galley more to enter.

I don't want to see settlers or military units being on curraghs.

I disagree that this change is damaging enough to the seafaring trait to not implement it. The extra move point carries through the whole game, remember, not just for curraghs. If people really want seafarers to be able to enter oceans, why not make another unit with the same name, buildable only by seafarers, that doesn't have wheels? Compromise.


And just for the sake of the argument, what is the highest sinking percentage, if not 100%? A 95% sink rate would make me think more than twice!

[donZappo]

If you just give it the load flag then it makes the AI think they're more valuable even when they're not. It will build the curraghs and have them sail around just like they were galleys. I think this discussion is boiling down to what the true purpose of seafaring should be in the game. Is it for monopolizing contacts? Settling island chains early on? Or is just a boon to early coastal exploration with the added bonus of having ships that are slightly faster throughout the game? +1 movement matters a lot more early on than it does in the industrial era. What should the true purpose of this +1 movement be?

[Catt]

This one's a tough nut to crack -- principally, IMHO, because AU tactics with naval units and human tactics are so different. My view of the implementation of "sinking chances" is that it was a deliberate design decision by the designers -- humans could risk the high seas but the AI would not -- in other words, early naval exploration was deliberately skewed to favor the human. Unless I'm missing something, there's nothing we can do to alleviate the disparity in power between human and AI with respect to "suicide ships."

It would be great if we could stimulate AI use of curraghs to make early contact along the coasts, but a proposal to nerf the human ability to risk sinking for a chance at contact seems to me to come down to trading a legitimate design decision (even if only made because it was too challenging to code intelligent suicide runs) for a more even playing field between human and AI. If we choose the "even playing field" approach, I suspect there will need to be discussions about some dramatic changes to artillery / bombard units, and particularly to lethal bombard of late game units such as bombers.

FWIW, I'd rather not completely nerf the inherent human advantage with opting for suicide runs (i.e., not eliminate ocean travel for curraghs), but my initial impressions of the seafaring trait's lowered sinking chances and increased movement allowance radically amplifies the power of the suicide option in the players bag of tricks. I only played the past AU mod with increased sea and ocean movement costs once or twice - not enough to form an opinion - but the lowered mobility didn't strike me as radically altering the gameplay intended by the original design.

Catt

[dexters]

Abit OT, but semi related...

I think the early naval AI needs to be tweaked a little. I've said this many times, but pre Frigates and Galleons, AI civs remain lethargic builders of navies. And there are a lot of missed opportunities in the first hundred of so turns where the AI has one galley landing 2 units per turn.

This is contrasted with the sheer madness that happens when it hits modern ages and even in Monarchy difficulty, you have AIs in standard maps building 20+ destroyers and assorted specialty ships like BB and Carriers. Heck, that's the size of my navy in LARGE-HUGE maps back in the PTW days, and that's when I'm well ahead and can afford to crank those ships out for my own vanity since I'm such a naval fan.

This may have something to do with the Escort-Transport pairing that triggers something in the AI. A solution, not so much for AU but perhaps for a bug fix is to randomly attach an Escort flag to galleys and caravels. Kind of like how some units, like Infantry can be classified as O or D by the AI. This may help solve the problem of lack of AI shipping.

[MrWhereItsAt]

Originally posted by Catt
...a proposal to nerf the human ability to risk sinking for a chance at contact seems to me to come down to trading a legitimate design decision (even if only made because it was too challenging to code intelligent suicide runs) for a more even playing field between human and AI.
Catt

Catt - forgive me here if I am misunderstanding the role of AU (I am new to it, and thus should perhaps be struck down for daring to question a vet Strat Forumer like you!), but is that not the purpose? To enhance the AI's chances with as small changes as possible. We both (and likely ALL) agree the human suicide ship tactics can be a huge imbalance, and this makes it worthy of attempting a fix. If the human loses this uncompensated advantage, then good!

Or does AU also have implicit in it's aims the need to obey the designers' decisions? If so, then we should be questioning the validity of changing ANYthing that isn;t clearly a bug.

To me, even though we cannot ensure the AI uses Curraghs to transport settlers, and giving Curraghs two-unit carry ability is too much, restricting Curraghs to coast squares and giving them a 1-unit carry ability is enough for me. The human's huge suicide advantage is gone, and there is a small compensating ability in place. And how many players will really use Curraghs to settle more than one or two city sites before MapMaking comes around anyway? Humans will want escorts for their settlers as well, keeping the attraction to Galleys.

That said, perhaps we should just plain out and out restrict Curraghs to coast and test it at that. Simple, easy and such a small change.

[Catt]

Originally posted by MrWhereItsAt


Catt - forgive me here if I am misunderstanding the role of AU (I am new to it, and thus should perhaps be struck down for daring to question a vet Strat Forumer like you!), but is that not the purpose? To enhance the AI's chances with as small changes as possible. We both (and likely ALL) agree the human suicide ship tactics can be a huge imbalance, and this makes it worthy of attempting a fix. If the human loses this uncompensated advantage, then good!

Or does AU also have implicit in it's aims the need to obey the designers' decisions? If so, then we should be questioning the validity of changing ANYthing that isn;t clearly a bug.

No forgiveness needed - the great thing about AU is that everyone's opinions carry equal weight (except maybe for the new uber-panel ).

From the AU MOd thread started by alexman:

Most modifications incorporated by the AU mod are a compromise between a) improving the AI, b) presenting the player with more options, and c) changing as little as possible. The level of this compromise is a delicate and subjective decision, so each proposed change is discussed by the Apolyton University community before it is implemented.

As you participate in more AU debates about prospective changes, I think you'll find that there is usually inherent tension between (i) making changes to improve the AI / weaken the player's tactics, and (ii) not changing the game in ways which eliminate or severely undercut tactics and strategies present in the stock game. If we only tried to improve the AI / weaken the human, we would need to make wholesale changes in many facets of the game, effectviely creating a "new game." If we only tried to change as little as possible, the AU Mod wouldn't be terribly effective at creating a more engaging game. Proposals, debates, game testing, and ultimately majority opinion will find the right balance between these competing goals.

My opinion on the curragh was that completely barring it from traveling in certain terrains would push too far towards the "weaken human" end of the spectrum (and deviate from stock rules by a fair amount) without a corresponding equal benefit of improving the AI / human balance, and that if a change was needed then perhaps increasing movement costs is a better middle ground. My view is that if we want to eliminate ocean travel for curraghs, wouldn't the same arguments apply equally well (or nearly so) to making a similar change to galleys and caravels?

As alexman points out in the opening post, there are two distinct potential "problems" to be addressed by the new presence of the curragh: (1) AI's failure to use them for routine scouting around coastlines; and (2) AI's failure to use them to suicide. I might have taken this off on a bit of a tangent by directing my arguments too directly to the suicide run issue -- if we could change AI behavior with problem 1 (sounds like this looks to be very tough), I think problem 2 is a much less important issue. Thing is it looks like we don't have a solution for improving problem 1, meaning all solutions are coming down to addressing problem 2 -- and problem 2, IMHO, is not limited to curraghs.

Your opinon may of course differ .

Catt

[donZappo]

But I thought a solution was created for addressing problem #1. If you give curraghs the load flag without allowing them to transport units, then the AI will use them and send them out to explore the coasts. This point has already been established. The problem is getting the AI to build the curraghs and suicide them, which is the real issue. I can see how this is exactly applicable the curraghs and galleys and I guess we need to address that issue if we start trying to specifically address the suicide-run tactic. I'm still hoping that someone can come up with a way to address this issue without barring ocean travel altogether. Maybe we'll get lucky and this problem will be "fixed" in the next patch!

That being said, would there be a benefit in just slowing down the ability of units to traverse the sea and ocean? These values could be adjusted so that seafaring gets 2 moves while everybody else will be crawling along at 1 move.

[Catt]

Originally posted by donZappo
But I thought a solution was created for addressing problem #1. If you give curraghs the load flag without allowing them to transport units, then the AI will use them and send them out to explore the coasts. This point has already been established.

Then I missed that bit and it's good news. I thought the various results of testing from you and others indicated that the AI might build more but that they tended to stay in port waiting for a unit to load that never would happen - didn't really digest the info in your post just before my first one

The problem is getting the AI to build the curraghs and suicide them, which is the real issue. I can see how this is exactly applicable the curraghs and galleys and I guess we need to address that issue if we start trying to specifically address the suicide-run tactic.

If it comes down to the suicide issue, then I fail to see how restricting curragh suicides is materially different than galleys -- they both come well before Caravels, and the power of suiciding increases dramatically on waterbound maps, in which case many human players would prioritize Map Making a bit in any event.

Maybe we'll get lucky and this problem will be "fixed" in the next patch!

That would be best result , but highly unlikely -- it's been around since the first release, and is well known to the Firaxis AI programmer.

Is there a view among others that the suicide tactic is sufficently more powerful with curraghs than with galleys to justify a change to curraghs but not galleys? Should the discussion widen to include possible changes to galleys as well? Is altering curraghs but not galleys a "middle ground" approach to slightly weaken but not completely nerf the suicide tactic?

Catt

[donZappo]

I was holding out a slim hope that it would be addressed in the patch since Breakaway/Firaxis went out of their way to reduce the sinking odds for seafaring but made it so that the AI will never use this to its advantage. I figured that since they changed the environment like this it *might* have a been an oversight to not have the AI suicide at all. Right? RIGHT?

I don't think that we should view galleys and curraghs seperately in this discussion. They both come early enough that I believe there is no significant difference in the timing of getting those contacts. Unless you research monarchy and republic yourself before going back and getting mapmaking, then you're not going to be that far behind in the tech race either way.

How important do you guys feel the sinking of ships is in the game? It seems to me that most players view it as just slowing down their progress across the water without having any game-altering effect. Then what is the difference between suiciding and just slowing down ship movement through the water? Or keeping suiciding and slowing down the ship movement as well. That last point doesn't sit well with me, though, since it just seems like a wishy-washy solution to the problem that wouldn't really solve the inherent problem. So would there be a big problem in restricting ocean/sea travel or just allowing it, but slowing it down significantly?

And one last note. If you give curraghs the load flag but no unit transport abilities they seem to behave just like galleys when in the hands of the AI. They explore with them, but don't wait around at home for some unit to load up into them. It's weird that it works that way, but the AI does act very strange sometimes!

[MrWhereItsAt]

Originally posted by Catt

My opinion on the curragh was that completely barring it from traveling in certain terrains would push too far towards the "weaken human" end of the spectrum (and deviate from stock rules by a fair amount) without a corresponding equal benefit of improving the AI / human balance, and that if a change was needed then perhaps increasing movement costs is a better middle ground. My view is that if we want to eliminate ocean travel for curraghs, wouldn't the same arguments apply equally well (or nearly so) to making a similar change to galleys and caravels?

Galleys perhaps, but not so much Caravels. Remember too that there is even more of a benefit to early contacts from suicide ships in C3C due to the far later ability to swap maps and contacts. Thus suiciding is made far more powerful to the human now, and has IMO reached the point where restricting Curraghs and Galleys from Ocean (if not Sea as well) is justified.

Caravels come around the time that these contacts can be traded, and it does not seem unreasonable to me to expect Caravels to get lucky and cross ocean. Although it is peripheral to the discussion, in RL Caravels (or near equivalents) successfully crossed the Atlantic with Columbus.

[donZappo]

Alright, I just ran a couple more tests to experiment with a couple of things. I made sea cost 3 movement points, and ocean 4 movement points to cross. I tried to cross a stretch of 2 sea tiles and one ocean tile to simulate what I view as the average minimum oceanic crossing. I sent 10 sea-faring curraghs, and 20 non-seafaring curraghs across this for comparison. I got the following results:

Seafaring: 6.3/10 ships lost per crossing (100% loss 2 times after the third move)
Non-Seafaring: 17.1/20 ships lost per crossing (100% loss 2 times, once after the second move!)

The reason I found this so enlightening was that it seems to present a situation where strategy is still preserved for the human while still providing a benefit to the seafaring civilizations and makes sure that The Great Lighthouse is still useful. In this situation, the non-seafaring human can still suicide their ships if they want to try for contacts, albeit at a large risk of resources. The seafaring human can suicide a bit faster (2 movements through sea squares) and has a reduced risk of sinking. Granted, the human is still the only one that will suicide their ships, but you could just view the AI as a "conservative" civilization if that helps to explain things. Heck, in a majority of cases I would probably turn into a conservative player with these values, but there would still be times that I would send a suicide ship out -- especially if I'm seafaring.

Additionally, upon constructing of The Great Lighthouse seafaring civs would now get 2 movements through ocean squares and non-seafaring would get 2 movements through sea squares. Both of those speed increases would come with the normal ability to not sink in sea squares making exploration/suiciding an easier alternative without making it overpowered.

I really like the feel of this test and think it falls firmly under the category of "presenting more strategic desicions to the human player" by causing the human to stop and think before they send the ships into the great unknown. Also, there would still be situations in which the AI could still cross small bodies of water safely, especially if they build the Lighthouse. Since this would nerf the seafaring trait a little, it could be balanced out by making the curragh a seafaring only unit. I don't know if that would fall outside the scope of the AU. It is a change that I personally would really like to see.

So, my proposed changes are:
Curragh - Seafaring only, maybe no movement allowed in ocean (or sea, even)
Sea - 3 movement/square
Ocean - 4 movement/square

I would also like to see the curragh be able to carry 2 units so that the human and AI could use it for early coastal and close island settling. After testing this last point a bit, I don't think that it is overpowered at all as any sizeable naval invasion probably would have to wait until the time Map Making is researched. By the way, the 2 unit transport is necessary to get the AI to load settlers on the curraghs.

These suggested changes could be debated a bit to try to iron out specific numbers for movement penalties, but I ask you to seriously consider this change as it does have a remarkably good feel to it when put into practice. As least I think so.

EDIT: Of course the later units could just be flagged to ignore sea and ocean movement penalties so they wouldn't be slowed down by these changes.

[donZappo]

Hmm, we appear to have no AU panel representation in this thread. Does that mean that this decision is going to be left entirely to "the people?"

[alexman]

My suggestion for galleys would be to take the same approach as the PTW version of the mod: 2MP to enter ocean.

For curraghs, I propose 2MP to enter both ocean and sea tiles.

The obove would weaken, but completely not eliminate, suicide missions, which I think are fun, and part of Civ3.

I'm still not sure about what's the best way to deal with AI's failure to explore the coast with curraghs. I will check back after I have done some tests.

[lockstep]

Suicide missions can be fun, but ocean suicide missions with curraghs ... completely out of whack, IMO.