A "real" AI cheat

[The Snug]

I definitely agree with Woody1 (for once! ) in that there's nothing in need of change. The patrol mission shouldn't become a one-size-fits-all decision.

Although the behaviour may seem weird, I'm yet to see a clear argument as to what the actual bug is; I believe there is no such bug.

And as Roland said I think we are blowing it out of proportion. It's taken this long since the game release for an experienced player to even notice the oddity. Again, I say leave it.

EDIT I might not be around for the next week so I may not be able to continue the debate for a bit.

Hope you'll be back soon so we can argue each other to death!

I'd say, however, that the reason you lack concern regarding this issue is that it simply doesn't affect your play style. Whereas for me, it's very damaging. It doesn't matter to you, because it will never affect you.

Myself, I'm a huge fan of sushido, which is the crux of my strategy on immortal level. Each food resource becomes more valuable than gold. On levels lower than immortal, massive modern naval warfare over food resources is just unlikely. But on immortal level, when the playing field is even, massive modern naval warfare becomes the norm. It is in the modern era that this anomaly becomes game-breaking.

Hence, for my play style and for my difficulty level and map preference and for my fish resource strategy, the proper functioning of this feature becomes paramount. This particular problem nearly caused me to lose my immortal game bcz it allowed the AI to demolish my fleet. Fortunately, my superior productive capability rescued me, but it severely delayed my victory.

Regarding this particular method, I think I could easily write a walk-through strategy on how to beat immortal level, but my methodology is severely hampered by this "bug." I'd sure like my strat to work on deity, but now I'm not sure if I could sustain the severe naval losses.

If this "bug" did not exist, I could beat immortal rather easily with my strategy, bcz then I could I protect my all important fish resources adequately and efficiently without incurring the obscene naval losses that this oversight permits.

The reason it hurt me, is bcz, although I had obtained a significant tech lead and possessed superior production, I could not build my spaceship, but instead had to build mass naval units to replace my remarkable losses (which, at the time, I could not account for how the AI was able to do it). And as I was drawn further into war, I eventually had to completely abandon building my spaceship in favor of going around conquering the world. First, I was obligated to killing the massive PA on the two islands next to me (who were the ones pillaging my seafood); by the time they were under control two other AI's had launched spaceships, so I had to go sink them. By the time their capitals were burning, I had to go burn Gandhi's capital to prevent his imminent culture victory. By the time I had accomplished that, another AI spaceship went up into the sky. I ended up just conquering the world, which is not what I wanted to do. I wanted to simply build my spaceship 400 turns earlier and save 16 hours of play. But, all bcz of this single anomaly, I had to go conquer the world. What do they say? If not for the lost horse shoe the kingdom would not have been lost?

 

[Balderstrom]

The best solution is to disallow pillaging when the unit has had a combat that turn already.

Since the Patrolling unit attacks the pillaging enemy, would not the best solution be to disallow retreat by pillagers?

This removes the XP gain bug. And doesn't significantly change gameplay.

In the examples presented previously, it would seem the Patrolling ship that successfully "intercepts" an incoming ship - should get XP, instead of the retreating unsuccessful raider. Much like how Planes gain XP from interception.

 

[The Snug]

Since the Patrolling unit attacks the pillaging enemy, would not the best solution be to disallow retreat by pillagers?

This removes the XP gain bug. And doesn't significantly change gameplay.

In the examples presented previously, it would seem the Patrolling ship that successfully "intercepts" an incoming ship - should get XP, instead of the retreating unsuccessful raider. Much like how Planes gain XP from interception.

That almost fixes it. Unfortunately, if the pillager wins the combat, it can still re-initiate a new pillage. Your fix limits the gross abuse by an attack sub, but doesn't stop the entire problem.

 

[DanF5771]

I've modified the above example to a more realistic case, just as it might actually happen in your game. Now you (Victoria) have 4 Fishing Boats on Ocean tiles that you can protect with 2 Ships of the Line. The improvements are threatened by Hannibal's single Destroyer that has 10 due to Navigation1+2. You also have 2 Ironclads in a Fort at the coast which are unfortunately too far away to help protecting your sea food.

With the current game mechanics and your SotL told to patrol this is what happens after you press <End Turn>:

Hannibal's non-Blitz Destroyer kills both of your SotLs and pillages the 4 Fishing Boats all in one IBT.

Let's look at what happens after implementing the proposed changes:

The Destroyer kills one of your SotLs, can not pillage again during its turn and thus moves away--you save 1 SotL and 4 Work Boats = a lot of hammers. After that you can try to secure the area by upgrading the Ironclads or moving other strong naval units into the danger zone.


I've saved the scenario as a WB file so you can also test this playing as Hannibal to follow directly how your Destroyer triggers the combats and attacks the SotLs on their coastal tiles.

I haven't considered the lack of transparency in MP games before, so trying to amend that, the pillage button is now inactive for a non-Blitz unit on protected improvements after combat and has the text "Requires Blitz" added.

With this example I hope I could point out how advantageous a properly working Sea Patrol function can be in saving a lot of frustration caused by those pinprick surprise attacks from fast moving navy (as sirsnuggles very well described above). As English is not my first language I apologize for any unintended misunderstanding/condescending sounding verbalisation.

Should anybody be interested to test the modified version I will attach the dll as well. So what do you guys think? Still way out of proportion?

 

[The Snug]

Seems to me to be precisely the way it was intended to work. And mirrors exactly what would happen on land (except that multiple resources can be protected simultaneously).

On land, in order for a unit to pillage a tile, it must first destroy all of the defending units. Obviously, on land, an attacker can only attack once per turn, so if there are multiple defenders, the attacker would not possess the opportunity to pillage and would also need to run away. Which is precisely what happened in the above visual scenario.

That was well illustrated Dan.

Perhaps we should put this in the "bugs" forum to get input from there. They're the ppl who study these things. I sent a PM to Solver, but since I'm sure he doesn't know who I am, he didn't respond to me.

 

[Wodan]

Since the Patrolling unit attacks the pillaging enemy, would not the best solution be to disallow retreat by pillagers?

Logically, therefore, we should have to allow retreat by the patrolling unit. You sure we want to go there?

 

[JujuLautre]

From what I read, the main issue seems to be the fact that a unit with a high withdrawal chance can get several XP by pillaging multiple times, am I right?

Then, the same way that siege engine can't attack when their target is at low health (and so, coincidently, cannot gain any more XP), why not make that the attacker does not get any XP if no fight was done (meaning that the attacker was at very low health and retreated without anything done) ?

Sorry if I repeat myself but it seems no one saw it

If nobody comments on that this time, I won't bump it again and consider it's useless, I promise

 

[DanF5771]

Hey JujuLautre:
IMHO it's not only the XP thing, but the general fact, that the Sea Patrol function turns ordinary naval units into Blitz-units. This is of course the more annoying the more the aggressor has.
Just imagine Gunships would be available with Flight and come with 8--there is no way to effectively protect your improvements from such units. Now imagine there would be an identical Land Patrol function which would indirectly gift those Gunships a free Blitz on top. Everybody would scream!

Changing the XP gained from withdrawal would mean modifying core combat mechanics of the game, I think this would be out of proportion. BTW, if you think going that route is the best solution for you, you can also change the XML-entry EXPERIENCE_FROM_WITHDRAWL in GlobalDefines.xml.

The disallowance of pillaging under the special circumstances instead of intercepting at another later point also prevents the AI from repeatedly trying to pillage with no effect. Now it just moves its units away.

 

[JujuLautre]

Just note that I did not mean get no XP from withdrawal in all cases, but just in cases like this one where the unit retreats after only one failed round. Not sure it's feasible though.

But all in all, the main reason why I proposed this is that I think there is no other easy way to handle this situation, let me explain:
- If we limit the "active" (pillaging) unit to 1 fight per turn including those resulting from pillaging near a patrolling unit, then either it's not possible to pillage any more after a fight and it changes the pillage system, or we make pillage possible ignoring the patrolling units, which would be silly
- If we limit the "passive" (patrolling) unit to 1 fight per turn thanks to patrol, then why should patrolling units be limited to 1 fight when the defensive units from regular fights are not limited ?

I personally am OK with the "blitz" behaviour: it makes the patrolling option nice (because you otherwise can't defend everything at once) but risky. Another "interesting choice" would say Sid I just think that in this case, the high withdrawal thing should be changed. That said, I also like the idea that this unit should not be able to retreat at all, because, in some way, the patrolling unit is attacking.

 

[Woody1]

Yeah, the more I read in this thread, the more I am convinced the pillage function is working exactly the way it is supposed to work with patrol.

Patrol shouldn't be some kind of special command that nerfs pillagers. It's simply a way to spread out your defensive area to multiple tiles. That's a huge advantage, but should also come with a disadvantage. The disadvantage is that pillagers can continue to attempt pillaging mulitple times per turn.

If you don't like the behaviour of patrol, then don't use it. Everything else in Civ has both upsides and downsides... why do you want to make patrol so special by nerfing pillage?

 

[r_rolo1]

Woody, we have been disagreeing since the beginning and it looks that we continue disagreeing

What are you saying is "Well , you have a function in game that allows 1 units to shadow cover multiple tiles. But that has to come with a slight price, that is to be vulnerable to lose 10 ships in 1 turn to a non-blitz unit and/or to be able to be milked from XP with high withdraw subs" . Sorry but that is not my idea of a fair trade or of a interesting choice" ( like JujuLautre said )... it is more my idea of "WTH, I'll never touch that button again in my life"

And woody, this is working as intended, I reckon. But working as intended != working well. Both Patrol and the combat mechanics are working as intended, but the interaction of both in some cases can lead to really wierd and unbalanced stuff. And this is the kind of misfeature that could had easily passed the QA ( others far more oblivious passed , like the Barrage stuff in 3.17 or the culture bug in 3.13 ).....

 

[DanF5771]

If you don't like the behaviour of patrol, then don't use it. Everything else in Civ has both upsides and downsides... why do you want to make patrol so special by nerfing pillage?

Because I feel that the downsides heavily outweigh the upsides especially in the late game with its fast moving and easily withdrawing naval units, so that establishing an effective defense with usual means becomes harder and harder + needs annoying micromanaging of units (upgrading/replacing) + overlooked loopholes can lead to very frustrating results. The AI is just better than the human in spotting those chances.

Also, the fact that there is a certain function in the game, but the AI seems to always "decide" that it is best not to use it should be a pretty clear indication that there is something wrong with it.

Finally and technically the proposed changes do not represent a nerf of the pillage system, because the rule that a non-Blitz unit coming out of a combat alive can not attempt to pillage an improvement which requires to attack once more already exists now, this rule is just enforced for the Sea Patrol situation. So it's only a (needed) buff of the Sea Patrol function.

 

[Woody1]

What are you saying is "Well , you have a function in game that allows 1 units to shadow cover multiple tiles. But that has to come with a slight price, that is to be vulnerable to lose 10 ships in 1 turn to a non-blitz unit and/or to be able to be milked from XP with high withdraw subs" .

1. You are overstating the "milking of XP" with high withdrawl subs. In the tests done here, they were giving attack subs Guerrilla III promotions for the 100% withdrawl, which subs can't get in the game. It is a straw-man argument. In the game, a highly damaged sub is likely going to die if it tries repeated pillages with patrol ships in the area. Probably not worth the XP.

2. If you can lose 10 ships in a row to a single enemy trying to pillage, then you are hopelessly outclassed and are going to lose your navy anyway.

And again, if you really think the advantages of a patrol order (spreading out your defensive area) are not worth the risk of an enemy battleship sinking all your triemes, then don't use the order. Plop your defensive units on top of your resources, and lose them all over the next few turns instead.

In the end, if you are outclassed by that much, you will lose all your ships anyway. If you like to lose them slowly, don't use patrol. If you want to take a risk and hope the enemy only has caravels against your triemes rather than battleships against your triemes, then use patrol.

You (and snuggles) are over-exaggerating the situation.

 

[scu98rkr]

If a fish resource is next to a city, can you patrol from inside that city ?

 

[DanF5771]

If a fish resource is next to a city, can you patrol from inside that city ?

No. This would require the pillager to attack the patrol in the city square--this is not allowed for naval units. Feel free to download the attached scenario and test stuff, it's really interesting.

 

[r_rolo1]

1. You are overstating the "milking of XP" with high withdrawl subs. In the tests done here, they were giving attack subs Guerrilla III promotions for the 100% withdrawl, which subs can't get in the game. It is a straw-man argument. In the game, a highly damaged sub is likely going to die if it tries repeated pillages with patrol ships in the area. Probably not worth the XP.

2. If you can lose 10 ships in a row to a single enemy trying to pillage, then you are hopelessly outclassed and are going to lose your navy anyway.

And again, if you really think the advantages of a patrol order (spreading out your defensive area) are not worth the risk of an enemy battleship sinking all your triemes, then don't use the order. Plop your defensive units on top of your resources, and lose them all over the next few turns instead.

In the end, if you are outclassed by that much, you will lose all your ships anyway. If you like to lose them slowly, don't use patrol. If you want to take a risk and hope the enemy only has caravels against your triemes rather than battleships against your triremes, then use patrol.

You (and snuggles) are over-exaggerating the situation.

1- Guerrilla was used ( by me ) just for fun. it wasn't needed, since flanking I + II + Tactics already give more than 100 % withdraw, as dan showed in the first test.

2- You can cover subs, you know?

3 - A unit with more than 100% withdraw ( acheivable in game with current rules ) will never die in a battle where he is the attacker. So , supose this: flanking I + II + Tactics ( 110% withdraw chances ) sub covered by battleships vs 10 destroyers ( hardly a trirreme vs scenario ). If the destroyers have patrol, the sub will battle all and withdral... with some luck it will kill the first. At least 10 XP per turn with no risk at all... compare with the non patroling destroyers scenario.

In fact this would work with any naval unit that can get more than 100% withdraw.... trirremes milking 10 XP per turn, if covered

I agree that this is not a big issue in SP, but only because of the fact that the AI is programmed to not use patrol. If it was programmed to use it ( as it should had been )..... "OMG, this patrol stuff is broken!" threads all over the place, I bet.

Sorry, but this is clearly broken, because it breaks the fundamental principle of Civ IV combat: units can only attack once per turn, except in very special ocasions ( read blitz )

 

[karadoc]

I agree with Woody1 that this issue is being exaggerated.

From my point of view, the sea patrol order tells your units to attack anything that tries to pillage the adjacent tiles. The pillaging ships aren't getting any kind of pseudo blitz upgrade, it is the patrolling ships that are initiating the fight.

In my opinion, nothing needs to be changed. The current functionality is fair. Sea patrol is still useful even with the current functionality. However, if something was to be changed, here is what I think could be done: since it is the patrolling ships that are doing the attacking, I suggest that the pillaging ships should not be able to withdraw (but that the patrolling ships can); also, each patrolling ship should not be able to attack more than 1 adjacent pillager per turn unless the patroller has blitz. These changes may not address the complaint about the pillagers being able to destroy multiple ships in each turn, but frankly I don't care about that. The sea patrol units have been ordered to attack the pillaging ships. It isn't the pillaging ships attacking. If you don't want your ships to attack the pillaging ships (because you think you're going to lose multiple ships in one turn, or whatever) then don't order them to sea patrol. Only order them to sea patrol if you really truly want them to attack ships that try to pillage adjacent tiles.

 

[DanF5771]

Sorry r_rolo1, but I have to disappoint you here.
GlobalDefines.xml has
<DefineName>MAX_WITHDRAWAL_PROBABILITY</DefineName>
<iDefineIntVal>90</iDefineIntVal>
so you can't give a unit a promotion which would push it over the 90% (Max is Sub with Flanking 1 + Tactics = 90%).
Still good enough for a lot of XP milk.

Woody1: Please tell me one clear drawback you would suffer from a regular game of yours with the proposed changes implemented.

 

[DanF5771]

I agree with Woody1 that this issue is being exaggerated.

From my point of view, the sea patrol order tells your units to attack anything that tries to pillage the adjacent tiles. The pillaging ships aren't getting any kind of pseudo blitz upgrade, it is the patrolling ships that are initiating the fight.

In my opinion, nothing needs to be changed. The current functionality is fair. Sea patrol is still useful even with the current functionality. However, if something was to be changed, here is what I think could be done: since it is the patrolling ships that are doing the attacking, I suggest that the pillaging ships should not be able to withdraw (but that the patrolling ships can); also, each patrolling ship should not be able to attack more than 1 adjacent pillager per turn unless the patroller has blitz. These changes may not address the complaint about the pillagers being able to destroy multiple ships in each turn, but frankly I don't care about that. The sea patrol units have been ordered to attack the pillaging ships. It isn't the pillaging ships attacking. If you don't want your ships to attack the pillaging ships (because you think you're going to lose multiple ships in one turn, or whatever) then don't order them to sea patrol. Only order them to sea patrol if you really truly want them to attack ships that try to pillage adjacent tiles.

Please read the thread/download the scenario and watch, please!

 

[ori]

1- Guerrilla was used ( by me ) just for fun. it wasn't needed, since flanking I + II + Tactics already give more than 100 % withdraw, as dan showed in the first test.

1) within current rules 100% withdrawal chance is only possible with worldbuilder - try giving a sub a warlord and addind flanking I, II and Tactic through normal promotions - it won't work. The game is designed to disallow promotion combos that would give 100% withdrawal chances... (90% is attainable - but really only for subs with an attached warlord - a really rare occurance)
2) the proposed "fix" really just makes patrol much more powerful since it not only disallows pillaging by the unit on patrolled resources but disallows pillaging period - basically making it much more advantageous to patrol with outdated units than before (hell even if there are only 2 massively outclassed patrol boats it prevents AI pillaging effectively - since as shown the AI will not wait a turn and waste movement points but rather leave the area).
It is a bug - but its effects are massively overstated in this thread and the proposed fix imho disbalances things much more than it does any good. As stated repeatedly the AI does not patrol well but that also means that this "fix" actually now just renders an AI exploit vs. the human player into a human exploit vs. the AI - I don't like it.
much for my 2cents - luckily I don't need to use any proposed fix and others are free to use it.
I do think that to really fix this one should go with the ingame description of what patrol is supposed to be doing - i.e. have the patrol boat attack and not the other way round (I know that this makes giving it the tile bonus difficult at the very least and causes other problems