WHEOOHRN Hiding

[r_rolo1]

@jdog

IMHO, and to use a poker analogy as you did, this is the equivalent to put a paper bag in the AI head instead of making it do a better poker face ( and it is not always the best idea to do always a poker face even in poker ).

About the second paragraph: I agree that probably the level of usage of the denials that humans do was not expected by the original coders and the position that today's usage by the top quarter Civ IV players of them ( and not only of the bribe ones ) is a little cheaty is clearly defendable and consistent in itself. Just remember that the logical end of the "make the AI look better by making the human player life harder" aproach is the position that bestsss defended in post #45, that is probably too much for this mod, to say the least. And to add, as I already stated somewhere above, the original coders in a similar problem decided to make the AI assume what the human stance is and that was not changed ( or even suggested as a change target ) by your code so far ( not mentioning that I still am not convinced that the human is always the biggest threat to every random AI controlled civ in every random chosen game situation in every randomly chosen map with every possible randomly chosen game options )

 

[TheLazyHase]

Also, the whole theory of "AI should not show hand full" is not more realistic than the way it work for now. Don't forget that a country that prepare to invade is very different than the same country with a big army to defend itself. Especially, a lot of people in the coutnry know that something is going on when they have to prepare the invasion, so the explanation of full hand being the diplomat that see that something is going on is somewhat realistic. They are hole in it, most notably the hundred year war plan, but so is the alternative.

Also, I support rololo on one point : knowing when a human will declare war is somewhat easy, or most of the time possible. The AI has a perfect poker face, so this will detract from my enjoyment of the game, because DoW will be sudden as opposed to planned (yes, even against human, I find DoW predictible ; it may be because I also play a lot Diplomacy where this is the basic ; it's also because I don't play on gamespy but with friend and some AI, so game are long enough to actually have peace period in them)

Another reason for which I believe that should not be changed is akin to the rocket jump phenomen in FPS : removing it change too much the game, in my opinion. Better AI nor UP are here to make that kind of change, as I understand it.

 

[MartinHarper]

Military plans are highly confidential, hence Pearl Harbour. You don't get to find that stuff out merely by having diplomatic relations. It requires espionage. Stealing Military Plans would be a fine espionage option to add in a mod. One suggestion: it is a mission you could only perform on an AI, and would tell you the location of their military units, whether they are planning a war, and what city/landing site they are planning to attack. It is modified by the normal modifiers. If the civ has built the Pentagon, it can only be performed in that city.

You can still judge that a civ is increasing its military, either directly or via the power graph. With good scouting, you can see the type of military units a civ is building, and see their promotion types. You can still see how good a civ's relationships are with other civs. This is all a far more realistic way of discovering war plans than "hands full".

The AI has a perfect poker face, so this will detract from my enjoyment of the game

It looks to me like BUG shows the war plan icon regardless of whether you can talk to the leader, so presumably it will also show the icon regardless of this change to the AI. Alternatively, you could use the chipotle cheat code.

 

[r_rolo1]

It looks to me like BUG shows the war plan icon regardless of whether you can talk to the leader, so presumably it will also show the icon regardless of this change to the AI.

Previous versions of it ( up to 4.0 IIRC) showed it in that situation , but the up-to-date one does not.

And "hands full" is not the same as announcing the war plans ( I'm getting tired of saying this ), but only a sign the civ is gearing to war. As you picked the Pearl Harbor RL example, I'll use it to show the diference:

The US had full knowledge that Japan was preparing a war up to December of 1941 ( mainly via the German embassy in Tokio, that had a American mole in, that correcly pointed out that Japan only had fuel to make a significant military operation until the end of 1941, due to the US sanctions ). The japanese diplomacy in that period clearly pointed as well that Japan was gearing to make a military operation against the French Indochina and the Dutch Indonesia to "protect" them of the British ( given that both France and Netherlands were being puppeted by Berlin ... as ANZAC-British troops had occupied the Portuguese East Timor without asking anything to Lisbon, further Brit actions of the same style in that area seemed likely ) and in the way get a more stable oil source, so EVERYONE knew that Japan was up to something even via the diplomatic channels. What was not expected by most of the people in Washington is that the real target of the Japan buildup was not primarily going to Indochina and Indonesia (they got there after Pearl Harbor ), but to Hawai try to take down the US Pacific fleet in one coup.

So, Japan clearly announced that it was going to do something ( the exact equivalent of the Civ IV "hands full" ), but the US governement was unable ( officially atleast... ) to take the right concusion of even who was the target due to the fact that it looked more likely that the target was other than them ( a exact RL equivalent of the game situation I described some posts ago )

 

[TheLazyHase]

Military plans are highly confidential.

The fact that a war is going on is not that cionfidential. In pearl harbor, Aerican were knowing that the Japan were to attack someone. ANd in almost all war, neither side was surprised when war break. They may be surprised by the attack plan, by the target, but not showing that you plan a war is no small feat, and very few people have achevied that.

 

[karadoc]

I reckon the AI should try to hide its war plans if it can. By announcing war plans to the world as is currently the case, the AI is ultimately more predictable than it otherwise would be and thus easier to defeat. It is a fault in the AI, because it's the AI's actions that are giving this predictability.

If you believe that humans are more skillful players than the AI, then I think it makes sense for the AI to try to play like a human does. So hide the damn war plans. Some people are saying that in real life there would be signs that a country is planning for war, and hence we should get warning about it in civ - well to that I say that civ is not real life, and again I point out that if you are playing against another human than you will not get any such warning. The goal of this mod is to make the AI better at playing the game. I don't think the goal is to make the game more like real life.

Moving units around the board, choosing techs to research, choosing civics, etc. are all parts of the game that are important for good strategy. But diplomacy is important as well. The change we are talking about is essentially about making the AI more skillful in diplomacy.

These changes will make the AI less predictable, but not less rational; they will make the AI more difficult to defeat, but will not give them any unfair advantage. To me it seems that although a lot of people have come to rely on the WHEOOHRN warnings, these changes should be made for the good of the AI. I'm sure there are other AI weaknesses that strong human players commonly exploit to great advantage, and we should make it our goal to correct them all. I frequently make use of the fact that the normal AI likes to throw away naval units by attacking with poor odds without strong reason. I use this AI weakness to my advantage, and it plays a role in my best strategies in defeating the AI. I also use the fact that the normal AI's war plans are predictable, and so I know when and where to beef up my defences. I exploit these weakness in the AI so that I can beat them. I don't see any great philosophical difference between the two weaknesses I have described, and so if you think one should be fixed then I would assert that the other should be fixed as well.

Better AI all round means making the AI more difficult to defeat without making them cheat to achieve that goal. So make the change.

 

[TheLazyHase]

Some people are saying that in real life there would be signs that a country is planning for war, and hence we should get warning about it in civ - well to that I say that civ is not real life, and again I point out that if you are playing against another human than you will not get any such warning.

You forget two thing :
* first, the realistic thing have been said for hiding the hand full. It's simply not a good argument because the realistic thing would be between the two opposite.
* second, human doe snot hide war plan to human. Predicting when an human will attack is far easier than for an AI, if you don't use the hand full sign. It's partly because the AI declare war randomly and does not know very well how to handle diplomacy, but realisticaly the Civ4 AI will never be able be good enough not to wage war randomly. So, for an human hiding hand full is a bit much in my opinion ; and in AI vs AI it change nothing since AI don't use hand full.

As Rololo said, don't forget that the AI have the perfect poker face ; nothing short of a SoD near your border can give away its intention, when it's a lot easier to do with human.

The argument "it make the AI play better" is about as ridiculous as possible. The AI don't change in any way, so it's simply not a better AI with hand full hiden. It may be harder to beat, but you could also give it one additionnal settler at the start, it's the same kind of fake difficulty.

And lastly, the AI should be enjoyable. I don't think that having an AI that declare war without any indication beforehand is more enjoyable than one that can give hint if you seek for them. You may disagree, still it's something to consider in that kind of change. Civ is about long-term planning ; random war outbreak will never be something I want to see more in civ4, and that's exactly what would add this change.

 

[phungus420]

It may be harder to beat, but you could also give it one additionnal settler at the start, it's the same kind of fake difficulty.

Actually the fact the AI declares it's intention to attack, especially with the BUG mod WHEOOHRN display, is the "fake difficulty" feature in the equation. It's a forced handicap in favor of the human, one which I don't see much point in keeping. I'd rather remove this "cheat".

 

[TheLazyHase]

Actually the fact the AI declares it's intention to attack, especially with the BUG mod WHEOOHRN display, is the "fake difficulty" feature in the equation. It's a forced handicap in favor of the human, one which I don't see much point in keeping. I'd rather remove this "cheat".

Because, clearly, you should be unable to plan anything that need peace, it's the way the game is built upon. It's semantic to say it's an handicap for the AI or that removing that is an handicap for the player ; in both case you want to raise the difficulty not by making the AI play better but by making the human be more miserable. Because the only real change is that you must be prepared to have a war anywhere, there is simply no other way to not have a lot of problem. That mean you can't rely on fast troop displacement to keep your border. That mean you can never trust the AI, simply because it can randomly wage war at any moment, even if you're above the war treshold in relation. Yay, clearly it's that the problem of civ4 AI, and not the myriad of way the AI does bad move.

Still, you don't talk about the real problem : removing hand full mean removing the only breach in the AI poker face, that even human don't have if you actually talk with them or keep a little attention. It's the same problem as the AI presented by TMIT that use very cheap diplomacy rule to screw over the player, and hand full is so used by everybody that removing it is simply out of the scope of better AI.

 

[Munch]

I implemented this change the other day and the outcome was predictable. My neighbour Tokugawa would refuse when I tentatively hovered over the declare war on so-and-so in diplomacy, stating rationally 'we just don't like you enough'. Sure enough, it was me he was planning on attacking, and while I could tell that he had an army, he didn't make it explicitly clear in his diplomacy screen that he was about to mobilise it against someone.

Personally I think this is an important gameplay change. As others have mentioned there are other ways to tell that an AI is gearing for war, by observing them, rather than relying on a broken diplomacy gimmick. I'm not sure that this change is the domain of Better AI as such, but I believe it is still an important exploit fix.

 

[r_rolo1]

@TheLazyHase and phungus

Believe it or not, this thread is not about finishing with the "hands full" hover I know,most of the posters decided to think otherwise, but the thread is pretty much about the order that the denials to be bribed to war are presented ( and if that change will actually make the AI "better" in any sense ). In that sense, I already gave my opinion: the logical output of this if taken to the extreme is to remove all the hover text , because they all give you intel on the AI in one way or another ( as bestsss pointed out in his post a page or two ago ) and not even that would make the AI smarter, just less acessible.

@ TheLazyHase

I never said that the AI makes a perfect poker face , but I agree with you that the AI making a poker face ALL the times is not in the line of what the original coders implemented ( otherwise we would not have hovers at all, just the red in the text indicatind that the AI did not wanted to talk about it ... far more easy to code ). My point is that the AI as it is today is too stupid to use the weapons it already has in this area ( that is, understanding what the other AI in game want to say with their denials and also create situations where they leverage their own denials ) , and because of that it can't do a poker face when making one is good for them and not do one when making a poker face is not the best option.

@phungus

Interesting you call of this a "human handicap" .Your point in keeping it or not is discussible , but if you really consider it as handicap-like, think on this: reducing the AI bonus makes the human smarter?

P:S: @ Munch

That was predictable, wasn't it? Getting Toku to friendly , as we all know, it is not exactly easy .....

I am still debating on calling this a exploit. True , it can be called a exploit against the current AI ( and even then ... ), but we could file a lot of stuff in there that is perfectly legit, as simply making a better MM than the AI The solution that BBAI normally have in that kind of situations is to make the AI more able to use what there is in game instead of making humans less able ( or completely unable ) to do what they could before the change. And obviously this has little to do with the exploit of not-so-well-thought game rules, like pre-3.13 diplo wins ( especially , but not only, via AP ) or Culture wins in No espionage pre 3.19 , so using that term might induce people to think that the situation we are describing and those kind of exploits have something in common besides the name

 

[MartinHarper]

That mean you can never trust the AI, simply because it can randomly wage war at any moment, even if you're above the war treshold in relation.

This change does not alter the AI's DoW% at different relation thresholds. You can continue to trust a Pleased Gandhi, for example. Additionally, if you can bribe an AI to go to war, it will continue to tell you when it has war plans, including plans against you. If that's not enough, you can use the chipotle cheat code or play Always Peace.

 

[jdog5000]

The change as proposed in the first post will be in the next version, with one alteration ... if the AI is already at war, then it will say it has its hands full as the first denial. If it's just at the planning stages for war, then it will check its other denials first and only say it has enough on its hands if it would otherwise consider your offer.

 

[karadoc]

Umm.. In my excitement I somehow didn't notice jdog5000's last post. I guess most of what I've said here isn't relevant anymore. I'll just put it in spoiler tags.

Spoiler :

TheLazyHase, I really don't understand how you can possibly say "human doe snot hide war plan to human." I don't know which humans you've been playing with. Humans don't warn each other when they are preparing for war. It would be stupid to do so (unless you were informing an ally of your intentions, or making a threat, or something like that.) Also, a strong human player will not assemble his attack force in plain view of the enemy if he intends to launch a surprise attack. Against a human, you need to scout to see their army. You need to think about whether their forces are building up, and where they might go when they are ready. This is the kind of information that tells you if a human is preparing for war. I believe it would make sense if you had to do the same thing for the AI. Why should we force the AI to be so stupid as to just tell you in words that it is preparing for war?

If a human ally asked me what they would have to give for me to go to war with a mutual foe, I might say "Sorry, I'm already preparing for a different war" But if an _enemy_ asked me to go to war against one of allies -- why the hell would I tell them that I'm already planning for war? I'd just say "no. That's my ally. Go away." The current AI seems to think it's a good idea to announce to the whole world whenever it is preparing for war. It doesn't just tell its friends, it tells the person that they are preparing for war against, and I think it's bloody stupid.

attitude: annoyed
"We will never trade with you, our worst enemy - by the way, we're preparing for war at the moment. So watch out!"

 

[TheMeInTeam]

You can probably just memorize pathing tendencies by the AI and then park units to watch it mass troops (works on everything but toku). It still doesn't make sense that the AI returns the EXACT SAME MESSAGE for BOTH CURRENT WAR and WAR PLOTTING. In light of that, I like jdog's answer especially as a trial.

I don't look at removal of WHEOOHRN in that context as making human life harder in the same context as Rolo. I see it much more along the lines of "removing something unintended", quite similar to the removal of wall chopwhip overflow although since we've got jdog at the helm here likely far better executed than what 3.19 did with it's baseball bat-edge to cut down overflow .

One must be careful though. If this forces players to figure out AI pathing tendencies + how reserve vs offensive scripts work, and that becomes common knowledge, we might actually wind up becoming *more* certain of AI targets and warplans...humans improving more than the poker face improvement of the AI!

I'm still in favor of tweaking this as jdog did though.

 

[Roland Johansen]

Telling your opponents that you're planning to go to war is a huge strategic error, so I'm happy to see a reduction in the frequency that the AI will use this diplomatic message. However, I wonder why it's left as a diplomatic message in cases where other diplomatic messages don't apply according to current AI programming (it's the last choice of various diplomatic responses). Why can't one of the other messages be applied when the AI is planning to go to war, none of the other responses apply but it's not obvious for the human opponent that they actually don't apply.

For instance, if in the future, this project expands the denial_no_gain into areas where the AI currently considers war but shouldn't, then the message could be used as a catch all remaining instances message. For instance, if an AI is using half of its production power to build spaceship parts or wonders then it probably shouldn't consider war with a civilisation that has an equal power rating. Or if an AI is on the verge of inventing a new military technology, then it probably shouldn't consider war until after the invention of the military technology. These situations aren't very recognisable for a human player so if they're bundled together with an AI actually planning a war in a single message 'denial-no-gain', then it will be especially hard for the human player to detect the AI plans purely from diplomatic messages.

 

[MartinHarper]

For instance, if an AI is using half of its production power to build spaceship parts or wonders then it probably shouldn't consider war with a civilisation that has an equal power rating.

Yes, a civ that is heading for a spaceship or culture win will want to be defensive, rather than start wars, as will a civ that is near to finishing a wonder. I think "We have enough on our hands right now" is the most accurate denial message in that situation.

If an AI is on the verge of inventing a new military technology, then it probably shouldn't consider war until after the invention of the military technology.

What if I'm willing to bribe the AI with that military technology, and more?

 

[Yakk]

They would still need multiple turns to exploit their military technology.

 

[DaveMcW]

Why can't one of the other messages be applied when the AI is planning to go to war, none of the other responses apply but it's not obvious for the human opponent that they actually don't apply.

If a human is watching turn-by-turn, any change in the AI's willingness to be bribed is a danger sign.

Lying about AI motives simply confuses humans in innocent cases, without adding any real security.

 

[Munch]

For instance, if an AI is using half of its production power to build spaceship parts or wonders then it probably shouldn't consider war with a civilisation that has an equal power rating.

I wouldn't be so sure; if the AI is racing towards a space race victory, chances are they might need to attack someone else to prevent them winning say, culture or space.