• We have updated our Community Code of Conduct. Please read through the new rules for the forum that are an integral part of Paradox Interactive’s User Agreement.

unmerged(29041)

Amnistiado por viejuno
May 12, 2004
5.496
0
I am conducting experiments on how AGCEEP works, and the situation of AI files. The following test was done using unreleased 1.39 AGCEEP. Not very different from 1.38 at least in terms of performance.

I have introduced several modifications to this copy of AGCEEP. But the only one that could affect this test is a modification to the HYW that forces a historic development:
-Henry always claims the crown of France
-The Dauphin always kills Philip of Burgundy
-Burgundy always signs the Treaty of Troyes


The tests consists in exchanging each and every AI file by the following one:
Code:
region = { } 
area = { } 
expansion = 0 
tradingpost = 50 
neighbour = 20 
enemies = 0 
traders = 20 
monopoly = 0 
war = -500 
ferocity = no
counterreform = no
combat = { } 
base = 0.5
front = 5.0
conquer = { 
	enemy = 1.0 
	supply = 0.1 
	distance = 1.0 
	owner = 5.5 
	notsupply = 1.0 
	base = 1.0 
} 
garrison = { 
	fortress = 5.0 
	strategic = 1.0 
	size = 5.5 
	supply = 9.0 
	war = 1.0 
}
Disclaimer: I am in no way proposing this AI for AGCEEP nations. It is an extreme case and the results might not fit our needs. I am just going to describe its effects so we can learn and improve the AI mess.

The test was run only until 1485 and my comments only go that far. As a result of the AI change, no nation ever declares war to another, regardless or cores or relations. I will insist on this: Not a single war was declared by any AI that was not present at 1419 or commanded by an event. The world is filled with absolute pacifists. Exploration and colonization are also abolished, but this is not pertinent to the test.
As the only wars are those provoked by events, this test becomes a tool on how AGCEEP performs in the absence of wars. The results are very interesting:
  • For areas filled with minors (HRE, Italy), the result is very good, as they now survive except when diplomatically annexed. Wars are very bad for minor survival.
  • For successful nations in history we have three cases. The first is those nations that depend on war to succeed. Russia and the Ottoman empire do need wars, and they should get them. Their neighbours however would be better as defensive pacifists, as they make better victims.
  • Nations that should not expand, like England, are perfectly maintained in the absence of wars, and also their neighbours like Scotland, that are protected from untimely agression.
  • Curiously enough, there are a bunch of nations whose performance has been heavily scripted in AGCEEP. This obviously includes Spain and Austria, but the most extreme case is France. I was most surprised to see that by 1485 France had acquired absolutely all her historic provinces with the exception of Orleanais, in the total absence of wars. I call this group the scripted nations, as events give them basically all they need. In general, Western and Central Europe are scripted, while the rest of the World is generally not.
  • This is the only game that I have played in over a year that has seen a completely historic Burgundy (until a particularly nasty event came along). Food for thought.
  • Provence and Brittany performed historically. However France soon started diploannexing some neighbours she shouldn't: Savoy and Switzerland.
  • There were very few defections, as nations in peace have a better control over their provinces
  • The overall level of armies is however very low, as nations do not build armies unless at war. It was funny seeing Burgundy get along with just 5k for 8 provinces.

If you are still with me, I am going to elaborate on some aspects:
1) The AI situation in AGCEEP is chaotic. It needs urgent fixing a lot more than a new map in terms of historic performance.
2) For scripted nations we have two programs. One is the script that will give them everything they need. The other is the cores, hitlist and aggressivitet. Obviously having both simultaneously causes a conflict. Either we de-script them, or we take the aggressivity out of them and let the script work unmolested.
3) Obviously the absence of wars will turn the mod into a great history simulation, but an awful game program. Who would want to play such mod? Not me. However it is possible that we can fine tune the war command in the AI so a nation will declare 90% of her wars to countries holding her cores or present in the hitlist, and I am going to conduct further experiments in to determine if such value exists. The general level of agression and the number of wars would go down quite significantly and the mod would perform better. This might not be acceptable to all, and should be discussed.
4) Some countries should be marked as not liable for diplo-annexation (Switzerland and Savoy in my test) and protected from being annexed by the AI. They are basically the neighbours of big nations and this is doable by a variety of methods. I did it with Navarra by timing her vassalizations. Again we need to discuss if we want to do it.

Since a pic is worth 1000 words, here is one (it will be up for a limited time).
AItest1.jpg

France already diplo-annexed Savoy. She is about to get Caux, Maine, Provence, Bretagne, Armor, Morbidan, Bourgogne, Picardie and Bearn by event. Burgundy will soon purchase Geldre, reaching her historic extension without going to war. For the first time in over twenty games, the inheritance of the Netherlands by Spain is within reach.
 
Interesting. AI scripting has never been my strong side, but I would much prefer a good AI script to solving issues with events. I think events is used mostly because their effect is easy to spot and test, while alterations in the AI requires much more detailed testing. However events are very heavy-handed and tend to punish the AI since it can't choose the better of two options in a given situation.
 
Fodoron said:
2) For scripted nations we have two programs. One is the script that will give them everything they need. The other is the cores, hitlist and aggressivitet. Obviously having both simultaneously causes a conflict. Either we de-script them, or we take the aggressivity out of them and let the script work unmolested.
Since RoTW isn't so heavily scripted, i suggest we go with the model good for RoTW, which even includes other parts of Europe for those who might say outside Europe is a different matter. The thing with France imo is over the top.

As for the rest, i am not in favor of using vassalizations, unless they are historic, to stop annexations.
 
Jinnai said:
As for the rest, i am not in favor of using vassalizations, unless they are historic, to stop annexations.
That is not what I said. If you want to prevent ahistoric annexations from historic vassalizations you must limit them to 9.9 years duration. That is what I did with Navarra. For countries that were not historically vassals like Switzerland you can make them activate a program that will break their vassalization to France or Austria every decade if the overlord is not human. Breaking vassalization only affects stability, not relations or alliances, and is pretty silent.
Sort of a protection program for historic reasons.

Regarding what to do with AI program, that can be done on a case by case basis. Some nations can benefit from running a script like France, I don't know, but most are better off without scripting their future.
 
I think that there are certain parts of the AGCEEP that are pretty much crucial to enjoyment of the game if the player plays a country affected by those parts and is attempting to play historically. The most obvious is the formation of the Ottoman Empire when playing as Austria, but the Spanish Netherlands is another, as is a strong (but not a rampaging diplo-annexer) France.

I think that we need to identify those events in the AGCEEP which the player wants to happen in almost every single game, and script the parts leading up to those events to ensure that.

In the case of the Spanish Netherlands, that would mean scripting Bugundy as a pacifist AI.

Those aspects of the game which aren't crucial for the enjoyment of the players of other countries (like Russo-Swedish conflicts, maybe), can do without being scripted (that is, CBs rather than war commands).


But on the other hand, would is be possible to have a scripted history-simulator as an option at startup? Having everything around the player go exactly historical (pacifist AIs, all wars scripted, maybe even A options only in events) would be a nice option for some games and some players.
 
A game without wars wouldn't really be desirable, but a game without all those ahistorical, completely meaningless wars that we have now and that actually make the game easier for a human by making it less likely for the historical majors (who have the potential to be tougher opponents because of their leaders and their beneficial events than historical minors) to form correctly might actually be a more challenging and thus better game. I think we should as a general rule prevent the AI from declaring war with -500 values and only use positive war values if the given country absolutely needs to declare wars (like the Ottomans or Russia) and we are certain that with "free" DoWs the country will not perform much worse than with scripted ones (not free province cessions, but events with the war command at the correct time).

I for one would certainly enjoy a game where the big historical conflicts, especially the Thirty Years War or the War of Spanish Succession that you right now never get to see in a 1419 game because until that point everything gets messed up by ahistorical wars and annexations, would finally take place with a certain probability.

As for the "annexation-blocker", this surely would be a controversial change, since both its benefits and its drawbacks are quite obvious. I wouldn't really like different treatment of the player and the AI here (and the check if the liege is AI or human would always require two events - one for the liege that checks if he is AI and has vassalized the country in question and one event for the vassal triggered from the first event that breaks the vassalage). The best thing here would be having the "annexation-blocker" as an option.
 
Last edited:
Reveilled said:
In the case of the Spanish Netherlands, that would mean scripting Bugundy as a pacifist AI.
That doesn't work. I have tried it and Burgundy is still eaten away by coalitions of wild minors and by France. If Burguny's expansion is scripted (as it is), then not only Burgundy, but also all the neighbours in the area have to be pacific, or at least not geared against Burgundy.

Twoflower said:
As for the "annexation-blocker", this surely would be a controversial change, since both its benefits and its drawbacks are quite obvious. I wouldn't really like different treatment of the player and the AI here (and the check if the liege is AI or human would always require two events - one for the liege that checks if he is AI and has vassalized the country in question and one event for the vassal triggered from the first event that breaks the vassalage). The best thing here would be having the "annexation-blocker" as an option.
I proposed it as AI only, because the player relies heavily in diplo-annexation to ahistorically expand below BB level, and that should not be prevented. But self restrain from ahistoric diplo-annexion is a 100% effective human-only "annexation-blocker". I don't think anybody enjoys France and Austria diplo-annexation of half of Europe, neither historicists nor expansionists.

It is actually easier than that. Austria and France, if AI only, activate an AI-only program in the protected countries early in the game. From then on, the protected countries will break vassalage from Austria or France periodically. If no vassalage, no event. If vassalage, it is broken and everybody lives happily. Small countries recover stability quickly, and is a small price for saving their lives. If a country abandons the program because it is targetted for historic disappearance, the events stop coming. Only countries that can likely get a land border with the offenders need protection. Milan, Genoa, Savoy, Switzerland, Bavaria, some German minors

In any case, an "annexation-blocker" would just be the icing in the cake. The AIs need some serious fixing, and that is a lot more important.
 
Fodoron said:
That doesn't work. I have tried it and Burgundy is still eaten away by coalitions of wild minors and by France. If Burguny's expansion is scripted (as it is), then not only Burgundy, but also all the neighbours in the area have to be pacific, or at least not geared against Burgundy.

Well, at least with German Minors, we want them to be pacifistic, right? The Holy Roman Empire remaning as such, as opposed to the Monster Empire of a Lucky One Province Minor, is another one of those things that I at least would like to see.

As to France, is Burgundy on her Hitlist? If so, we can remove her, if not, then I assume no-one else is (or France would target them instead), and we can temporarily put France on a Pacifist AI until a time when there is more war going on. In any case, a France of historical strength is another one of the things we want, so controlling an AI France should be a priority anyway, methinks.

It is actually easier than that. Austria and France, if AI only, activate an AI-only program in the protected countries early in the game. From then on, the protected countries will break vassalage from Austria or France periodically. If no vassalage, no event. If vassalage, it is broken and everybody lives happily. Small countries recover stability quickly, and is a small price for saving their lives. If a country abandons the program because it is targetted for historic disappearance, the events stop coming. Only countries that can likely get a land border with the offenders need protection. Milan, Genoa, Savoy, Switzerland, Bavaria, some German minors

In any case, an "annexation-blocker" would just be the icing in the cake. The AIs need some serious fixing, and that is a lot more important.

Should the annexation-blocker include a revassalisation event for France or Austria?
 
When deciding which AI files to make pacifistic and which not, I'd propose assigning countries in certain periods of time to one of the following three groups:
  • Pacifist: The country didn't declare any wars, or at least none that can be represented well in EU2 (i.e. minor feuds between HRE states that would lead to ahistorical province annexations in EU2 don't count). These countries have high negative war values and don't get any events with war commands.
  • Expansionist: The country in this period of time expanded gradually in one or several areas through relatively small, local wars. Good examples for this are the expansion of the OE on the Balkans in the 15th century or the Mughals in India. Here, it is preferable to let the AI pick its wars freely, i.e. give it a positive war value, place the countries in the desired expansion area on the hitlist and give it cores on the area.
  • Large-scale wars: the country during the timespan in question participated in a big war, usually in a large coalition, that is of some historical significance and prominence and was usually waged for a very specific cause (like people being thrown out of windows or late monarchs leaving a testament that lots of people are dissatisfied with). In order to make sure that these conflicts that I think most people would like to see in their games happen, the involved countries get a negative war value (for the respective period only of course), are, if this is important, put into the right alliances by the "Arras-trick" (vassalising the country that is supposed to switch alliances to its new ally and then immediately breaking the vassalage and restoring the lost stability) and declare war at the appropriate time by event command
 
I agree with the two first groups, but I'm not sure for the last one.

IMHO, large scale wars could (should?) be separate scenarios starting just before and with accurate setups and alliances. Otherwise, it could end in very weird situations.

We can model end of HYW because it is at start of 1419 scenario. I'm not sure it could be possible without problems if scenario started in 1300 or so.
 
YodaMaster said:
I agree with the two first groups, but I'm not sure for the last one.
Neither am I for the same reasons. Dragging nations toward historic alliances when the game conditions could be so different creates unforeseen problems, not to speak of a player suddenly deprived of his allies that he could have long term plans for.
 
Twoflower said:
When deciding which AI files to make pacifistic and which not, I'd propose assigning countries in certain periods of time to one of the following three groups:
  • Pacifist: The country didn't declare any wars, or at least none that can be represented well in EU2 (i.e. minor feuds between HRE states that would lead to ahistorical province annexations in EU2 don't count). These countries have high negative war values and don't get any events with war commands.
  • Expansionist: The country in this period of time expanded gradually in one or several areas through relatively small, local wars. Good examples for this are the expansion of the OE on the Balkans in the 15th century or the Mughals in India. Here, it is preferable to let the AI pick its wars freely, i.e. give it a positive war value, place the countries in the desired expansion area on the hitlist and give it cores on the area.
  • Large-scale wars: the country during the timespan in question participated in a big war, usually in a large coalition, that is of some historical significance and prominence and was usually waged for a very specific cause (like people being thrown out of windows or late monarchs leaving a testament that lots of people are dissatisfied with). In order to make sure that these conflicts that I think most people would like to see in their games happen, the involved countries get a negative war value (for the respective period only of course), are, if this is important, put into the right alliances by the "Arras-trick" (vassalising the country that is supposed to switch alliances to its new ally and then immediately breaking the vassalage and restoring the lost stability) and declare war at the appropriate time by event command
That realy doesn't cover those that were agressive, but not expanisonist and were not involved in anything outside of regional wars (though important ones than need to be in).
 
YodaMaster said:
I'm not sure with "annexation-blocker" neither. I understand it is for AI only but remember vassal-breakvassal commands are really buggy.
Well this is just a thought, not even what the thread goes about, and not related to my test as it does not depend on the AI. We can just think about it and discuss it on a later time.

Norrefeldt, in my opinion it is not just the minors who need to be cooled down, from their suicidal instinct, but in general everybody that does not need to expand through war.

My proposal would be to get rid of 95% of the AI files. Set a couple of pacifist AI files (counterreformed need their own), with a war setting not a -500, but at another value that testing determines low enough so countries having it very rarely declare war, but that the occasional war is not completely ruled off. Colonizers need especific AI files, and so do it countries that need to expand by means of war, or countries that contain revolters, as otherwise they do not try to reincorporate them. For their own good, minors in minorlands (HRE and Italy) should be ultrapacifists.
When needed, a country moves from pacifist to expansionist or the opposite by getting a specific AI by event, or returning to the Pacific Pool.

We must be aware that by becoming more historical, the mod also becomes more predictable. Perhaps Fantasy should disable the historic AI set-up.

And the HC needs to discuss what to do with heavily scripted countries, France being the prime example. I am not sure what to think of that. After all we are watching a movie, but in the other hand it works quite well.
 
Jinnai said:
That realy doesn't cover those that were agressive, but not expanisonist and were not involved in anything outside of regional wars (though important ones than need to be in).
EU2 cannot cope with agressive nations that do not take land, and if you release a wolf between sheep he is going to have a field day. When weak he will be safe from agression, and when strong he will eat left and right.
 
Norrefeldt said:
I do think we should cool down minors that historically didn't declare one single war.
I think we should very much reconsider their inclusion since they will typically get into war one way or the other. Unless they really played a significant role what does their inclusion add to gameplay?
 
Fodoron said:
EU2 cannot cope with agressive nations that do not take land, and if you release a wolf between sheep he is going to have a field day. When weak he will be safe from agression, and when strong he will eat left and right.
However that is what Rajputana is. What land was taken was vassals, mostly it was just tribute and to not have them DoW would be 100% unhistoric as would be removing them (their military might is underresperented by the land they control). They did not really take extra provinces, but were always agressive both internally and externally and they were too big and important a force to be abstracted.
 
Sute]{h said:
I think we should very much reconsider their inclusion since they will typically get into war one way or the other. Unless they really played a significant role what does their inclusion add to gameplay?
War alone should not be the guiding criteria. IMO AGCEEP has given it too much premiere importance. AGCEEP is not a wargame. It has wars modeled, but its not Europa Universalis: Total War
 
A problem with aggressiveness is that the nations often dow the wrong guys. A Poland is often much more eager to wage war on HRE minors then on Russia like it should, just like Spanish and French will rather obscurely kill Scotland then go after each other - AI has been programmed to pick on the weak, which is not always historical. Historically nations made war on each other to preserve the balance of power - something to wish for EU3, probably...

A problem with lack of aggressiveness would certainly be a lack of consequences to an ahistoric development. My Bremen may conquer half of Germany, yet the silly Prussians, Austrians, British and French will follow their scripts, ignoring my empire and allowing me to defeat them one by one.