When does the AI choose for war?

[Kensai]

This is probably something only one of the Philippes can accurately respond as I guess it is hardcoded in the AI's (Athena's) logic. Albeit my two-year-plus experience with PON, I have not been able to find a bulletproof pattern. When does the AI decide for war, if a declaration is not scripted? I am suspecting that there is some kind of magic formula that checks on

• relations between nations (F9)
• military power ratios (F10)
• existence of scripted CBs ($diCBShort, $diCBLong, etc) since it cannot forge its own

So, in a working scenario, if an AI nation "senses" that its power ratio over another nation is high and it holds a valid CB against it, if the relations are bad enough (with higher probability the worse they are) it can opt to declare war. Am I interpreting this right?

[Kensai]

I suspect something similar for the peace process. Obviously here should play a role some other parameters: including National Morale (NM) and Victory Points (VP) lost. Relations are stuck to -100 in war so they should be out of the formula, but perhaps military power ratios still have a use (albeit more flexible) to make the AI understand when it is fighting a hopeless war or not.

[loki100]

Heres a guess. In a crisis, the AI will risk war if it thinks the (continental?) power ratio is in its favour. In my AAR I've seen one AI-AI war trigger off a crisis (Prussia vs Britain) and I think Prussian estimated that Britain was weak. I've seen Germany play the mobilise card against me when I had few troops in Italy (all were in a war with the Ottomans), so again the force ratio was really in its favour. Finally I've just seen the US AI use that card (which if reciprocated guarentees war and risks it in any case) against Britain - now badly weakened by wars with Italy and of course very weak in N America.

I've seen Russia attack the Ottomans with no script once, so obviously a CB for territory and a judgement about relative power - in that case I was beating the Ottomans in my own war with them.

So that does suggest that weakness is a major part of this judgement, but I think it is using land power on the continent - so its not just F10.

I've had no success with telling the diplo AI to look for war, so I think that power ratio makes it stop. Here of course, defensive alliances are so important in shifting the perceived balance of power and since they are immortal (such as Italy-France) then that probably makes say Germany unwilling to do war with France (I had no 1870 war for example).

So F9 probably matters but power ratio is the key.

What we need is a diplo AI that allows mutual support (such as Austria and Germany hunting together after say 1880) or a degree of idiocy to reflect misjudgements. On the other hand, Europe 1850-1914 was pretty peaceful, so we don't want to see the AI pilling into war after war as it does in Victoria?

As to peace, I've seen pointless wars just drag on (Britain-Prussia), I've seen the AI end it almost immediately and I've seen the AI end when it was clearly losing (ie Britain gave the Boers a colony), so I think here it is ratio and progress (the British-German war saw no action)

[Matto]

A never seen AI join to war where has defence treaty ...

[Kensai]

I think it is using land power on the continent - so its not just F10...

Hmm, so the AI has access to some kind of metric that the human player does not. That's ok for me, generally, but ideally it should not know the exact consistency (land or naval) of the rival power, otherwise it will always initiate advantageous wars. There is already the AI tactical clairvoyance setting if someone wants an all-knowing rival AI, does this happen at a strategic level as well?

What we need is a diplo AI that allows mutual support (such as Austria and Germany hunting together after say 1880) or a degree of idiocy to reflect misjudgements.

That would be cool indeed. Perhaps Athena should be introduced to a level of less sensible choices, some kind of uncertainty or paranoid behavior, something to add to choices not really making sense. And I think there should be a dice, rather small but there, that adds the probability that a nation drops an alliance over time.

On the other hand, Europe 1850-1914 was pretty peaceful, so we don't want to see the AI pilling into war after war as it does in Victoria?

Indeed. World wars should be pretty rare, local remote conflicts quite frequent. WWI happened because of all the interlocking defensive alliances firing in a chain reaction, something that can be awesomely abstracted in PON in its current state. Hoping for that 1910 DLC!

A never seen AI join to war where has defence treaty ...

Perhaps that power ratio formula is not high enough? In theory the AI (to join or not) should use again the formula of war, but with the important change that this time the alliance adds up its sum of power against the power of the enemy. So, by F10 standards, if Austria and Germany are allied and France declares war at Germany, Austria's decision to join this war should be a big fat yes, given that the power ratio will include BOTH Germany and Austria against France, which should be quite high. All this should happen unless relations are too good between Austria and France and the AI decides for the VP penalty.

If the AI NEVER joins, no matter what, we might have a bug.

[loki100]

Hmm, so the AI has access to some kind of metric that the human player does not. That's ok for me, generally, but ideally it should not know the exact consistency (land or naval) of the rival power, otherwise it will always initiate advantageous wars. There is already the AI tactical clairvoyance setting if someone wants an all-knowing rival AI, does this happen at a strategic level as well?

my guess is it uses the relative force algorithm so yes it must know more than it can see. It must be using something like:

SelectFaction = $BUL
RegInit = 1
SelectSubUnits = Theater $Balkans;FactionTags BUL;Domains $Land

SelectFaction = $SER
RegInit = 3
SelectSubUnits = Theater $Balkans;FactionTags SER;Domains $Land

SelectFaction = $MON
RegInit = 4
SelectSubUnits = Theater $Balkans;FactionTags MON;Domains $Land

SelectFaction = $RUM
RegInit = 5
SelectSubUnits = Theater $Balkans;FactionTags RUM;Domains $Land

SelectFaction = $GRE
RegInit = 6
SelectSubUnits = Theater $Balkans;FactionTags GRE;Domains $Land

RegInit = 7
RegSum = Registry 1
RegInit = 8
RegSum = Registry 3 4 5 6
RegEval = 8;<;7

but on a continental not a theatre scale. Now that sort of algorithm just can't work if the AI is only using what it can 'see'?

A never seen AI join to war where has defence treaty ...

me neither, I've seen Germany abandon Austria twice (ok both times when it was at war with GB) and France/Belgium abandoned me when at war with Germany

[Kensai]

loki100, this is a scripted event. I am talking about generic situations. Obviously scripted events should have their own extra conditions for immersion and approach to the historical situation.

As to peace, I've seen pointless wars just drag on (Britain-Prussia), I've seen the AI end it almost immediately and I've seen the AI end when it was clearly losing (ie Britain gave the Boers a colony), so I think here it is ratio and progress (the British-German war saw no action)

Perhaps unseen stuff is happening in trade boxes. The AI sends a lot of ships there, especially in later stages of the game where more nations have large navies. I still find the damage and attrition in MTBs rather low, especially in war. Perhaps even some more warscore should be lost or gained in that places to abstract the seriousness of an economic trade war. If two nations are far away with no intention to invade one another, some wars do indeed drag a lot.

Otherwise, Athena might resolve this again with a peace dice, if a war is dragging for too long without significant warscore changes.

[loki100]

loki100, this is a scripted event. I am talking about generic situations. Obviously scripted events should have their own extra conditions for immersion and approach to the historical situation.

its not, its just an example of the sort of script you can write. Here's one I set up in my own game to try and make Germany and/or GB more aggressive:

SelectFaction = $CMN
SelectRegion = $Anglia
StartEvent = Event_nam_CMN_Anglo-German alliance|1|2|NULL|NULL|$Anglia|NULL

Conditions
MinDate = 1906/01/04
MaxDate = 1920/12/31

SelectFaction = $GBR
CheckAILevel = 1

EvalIsAtPeaceWith = Ger

SelectFaction = $FRA
EvalRgnOwned = $$Alsace
EvalRgnOwned = $Lorraine

SelectFaction = $GER
RegInit = 1
SelectSubUnits = Area $Continents_Europe;FactionTags GER;Domains $Land

SelectFaction = $ITA
RegInit = 2
SelectSubUnits = Area $Continents_Europe;FactionTags ITA;Domains $Land

RegEval = 1;>;2
RegDeactivate = NULL

SelectFaction = $ITA
EvalVP = >; GBR

SelectFaction = $ITA
EvalVP = >; GER

Actions

SelectFaction = $GBR
AddDiploItem = GER;$diShowSupport;CURRENT
ChgFacRelationships = GER;10
SelectFaction = $GER
ChangeFacMorale = 10
SelectFaction = $FRA
ChangeFacMorale = -5

SelectFaction = $GBR
AI.SetDiploItemBias = GER;*passage*;200;*Military*;200;*DipSupport*;300
AI.SetDiploItemBias = ITA;*War*;200
AI.SetDiploItemBias = FRA;*War*;200


SelectFaction = $FRA
AI.SetDiploItemBias = GBR;*War*;300;*DipSupport*;0
AI.SetDiploItemBias = GER;*War*;300;*DipSupport*;0

EndEvent

despite notionally increasing the bias to war from 100 (base) to 300 its made no difference. But it is that code the AI uses for force evaluation.

Perhaps unseen stuff is happening in trade boxes. The AI sends a lot of ships there, especially in later stages of the game where more nations have large navies. I still find the damage and attrition in MTBs rather low, especially in war. Perhaps even some more warscore should be lost or gained in that places to abstract the seriousness of an economic trade war. If two nations are far away with no intention to invade one another, some wars do indeed drag a lot.

Otherwise, Athena might resolve this again with a peace dice, if a war is dragging for too long without significant warscore changes.

don't forget the AI doesn't ships, so by late game its still using what is left of its start fleet + maybe a few that happen by script. In my game, Germany had no ships at all, so couldn't harm GB anywhere.

[Kensai]

I have found the hard way that the Biases are not working adequately, at least not while I was trying to fix the Garibaldi chain and make Garibaldi storm Rome. I was using the regional interest ones, but I guess the diplomatic ones are working in the same way. Perhaps they need to be relaxed a little bit, otherwise at least in PON, they don't reach their threshold levels for the AI to take action.

[Christophe.Barot]

or generalissimo

http://www.ageod-forum.com/showthread.php?33766-PON2-Subforum-help-and-hints-requested-for-volunteer-developing-team

http://www.ageod-forum.com/showthread.php?33766-PON2-Subforum-help-and-hints-requested-for-volunteer-developing-team&p=300044&viewfull=1#post300044


very quickly

fundamentally agree with all that was said :

we don't have enough clues (or very little, Pocus told me AI took alliances, and even support in consideration) of how AI /

- makes alliances
- go for war
- decide for peace

we know that there is a random factor, that relations matter A LOT, that alliances matter, that support probably matter

there is some strength evaluation but we don't know at what scale it is computed, global, local, what percent

it seems vicinity matters (Rumania will be wary of a Russian ready to jump o, Ottioman in Balkans, whatever policy Russia can have)

it seems size and success matters - I note by guess and experience sort of a "badboy" factor - begin to be big, and Montenegro will be wary of your Russia

alliance snowballs (I love you cause you're my ally, I won't ally you cauuse ... you're not my ally)

- ditto bad relations (basically, AI won't make peace with you, even after an eventless war, cause I hate you cause ... you're my enemy) this imho comes for forever alliances plus fact allainces draw towards maximum relations and not just a differencial of equilibrium (balance) level - or a too high one (rubberband towards 100, not towards natural relations + 10 or + 15 ..)

we know that now peace has a slowly progressively higher chance to happen with time (but we don't know on what bais)

I have no hint national objectives would play a role in AI decision (while they should, with safety and balance of power, lead strategy, AI geopolitical analysis is a part of AI which couldn't be implemented)

it seems only land forces are considered (or are global ones) and AI don't cionsider naval threat and matters (guess Corea and Channel are more or less treated as rivers, minor obstacles, but really? i don't know Athena enough to be sure, but clearly, the naval parameter is underestimated and underplayed by AI)

we've had during tests an awful time where AI went berserk and huge lots of aberrant weird wars erupted, most were "cause we're neighbour" type, we had Britain invading USA (till middle west), Germany and France attacking Belgium (not at same time) Spain-Portugal war (why not) and even a Holland-Hannover war. all this still with neverending wars and absence of third parties reactions of course. nothing of it even seeming historical the little bit. I was happy when Philippe T and Pocus put an end to this chaotic nonsense and introduced in next version an AI Bias.

then AI was somehow passive, fact is building from scratch (PON is not like ACW ROP WWI or Napoleonics, you don't have 2 fixed sides - this is what makes time covered by PON unique - EU like - and can't just recycle Athena to another technical fighting environment) an AI led by objectives, alliances, evaluation of comparative and absolute values of objectives (besides fact that VP comes more from industry than objectives - VP balancing subject) DEMANDS MONTHS of FULL TIME round the clock work - and time is taken by debugging, fixing values, balancing game

problem of course is we have not the basic values or algorythms so we apply modifiers on basic values we don't know of ....
this is not the most efficient way to work

I guess threat evaluations could play a role

we could also tinker with areaInterest and local Interest but diplomatic behaviour is not military behaviour (is better applied to have history like behaviour in crimean war) - I saw local interest and tinkering play a role, in my russian game, all of a sudden, half the british army or more was stationing in Ottoman empire, serving to nothiong (they didn't dzeclare war, and starved when Russia or minor balkanic took provinces) - this is obvious result of a commendable effort to involve Britain in defence of the straits, but the effect produced - some sort of NATO-like peacetime garrison, is to fix imho

national morale does not seem to play a part either - I may be wrong

lots of this needs correction of course

we have no clue how much those factors play a role, and we badly need the quantifications, basic values, algorythms - we apply modifiers (AIbias, others) on basis we don't know of

it is excellent that players don't have them (what about historical feeling, fog of war, balance between player and AI) but scripters/modders/codders must have them - else we can't work properly, we have no clue if we make dramatic changes, or won't change situation at all

unscripted crisis are another subject ..

all this would better be in improving PON than in general forum, could we have it moved there

[Jim-NC]

I have actually seen the AI start a war. In the PON redux game, France decided to attack Britain (not sure why, but it did). Britain had a DA with America (which the AI refused to honor), I believe a DA with the Ottoman Empire, and support from several countries. It end up that we had Britain/Ottoman Empire/China fighting AI France.

It was interesting, in that as France had been human controlled, it had a more modern navy (it had ironclads, and some 1870s ships). France sent it's navy out to visit regions (1 fleet sailed around the British Isles for example). It didn't sent the whole fleet into the MTBs. So there maybe a calculation between the navies to determine what the AI will do. There were 2 large fleet battles (1 in the med, and 1 in the channel area).

So don't say never. I seem to remember in the past seeing the AI do the 1 and done war. It declares war, and then the next turn offers peace (which the other AI accepts). No fighting, just the avoidance of a PP penalty. We banned that sort of thing in the multiplayers.

[Christophe.Barot]

I have forced wars to test game in my present game, twice :

- French-Italian / Austrian risorgimento war - chain seemed broken, I wanted to test what happened if Austria actually declared war, was a script, where I set a war between them, just see how the remainder of chain of event would behave once initialk war was triggered (had red Austria was such a chicken, never declared war - not this time either), I was the prussian, so knew i wouldn't strike France in the back

well, once war was declared, that was the French who played chicken, with good reason - as a sudden Austria mobilized (France had no such option, solution for keeping balance could be removing this option for Austria in early game) and got twice the size of french forces (Italy was a small addition) France remained prostrated, the few meetings Austria had with SOME Italian and French forcres were one sided - Italy begged for peace (and got it) almost immediately, France waited a couple of months, without attempting anything, begged peace, got it

second time , ditto script, wanted to help/push Russia in 1877 war, just to get a balkanic game (no Russian war, no Bulgaria, no Balakn chain, no Berlin/San Stefano - no Gortchakov and Andrassy tasting my Varzin cigars, beer and schnaps, scheide - couldn't play the "honest" (cough cough, sorry, the cigar) broker - that was the war which broke, Russia had more forces than Ottoman in her vast Empire, buut Ottoman were highly concentrated in Balkans - when they saw the Ottoman main pile, Russia begged for peace, that the Sultan immediately granted - I don't know if they were half blind against blind fight, but one can't force a Donkey who doesn't want to drink, even a Russian one - morals, it seems very difficult to force AI to make daring but reasonable actions - better be more cautious, sure, but Russia backing from ottoman, weird (especially as Russia was allied to both me and austria, we could have thought she wouldn't fear Britain (which played the great game in Khiva ..) )

a huge hinsight into AI evaluation routine would help, even if all AI geopolitical analysis and reaction must be scripted ... something is rotten into the kingdom of denmark (to be, or not to be a modder, sigh ..)

[loki100]

Worth noting the AI tends to amass new forces in its capital, building a monumental stack which it doesn't break down. So I'd guess your 1877 Russia did have a large army but it was all in St Petersburg (too large to put it all on train) and didn't fancy the walk?

In my game I did the same with 1877 (apart from the cigar theft for the same reason) and it was an easy Russian win but by then the poor Ottomans had lost the Crimea war and been beaten twice by Italy.


But I think there is a lot coming to play around the AI reluctance to war, and one is it doesn't seem to build regional armies - no problem for most of Europe but a big one for Russia

[Christophe.Barot]

a) Heres a guess. In a crisis, the AI will risk war if it thinks the (continental?) power ratio is in its favour.
So that does suggest that weakness is a major part of this judgement, but I think it is using land power on the continent - so its not just F10.

b) I've had no success with telling the diplo AI to look for war, so I think that power ratio makes it stop. Here of course, defensive alliances are so important in shifting the perceived balance of power and since they are immortal (such as Italy-France) then that probably makes say Germany unwilling to do war with France (I had no 1870 war for example).
So F9 probably matters but power ratio is the key.

c) What we need is a diplo AI that allows mutual support (such as Austria and Germany hunting together after say 1880) or a degree of idiocy to reflect misjudgements. On the other hand, Europe 1850-1914 was pretty peaceful, so we don't want to see the AI pilling into war after war as it does in Victoria?

d) As to peace, I've seen pointless wars just drag on (Britain-Prussia), I've seen the AI end it almost immediately and I've seen the AI end when it was clearly losing (ie Britain gave the Boers a colony), so I think here it is ratio and progress (the British-German war saw no action)

a) concur, but not even, AI usually in present setting will barely risk war at all (it's better than former situation when AI got for war for sake of it, but not optimal)

b) defensive alliances seem to influence AI towards more caution, [color="#40E0D0"]while their own alliances don't seem to embolden her[/color] AI will be deterred somehow by defensive alliances seen as a block, but it may depend on relations - one cannot rule out that his own strength and success can make a long term difference in AI behaviour and indirectly dramatically modify balance and behaviour :

- a strong Russia allied to Austria (central asia rival is Britain, Balkan enemy iearly stages s ottoman, no need to fight on several fronts) and Prussia will possibly intimidate Prussia and prevent agressive moves against Austria, therefore weaker prussia (happened to me) and no unification (i never threatened Prussia)
- a strong Germany may make Russia feel unsecure (wonder why, i only cross her territory to smash rebels, then general and troops make home by Pullman on Russian railroads to Breslau or konigsberg - even Warsaw actually, took it from rebels, was not automatically transferred to my ally - no way to give it back, wasn't even asked, is a german objective, "good move" even if unintended, and illogical, hence a script proposed) so Russia shy away from Balkan venture (verdammt, I even restrained from allying to Ottoman to let Russia a free hand, no way ... no longer)

here is a hint : http://www.ageod.net/agewiki/EvalNatThreats apparently this script orders tells us something about AI "reasonong" :
at war and threatened (morale is 50 or lower, or enemy land units are worth 1.25 or more own's value).
at peace but threatened (nations with hostile relationships (-25 or below) have land units combat value worth 1.25 or more own's value).

looks like size of army and relations play a role (presence or absence of coveted objectives, or annexable provinces, not !)

c) unless we have a coder with hours of available time for this (which isn't a priority, priority is reduce AI running time) we'll be reduced to scripts (correctly written and extensive set of scripts, especially with actors, choices for players, may do wonders ... BUT ...
we'll need the basic values and algorythms ..
and I fear absence of geopolitically developed diplmomatic AI will need LOTS of scripts to compensate, which will load the computer heavily (computer can stand them, but they'll slow runnin g time again)

idiocy factor is an important historical factor : it especially intervened in crisis - hubris :

basically it is the reverse of Diplomatic Genius (Cavour, Bismarck) who, rather than "dominating the crisis", created it ahead, creating such situation before crisis happens so enemy cornered himself into an unpalatable or erroneous choice - declaring war unknowingly that France was roaring behind one's shoulder and had a SECRET alliance, or provoking France to save an (intact) honour, while having taken all countermeasures (divulgating French schemes about Belgium to British, ensuring that Austria knew Russia was watching their moves, and even that Austria had nothing to gain to a war any longer

idiocy factor (Menchikov at Constantinople, Austrians versus Piedmont perhaps (in 1866 they had no choice), The "Sphynx of the Tuileries" taking his feet in his own devious schemes - after, hubris was rather the chamber of deputies than Napoleon himself, sick , Nicholas II in 1904 against the "dwarf yellow monkeys" (indeed he spoke about them that way), the Kaiser trying to impress British with german fleet (indeed they were, but not with anticipated results), or overplaying his hand with Russia (OK, he had to keep Austria under his wing, still ..), or pushing Austria to maximum demands on Serbia )...is a case where all cornered themselves

question is how to take into account the "idiocy factor" in a game is a recurrent one in designing (who would launch a 2 front war, rush into belgium when panzers close to Sedan, or charge with bayonets when you face trenches and machineguns, else ...) - and not that simple:
- if you force a player to an idiotic play, he'll feel usually frustrated (I don't rule out doing it, but I expect that) - best and most simple is creating automatic situation (war, your nation demand it, our national morale is > 120 ! too bad ! period !) as you get the wanted result
- If you let the player free, but apply the idiocy factor to AI only, you properly sink the AI (which doesn't need that- game is complex)
- if you don't plan anything, you remove usually a lot of context, constraints which plagued the ruler (don't think Tirpitz pangermanists Kolonialverein and Konzerns had nothing to do with fleetbuilding and that it was just an idiotic monomaniac spontaneous passion for yachting in battlecruiser category of the Kaiser ..)
- best is usually to build an encitement, positive or negative, sort of a "proposal you can't turn down" (Godfather style), really ...; either you make it attractive, or you make alternative unbearable, or you make it unavoidable, or .. (for some, a one out of 1 chance drawing a French "elan" event, where French morale boosts to 150 may do the trick, challenge and so on .... or losing 30 000 prestige points may be worse than Alsace-Lorraine, with a tiny possibility to keep them ..)

d) I have seen both, two factors imho, random, and duration of war (2% more per turn If I remember Pocus well, in a patch). I must also tell AI consider all wars are absolutely independant, she'd offer me wonders when I crush her mercilessly, but wouldn't even try to make peace with my own allies ... (see Austria versus Russia in Vienna, and Serbia) absence of alliance play must absolutely be fixed

[Christophe.Barot]

loki, you've a typo in your script, just in case

EvalRgnOwned = $[color="#FF0000"]$[/color]Alsace

[Christophe.Barot]

it goes adverse to my former remark, in my second game, all of a sudden, AI ran amok and 2 big wars erupted

one between [color="#FF0000"]Austria [/color]and [color="#40E0D0"]France[/color], for a reason I can't explain (no crisis, not even by module), no common border, sure relations were very very low, sure Austria (like Russia) is my +100 relationships ally (Italy is a common ally of me Prussia and France) and war erupted (I realized when I lost 900 VP for not declaring war (DOW requests should be more "underlined", either by a pop up, or at least by a red color, so plmayer don't miss them, playing mechanically, those important messages drowned into the hundredths of trade ones (redispatching and powerful iondustrial Germany is now the n°1 hub in world trade) - basically Austria and France don't get along at all (as a Player, gee, as reviewer of balance and scripter/ free volunteer "developper", it's OK but I'm less enthusiastic - Austrian hostility makes little sense, goes beyond supporting needed ally- Austria has nothing to gain)

one following a random crisis between [color="#008000"]Russia [/color]and [color="#FF0000"]Britain [/color], which followed one for Afghanistan (an accurate historical crisis spot, i noticed with pleasure), and had resulted, besides the usual random prestige redispatching, to an actual in game consequence, a total drop of relationships (aroud - 99 - 100) or skyrocketing rise of hostility, depending on point of view. Result is that Britain, in spite of my full support to Russia, declared war to her;

the good thing is AI isn't sleeping

still this, imho, before more information I'm expecting, indicates a [color="#FF8C00"]diplomatic system mainly based on relationships[/color] (which, imho, is a bit tautologic, snawballing, the worse the relations, the stronger the tendancy to worsen, is is not based on global situation, field analysis, but pure relationships - even if I guess there are rubberband mechanism as well as influence of at least national provinces owned by competitor) [color="#FF8C00"]rank in VP[/color] (sort of badboy) - as I see alliances are way more difficult to get once you are ahead, or get close, and regime; [color="#FF8C00"]vulnerability[/color] is also an attraction to war, but should not that way.

seems threat is evaluated as a combination of relations, victory and power, one, other , or both, while alliances seem little or not considered in threat evaluation, and gain sometimes not at all.

the fight over Kabul makes sense, theorically, but NOT if one consider that
a) Britain WAS the owner of kabul - a war would have made sense had Britain to chase Russians from there, not as there was some British occupation of Khiva (not a British goal) so[color="#FF0000"] no gain of objectives to expect[/color]
b) Russia had allies, British was [color="#FF0000"]isolated diplomatically[/color] - not only this but Russia had POWERFUL allies - namely me (Austria won't cross Russian steppes towars Hindu Kush - nor will I ... but I'm building a huge MittelAfrika, whre I just ousted Britain from Nigeria, am a competitor of Britain in East as well as in South Africa)
c) [color="#FF0000"]I am ahead of Britain[/color], not only in terms of VP but also in terms of [color="#FF0000"]military threat[/color] - not only do I possess the biggest army in the world (after a mobilization, seems Britain catched up a bit, at 87% or so of my miltary might, but I'm still ahead) but also I am the only one with a modern fleet in game (except Japan, but thgis is a script)

war happened immediately after I joined my Austrian ally - an alliance is something to protect - therefore I was at war with France (in order to help my Austrian ally, I would not have attacked preemptively, but weakening future revanche while respecting alliance made sense - ditto a bit thrill of action for a player is tempting - Austrian war on France made less)

so - analysis :

Britain made an attack on a weak target, by free (even if serious) hostility and opportunism (AI programmed - efficiently - to make war on (determined by relationships) enemy - while his allies are distracted - (me with France) -but :
- there was no gain to expect (except weakening Russia, but it will weaken Britain too, so it is detrimental to her)
- Russia was NOT a threat
- there was a really serious threat (Germany, me) in which hands Britain played

while such a crisis, coming from a [color="#FF0000"]misjudgement [/color]of situation, and resulting in a war, is [color="#008000"]possible [/color](such a war was close to erupt several times, due to St Petersburg cabinet as well as Whitehall paranoia), such a choice is a misjudgement of AI, based on wrong parameters, relationships and weakness of a secondary competitor

[color="#008000"]in 19th century diplomacy, British cabinet would have :[/color]
[color="#008000"]- concentrated troops[/color]
- launched a [color="#008000"]warning[/color] to Petersburg
[color="#008000"]- perhaps mobilized[/color] (mobilization was rather [color="#FF0000"]uncommon [/color]in very liberal Britain)

-[color="#008000"] closed relationships with France[/color], the only one nation not in Bismarckian system, while Germany showed support of Russia, built a fleet, and clearly went across British expansion - or at least not distract herself by a peripheral vain conflict with Russia (would have British cabinet considered france too weak to ally with)
- [color="#008000"]tried to improve relations with Russia[/color] (Germany being the main threat) to let each other [color="#008000"]free hands[/color] to face her

- which was [color="#008000"]standard practice in 19th century[/color] - a too dominant nation rises - other nations try to patch up differences - without necessary jumping to the main threat throat, but just to[color="#008000"] avoid diversionary fronts [/color]

clearly, present AI is not yet instructed to have such behaviour, so needs some instructions we should fix :

a) by giving AI the good parameters, [color="#AFEEEE"]identifying main threat[/color], having an [color="#AFEEEE"]objective led strategy[/color], looking for [color="#AFEEEE"]alliances[/color], [color="#AFEEEE"]patching secondary fronts[/color], and sending [color="#AFEEEE"]warnings[/color] to protect balance of powers (threatening weaker competitor is common practice, and usually results in counter threats and warning of other powers - Germany in 1875, Russia leading to crimean war - jumping on weaker power is unheard of, except strong internal trouble)
b) by placing [color="#FF8C00"]relations [/color]at their right place, not the main drive and determining factors but [color="#FF8C00"]resulting [/color]from interests and actions, as such a[color="#FF8C00"] factor of random blunder or of hostility making the difference in case of real crisis[/color] ([color="#FF8C00"]as national morale[/color] should be) deciding l[color="#FF8C00"]evel of tension[/color] in case of crisis, as such an [color="#FF8C00"]"idiocy factor"[/color] perhaps but [color="#AFEEEE"]not a strategy goal or drive[/color]
c) by scripting behaviours to supply AI with geopolitical analysis - still those cannot substitute fully to an AI "goal-threat" analysis (who wants my [color="#AFEEEE"]objectives[/color], who can[color="#AFEEEE"] annex my provinces[/color], who [color="#AFEEEE"]among them[/color] has a [color="#AFEEEE"]strong military[/color], who can [color="#AFEEEE"]help me[/color] against them)

[Kensai]

There is no chance the AI started a war without a good reason. And when I say good, I mean within the rules of conflict, as the AI understands them. The AI cannot forge a CB, it is not programmed to do so. Therefore, if a war erupted:

• It was a scripted war.
• It was a scripted or infringement CB and the AI took opportunity for whatever reason (probably the power ratio and bad relations).
• It was a bad crisis that spiralled out of control.

These are the only possibilities as far as I can understand Athena. Thus the war between Austria and France is rather suspicious, what year is it? Perhaps it was an event that you missed that lead to their hostility. Regarding the war between Russia and the United Kingdom, well, a crisis can always lead to bad results if rulers with high Imperialism and Diplomacy are in charge, as this changes the available crisis cards they have on their deck. So perhaps, as in real life, the British rulers decided for something that they did not really desire, but their pride got the better of them. I remember back when I played Shogun Iemochi with his pathetic Imperialism value of 2, when a crisis erupted I could only press for the weak decisions. He was cowering for the most powerful ones.

Btw, I find the concept of crisis cards according to the rulers' attributes an ingenious mechanism. I hope it is worked upon for future titles as well.

[loki100]

Kensai

there is another source of CB - the one that just pops up as 'bad relations' and gives a 6 month CB (often against a country that didn't exist). That may explain this Franco-Austrian war, esp with bad relations and that Austria has a much larger army. But this is neither scripted nor the fall out from a crisis - I actually think it was the logic behind the Anglo-Prussian war in my AAR as I never saw a full crisis report pop up

[Kensai]

I am not aware of that. Never experienced a CB that pops up simply because of bad relations. Actually, I find it almost impossible there is something like that as there is no CB whatsoever with a duration of six months in the files (DIDuration = 12).

[loki100]

I am not aware of that. Never experienced a CB that pops up simply because of bad relations. Actually, I find it almost impossible there is something like that as there is no CB whatsoever with a duration of six months in the files (DIDuration = 12).

I saw one in my game last night - Austria vs Ecuador, happens all the time, but they rarely lead to much. But it does set a short term CB that could then be used if the AI evaluated it in terms of animosity and relative power relations

[Kensai]

Unless you have some kind of experimental version of the game, I am telling you this is almost impossible. The Short CB gives a whole year of CB (DiDuration = 24). All the other CBs have no link to such a feature and even if they had, the durations are different: a temporary and infringement CB is for 3 months, a short CB for a year, a long CB for 30 months, and a permanent CB, well for 9999 turns, thus for the duration of the game.

Are you sure it wasn't a simple infringement CB that escalated into war? These happen a lot and give recurring CBs because of disputed territorial claims.


PS. I haven't installed the latest version of Pocus' PON because of the reported instability issues, so unless what you describe is a feature added in the latest patch, I am pretty sure it must be something else. Unless, obviously, something that I have never come across. But that begs the question, where the hell are its describing files?

[loki100]

It does, all the time 'because of tensions X has a CB on Y', often one of X/Y doesn't even exist. Its a regular part of the diplomatic cycle, I've had them sometimes but never really made use of them. I mean I know my game started in 1.02, 1.03, 1.04 and back to 1.03 but I've seen these from game start, here's an example:

[Kensai]

Ah, yes indeed. In my mind I had this covered as a subcategory of the third option I gave (crisis). Of course, this is a crisis without a crisis interface. Am not sure if it gives a temporary or short CB, in any case it would be either 3 or 12 months, not 6. There is no such thing as a six month CB, unless you have modded the files.

[Christophe.Barot]

year was 1880, relations between Austria and France were poor (actually France relations with almost everybody but Italy and USA were poor (weirdly, neutral with me, isn't it ironic) - which is part of the [color="#FF0000"](excessive imho, while sound in principle) billard effect of relations[/color], if your GP is very active to get good relations (trade treaties, state visits, support, alliances), then by compensation every other GP begins to have bad relations with all other - [color="#FF0000"]communicating vases, null sum [/color]

also [color="#FF0000"]making a separate peace kills relations[/color]... while [color="#FF0000"]game does not allow yet for gloabl peace with alliance[/color]

no scripted war between France and Austria in 1880 of course, Plombieres is well over

British - Russia war following a big crisis makes sense, but only as a blunder - a local zealous officer provoking a border incident, and, well, British believing Czar is on to a counteroffensive in Turkestan, plausible, no way to be a cold blood decision, but historically possible as a blunder, yes, as Central Asia was a hot spot leading to paranoia, ditto a war could have erupted in Fashoda between france and Britain, or between USA and germany in Philippines (little known episode) or venezuela (a crisis between germany and venezuela has no sense, too much disparity- certainly shouldn't redispatch prestige, is duck shooting (ditto between french and Peuls) , crisis - real one was between USA because of doctrine Monroe, and Germany

- conversely Austro-French war i n 1880 makes absolutely no sense historically - I can send you files of 1880 if you (Kensai, Loki) want to test, but I guess some things happen randomly - honest, was somehow bored by seemingly absence of major events (lol, just changed) already folded situation (Germany had won in 1878 - I am absolutely sure nobody can stop Germany USA Britain to win by micromanaging, it's possible with Russia too (but comes later) - must be possible with France - with others must be tricky - no standard unification nor normally colonization for Austria, needs a solid control on Balkans Italy and Germany, dunno if it is enough given the huge weigh of economy in VP, which is (too, imho, regarding objectives) predominant - and Italy and japan, even mictromanaged by Loki who did wonders, are too weak structurally to compete with biggest powers - coal, demography, indeed that is the way it should be, thow se countries should be able to win by VP conditions, not absolute power - too much structural disparity ) that I pressed the "go "those years, and certainly missed a crucial message, just noticed I had an unusually strong Austrian presence in germany

seriously, folks, the natural game of Austria is to press Germany for military action (war, or just military pression) on Russia, France is NOT Austria concern, ditto ask germany to press diplomatically Italy for neutrality, France was Austria enemy before german and Italian unification, when she threatened german minorsn, and helped Piedmont after, was just cause she was her ally enemy - no common border

those random CB make little sense (even without the nonexistant country, which, for credibility, should be tested, a CB between britain and nonexistant Ukraine or Bohemia is nonsense) - they assume a diplomatic incident will degenerate into a war without grounds - totally false, historically [color="#FF0000"]all crisis which happened[/color], would they be engineered by mastership, blunder and idiocy, or uncontrolled factors (revolts) [color="#FF0000"]had material grounds[/color] (greed or fear) :

1877 RT war - caused by uncontrolled revolt (Bulgarians) - and in a ground where Russians had big ambitions - VP if client control objectives (have or should be) objective Constantinople, a permanent Russian goal (note that , as common german allies, they had Austrian neutrality at start, a rare thing here) crisis when they close to constantinople - British send fleets, Austrian send diplomatic note, threaten to mobilize - then congress where everybody counts his allies, in spite of his pretence Bismarck was all but a honest broker, as he encouraged Britain and Austria undercover

2d Italian liberation war, and 1870 Cavour and Bismarck maneuvered opponents (+ idiocy factor at least from French, more chamber of deputies than ill napoleon, before 1870, those are french, not Wilhelm II, who show hubris), BUT obvious stake (lombardia, german southern states, alsace lorraine came later) AND trigger (Spanish Hohenzollern, Italian revolts), needs both

crisis did not happen because benedetti was "insulted" - had a Russian embassador been insulted, there would have been a cold, but both courts would have patched it, no crisis - there was a crisis because there were grounds to a crisis and french were nervous with a reverse spanish alliance (- paranoia - plus rivalry for influence in Spain, somehow, Napoleonic union latine pipe dream) then it degenerated

ditto for Menchikov, "insulted" by Sultan refusal (of generous Russian proposition to hand his power to the Czar ?) - was just a classical takeover ultimatum, Like British in Egypt, byt ottpoman were a bit stronger, and supported - pretext was given by the Monk quarrel, but any Chritian revolt, or a merchant bullied, would have done too

ditto for Menchikov, or 1875 war in view, or Boulanger, there is not a demand to have a change of ministries or more friendly attitude by chance - in both cases, there are solid grounds - offensive for Russia, who has ambitions in Ottoman (namely making ottoman a Russian protectorate) defensive but agressive for germany (who wants to prevent French rearmament and revenge) - state is considered as extremely weak (not too much or he gives up, unless support, but crisis is with protector) - Ottoman of course is not an European army, and France had not recovered from defeat - and in both cases, either initiator backed down (germany) cause of international support of victim, or got a war (Russia) there are no such things as a crisis abour hostile belgian or swedish ministers - no ambition towards those countries - Bismarck or Giers couldn't care less - crisis happen where there is a barril of powder , not randomly (else, they just don't escalate)

what I mean is that there was no generic random historic crisis (they happened in identified hot spots, with conflicts for objectives) - nor did they result in plain uncontrolled prestige transfer (which could ruin the game for all if were more important what good is playing well, if a competitor gets more prestige in a random dice roll) crisis were interactive (they sometimes lasted months - 4 6 game turns - with call to allies, positions, escalations,- which could result in a really very interesting diplomatic game, but would require to much resources to redraw i'm afraid - and resulted in war to avoild loss of prestige, by choice of cornered loser ( sometimes in a loss of prestige, but by choice, when stake wasn't too important, relations indeed not too bad - they played a role, yes, but not main one - and national Morale not too high - fashoda is an example - unpleasant to back down, but Kitchener was stronger, and germans held Alsace Lorraine) which means that statesman player "controlled" them, had a choice, and lost control by mistake

all this to point out that system, driven by relationships, could take some improvements, and that [color="#008000"]driving diplomacy by objectives, alliances, military force and possibly national morale is needed to get close to 19th century reality and get its diplomatic flavour[/color]

alas, failing resources (setting such a huge combined game was already an herculean task - so was tuning it) such an AI analysis couldn't be set - so we have standard game partly satisfactory solutions, with sometimes very undesirable effects

as diplomatic AI can't possibly be deeply remade, so we know in what sense we have to make scripts, to set geopolitical accurate historical choices not considered by AI

still I think incorportating SOME basic aspects of reasoning ([color="#008000"]tests on conditions - conflict of objectives, threat based on objectives and possible annexions, else hardcoded [/color]- there are less than a dozen GP and japan and austria don't care each other) [color="#008000"]in AI analysis/test[/color] , if possible, seems to me [color="#008000"]highly desirable[/color]

barring that we'll have, I fear either a very rigid game (more than needed, as RT RA - balkans AI - north italy-alps adriatic GF south germany alsace lorraine RJ AP before unification korea manchuria and even BR BF - colonies BG fleet supremacy- frictions are absolutely structural so room for sghifting priorities, but not to random opposition ) or unsatisfactory reactions from AI

[Christophe.Barot]

Are you sure it wasn't a simple infringement CB that escalated into war? These happen a lot and give recurring CBs because of disputed territorial claims.


PS. I haven't installed the latest version of Pocus' PON because of the reported instability issues, so unless what you describe is a feature added in the latest patch, I am pretty sure it must be something else. Unless, obviously, something that I have never come across. But that begs the question, where the hell are its describing files?

I play with the 1.03 e - too much lag with B F1 F4 else (already 95% of my playing time taken by adjusting production and sales - micromanagement one tells ..) so not untested features

absurd CB happen alll the time - diplomatic insults giving CB randomly - provoking nothing - unexistant nations and rebels should get out of the view (rebels are always at war), - of course conflict for colonial grounds (me with Britain in Nigeria for example) or provinces (Alsace-lorraine) make sense and should be kept - I wonder if those random CB are wad (reminds me chaotic wars of former versions)

of course, unorganized states default of payments led to intervention (Mexico, Egypt, venezuela, Nicaragua, Ottoman, persia, China ..) but Austria !!! (would make sense with Serbia, not Guatemala) .... guess we should dispense of generic sandbox imho (which is already partially done I guess, never saw a crisis between US and ol world nations, nor japan (would make sense with Korea and china, nobody else)

[Kensai]

Austro-French war in 1880 makes absolutely no sense historically

Heh, remember you are playing in an alternate reality and things diverge from the historical timeline. You have to create stories for yourself to explain things. I do it all the time, it's part of the fun!

Athena may be difficult to understand, but she would not go to war without a good (logical) reason. This is the essence of computer intelligence. We, as humans, are here to interpret it, in case we have not programmed it ourselves. Thus, to repeat, if relations were bad and there was a nice difference of power ratio, perhaps the AI thought it was the perfect opportunity to wedge war. Unfortunately, the AI in most of these games consider relations only between two-nations, not the biggest picture of geopolitical gain or loss. But this is something most games cannot abstract accurately yet, not a problem of AGEOD games only.

[Christophe.Barot]

Agree, Kensai, but I'm not speaking here with player glasses, but with designer ones (or , more modestly and accurately, reviewer and helper) problem is not imho if design was perfect - none is - or if game is good (it is a lot , and has supreme merit to be there) but if it has wanting sides we can improve - imho diplomacy could be improved, and, IF possible, should

for obvious reasons, huge programming task of incorporating geopolitical considerations couldn't be achieved in available time span with available resources (too close a deadline, too huge and time consuming the debugging and tuning, and Pocus has to sleep a few hours a night, however gifted and hard worker he can be - don't forget Hofstadter law)

so "we" had to fall back to simpler available system I guess - that is relationships, which are something, but a is a resulting and (the reviewer/ historical oriented researcher is speaking) has too drawbacks imho :

- a) it is historically inaccurate, which is detrimental to "feeling" (I mean nation long terms policies, while influenced by anthipathies, were mainly based on objectives and threats)
- in this example Austria not only [color="#008000"]follows faithfully an ally- OK[/color] - but shows [color="#FF0000"]overzaelous[/color]
- something which could be avoided either by AI routine of simple interests analysis or by hardcoding it
- no war [color="#008000"]except natural enemies[/color] (due to german/Italian interests, france still qualify, except with a script/code modifiing it later with unifications which would unqualify her) - or [color="#008000"]following an ally[/color]
- or interests - For Austria, Balkans rivalry - [color="#008000"]objective driven[/color] (with a combination of objective/annexable or not) - Balkan rivalry with Russia, Belgrade, Bucharest Sofia, Italian objectives (including Trieste Trent), and in Germany (Munich Saxony hannover as not annexable objectives) - note it can be hardcoded - Aibias is done for that, but is then more rigid (a Russia not owning Kabul and not closing Constantinople is not a british threat)

I'm not arguing there, just exposing concepts to have AI stick to real world history routines (Bismarck or Gladstone analysis, simplified in my way for gaming reasons of course)

b) gamewise it creates a snowball effect - no blame intended - just feedback

ditto the behaviours of not declaring wars, but withdrawing for it, creates by game mechanisms, unwanted relationships consequences which we should adress


difficulty for Athena is of course to incorporate these factors, here's why I expose them (after, i'm not competent to determine how heavy is coding work, and even less to code myself)

therefore, my diagnosis is :

1) we should try to improve things by script, strongly influencing game (but it may load or rigidify game a bit) as much as we can

still AI wrong reactions - as Athena won't invent criteria (Sweden neutrality, Monroe doctrine, Tordesillas in EU) if we don't feed her with routines and relevant factors and criteria, especially in a multilateral game - ACW ROP aren't - may still pose problems script won't all solve, so ..

2) we should inform coders of AI improper behaviour, pointing out problems, suggesting analysis (snowballing relationships fe) and possible solutions (tests fe), proposing accurate factors of test provided by game (hardcode categories in last resort, objectives, alliances, will develop it ..)

that's what I try to do, with Generalissimo, Loki and you as primarily targeted public, as working on project

after, what can be done for AI ...matter of resources and time, as always - but goal is to have game looking more historical, so sorry if it sounds like criticism, it's intended a constructive one

[Christophe.Barot]

There is no chance the AI started a war without a good reason. And when I say good, I mean within the rules of conflict, as the AI understands them. The [color="#FF8C00"]AI cannot forge a CB, it is not programmed to do so[/color]. Therefore, if a war erupted:

• It was a scripted war.
• It was a scripted or [color="#FF0000"]infringement CB[/color] and the AI took [color="#FF0000"]opportunity[/color] for whatever reason (probably the [color="#FF0000"]power ratio and bad relations[/color]).
• It was a bad crisis that spiralled out of control.

These are the only possibilities as far as I can understand Athena. Thus the war between Austria and France is rather suspicious, what year is it? Perhaps it was an event that you missed that lead to their hostility.

in former versions, during playtest, there were random sandbox CB forging from AI, lots, led to absurd chaotic wars - criterium was neighboorhood usually, France atatcked belgium, Prussia attacked Belgium, Spain atatcked Portugal, Britain attacked USA (Canada neighbourhood I guess) - I sighed of relief when this was abandoned and things reverted to "normal"

actually it was infringement - still [color="#FF0000"]can't spot any province where infringement could take place[/color], that's what makes war puzzling

[ATTACH]26223[/ATTACH]

here's the Italian situation - took 3 turns back - Britain was not yet at war, hence my still good relationships - dropped as soon as Britain declared on Russia (after a crisis resulting in spectacular hoistility - first time I see such a crisis - the - (minus) 99 in GBR/RUS relations comes from Crimean war, never improved, and was worsened by indirect "billiard" effect - it preexisted to crisis

[ATTACH]26224[/ATTACH]

France diplomatic situation is abysmal, except with Italy (once I had scripted a French-Italian alliance to correct Plombieres failure, relation snowballed) :
- bad relations with Russia due to crimean war
- bad relations with Britain due to France making peace - obviously separated peace, no alliance peace exists yet (Britain didn't send a single soldier it seems, still hates France for having made peace)
- bad relations with Austria due to Risoregimento (France surrendered little after Italy, still abysmal relation persists)
-HAD correct (positive) relations with US (I suspect regime nature plays a big role in sympathies - fully acceptable, but still excessive - didn't prevent France allying Russia, and an alliance between liberal Britain and Authoritarian Germany was absolutely a possibility (was turned down by germany as Britain wouldn't have helped on land, while Germany would have had to confront Russia in addition of France, but is another story)
here too, billiard effects were excessive (excellent US /GBR relations, plus separate peace - France did nothing to offend USA)
- surprisingly moderate relations with me - of course I have -75 since war, but I was at +6 before war in spite of QAlsace-Lorraine (that France is relieved to get peace and cut her losses makes sense, but this was a greatly exagerated effect, such gratefulness is excessive I keep Metz and Strasbourg) and compare with Russia and Austria, and even Britain

have a look

[ATTACH]26225[/ATTACH]

[color="#40E0D0"]so in resume :[/color]
- there should be a rubberband to draw relations, out of crisis and other actions or situation (national provinces owned), towards a balance point
- balance point should be influenced by alliances, provinces lost and objectives coveted (conflict) ... but moderately, alliances should not set reference at +100 like it is now, provinces lost should be a low reference ppoint
- having alliance peace is a MUST - barring that, strongly decrease effects of separate peace - results are effective
- reduce duration (10 years = 240 turns, 4 years = 96 turns) and cost of alliances
- some AI analysis in terms of objectives and other would be nice
- decrease effects of regime, increase effect of owned provinces (to tune) or coveted objectives (ditto, to tune)
- decrease effect of peace relief - especially if I get provinces - patching is OK, gratitude is too much, is a bonus to full scale invasion before making peace
- examine use and necessity of CB without teritorial dispute or minority revolt .. I see little, if any , historical examples

(edit) added - consequences of badboy (seems to perceive effect) and mobilization to review, principle has a point, generalization creates weird effects :
Rumania/Russia Transvaal/Germany are not scared by a strong proposed protecttor, the stronger, the better
allies or nations not targeted (criteria : objectives and annexable provinces) are not scared by mobilization or massed armies (obviously) targeting somebody else

[Kensai]

Some good points.

Regarding the infringement reasons, every now and then you should see messages for all nations regarding disputed regions. These do seem to appear every 3 months (6 turns) as the infringement CBs refresh. As assistance, I use a lot the little filter buttons that show de jure and claimed regions on the map.

Regarding your "resume" proposals, most are good, but can be difficult to implement.

We really don't know what goes behind relations as you don't have access to other nations' decisions. For example, if a nation is keeping its tariffs high, eventually every other nation will hate it. Perhaps the AI is not balanced enough to make the best decisions regarding tax levels, I have seen that in my Game of Victorian Thrones where the AI nations always keep most taxes towards the max, even if they have plenty of money. Also, if armies are kept at the borders, which happens a lot in continental Europe, crises might erupt that make relations take a further dive.

War and peace proposals can only happen vis-a-vis. Call this a limitation of the game engine, but it doesn't seem you are able to sign separate agreements en block, as an alliance. There might be some workarounds, for example if an ally surrenders then the other ally should get a hefty VP-NM penalty so that (unless winning considerably in its own right) it becomes more convinced to sign the end of the war too. For the historical wars of the game, and there are many, after we make a complete review of all the can possibly go wrong, I believe we should author war ending events to help the AI. There are more than enough instances for complains that wars (like the Crimean) would reach a standstill. Actually, I think there should be some kind of "leaking NM" for long wars so that nations eventually stop them. I believe it is already implemented but way too weak. Wars over long terms have to stop unless you make miracles to content your population.

Indeed, a max duration for alliances should be in. Then, perhaps, a decision to continue or abandon them should come up with an increasing "cost" (haven't thought in what, yet) to keep them. This way perpetual alliances will be out of question. Regarding NM, VPs, and generally contentment, there are a lot things that these parameters could balance to show winner-loser. Effects of regime are nice, actually I would suggest we go to the other direction, tie decisions more to strong-weak rulers. A weak ruler should be a relatively passive period for a nation, a strong ruler a period for energetic decisions and war. This is historically accurate, otherwise the human player will gang-ho his way through the entire game.

Regarding CBs without territorial dispute or crisis, indeed, I don't think they happened a lot in this era. The current CBs are ok. Perhaps as an addition, the AI could be taught to forge its own CBs as a rare occurrence, when everything is fine.

Last, the badboy effect: this could already be abstracted through worse relations with the rest of the world and falling NM. Perhaps bad relations should have catastrophic rises in costs of trade with the nations the relations are bad, in order to have a difficulty to survive.

PS. All the above have been more or less my own remarks in these last two years I have been playing the game. Only a v1.1 can address them, but unless you hear official communication don't hold a big basket. That's life, we can enjoy the game as it is.

[Kensai]

For example, if a nation is keeping its tariffs high, eventually every other nation will hate it. Perhaps the AI is not balanced enough to make the best decisions regarding tax levels, I have seen that in my Game of Victorian Thrones where the AI nations always keep most taxes towards the max, even if they have plenty of money.

Hm, I stand corrected. I think I have a theory of why this is happening in my game. Given it was a PBEM with all major nations occupied by humans, the economy was supercharged and the internal market more than satisfied. I believe that if I pass the script to lower contentment and rise militancy, the AI will be persuaded to lower taxes. As of now, it keeps them ultra high because it can, people get what they need so there are no complains.

Will let you know how this goes. But right now it is a problem as the AI nations accumulate huge amounts of money and capital.

---

Btw, did I tell you how much I love crises when they make sense?
(if anyone has trouble reading the crisis metrics, let me know for explanations)