@Gnivom and devs - question and comments regarding AI counterespionage/espionage behavior

  • 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.

bbqftw

banana vendor for unhuman entities
2 Badges
Jan 18, 2014
5.394
6.188
  • Europa Universalis IV
  • Europa Universalis IV: Mandate of Heaven
Several experiences in recent campaigns have led me to hypothesize that the AI is 'psychic' regarding your claim fabrication - that is, that it greatly prioritizes counter-espionaging you while you are building a spy network in it, even before any public reveal that you are building spy network (e.g. claim fabrication made, or other espionage action).

Of course, that's kind of difficult to actually prove, so here's general observations consistent with said hypothesis.

1) Very often AIs will begin to counter-espionage during fabrication

If counter-espionage debuff was present at the beginning of the spy network buildup, I would understand. Fine - that is a calculated risk. Both human and AI have limited diplomats (I think?), if we're counter-espionaging people that aren't even building spy networks on us, that's essentially a wasted diplomat that could have been building relations for alliances, or annexing vassals, or doing other productive things. But the equation changes if we knew that people were fabricating, since that may affect our diplomat allocation priorities.

But right now, that's very often not the case - you can get through 3 months or so of claim fabrication before the counter-espionage debuff begins, which leads to high suspicion that even before you fabricate a claim the AI 'knows' you are building spy network and is acting on it.

2) AIs will counter-espionage you even when associated with neutral attitude

I could understand outraged, threatened, or rival AIs randomly counter-espionaging. But neutral? That is interesting, and hardly seems like a good allocation of diplomat-time in most cases (mind you, its making a reasonable play since I intend to instantly fabricate after annexing Assam, but still, I would expect it to change to threatened in response if it was a threat calculation that determined counterespi, not "is the player building a spy network on me"):

X4J6M9K.jpg

X4J6M9K.jpg


Ultimately, whatever, its a small thing. Its not going to slow me down that much, and its hardly gamebreaking. But it would be nice to have dev confirmation that yes, the AI does know exactly when we are building spy networks and will prioritize counter-espionage. Might be a stealth buff to the rather subpar religious these days :) as deus vult cares not about your lack of claims.
 
Last edited:
Second I'd like to ask why AIs seem to prioritize taking the espionage idea group so much. This idea group is, for the large part, absolute trash, useless, even more useless for AIs, so I am curious when nearly 40% of tags sampled take this group in first 4 (maybe a biased sample). Its not just OPMs, GPs do this too (espi Prussia? really?) and I'm curious why this is if indeed the idea group weightings were based on player idea group elimination threads.

However, combined with psychic counterespionage (if that's a thing) it does now severely slow down spy network fabrication, though of course there's always the trusty spymaster to beat down cheeky espionage using scum. But such a thing merely delays their demise (at least in my case, 10 patch versions have given me a slightly irrational desire to exterminate every espionage taking tag), its basically 'lose slower' while something like diplomatic may actually turn their situation around.

(Side note: I propose on VH difficulty, AI only takes following idea groups:

* Administrative
* Economic
* Humanist
* Religious

(with either religious or humanist strongly picked in any situation with sub-80% religious unity)

* Diplomatic
* Exploration
* maybe Trade

* Defensive
* Offensive
* Quality
* Quantity
* Plutocratic is okay too, I guess

If you want to go ultra hard, every AI takes defensive first and takes military ideas at 3 5 7 as well)
 
Several experiences in recent campaigns have led me to hypothesize that the AI is 'psychic' regarding your claim fabrication - that is, that it greatly prioritizes counter-espionaging you while you are building a spy network in it, even before any public reveal that you are building spy network (e.g. claim fabrication made, or other espionage action).

Of course, that's kind of difficult to actually prove, so here's general observations consistent with said hypothesis.
...
Ultimately, whatever, its a small thing. Its not going to slow me down that much, and its hardly gamebreaking. But it would be nice to have dev confirmation that yes, the AI does know exactly when we are building spy networks and will prioritize counter-espionage. Might be a stealth buff to the rather subpar religious these days :) as deus vult cares not about your lack of claims.
No, not difficult at all.
Unless I grossly misunderstood something, it is not even hidden...
Untitled.png
 
Don't know about that espionage stuff but about idea groups - its time to finally buff useless and weak idea groups - rather then to railroad AI (very hard on not) to pick only good groups. They are fixing Aristocratic, they should fix other groups as well. There were plenty of good suggestions in Suggestion forum. AI picking the same groups as player "fix" was just a sad case of avoiding serious balance issues..
 
Bumping this. Had a country that was not counter-espionaging me immediately counter-espi before the spy network even ticked once...I'd like some confirmation that I am not crazy. @Gnivom

No, not difficult at all.
Unless I grossly misunderstood something, it is not even hidden...
View attachment 268762
Those are the actions such as sabotage reputation / sow discontent actions which reveal the presence of spy network, not the buildup, as far I know. However, until your spy network buildup is caught or you make a claim, or you make an espionage action, its not known whether spy network is being built. So basically the AI is acting off of information that is not available to player.
 
Another concern with this behavior is part of it is straight up AI throwing.

If the AI is counterespionaging within days of you building up a spy network, then it presumably has an idle diplomat. It could be gathering useful alliances with that diplomat, but it is not. Instead it dedicates the diplomat to delaying their time of destruction by 20 months (and smart play right now against someone you can't destroy in single war is not take the claimed province so you can kill them after the truce timer without needing to fabricate the next claim), presumably to spite the player.

Reminds me of the days when OPM Augsburg would randomly build spy networks in your country just to assassinate your leader. Good thing that mechanic was removed.
 
This idea group is, for the large part, absolute trash, useless

Is it? I take it every game. You don't like extra siege ability? :)
 
I see this too. I thought it strange that random small nations had set me as counter espionage so I slapped a spy on and watched what happened. You get utter nonsense like neutral Madagascar Fetish insta-reacting to a colonial Asian trying to set its first claim off a nearby island.

If the AI is counterespionaging within days of you building up a spy network, then it presumably has an idle diplomat. It could be gathering useful alliances with that diplomat, but it is not. Instead it dedicates the diplomat to delaying their time of destruction by 20 months (and smart play right now against someone you can't destroy in single war is not take the claimed province so you can kill them after the truce timer without needing to fabricate the next claim), presumably to spite the player.

Maybe this is paradox's idea of a good use for that "free player only diplomat" the AI gets?
 
It feels like the AI is bullying the player. I felt like getting stuck in a lot of multiplayer games as well because everyone nation i tried to fabricate on would just counter it. Making a single claim takes forever.
 
Last edited:
Is it? I take it every game. You don't like extra siege ability? :)
I like actually having admin points more, among other things.
 
There's a long holiday right now but I've poked @Gnivom about this thread for when we're back in office. :)

They removed extra relation slot reserved for the player, not the extra diplomat. And it was done in 1.21, extra diplomat IIRC is still in. (Could be wrong though, haven't checked it lately.)

There hasn't been a reserved relations slot since sometime in 2014. What you're seeing in recent patches is not related to that.
 
There's a long holiday right now but I've poked @Gnivom about this thread for when we're back in office. :)



There hasn't been a reserved relations slot since sometime in 2014. What you're seeing in recent patches is not related to that.

One should not post straight after waking up. I don't even know why I wrote that :D Because change in 1.21 is tied to the 'over the limit' of the relations etc.
 
Strictly circumstantial evidence but I can build spy network without triggering CE by AI until I claim the 1st province. After that point sometimes they start CE immediately and other times it can be 2 or 3 claims. I wonder if threatened attitude, AE, negative relations factor into the AI's CE focus. I know you are a much more aggressive player than I am which means you could be reaching that threshold much quicker. Because I do notice AI being much more vigilant once I've triggered that outraged attitude. Also there may be cases where the less tags around means less use of diplomats in general, so they could have a "spare" always around, ready and able to thwart your fiendish plots.
 
Well spotted. I was not aware, but the AI only considers counterespionage when they are being spied upon. This seems like a minor bug, but since my philosophy is that the AI should cheat as little as possible, I will fix when I get time.
 
Why propose humanist ideas for the AI on very hard?
You dont see many AIs ever having issues with rebels on very hard -> humanist is useless.

Whats left without all that unrest stuff?
- 2 max promoted cultures -> useless for all small AIs and large ones are empires so high chance that it is useless too.
- 30 improves relations -> relatively useless too, because AE for AIs on very hard is very low.
- -10 idea cost/harmony -> AIs get ideas cheaper anyways, i seldom see them short of monach points -> useless.

The one diplomat from espionage is way better then thouse 3 things together for the AI!
I would go so far as to propose, that the AI should never pick humanist on very hard.