Auto siege feature allows to "see" in fog of war

[FrogCrusher]

I am sure it well known by many of us but I just realized by using this feature that the alert appearing for army refusing to engage during auto siege is aware of army in FoW. And the best thing is that when you click on the alert, it highlights the province where the army is.

EDIT : I did some more testing and it is not reliable at all
See https://forum.paradoxplaza.com/foru...s-to-see-in-fog-of-war.1564193/#post-28708324

 

[necro84]

It shows how popular is this option if nobody noticed it before

 

[vaLor-]

I am sure it well known by many of us but I just realized by using this feature that the alert appearing for army refusing to engage during auto siege is aware of army in FoW. And the best thing is that when you click on the alert, it highlights the province where the army is.
So now, at the beginning of a war without knowing where is the enemy army, I use 2k in autosiege and look at the alert

TIL, never used this feature because I like controlling my armies. Maybe I will now

 

[TheMeInTeam]

How did it go? Something like "AI doesn't cheat with FoW" plus some explanation of how there's an exception but only for stacks that it saw, with a chance to forget lol?

 

[FrogCrusher]

I edited my first post.
After some tests, it seems not reliable at all.

In fact, it worked very well during my last wars in my last game (past 1650) then I thought I discovered some things. But at game start, I didn't manage for it to work. There is still someting with units seeing in FoW as for exemple it tags ports with boats in FoW as too dangerous provinces. But for land troop, it is very random in fact.

 

[strqfzd911]

How did it go? Something like "AI doesn't cheat with FoW" plus some explanation of how there's an exception but only for stacks that it saw, with a chance to forget lol?

I suppose Paradox has been a little tricky with how they describe the rules of FoW for the AI. Technically they aren't lying when they say it obeys the same rules as the player, but since the AI dosen't visually see armies like the player does this fact is kind of meaningless... The problem here is that the AI uses a threat evaluation system that applies to every province they have military access in, regardless if its in fog of war or not. This means if the threat evaluation from that province is above what they've been coded to handle they'll stay away from it, likewise opposite if the threat evaluation is below their coded balance. This is why you often see cases of the AI appearing out of fog of war to stackwipe one of your smaller siege stacks you weren't paying attention to, before darting back into allied territory when you send a larger stack to kill them off.

 

[TheMeInTeam]

I suppose Paradox has been a little tricky with how they describe the rules of FoW for the AI. Technically they aren't lying when they say it obeys the same rules as the player, but since the AI dosen't visually see armies like the player does this fact is kind of meaningless... The problem here is that the AI uses a threat evaluation system that applies to every province they have military access in, regardless if its in fog of war or not. This means if the threat evaluation from that province is above what they've been coded to handle they'll stay away from it, likewise opposite if the threat evaluation is below their coded balance. This is why you often see cases of the AI appearing out of fog of war to stackwipe one of your smaller siege stacks you weren't paying attention to, before darting back into allied territory when you send a larger stack to kill them off.

What the AI is able to react to in terms of information is the closest proxy we can use for what it can "see", and the game has limits for that. The question is whether those limits are reasonably close to the player's. When the AI can "see" movements that would clearly be in FoW for a player, or when it can "see" what is valid pathing for enemy troops vs not (WRT ZoC --> information about opponent return provinces can only sometimes be inferred by humans), it is disingenuous (at best) to claim these are scenarios where the AI is still "obeying the same rules as the player".

What information is available to the player is part of the game's rules! You reveal fog based on unit placement. I don't think many players would consider someone using a map hack to reveal fog in MP to be fair play...instead I suspect the vast majority of the community would (correctly) consider that to be cheating. The AI is using information for "threat evaluation" when a player in the same scenario would not have access to that information. Aka, it is using information that should, according to the rules, be "in the fog".

Note that this might even be the best way to make the game work. It is a bit unfortunate that how it was represented to us was misleading at best though. Even at the time, experienced players immediately raised eyebrows when a dev posted that the AI doesn't cheat FoW, because those years ago we were using its sight cheats to get it to lift sieges (due to being under "threat") using units it shouldn't "see". It was especially easy in patches where the AI was less "sticky" to sieges and didn't value progress much.

 

[FrogCrusher]

So, I did another test, and it seems that when a big army is attacking a country, the defender is in sort a "freeze" mode and then it works pretty fine.
See the screenshot below. I'm Ming and I attacked Mongolia. The 9k Oirat army is clearly in the FoW but it is detected by the auto-siege.

 

[I like wars and maps]

It shows how popular is this option if nobody noticed it before

Paradox sold a feature wich is inferior to the one in Imperator for money