AI is still walking through forts in 1.19

[durbal]

I'm not going to claim I understand the rules of forts at all, but just based on playing lots of games and trial and error I don't know why you'd expect that fort to block anything. From your territory there are quite a few fort combinations I'd expect to do something such as:

1) Hoya and Braunshweig, Cassel or Thüringen should stop movement to Luneburg from west to east
2) Braunshweig and Hessen or Nassau should stop movement to Thüringen from west to east
3) Thüringen and Hessen or Nassau should stop people moving to Braunshweig from south the north
etc. etc.

...and depending on the military access of the war participants I wouldn't guarantee any of these either. Best off deleting your forts in such an awkwardly shaped empire.

That said, this obviously doesn't help in clarifying the rules at all. And I could have picked any number of posts in this thread and posted a similar thing - nothing personal at all. People should just stop expecting single forts to do anything. [Fort - province - province - fort]^n along your border in a nation at least 2 provinces deep is the way to use forts without getting bajanxed by mystery rules. Expect this and only this to work and your fort headaches will disappear!

I included a savegame that will allow you to see what occurs. It's the best one with this problem that I've ever been able to capture. The AI will walk to siege the fort in Braunschweig from the north. If you move to attack them while they are sieging, they will be able to move to Cassel or Thuringen -- DIRECTLY THROUGH THE FORT.

This is most definitely (as was the case with the Mazandaran savegame I posted earlier) caused by the AI being able to move through forts using the 'shortcut' rule. Forts in narrow access points do NOT block the AI from moving through them if the AI can take a much longer path to get to that same province. That's the problem and it's significant since it makes chokepoints LESS useful than placing forts away from chokepoints (mountain passes, straits, etc.). That's nonsensical.

 

[durbal]

The AI walks through forts in every single war. The shortcut rule allows them to do this. I don't think I've fought a single war where the AI did not do this at least once.

How many more savegames do I need to post here? At least acknowedge that the problem exists.

 

[Gratak]

The AI walks through forts in every single war. The shortcut rule allows them to do this. I don't think I've fought a single war where the AI did not do this at least once.

How many more savegames do I need to post here? At least acknowedge that the problem exists.

The player walks through forts in every single war. The shortcut rule allows them to do this. I don't think I've fought a single war where the player did not do this at least once.

How many more savegames do I need to post here?

 

[durbal]

The player walks through forts in every single war. The shortcut rule allows them to do this. I don't think I've fought a single war where the player did not do this at least once.

How many more savegames do I need to post here?

Load my GH save with Mazandaran. Load my Hesse save. Tag switch and do the same thing the AI does and report back.

And even if the player were able to do it: why should anyone be walking through forts?

 

[Gratak]

Load my GH save with Mazandaran. Load my Hesse save. Tag switch and do the same thing the AI does and report back.

And even if the player were able to do it: why should anyone be walking through forts?

The problem is that you would have to give the walk order the exact time the AI did. And I totally agree with you that the "sometimes you can walk through forts" rules are stupid and they should change it to something one can understand. I just disagree in you saying all the time "it's an AI bug, shoud be easy to fix".

 

[Zelius]

Load my GH save with Mazandaran. Load my Hesse save. Tag switch and do the same thing the AI does and report back.

And even if the player were able to do it: why should anyone be walking through forts?

I tried your saves.

In your GH save, the Timurids are already moving, so I can't tell whether there's a bug or AI cheating, but I present to you, The Odd ZOC Rules:
Player Ottos doing a loop through Constantinople anti-clockwise:

Spoiler

Player Ottos failing to do same loop clockwise:

Spoiler

In your Hesse save, Saxony just sat there and waited to get pwned:

It could be one of the many cases of buggy AI behaviour that are resolved by reloading, and thus hard to reproduce...
For reference I am running unmodded 1.19.2 with all the DLC, which my game tells me is incompatible with your saves. Perhaps that has something to do with it?

 

[Gratak]

I tried your saves.

In your GH save, the Timurids are already moving, so I can't tell whether there's a bug or AI cheating, but I present to you, The Odd ZOC Rules:
Player Ottos doing a loop through Constantinople anti-clockwise:

Spoiler

Player Ottos failing to do same loop clockwise:

Spoiler

In your Hesse save, Saxony just sat there and waited to get pwned:

It could be one of the many cases of buggy AI behaviour that are resolved by reloading, and thus hard to reproduce...
For reference I am running unmodded 1.19.2 with all the DLC, which my game tells me is incompatible with your saves. Perhaps that has something to do with it?

This loop working in one direction but not the other is damn strange... I've found that in almost all situations you can walk towards your own unoccupied lands. Does not work if you are coming from allies (that are not in the war) and it seems it also does not work over straits. That would at least be the only reason clockwise could fail...

 

[TheMeInTeam]

The AI's rules are not the player's rules.



Maybe they're in another war?



Nope.

Is palatinate giving MA to someone they're fighting, or someone on either side of this war?



Nope.

I also observe different ZoC rules for player occupied forts and AI occupied forts. Worst of all, I have longstanding experience that suggests it's not wise to trust the mechanics are WAD, or that they're fair, or even that they're as-represented. Such is the consequence of observations against that trust.

 

[Zelius]

The AI's rules are not the player's rules.



Maybe they're in another war?



Nope.

Is palatinate giving MA to someone they're fighting, or someone on either side of this war?



Nope.

I also observe different ZoC rules for player occupied forts and AI occupied forts. Worst of all, I have longstanding experience that suggests it's not wise to trust the mechanics are WAD, or that they're fair, or even that they're as-represented. Such is the consequence of observations against that trust.

Did they retreat to that province? As I think you already know, there is a (bug? necessary feature?) that the retreating army will not be black-flagged if they lose MA during the interval, like so:

After the retreat began, I used console to change ownership of one province from Karaman to Ottomans.

It does not seem unfeasible that Trier or their allies would've had MA to Palatinate and then cancelled it or had it cancelled en-route.

..

I've never stated that ZOC and AI pathing in general has worked perfectly over the patches (while testing ZOC I found that ordering an army from A to B sometimes made it take a longer route (time-wise) than could achieved by manually setting a path, for example), but I do take exception to the suggestion from some that the developers have intentionally written a series of 'special rules' for the AI to cheat - i.e. code that if found it would not be removed - and then denied it on the forums. That position isn't helpful in my opinion, since it can lead to only two things:
a) rage at the game for cheating instead of trying to understand the mechanics (admittedly it is complex and never explained anywhere in-game);
b) shrugging off shady behaviour as integral AI cheats, thus diverting attention away from bug-reporting and fixing.

 

[Dracolithfiend]

If there are still issues with AI marching through active forts in 1.19 then I would love a savefile where this happens so we can investigate and make further fixes.

Sadly, screenshots do not help in this regard, there is too much that doesn't get shown with just a screenshot.

I can almost certainly get you one after work. I know a spot that the AI reliably glitches through.

 

[TheMeInTeam]

Did they retreat to that province? As I think you already know, there is a (bug? necessary feature?) that the retreating army will not be black-flagged if they lose MA during the interval, like so:

After the retreat began, I used console to change ownership of one province from Karaman to Ottomans.

It does not seem unfeasible that Trier or their allies would've had MA to Palatinate and then cancelled it or had it cancelled en-route.

..

I've never stated that ZOC and AI pathing in general has worked perfectly over the patches (while testing ZOC I found that ordering an army from A to B sometimes made it take a longer route (time-wise) than could achieved by manually setting a path, for example), but I do take exception to the suggestion from some that the developers have intentionally written a series of 'special rules' for the AI to cheat - i.e. code that if found it would not be removed - and then denied it on the forums. That position isn't helpful in my opinion, since it can lead to only two things:
a) rage at the game for cheating instead of trying to understand the mechanics (admittedly it is complex and never explained anywhere in-game);
b) shrugging off shady behaviour as integral AI cheats, thus diverting attention away from bug-reporting and fixing.

Doubtful, I tried to follow them for stackwipe ~1 day after they began the movement. AFAIK that black flag bug usually happens with separate peace deals. though I suppose it's possible in principle and hard to check.

 

[durbal]

This is extremely annoying in certain areas and it's not at all hard to reproduce. Save games have already been provided, but if you need to see this occur time after time just start up a game on the Italian peninsula and watch the AI repeatedly just walk right past forts in places like Rome and Napoli where there is absolutely no other reason for them to walk through these forts.

This really should be a priority since it absolutely ruins some wars and it's been occurring for so long.

I can't overstate how game-breaking this is. Taking critical forts (and even having them) becomes meaningless when the AI just walks right through them whenever it feels like it. I just watched a war where Hunyadi beat back the entire Ottoman army to the east side of the Bosporus and began sieging Edirne. So the Ottomans move onto Constantinople (owned by their enemy) and begin sieging. Then the Ottomans just walk right through the fort in Constantinople to get back to the Balkans and move onto Pest. C'mon. Whether or not the follows the rules as they are coded, this kind of behavior defies all logic.