AI stops all his movement on reload and other AI glitches

[unmerged(36024)]

I'm playing EUII 1.08 (5.09 beta)
1. When you reload saved game, AI stop all of his movement, thinks day or two then relaunch his troops (usually in the same direction)

Well, can AI think about moving units without stopping 'em?

2. AI regularly (even without reloads) change it's orders without reason.
Example 1: 4-Province Karaman was at war with Venice and Cyprus, make peace with Venice and still at war with Cyprus, which control Adana. Cyprus have no troops on mainland. Karaman is at war only with Cyprus. During 30-year period I look at 25'000 army of Karaman marsh into Adana, stay there for 2 months and then leave it. For me - it is a bug.
Example 2: Austria and Hungary have a war with Ottomans, kick some butts and besiege 2 ottoman provinces. Then Wurzburg (Austria) revolts. And 80'000 army of Austria and Hungary leave war with Ottomans and go to Wurzburg... Ok but when they reach Wurzburg, they stay there for 2 months, loose some man due to attrition and again go against turkeys... and so on, and so on (now they cross Hungary 4 or 5 time). Ottomans now besiege 2 of their provinces and I scream from this idiotism of AI and think about how to get MA on Austria to send my 5'500 infantry army to relieve Wurzburg (i'm playing Novgorod) only to stop this madness.

So can AI split it's forces between several goals and be more stable in it's plannings

 

[unmerged(6147)]

Yes, such things are sad indeed.

It's rather surprising that Johan has not been able to fix cases like the Würzburg one. IMO the algorithm should be

If rebs have taken control of a province of your own:
a) chrush rebs
b) besiege with min no of men needed, do something useful with the rest

I wonder what the algorithm is in reality. Probably the same as for all sieges, i.e. siege with a huge army. And that then is the basic error in the present algorithm, it does not differentiate between the two situations.

Of course, a small sieging force means an enemy could interrupt your siege but that will happen very rarely in a case like the Würzberg one.

The Adana thing is even worse. Perhaps the evaluation for the two alternatives (siege and not siege) come out the same and if so the program is coded to choose the alternative that is different from what presently is the case. Anyway it appears to be the wrong solution. Another alternative is that it is a cavalry only force and the well-known bug of "siege with cavalry only in plains is always abandonded" (How do you spell that? ) hits.

Many years ago I asked Johan about the siege algorithm. I said something to the effect that if I coded it I would do it like this

a) calcualate how many men needed (i.e. 5000 per fortress level)

b) add something (a little ) for attrition and to protect from enemy attacks

And do you know what he answered... That the present algorithm was something similar to that....

It appears the b) part of the algorithm needs a serious overhaul and it has been for as long as I have played this game

 

[unmerged(36024)]

1. Adana is a mountain.
2. May be AI amass troops for storming the castle, but having low tech can't do it?

Anyway it's a bug for me.