AI overland stack/unit movement strategy

[Nelphine]

Can we make turn based on difficulty? 55-5 per difficulty?

[Arnuz]

Leaving less units in garrison for more aggressive personalities I like, mostly because it makes even more extreme the t40 cheating advantage problem and should basically force you to fix it Seravy

[Seravy]

Can we make turn based on difficulty? 55-5 per difficulty?

Maybe but it's not easy to do and confusing for the player so I rather not.

Checked the raze decision. Each unhurt (total damage<5) fantastic creature adds 1% chance for each 144 cost. This number seems to be based on their old costs. Changing this to 50 is necessary to reach the intended +100% razing chance for 9 great drakes.

[Seravy]

Leaving less units in garrison for more aggressive personalities I like, mostly because it makes even more extreme the t40 cheating advantage problem and should basically force you to fix it Seravy

Considering we already concluded the cons outweigth the pros on changing the early game, forcing that direction anyway might not be the best thing to do

[Seravy]

So...amount of units for each personality.

Peaceful - 3. If there is anything that needs more than the default, it's this one. Helps with the "allies with everyone" problem too, as there will be fewer strong units they will send after the player.
Lawful - 2. Can't make it based on war.
Chaotic - 2? We don't want the random element to bring an excessive amount of units? Otherwise, 1 also makes sense. (No, can't make it random)
Aggressive - 1. Makes sense. It does mean we no longer have a "default" personality that has no unique trait associated, though, as this used to be that one. Fortunately, the garrisons aren't very relevant in diplomacy.
Ruthless - 2. If the wizard triggers the Jihad chance, it'll use 0 anyway and if not, it should have the default so be able to stockpile enough units to have the surprise "where did these doomstacks come from, hey I thought those are staying in the city!" factor. 3 might even be a good idea to increase that effect, tho if they do not trigger jihad, that backfires.
Maniacal - 3. As these wizards rely on cursing, they have less need for doomstacks and more for self-defense. However if they did not get a curse (for example mono-Life, mono-Sorcery), this can backfire so we should also consider 2.

[Seravy]

While implementing the new intercontinental target system for ship attacks, I realized the feature that makes the AI avoid attacks with smaller stacks has a huge negative impact on this. Ships don't stack up very well, so if stacks below 4 ships have reduced "value", that prevents naval attacks from happening most of the time.
We might want to counter this effect, I think the AI is far more effective if it's reckless on naval attacks.

[Nelphine]

I don't know if I've ever seen ruthless trigger jihad, so I'd rather not increase them to 3 for that reason.

Mono life and mono sorcery are both strong late game, so I'm fine with maniacal staying 3 in those cases.

[Nelphine]

I don't know, I've never seen reckless ship attacks be terribly effective. Typically there are two things in the seas: other ships (which they don't need to attack recklessly to be effective against) and doomstacks (that no amount of ships does much to.)

And life/nature/death are not great at attrition against intercontinental doomstacks (when over water; nor is sorcery when against fliers), so its not even helpful in most cases to harass doomstacks.

[Nelphine]

Reducing the cost to razing ratio helps with having uncommons and rates raze cities too, so I like that change for note than just matching the 9 great drakes.

[Nelphine]

Leaving less units in garrison for more aggressive personalities I like, mostly because it makes even more extreme the t40 cheating advantage problem and should basically force you to fix it Seravy

Considering we already concluded the cons outweigth the pros on changing the early game, forcing that direction anyway might not be the best thing to do

We concluded increasing the production while reducing spellcssting was a bad idea.

We didn't conclude overall cheating shouldn't be reduced.