Test games played

[Seravy]

Once again the AI struggles with mana costs a single turn into the war, while having off the chart army strength (like, 8 times mine) and tons of stacks.
Maybe we should lower the AI gold bonus (do they even need ANY gold bonus? They might not be as good as the human but they at least know to spend more on adamantium etc cities, and they already have a production bonus for generic production...) and raise the power bonus (but this risks all the extra turning into research and skill if they weren't at war, like the Myrran).
The AI is playing dwarves though which means they are bad at magic power and strong at units, so that pushes it strongly in that direction and might be the only source of the problem. (last game the AI had mana problems too, but recovered from it soon after losing the initial wave of stacks entering my plane, and forced me to run low on mana as well, equalizing things. This game I never reached low mana as I was able stock up much more due to delaying the war, and due to the AI actually starting from half my amount (dwarves->less amplifying towers->less casting skill->less stored mana)).

The AI has Counter Magic from their 2 found Sorcery books. Considering battles are already hard to win as is, this often pushes otherwise winnable battles into losing ones. I lost several cities to it already. I don't remember if we already decided Counter Magic is good as is, or it's still listed as potentially overpowered, but in battles where the only spell I can use is Animate Dead (40MP due to Divine order, so 50% to counter), it's really powerful.
...when the AI has no mana, I can even kill archangels. I can wear down one with bow attacks and spells, then animate it to kill the next and keep doing that. Doesn't always work but horsebowmen are fast enough to sometimes buy enough time.

Necromancy feels overpriced for 1000. It's clearly useless once I have the superior Animate Dead, and too expensive to make before that. It's also extremely situational, it's not that often when I use a stack that has a hero in it, excepts relevant losses, but can guarantee the safey of said hero at the same time. Plus the hero needs levels for the ability to do like anything at all.

I think the AI should cast Endurance more as a standalone spell category when certian conditions are met :
-AI starts with dwarves, past a certain turn (hammerhands with move 3 are scary, move 2, not so much, even if they have tons of other buffs)
-AI knows Hydra (really nice if it moves 1 more and has the highest target priority for buffs)
-maybe if AI knows Death Knights (they can reach on turn 1 with the buff, not otherwise)
-maybe if the human has Entangle
Don't think anything other cases benefits particularly critically? (sure, it's a good buff on almost any very rare or good unit, but something that stands out)
Overall, the AI can't stop my demon lords from destroying their cities, and while I did lose several of mine, I am reclaiming them and the AI will soon run out of stacks on Arcanus (if nothing else, zombies eating away their mana and smaller stacks while multiplying makes sure of that) so winning seems inevitable - we already know the 9 demon lord stack is capable of banishing the wizard if I happen to go for it, too. (they have a soul linker but it dies to doom bolts on turn 1 before archangel can reach the demon lords, so it's actually easier than the 9 archangels...)
Also it seems a single pegasai or shadow demon can kill an entire stack of dwarf units when the wizard is out of mana so blocking towers will be trivial on the long term...if the game even lasts that long.
There are no lairs left on the map for the AI to find new spells from, so I declare this a win.
(they also missed Consecration, so I could pestilence their empire for the win if I wanted. It's even 25% cheaper due to Divine Order  
oh and missed Enlightement too and dwarves aren't good at research so I could even do Spell of Mastery.)

As for the difficulty, idk. I used an abnormal strategy (razing all of arcanus) and I would have lost if the Myrran AI declared war immedaitely due to it.
So hard to call this a real win, I was betting on a ~40% possibility by doing that, and got lucky. I also got lucky on Arcanus oponents (they were fighting each other, and used no nagas or other units actualy dangerous. All 3 were using sprites only!)
This was clearly on the lucky side of the spectrum (good start, good opponents, lucky on Myrran) but even lucky games are part of the average.
Still since our definition was 100% wins with this sort of wizard, we need to win unlucky games to really have conclusive evidence.

This is the result of  like 3-5 turns of war, without even dispelling their globals :
   
There are no more archangels left on Arcanus except a few in one of my lost cities. Even if she had more, I can easily kill them due to no combat spell support, and abusing the OP demon lords. So I don't expect to lose any more cities.

[Nelphine]

I really really think counter magic and dispel magic should be based on base spell price. In particular, things like specialist and 10 books making your spells weaker is .. bad.

But even with that, I would keep counter magic on the watch list.

Endurance is unbelievably good. It should be cast on every unit in every full 9 unit doomstsck, and every transport, as top priority.

[Nelphine]

In terms of razing cities, the whole point is to use it in order to avoid garrisons in order to kill the enemy faster than they can react to. You do it so that you need exactly one offensive stack, and you kill the entire eneny empire. You don't need economy at that point.  Everything else that should go into offense goes into defense against the next ai, who you're not even at war with yet. Which means, you can get away with far less economy that you would need to wage a war holding cities.

Admittedly, I generally don't raze the first opponent I fight - usually one doomstack is enough there anyway. But I've taken to treating the 2nd and third ai as the same opponent, and only declare war once I can destroy both. (On myrran I usually treat both ai as the same one, and then the third and fourth as the same one.)

[Sapher]

Uploaded part 2 and part 3

[Seravy]

Watched all 3 parts, some thoughts.

Werewolves seem borderline overpowered. They either work or not but when they do they can easily clear the whole map,if you have enough of them. Adamantium certainly makes a difference, and 50% more of them from spellweaver too, but I think without those there are no problems so I don't think we need to change the wolves. (or at most, by reducing their defense by 1). Normally wolves become obsolete way before reaching the final enemy, often even before reaching the second last, although adamantium or certain buffs can extend this significantly.

Life Drain allows converting all your gold (if alchemist) into casting skill at a rate 2.5 times better than just directly spending power on it, if the target is 4 resistance. (or 40% better than being an Archmage). Again the spell is good as is (you have a range multiplier in most battles, there are a limited amount of enemies, and you don't always have alchemy), but the specific combination of having alchemy and a regenerating troll neutral at 1x range breaks it. Don't think there is anything that can be done about that but it's clearly gamebreaking. Lairs could be abused similarly but no lair monster can have 4 resistance, lowest is Hell Hounds at 5, but those are move 3 with fire breath so it's impossible to avoid engaging them with just a draconian spearmen. So it's specifically just troll neutrals and nothing else. Which means, if we care to, we could prevent this by making troll unavailable for neutral cities closer than 15 tiles to the human's starting city. (Note that it's an entirely luck based "strategy", you either get a near troll neutral for free 2.5 times the casting skill, or not. (Also this effect was cumulative with spellweaver and having adamanitum, amplifying the whole strategy to 2.5 times the default SP generation, which, even considering skill being nonlinear, is a 58% boost, or 137% together with Spellweaver.))
Considering the AI can't do this strategy at all (in fact can't gain skill from Life Drain and Syphon Life against nonhuman targets), we might want to actually disable close troll neutrals. (I'm not a fan of that for race diversity reasons though. It pretty much means you can't have troll cities in the early game if you haven't started with them, and that's sad.)

Hard not to notice that 2+2 opponents as Myrran is a huge benefit for early strategies, as there is one fewer player to beat (so over 33% less stuff to kill) on their plane, while also having a weaker final player than usual. It might be unavoidable to make Myrran into 3+1 opponents like Arcanus, even though I don't like that idea.

Spellweaver boosting werewolf production isn't very good for balance either, although it isn't cumulative with Conjurer at least. Moving the spell to the summon category could easily change that but I like Death having this spell  to benefit from Divine Order and don't want to move it. There is a chance we replace Spellweaver in the future (although I'd say it's a 50-50, unlikely to come up with a retort really worth including when we couldn't for so much time) which is another reason to avoid changes based around it.

Ship movement abuse also contributes a lot to early strategies winning, either by letting the unit move an infinite distance (carried by a chain of ships), or attacking twice (in the game I played, the wraith form ship carried the wolves into a second battle quite many times). Unfortunately, there is nothing that can be done about this as the existing ship rules allow for doing it - it's not a bug, it's a legal form of movement. This is another thing the AI can't and will never be able to use. I don't really see how the rules could be redefined without changing the intended functionality (ships carrying units without spending movement on those units) and keeping the system intuitive.

[Nelphine]

I'd just ignore the troll abuse. If people want to use it for lunatic that's fine; that's one reason why lunatic should have excessive resource boosts even if lower difficulties get changed. For most games and strategies, it's not worth the loss to varied gameplay.

On ship movement - who cares if it was part of the original rulesets. There's no intuitive reason why it's different from windwalkers, and there's no balance reason for it being different from windwalkers. And even without excessive abuse as you describe, just getting to move again after the ship is something the AI can't replicate, so by itself that's realistically a minor abuse.

I'd much rather ships and windwalkers acted the same.

Werewolves I think are fine. Multifugure adamantium we all know is a huge problem and still is. No reason to change anything for that unless you're going to address the actual problem (same as slingers, bezerkers, pikemen, etc)

[Seravy]

On ship movement - who cares if it was part of the original rulesets.

Sure but that means we need to define a new ruleset. One that works, and is better. Which I think I would have already done if it was that easy to do.

and there's no balance reason for it being different from windwalkers.

There is a huge one. Wind walking allows the units to fight. So the stack, while using a wind walker, is safe in combat. (at least if they are ranged. Otherwise, not so much. But even for melee if you can cast flight in battle, they are able to participate.)
For ships, the unit has to be able to move specifically to avoid getting sunk on the sea. If the ship used up the movement points, the units on it won't have any, UNLESS they are faster than the ship. So transporting cavalry would still be safe, transporting bowmen or hammerhands would not. Now that makes even less sense. I'd say at least 50% of ship uses by human players fall into the category of "make ship, load troops, unload on the other shore, AI sinks ship next turn". That sort of maneuver would become impossible, which, paired with the AI naval superiority that happens in most games, means the player is landlocked to whichever continent they started on and if it isn't large enough, lose the game.
Of course we can decide this is better than the existing system and go for it, but I don't think we should. Even if the AI actually plays by these rules due to its limitations (cannot unload troops without passing a turn), they are not very suitable for human use.

There is another difference btw, wind walkers can leave stacks and move alone. Ships cannot, they forcefully drag the entire stack with them. As is you can move the ship, then take a step backward with the other units if you want the other units to stay in position (as long as they had at least 0.5 movement left). If it was like wind walking, you could not unless all the land units had at least 1.5 moves left instead.

[Nelphine]

No, I want to make them the same. Meaning, ships would stop dragging units along if the unit was a) patrolled and b) on terrain that it's normal movement can handle.

And that 'most common scenario's has one more component - usually the units only have to move 1 or 2 tiles on the ship to reach the next continent. So even with this change, most units will have enough movement to make it work, even if the AI sinks the ship the next turn.

Yes windwalkers do combat things to. Not relevant enough, because combat only matters in that common scenario you described. And either, the units make it so don't get attacked, or they don't make it, and the ship sinks anyway. Actual combat ability means almost nothing.

[Seravy]

So even with this change, most units will have enough movement to make it work, even if the AI sinks the ship the next turn.

Most units are move 2.
1 move to step into the ship, 1 move on the ship, out of movement.
Move 3 units would be able to travel a single tile, but two, forget it, you need at least cavalry for it. (move 4 or 5 units are fairly rare, actually. And most are already flying.)

Besides, even if this wasn't the case, faster units being able to leave the ship while slower units cannot makes no sense whatsoever.
Further adding to the confusion, if the ship can still move after reaching the shore, the unit would have 0.5 moves left, even if it exhausted its movement points, so it could move, even without the ship. But if the ship used its last move, the units also can't move. 

So this idea is rejected. It's just too counterintuitive. If anything, wind walking should work like ships, not the other way around, but it's a rare enough ability that I rather not mess with it just for that. It was pretty hard to make it work without bugs at all. But wind walking movement is definitely the worst possible movement ruleset I've ever seen in any game.

And either, the units make it so don't get attacked, or they don't make it, and the ship sinks anyway. Actual combat ability means almost nothing.

Except that's not how the AI sees it. If carried by a windwalker or floating island, the entire stack is counted in the defending force. On a ship, only the ship (and swimming/flying units) is. I made sure of that when implementing sea and intercontinental attacks.

[Nelphine]

Seriously? Windwalking works exactly how it's expected. Every time I use ships, I lose movement - carrying units I don't want to. Windwalking matches other games to a T - it's intuitive. Ships are boggling.

Anywho, you and I disagree that's fine. But don't say ships are full of abuse without acknowledging that the ruleset makes them that way (I'm fine with you stating that ruleset, including the abuse, is better than the alternative, even if I disagree - but the ruleset leads to the abuse, and you have another ruleset already in the game that doesn't lead to that abuse, is intuitive, and is balanced between the human and AI.)

As for how the ai sees it, that doesn't change the outcome of transporting units via ship. They'll still see it the same way regardless.
And no, 3 speed units are fine, which are the most common. The ship is built in the city with the stack already garrisonung the city, and the real garrison waiting outside, so no need to waste movement moving into the ship. Otherwise, the ship still gets sunk too early according to your 'only lives 1 turn' theory. If it doesn't (which in my experience is actually the case), then typically, the ship can move and pick up and drop off several turns before getting sunk. It's usually on the way back to get more units that it's sunk - not en route with the first stack.