Sleep and zombies, or zombies in general

[Bahgtru]

Can't you see that you're calling werewolves anything but fair or balanced though in that sentence? Otherwise, why would they be the exception? They're not the only strong unit with low hp...
For example: I remember talking about +1 but right now they're +2, I don't know if that's the version.

[Seravy]

meanwhile edited my post to add this part :

...also it's bad design when a problem the player has to solve, ends up solving itself, and death wizards summoning the wolves they are weak against once turned undead, is exactly that sort of thing. Like how you don't make monsters weak to poison drop poison daggers unless your goal is to make an easy, braindead game.

Undead wolves are the counter to death wizards in the early game. Death wizards summon werewolves in the early game and you have the means to convert them in the early game. So Death wizards self defeat themselves and the original correct strategy to beat them (research Blood Lust, put it on anything good, werewolves are ideal but you can use other things) will no longer matter.

Oh and btw, you don't need to put the wolf to sleep to feed it to zombies. Put a cloak of fear on your zombie, good enough.The zombie might even get lucky and chew through an entire stack of them without getting hit once.

[Bahgtru]

Maybe death AIs shouldn't use WWs when the player has a death book count then? You have other similar rules in other cases I believe. Or not use lycanthropy until it also has bloodlust in that case?

Sorry if this seems to be insisting.. It's just that raising wolves IS fun, but not necessarily wizards - neutrals too. And finding creatures that you can't raise sucks as an undead strategy. So any creature that can be summoned improves the strategy, and WWs can happen in lairs - they'd be the one death lair raisable creature, which is why the decision to make it impossible sucks. This even when you can summon your own, but moreso when you've not reached the spell yet. So in short, it's all about the fun.

[Seravy]

Well if it is that much fun for you, I recommend changing the unit tables using the tweaker to remove the immunity from future versions. I can't really imagine having a 95th type of unit I can raise when I already have 94 others, as a large enough increase in fun to care. Especially if it's one I can summon.

[Bahgtru]

Eh, the issue is that when you find them then you don't have other units to raise, in that match. The total number is not as meaningful as that.

Also I would have the AI self defeating, as you explained. How difficult would the change be?

[Seravy]

It wouldn't be difficult but it would make the AI weaker so I'm not doing it.

[Bahgtru]

Uh, teaching to prioritise bloodlust on WWs when the player's using death makes it weaker?
While doing it, you could also teach it to prioritise it against sorcery, to gain illusion immunity.

[Bahgtru]

How hard is it to get a werewolf with ghouls? If it's too easy, that still ends ups with the ai making werewolves, that the human then collects, and then the AI who made the werewolves has defeated itself.

Isn't that exactly the same with - say - hellhounds though? Aren't you calling undead raising broken per se?

[Nelphine]

No. AI is BAD at creating undead in tactical combat.

That means if you make werewolves, and they can be turned into undead, the death AI cannot turn your werewolves into undead.

However, since werewolves aren't death immune in this scenario, the death ai can still kill your werewolves.

However, we reverse the scenario. Now the AI summons the werewolf. The human isn't bad at creating undead, so the human uses their ghouls to kill and raise 2-3 lone werewolves.

But now those werewolves are undead. So the human sends them back against the AI. And now it is death immune so the AI cannot in turn kill the undead werewolf.

Werewolves are designed to be one of the strongest early game units. They beat everything else common and uncommon on damage and hit points.

All they have is bad resistance. 

So no, they aren't the same as hell hounds, or any other unit. Werewolves are designed to beat those units. If your ghouls raise those hell hounds, the ai werewolf still beats them.

 And the undead hellhounds can still be killed by the chaos wizard. Hell hounds can be killed by fire bolt. Werewolves can't - they have too many hit points.


If werewolves can't be turned into undead, then when the AI makes them, the human can't immediately steal them one or two at a time, and get stronger versions. Instead the human is limited to the same methods of killing the werewolves that the AI has available to it.

[Seravy]

Uh, teaching to prioritise bloodlust on WWs when the player's using death makes it weaker?

No I meant to not research werewolves would be a bad move.
The AI doesn't have early researching tactics implemented for Blood Lust (and that thing isn't conditional, it's "if turn<X and spell=werewolves, push research to Y% on the bar", that simple, it goes into the power distribution procedure, not the research choice.) and normal research production without that is extremely slow at the beginning of the game. That would delay wolves to the point where it's already meaningless to prioritize the tactic.

The human tends to do "I need this spell now so I max research", but the AI doesn't have such intelligence. Unless it's explicitly coded they need to push the research bar up, they'll follow normal power distribution rules that are based on turn count, objective, and overall game state (who is stronger and in what areas, how much of each resource the AI itself needs, etc) - they will pick Blood Lust after werewolves, but won't spend on RP or build more libraries for it.

Then, the AI doesn't use "on demand" casting of Blood Lust - they might use it more, or less than werewolves, "less" seems far more likely, again depends on personality and game state. So they'd still have unbuffed werewolves. (I already explained this recently)

These two are enough to refuse the idea... but actually, there is one more limiting factor for the AI. To pick Blood Lust for research, it needs to be visible in the book. So it had to be the AI's second guaranteed uncommon pick - first was the werewolves.
Looking at the list it might be a good idea to actually change that pick though - if the AI has 8 books, guaranteed Black Prayer is not meaningful as it's not an early research choice. But that still leaves several other uncommons to choose from - I guess I should look at their research order and choose the spell they would want to have first. In fact, might as well do it for all realms and see if they properly have different choices for 8+ and 7- book uncommons, as the meaning for each is different.

So, in Death 
research order is Werewolves (already picked), commons, Possession, Reaper Slash, Syphon Life, Black Prayer, Night Stalker, Shadow Demons, Blood Lust, Drain Power, Wall of Darkness.
Additional relevant rules that modify this order are:
-Delay werewolves if it's no longer the early game (is useless)
-Blood Lust first if Werewolves known
-Shadow Demons first if human plays Death (this is there against human using werewolves, but it's)
-Reaper Slash, Syphon Life delayed if no contact (more), or no war with human (less)

I thought we had some sort of "earlier blood lust if human plays life" rule but I'm not seeing any so I guess we decided to abandon that idea. No, we assumed it happens by default due to the werewolf based priority rule?

So yeah picking Blood Lust as second guaranteed at 8+ books seems better than Black Prayer, as it already guaranteed the werewolf rule will trigger.

Nature
This has more choice rules :
; Nature guaranteed UNCOMMON #1 : Cockatrices if Focus Magic
; Nature guaranteed UNCOMMON #2 : Transmute if Warlord, Myrran but not if POOR minerals or alchemy
; Nature guaranteed UNCOMMON #3 : Giant Spiders if (not Myrran or human Myrran) and not Cockatrices
; Nature guaranteed UNCOMMON #4 : Crack's Call if no more than 1 death and chaos book - no effective removal/counter to heroes!
; Nature guaranteed UNCOMMON #5 : Great Lizard if not Shadow Demons
; Nature guaranteed UNCOMMON #6 : Change Terrain
; Nature guaranteed UNCOMMON #7 : Land Linking

Research order is Spiders, Land Link, Ice Bolt, Transmute, Cockatrices, Lizard, Crack's Call, Change Terrain, Catapult, rares
Relevant rules for order are
-Ice Bolt delayed if Fairy Dust already known (almost always)
-Land Link faster if Spiders known
-Cockatrices right after Land Link if Focus Magic

Cockatrice choice looks fine, it's prioritized high enough to get chosen for research. Transmute likewise. Spiders too. Crack's Call should be 7 or fewer books exclusive, it's not an early research choice. Lizard too. And Land Link should be before Change Terrain as it's just a plain better spell on top of being research prioritized.

Sorcery
This picks Aura of Majesty and AEther Binding, if playing 8 books. They are also the first, and third choices in research order, so that's matching. (Water Elemental is second, if it does show up before AEther Binding, the AI will go for it, but very unlikely.)

Chaos
Picks fire storm, gargoyles, and in the unlikely case it already had wolves or spiders then lighting instead of the gargoyles.
In research order, Gargoyles are first, then Volcano, Chaos Channels, Fireball, Lighting Bolt, Fire Storm, Fire Giant, Chimeras, Surge, Immolation.

Modifier rules :
-Fire Bolt delays Fireball
-Lighting is delayed if no contact or war with human

Gargoyles choice is matching, but Fire Storm isn't really. At 8+ books it should prioritize something else, that's earlier on the list. We want Volcano to remain random chance as it's not fun to play against so making it guaranteed is a bad move. Chaos channels...why does the AI even research it this early? Oh yeah low cost and chaos uncommons aren't that great. It's probably nice to have early sometimes for that surprise flight factor but I don't see why this should be a guaranteed spell. The combat spells are useless if no war happens, while Fire Storm is too it's expensive so getting it in advance is better than hoping it gets done while being killed. So I don't have a better idea than leaving it at Fire Storm, although changing to Lighting Bolt could...eh, no it gets delayed, even if chosen. Ok, this can't be improved.

Life
Steam of Life and Altar of Peace if 8 books. Dropping Altar of Peace if Chaotic but that's a corner case not worth optimizing for - in this case they get Prayer.

Research order is
Altar of Peace, Stream of Life, Prayer, other spells.

Good, perfect match. Well, this took a while but we did find two places where we could improve it.

While doing it, you could also teach it to prioritise it against sorcery, to gain illusion immunity.

Good idea but as you can see they already prioritize picking it for research against everything, assuming they had the wolves. The weak link is providing the RP to finish the research and casting the spell enough times, both of which we can't and/or don't want to modify.