Retort balance, hopefully last time

[Seravy]

At most down to groups of 3 wolves, but preferably no less than 4.

Yeah but this isn't supposed to work (or at least be risky enough not to attempt it).
I mean, sure, it should work against 2 swordmen or something like that, but there should be some threshold above which it should not that's still not a doomstack but a common one the AI has plenty of.  Also there should be a major risk of the stack getting attacked and killed by an actual larger stack while split up. (nagas move 3 tiles, so 4-5 nagas attacking from outside the range of your vision, killing a group of 3-4 wolves, should be a thing. The AI can summon a stack of that in like 2 turns on Lunatic. By turn 30 when you have your first stack, they should have 60 Nagas roaming around. Per player.)

So I still don't understand how that can work. Mobility should also be a major roadblock - while Nagas move 3 distance on most terrain (they'll find a river, swamp, ocean or tundra tile to avoid getting slowed down), and random gnoll enemy stacks move 3-5 on land, wolves struggle to take more than 2 steps a turn and obviously they can't kill any unit standing on water either. So those nagas, even if you see them, will go and attack your undefended cities and the only way to prevent that is garrisons as they directly attack from the direction of the sea.
And it's not just nagas who do that either - Nature has sprites, (pure) chaos has gargoyles, and death, well it doesn't have much but even death sometimes has wraithformed stacks. Normal units are less of a problem - only Gnolls move too fast to reasonably catch but that's still a 1/9 chance per enemy player, although nomads might have some horsebowmen stacks that move 5 as well.  Life poses the least threat - in their case the risk is bumping into what looks like a weak stack, only to find out it it has so many buffs it actually beats the small group of wolves. (Have had that happen when I attacked a random low level hero for example, but klackon or gnoll units can be dangerous too.)

Berserkers I understand - 1 buffed berserker can kill an entire enemy stack and they have pathfinding too. But it comes that much later and costs that much more than the wolf it isn't a serious problem. (I'm playing Life berserkers right now and I couldn't really afford killing everything. I tried and I did so for the first wizard, but on the second and third they already had way too much units built up for it to be possible.)
But for wolves or other strategies, splitting up shouldn't be all that effective. I do that when I can and it almost never helps, the enemy just outmaneuvers the wolves anyway.

...is it the "no random attacks, wars or rampaging monsters before turn 40" limitation? I can imagine that providing 100% safety if you do manage to get things started 10 turns earlier due to extra retorts - the first wizard will be busy getting killed by your units but nothing else can attack you. Still on turn 41, things should start hitting the fan...

[Nelphine]

Wraithform trireme keeps your wolves very mobile. Still can't deal with water units, but those intercontinental units still move nearby on land regularly. AI don't go way off track to get to you until much later in the game. You can identify within about 4 squares, maybe 6, where all the AI units will travel. Then you just attack along that route.

And no, by turn 50, 4 werewolves is far stronger than almost any stacks the ai has except doomstacks or developed city garrisons. Taking out a magician and 2 swordsmen is no problem. 2 halberdiers in addition might make you use some spells like summon zombie to assist, but they still die without (generally - confusion is the only real exception) any losses for the werewolves.

(Note that I'm not talking so much about lunatic. We don't balance around lunatic for a reason )

[Seravy]

Well, werewolves are supposed to be a Master difficulty strategy if played perfectly - like most of the 8 book uncommons - so if it beats the game up to there, and not lunatic, that's fine. Especially as it doesn't beat the game by itself, you still need to transition into another strategy after the first 1-2 wizards.
A 4 stack of wolves does cost 420 MP (without retorts) so it's anything but a small stack of units. You could have almost 8 bears from that, nearly a full stack of the strongest common creature.
If you could do the same with 1-2 wolves, then we'd have a problem, if 4 are needed, not so much.

Wraith Form ships are great, I used them too. Still, it didn't really help when I faced multiple stacks of 2-4 gnoll units that could each reach my freshly conquered city, with stacks of nagas also coming from all directions. I guess that was a "worse than average" game though, fighting two people with nagas plus one of them having gnoll and barbarian units coming on top of that. I still won the war, but I got slowed down having to play whack-a-mole with all those things, losing and reconquering the cities, so by the time I conquered the continent and managed to beat those two wizards, the third one won the game, so far ahead I couldn't hope to catch up. I don' t remember the other test games but that particular one is hard to forget.

[Bahgtru]

Guys don't go on about later phases, this is all about the first 20-30 turns and the snowball effect. Wraithform triremes or full stacks of bloodlust wolves is just out of the picture. For the first turns, you just don't defend. As also Sapher and Nelphine show you keep your territory clean instead. If you don't believe me just open the save files in the zip. This is just the ABC to beat lunatic: leaving units in defence is a waste, and should only be done for unrest duty, so you need to know how much you can split your units and still not lose them. To kill roamers quickly, you need small stacks that are both your defence and your attack. Now that this is clear to you and out of the way, the only relevant bits remaining are these two:

why should the wolves be stopped by confusion if they are meant to be unstoppable?

They aren't. The AI can cast it once, maybe twice. So each battle you lose one or two wolves but win the battle. So you can keep conquering - but it'll cost you, and this will slow it down.
However it does eliminate the possibility of attacking the sorcery wizard with fewer than 4 wolves.

Perfect. I agree wholeheartedly: this makes using small stacks to defend while attacking difficult. Now pray tell... Why the heck exactly the same should not happen with a death AI? Why should the death AI be fucked because I have death immunity and laugh at black sleep, zombies and ghouls? (and in your answer keep in mind that we're talking about turn 1-30) (lol ghoul poison, don't even try)

Weakness just reduces the force a bit, black sleep changes the numbers completely

Because halving two units is completely different from eliminating one. Oh wait. If I have two units that deal no damage, I'm in a worse situation than if I have one but it deals full damage.

No seriously, 2 weakness < black sleep on WWs, please accept this simple fact. Each weakness drops the tactic value of the unit by 1/3 in my eyes, in the late game. But at the beginning it basically does nothing at all with swordsmen, and even halberdiers can be dealt with. (Nelphine I don't understand your point with alchemy - I'm not comparing tactics, I'm putting WWs against AI units)

So in the end: there's a problem with WW death immunity evidenced by pawning the game at lunatic, what are we going to do about it?

Nothing. We don't have options anyway, and one, reasonable, win is not grounds for nerfing a unit.

"Reasonable"? That snow-balled me onto a win with almost mono-death against an AI with exorcism? Are you serious?
OK, I'll repeat that victory then. Give me some time though, I'm busy.

As far as options go, yes we do: your own idea - undeath immunity - modified to take account of the feedback received.
That's "reverts to original on death": gives some undead but not the broken WWs that would break the game, and even leaves some of them for fun (the neutrals) while giving lots of flavour.

...is it the "no random attacks, wars or rampaging monsters before turn 40" limitation? I can imagine that providing 100% safety if you do manage to get things started 10 turns earlier due to extra retorts - the first wizard will be busy getting killed by your units but nothing else can attack you. Still on turn 41, things should start hitting the fan...

Thanks to additional resources through conquering I could beat a 2 life books wizard with exorcism just to give you the idea of how much shit hit the fan... I did lose 1 important city, but caught up as soon as I could re-organise myself. This is lunatic, as a reminder.

Again, I'm not saying they're totally OP, and I agree that they should beat the game at master. I'm only saying that death immunity makes starting against a death AI a walk in the park at lunatic, so the solution needs to be improved. "Reverts on death" does exactly that while keeping the game safe against brokenness.

[Nelphine]

I'm pretty sure reverts on death is out of the question. We don't have extra unit buffs available so I don't think there's anywh3re to store the data of what they originally were.


My point about alchemy referred to an alchemists guild for magic weapons on the AI halberdiers (or the AI has the alchemy retort).


In the first 20-30 turns, I don't see it. It's a minimum 11 turns just to research werewolves, plus at least 3 turns (more likely 4) to summon the very first one.

Doing that, and reliably getting enough lairs to guarantee enough income to summon even 5 by turn 20 seems out of the question. And with less than 4, you're still not going to even beat ghouls - poison totally works when you have too few. Or it did in my game.

[Bahgtru]

You swing your multiple full stacks out, and sweep up all the incoming units while you conquer.

Yeah but this is exactly what you can't do with nerfed wolves. You can't split the stack up to that extent. Almost every nerf we did recently is trying to prevent that. So how is it still working then?

Again, I've made it work vs a death AI because of death immunity. Fix this oversight with "reverts to original form" and we're good. That's all I'm saying, not to nerf them further.

I'm pretty sure reverts on death is out of the question. We don't have extra unit buffs available so I don't think there's anywh3re to store the data of what they originally were.


My point about alchemy referred to an alchemists guild for magic weapons on the AI halberdiers (or the AI has the alchemy retort).


In the first 20-30 turns, I don't see it. It's a minimum 11 turns just to research werewolves, plus at least 3 turns (more likely 4) to summon the very first one.

Doing that, and reliably getting enough lairs to guarantee enough income to summon even 5 by turn 20 seems out of the question. And with less than 4, you're still not going to even beat ghouls - poison totally works when you have too few. Or it did in my game.

"revert to original" can mean anything, just make that a random swordsman for example, that's going to give you a garrison unit for unrest that'd be perfectly fine to keep a player who's invested in undead creation content if not happy. If there's enough space, make a swordsmen of a race that that AI has a city of, and we're actually even precise 80% of the times...

Yes if you find the alchemy city you have some problems but you can account for that as you see it. Put your various small stacks together and you should beat it. I've managed, maybe it doesn't happen always but the question remains: why should it be so much easier against a death AI than a sorcery AI?

[Nelphine]

Why not? Why is it so much easier for life to kill chaos than nature?

[Bahgtru]

But should a realm be tailored so that it sucks in the AI's hands against itself in the player's hands? That feels completely wrong.

[Nelphine]

Life and sorcery are the same. AI is bad at buffing effectively. Arguably nature us the same because AI is bad at focusing stacks.

[Seravy]

Perfect. I agree wholeheartedly: this makes using small stacks to defend while attacking difficult. Now pray tell... Why the heck exactly the same should not happen with a death AI? Why should the death AI be fucked because I have death immunity and laugh at black sleep, zombies and ghouls? (and in your answer keep in mind that we're talking about turn 1-30) (lol ghoul poison, don't even try)

Because if every enemy is the "bad, takes too long to beat with too much losses so smart players pick the other one(s)" kind, the strategy is no longer playable?
It isn't the case though - it's strong against Chaos and Life (before exorcism), weak against Nature (flying sprites are hard to deal with in larger groups) and Sorcery. If it's also strong against death we are good vs 3, bad vs 2 which is ideal - AI's can play more than one realm so a 3+2 distribution still puts roughly half AI wizards in the bad category. (only needs a few nature or sorcery books to have confusion, nagas or sprites)
Note these are "soft" weaknesses - you don't lose against those enemies but it makes the tactic slower, more expensive and less effective. If it was a "hard" weakness, we'd want no more than 1 realm that falls into that category.

Although honestly I still think the Weakness spell should have the exact same effect as Black Sleep. Those small stacks, when potentially reduced to only 3 attack strength, should be at risk of getting wiped out.

"Reasonable"? That snow-balled me onto a win with almost mono-death against an AI with exorcism? Are you serious?

You were beating Exorcism because you had Resist Magic. That has nothing to do with snowballing, it's pure luck. Anyway, at least compared to what Sapher has been showing us, the game still going in 1408 is reasonable
(Yes, it does count as a very good win still, but it's not so far off the scale to be unacceptable. Every once in a while this outcome is still ok.)

I'm only saying that death immunity makes starting against a death AI a walk in the park at lunatic, so the solution needs to be improved.

If, and I still have doubts, but if this is indeed needed, then the solution should be to improve Weakness (make it -4 attack, at only 2 attack/figure the wolves should no longer be able to fight even weaker halberdiers.) as it fails to serve the purpose it was meant to.

I'm pretty sure reverts on death is out of the question. We don't have extra unit buffs available so I don't think there's anywh3re to store the data of what they originally were.

Now that I think about it we could store it the same place we stored settler race - in the "road building target" byte. Wolves also can't build roads so they don't use it. It does feel an excessively complicated solution though for a very simple problem (the Weakness spell not being effective enough).