Current Impossible Strategy

[Nelphine]

Confusion is by far the biggest problem. Bless solves a lot of other problems. Pouring out bezerkers every single turn forever mostly makes up for it though. If you're having too many problems, getting 2 or 3 cities pumping out bexerkers should make up for it.

The extra speed from an endurance warship, for me, is huge. Its the fastest non cheesed hero unit in the game. Being able to move 5 overland is what let's you crush impossible AI so effectively, because you can attack anything you want.

It also has better sight range than a trireme I think.

In combat, the extra stats are super helpful to keep it alive, and ranged attacks mean you can do something (even if not a lot) without risking it.

Getting early nodes is vital. When you have higher casting skill than an impossible AI, it gives you a lot of freedom in combat - healing, dispel magic, black sleep, life drain, whatever is appropriate.

[Nelphine]

I'm also thinking 2 life 6 death is probably the best books. The extra production is great, and spells like black sleep help a lot.

[Nelphine]

My most recent version of this, I got stream of life in the .. Second lair I cleared? So my cities only needed one troop for garrison. Which increased my production a ton. Then I picked up demon lords in 1405 and had my first summoned before 1407. I defeated the fourth AI in October 1410.
I had a little over 550 overland casting skill when I won. (And I stopped putting power into casting skill in 1408 due to increasingly high mana demands.)
Including treasure found, I ended with:
Alchemy, Warlord, Omniscient, Spellweaver (found)
4 life (2 found), 7 death (1 found)

However, I did keep doing a fair amount if research (my mana demands hadn't got high enough to cap my power on pure mana). As of Oct 1410 I only had 3 very rare and 2 rare spells left to research.

[Nelphine]

As a further note on this, I'm finding I simply can't keep up with all the overland spells I want to cast. With the bezerkers (and whatever very rares I end up using), I can take cities faster than I can buff them. I can't keep unrest down via summons or stream of life, I can't get prosperity or inspirations cast, I can't max buff individual stacks of units. I constantly am building new things, summoning more spirits, summoning creatures to guard nodes, buff new warships, do the most important buffs on my attack troops, throw the occasional city buff where I can, disenchant area whenever an AI is death and at war (once they get even one curse, this becomes fairly constant). The pace is simply too fast on impossible to play my normal 'perfectionist' game. It becomes 'lordy I hope I remembered all the things I meant to do'.

One big problem is that this really really emphasizes how poor the AI offensive is. Due to how huge the AI becomes, very rapidly, I have to constantly stay on top of them, and try to hit them as fast as possible. But that means, when the AI is taking cities away from me, it means virtually nothing because AI can't keep up any actual offensive push. Since I'm either taking a city, or attacking a node, (or losing the full stack) almost every turn with every single bezerker stack (and I'm building new stacks roughly every 5 turns, so multiple stacks per year), the AI would have to be able to take 2+ cities away from me per turn to be a remote threat.


This also really emphasizes the problem of AI target decision making. Due to the whirlwind pace that I have to play to stay ahead of the AI, I'm constantly leaving cities as close to unguarded as possible (worst case, I lose a city, but that virtually always means that my main bezerker stack also took a city, which puts me ahead - the AI ends up with a city that was sacked twice, while I end up with a city that was only sacked once). Which means, I've seen the AI take 5 different stacks of 1 or 2 units, and instead of combining them and attacking the following turn, all 5 stacks think they are strong enough to attack me - so I can use my superior casting skill to a) defeat each stack without any problem and b) I can virtually always turn 1-3 units per invading stack into undead, which lets me acquire absolutely massive garrisons without ever building anything.

So far, I'm not really proposing any kind of solution, nor even am I sure this is actually a problem - it's more of a requirement, that if you can't play this fast, then the impossible AI resource bonuses will overwhelm you.

As a note, I have lost using this strategy once out of 7 games. It certainly isn't an automatic win - a lack of any nodes at the beginning that are weak enough for cavalry to take AND being unable to spike the first AI capital with the cavalry means you don't have enough snowball momentum to get ahead of the AI, and you quickly get overwhelmed (since you commit to early military, if you don't acquire an entire AI empire, you'll be extremely behind in cities; and without any nodes, you can't hope to keep up power wise.)

(In this game, my bezerkers, if I throw full buffs on, are melee 16, thrown 11, +4 to hit; armor 9, 6 hits per figure, +1 to defend, 8 resistance (+bless +resist elements +wraithform). Their thrown attack, assuming its against a behemoth with iron skin, then the thrown attack does 12 damage, which isn't much and the behemoth gets to fight back (then the regular attack does an average of 33 damage, and the behemoth gets one shot even with iron skin). Of course, I throw Prayer and Black prayer as well, so the throwing attack against the behemoth actually does 20 damage. Against something 'weak' like a great drake, the throwing attack does an average of 36.6 damage, so the great drake dies before it even gets to fight back. And Seravy was worried about first strike heroes....

Guess what? Confusion is still the worst thing ever. Dispelling wave is annoying, dispel magic is annoying, but the raw stats on the bezerkers are so high that those aren't really a big deal - which is also why I rarely actually buff my bezerkers to anywhere near the level described. Confusion on the other hand? Can singlehandedly take out entire stacks. The vast majority of my combat spells are raise dead and dispel magic. Great Wyrms are also annoying, but I've never seen the AI get very rare summons in these impossible games. They do start getting rare summons around.. 1408 or 1409, and I haven't actually fought a sagemaster or anything. So, great wyrms controlled by a sorcery wizard would be a nightmare.)

[Seravy]

I've just recently discovered the smallest ground stack the AI can consider using for an attack is 3 units, so the 1-2 units attacking should not be possible, unless those units had water movement. If this continues being a problem, we can raise this to 4 units. Unfortunately it also means if the AI has fewer units but could attack an empty city, they do not.

[zitro1987]

During a game playing a chaos wizard, I conquered a few Barbarian towns from a wizard with death spells. I recruited lots of berserkers to protect my towns, but forget confusion - possession is so brutal under the right circumstances I had to move my berserkers away in place for elven lords and pegasi.

I agree on the need for aggression, whether aggression towards nodes/caves or against other AI. However, I often find myself timid and on the defensive whenever nearby AI is tense/restless, possibly explaining why I generally can't win 'extreme', let alone 'impossible' in this mod.

[Nelphine]

Ah, seravy, definitely has water movement. Lizardman and/or prolific wraithform and/or prolific water walking. On impossible in the early years these two spells get spammed a fair amount.

However, that also explains why the AI doesn't take empty cities occasionally.

Any chance of making an exception for an empty city?

Also, as I said, so far I'm not trying to offer a solution to this. Its definitely getting into extremely complex target decision making. And unfortunately, its definitely related to 'they have lots of little stacks near each other, obviously they should group up', but I'm aware that's one of the specific things the AI doesn't understand, is the location of other stacks, their own or other wizards. It also gets into 'if I didn't have such high casting skill, and specifically, syphon life and summon zombie, this tactic might still be a good one to use against me', again, way too complex to propose any sort of immediate solution.

Hum. I have some thoughts, but it really requires understanding stack creation perfectly, which I certainly don't.

[Nelphine]

Zitro, yeah I specifically have bless for use against death wizards. This tactic simply doesn't easily change to other realms (though all 3 other realms bring very helpful upgrades - flame blade is amazing for barbarians, nature is amazing for omniscient, sorcery has resist magic and flight - I don't think any of them are as good as the 12 picks I'm currently starting with.)

[Seravy]

A question. You said the AI never reaches the point where they can use very rare creatures offensively (let alone in effective 9 stacks).
Don't know if it can be done but, If there was an additional stackbuilding procedure that gathers 9 of the AI's best flying summoned units into a single stack, disregarding garrison requirements outside the capital, would that help at all?
Does a stack of 9 Great Drakes (or angels or chimeras or whatever they have) help the AI more than having those distributed as garrison to protect the AI's cities? Do you think the human player would intercept and kill these units one at a time while they head towards the gathering point? Does such a stack even work against a human player? (I would probably aim for killing 1-2 creatures each battle until the stack is no longer powerful.)

[Nelphine]

This is a very slippery slope. First off, you wouldn't limit it to just summons.

But this is effectively how I play. My weakness is that a) my defenses are complete garbage and b) I have very few offensive stacks relative to the size of my opponents empire.

The AI would absolutely be much stronger if it started doing this, but not being able to do this kind of thing is one of the biggest reasons why its fine that the AI has such a gigantic numerical advantage.

(In another game I play, the original warlords, the AI does know this strategy. And it's brutally effective. But, the human can see the stack forming, and can see where it is no matter where it is on the map, because there are no viewing distances. With CoM, the majority of games, this stack would be undetectable until it slammed into the humans empire.)

I would suggest, this is the kind of thing that should be done as a side project. Its effectively the next stage in CoM AI development, and if you go that route, you'll probably find other tweaks you'd like to make as well - but increasing AI will require potentially major resource bonus changes to balance it.