Life realm AI in the early game

[Nelphine]

I think for mono life we should override the turn 30 research. The point of not doing research is that you're summoning things. Yes you do some things here and there that aren't summoning, but summons is the main point specifically to not be wiped out pre turn 40. They shouldn't research uncommons until they've got the top 8 commons though, although I understand you've said this would be difficult.

If we embrace that mono life just doesn't worry about pre turn 40 then I don't think they need to avoid research or skill power either.

I firmly believe mono life should start with holy armor and that it should set their priority to use unit buffs to high. However, we don't want this on multi realm wizards with life. So I would make holy armor the 5th pick.

Before that, they should have heavenly light and just cause, because they can use those pre turn 40. (Which one is number 3, which one is 4, is dependant on which you want triple realm life to get).

All life wizards should pick up heroism for heroes - that should stay as number 1 for AI. Endurance is also great for all combinations of life realms, so that is number 2. Holy weapon is 6, bless 7 (it could swap with holy weapon), healing 8, star fires 9, guardian spirit 10.

[Seravy]

Okay, so what should the Life AI's power distribution strategy for turns 1-30 be?

100% mana before turn 10 (to ensure they can at least cast their heavenly light or whatever else we make them start with right away), 50% mana and 50% research from turn 11-30?

[Nelphine]

Probably? I don't know what the settings ate normally (as in past turn 30) but I could see it just going to those settings on turn 11. Unfortunately this is one area I know very little about when it comes to the AI, because it seems to work beautifully how you have it set up, so I'd rather bow to your experience for exact numbers there.

[zitro1987]

5 starting common spell proposal:
*heroism
*holy armor (useful early)
*guardian spirit (the more military-strength version of it)
*heavenly light
*Star fires
other 5th options: holy weapon or just cause (but have just cause cast around turn 30 or so)

Why just cause is cast so early with AI has a town or two? not a good investment early. The AI should spend its precious mana with heroisms+holy armors/holy weapon on swordsmen and guardian spirits (and heavenly lights)

Healing is not a great AI choice. We're pretty much all agreeing they struggle at the new state with its initial spells and expansion and combat spells also cost mana. I'd rather give them star fires, which can severely damage/kill some of your early summons like ghouls and sprites.

uncommons: why those 2 spells? especially in place of important, cheap commons? Have them research some commons first. then have them pick instead:
*Prayer (now there's a combat spell!)
*maybe astral gate is ok.

Stream of life is a poor choice to rush-research by AI. It's a very expensive spell when they should rather spam guardian spirits or cast heroisms and other buffs.

[Seravy]

Why just cause is cast so early with AI has a town or two?

To get 10 (or 20 if famous) fame hero offers. They are vastly superior to basic heroes and if they got a bad hero it will not be dismissed to free up room for better.
I'd say it's important to have it done before turn 30 (earliest hero offer) but it doesn't need to be the first thing to cast anymore - heroes no longer appear on turn 1.

Uncommons - because the existing research system was not flexible. I've started to rewrite it, which will allow researching the important commons first, but this will take at least a whole day to finish. (for one thing, I have to sort all 200 spells out to define a preferred order of research)

healing - because buffed "super" units are utterly useless if you can't heal them. No matter how many common buffs you have on the unit, it will take damage from fire bolt, fairy dust, etc. Not a lot of damage, but for a swordsmen, even with buffs, 2-3 of those can mean getting killed - and that's not acceptable when it had 200 mana of enchantments on it. Also, to keep heroes alive, which is no longer necessary before turn 30 but still important after.

Star Fires might be a good idea - the only problem is if a normal unit attacks, it does nothing, while the healing might do something. The AI won't get replacement swordsmen until around turn 30 so it has to keep alive those few it has - if it even has enough (on high difficulty only) - and due to their low numbers, they'll have a high concentration of buffs, making healing on them a very good investment.

Stream of Life is expensive, but - and I say this the fourth time I believe - the AI has to be prepared for a game state when they are not in a war, as that is likely going to cover 50% of the cases or more. (They might not meet the human, they might be on Myrror, they might not get attacked by them, etc)
I'm considering to reduce the casting cost to 180-200 though, I'm having a hard time to justify a spell that has no immediate benefit at uncommon costing 300 even if it's very powerful on the long term. For 300 I can have an army that can conquer me an extra city.
However, it's a lot better for the AI - the extra gold they get from no rebels allowing maximal tax rate is going to be significant by the time the research and casting is done. Not only do they get higher gold fro taxes than a human, but this also doubles the population growth advantage they receive. Of course, once the new research system is done, that time will be after researching the essential commons, not before. (and after Astral Gate, too - dropping settlers or units on myrror is even more important than maximal taxes - and can make the AI win even if they got otherwise obliterated on Arcanus by the human)

I most likely won't change the Guardian Spirit, at least not significantly. I rather have 1 out of 20 AI wizards start weak than to give an out-of-role spell to the realm. A small reduction in cost is fine - it's way overpriced for 80 and I also never use any - but I don't think I want the AI to spam them, or have them strong enough for that to be a working strategy.

So I guess the starting spell list will be
Healing
Heavenly Light
Just Cause if not monolife ONLY (risking to miss such a good economy spell for a dual wizard is a bad idea, and they won't get all 10 commons)
Heroism
Endurance
Holy Armor - only monolife will be able to reach this far by skipping Just Cause.

The order doesn't matter for Life wizards but I'm not sure what order to have for multicolor - Healing, Heavenly Light, Just Cause and Heroism are all critical, game changing spells. Probably best to keep the existing order.

[Nelphine]

Healing is not critical or game changing. Go play a game where you have 3/4 of your battles as strategic combat battles. Now tell me how useful healing is. AI spend the majority of their time NOT fighting humans. Give them spells that help with that.

Now, lets worry about that 1/4 of battles where you might be allowed to use healing. If you're at war with a human and you have 3 units heavily buffed (say, klackons halberdiers with endurance/holy armor/heroism); or you have a city with 9 klackon halberdiers. Which one will the human attack? If the human can beat the buffed halberdiers, then healing will do literally NOTHING to stop the human from beating the 9 unbuffed halberdiers. Healing MIGHT help the buffed halberdiers - but most likely, if the human can do 54 damage to the buffed halberdiers in the first place, then an extra 5 (or 10 or 15 or 20) damage from healing is not going to make a difference. And as we've agreed elsewhere, the HUMAN chooses what to fight most of the time. Which means the human will pick something it can beat, even if the AI has healing.

Buffing makes it so the units can't be killed at all. Healing prolongs the inevitable. The only reason the human can make such good use of healing is because they can make one unit monstrous, but not quite invincible - such as hero tactics - and that hero takes 1-5 damage over a whole fight, and you want to heal him to keep him going on to the next battle. Or, you're taking your hero and fighting things like great wyrms; you can survive one melee attack, and you can fly (or whatever) so the great wyrms can't hit you back. You use healing to top up so you can kill the great wyrm with lots of dead time in between. The AI cannot possibly reproduce this.

Healing is NOT critical or game changing for the AI. Giving them healing just means they'll waste mana on it in combat with the human, which means they don't have that mana to use for actually important spells like buffs.

[Seravy]

Healing is not critical or game changing. Go play a game where you have 3/4 of your battles as strategic combat battles. Now tell me how useful healing is. AI spend the majority of their time NOT fighting humans. Give them spells that help with that.

If every AI follows this logic and ignores spells that work on the human player, then they'll lose against the human player badly because HE will have those spells while the enemy will not.
Besides, we are talking about at least partially Life AI here. Which increases the chance of a personality that avoids fights with other AI wizards and allies with them instead.
I don't think having Healing or not significantly affects the AI's chance of developing into "the" strong wizard that fights it out with the human player at the end.
On the other end, ask yourself the question. Which is scarier, a wizard that has Sky Drakes with Iron Skin on them...or the same plus Healing? A wizard that has a strong hero you can barely damage but can kill through attrition...or the same plus Healing so the hero NEVER goes down? Healing is at the very least more cost-efficient in hit points/mana than any direct damage spell in the game, and this efficiency grows along with the strength of the target. And there is one more thing - a lot of good heroes can cast it to heal themselves. Chosen One included. Who is also a battlemage, so he/she'll just heal back any damage you did by attacking it. I lost several cities to that once.
I don't use healing for purposes you listed. I use healing to TANK the damage from the enemy direct damage spells to my heroes or other expensive units. Each turn the enemy casts Reaper Slash on my hero? Good, I cancel that out by Healing for one third of the MP and we'll do it every turn until the AI is out of casting power. (I might need to cast it twice if the hero doesn't have high armor...but low armor heroes are casters so they'll do it, no problem.)
The AI can do this. Try nuking their good hero (or great drake) with Fire Bolts when they know Healing. It'll fail. (if it doesn't, report it as a bug) Without Healing you can kill even the strongest AI unit by repeatedly casting direct damage - they have no way to cancel the damage and you only lose...what, some spearmen and 60MP spent on Fire Bolts? For killing a Great Drake? That's a bargain. But if they have healing they are immune to this tactic.
Healing doesn't heal 5 HP per battle. It does so every turn when it is cast and if there is a very strong unit in battle, that is often the best thing to cast every turn. To use your example, for each 3 healing I cast, my hero gets to kill an additional enemy very rare creature.
Healing effectively means immunity to low end direct damage, and in the right circumstances it works even on stuff like Doom Bolt (by right circumstances I mean being able to cast it twice a turn mainly).

Buffing makes it so the units can't be killed at all.

I wish. Swordsmen aren't that good. And that's all the AI has until way after the point where this choice matters on monolife - they can research it anyway by the time they have halberdiers, so those are not relevant to picking healing. What's relevant is multicolor wizards (who might not have it for research otherwise) and whatever the AI might target with Healing prior to turn ~30-40 (their 3-7 buffed to max irreplaceable swordsmen).

With that all said, the question isn't really about including or not including Healing anyway.
The last 4 spells in the realm (Star Fires, Guardian Spirit, Holy Weapon and Bless) are not important for multicolor wizards, and not important to have as a starting spell on Life wizards (okay, Holy Weapon and Bless are decent but their RP cost is lower so researching them is far better). So there is no reason to leave out healing - there is nothing better to include in the top 5. (unless you want to argue Holy Armor is more important for 2 realm wizards  than healing, which I doubt. Those wizards won't rely on normal unit buffing tactics.)

The question is, the order.

Healing can be anywhere from 1 to 5 on the list, right now it's the third. But it's really hard to tell if it's better or worse to have for a multicolor AI than Heavenly Light (4th) or Endurance (5th). I'm leaning towards better because it's a combat spell - it doesn't have to compete for overland skill with other AI spells, yet opens up a new ability for the AI in combat - one that's definitely situational but potentially gamewinning and costs absolutely nothing to have around - if the situation does not come up, the AI doesn't need to cast Healing - they won't do so unless they judge that's the best move they have.
Then we have Just Cause (2nd for multicolor) and Heroism (1st) which might or might not be better, aqain. I'd say they are, as one is a global major economy boost, and the other, well, the other means the AI's capital has Soul Linker +2 instead of +0 in many cases. Which is huge. So I guess Healing is fine being the third pick.

Edit : there is one possibility I see for changing the position of healing - Swapping it with Heavenly Light at 4th place. It might be even better than healing. The Ai ranks Healing very high in trades, so it had a good chance to obtain it - often from the human player who foolishly trades it away because they can get some decent uncommon for it (I almost lost a game to that trade once!).

[Nelphine]

Endurance and heroism should be the best multi color life spells. For mono life, I'd give the AI (and use myself) holy weapon and bless before healing. The situation you describe, being immune to low end damage, is the situation I find unusual in my games. Obviously you and I have very different experiences. This is another area I'd like to get other opinions about.

The question isn't about an iron sky drake vs an iron skin sky Drake with healing. Its an iron skin sky Drake with healing vs iron skin sky Drake with endurance. Endurance is flat out better in all situations involving a very rare.

For virtually any situation, endurance is flat out better than healing. Endurance is good for every single realm - and it gets better the more times you cast it. Endurance should be the number two choice for life spells for the AI.

Holy armor should be in the top 5 because mono life should start with it. Every single unit they own should have it, so they need to start casting it as early as possible. And it should be number 5 because multi realms don't need to start with it.

That leaves healing at number 6. Unless maybe just cause should go to 6? But they need it for hero summoning.

[Seravy]

For mono life, I'd give the AI (and use myself) holy weapon and bless before healing.

Since all spells are guaranteed for monolife, picking a 50 RP spell over a 100 RP one is a waste, considering we already agreed on making the AI do research from turn 10.

(I wouldn't pick Holy Weapon ever, but I would pick Bless. Not if I'm planning early research though.)

Its an iron skin sky Drake with healing vs iron skin sky Drake with endurance. Endurance is flat out better in all situations involving a very rare.

How so? With Endurance, the Sky Drake will take 1 less damage per attack (10% of 10 shields) - that's nice, but unless it's attacked over 5 times a turn (and survives all 5 despite taking damage from each of them), that's less damage absorption than casting healing on it.
To Def effectiveness is capped at 15 shields - so Endurance will never prevent more than 1.5 points of damage per attack which is rarely better than 5/turn. And this is the worst case scenario for Healing - if there is a magician or caster hero in the battle, healing can be used more than once to prevent 10, 15 or as many damage as needed.
Furthermore, Endurance costs 70 overland casting skill - so using it means 7% fewer Sky Drakes overall, while Healing is "free" - it doesn't use overland skill. Admittedly, 7% is insignificant, but there are two problems - the target might be a cheaper creature which makes 70 far too expensive, it's 25% of the price of a Chimera!
And the big problem - endurance won't be cast anywhere nearly enough times. By the time the wizard has rare summons it'll be having about 10 buffs. Assuming each has equal priority (they don't, the rare ones will be at 3-4 times more than Endurance) that's a 10% chance to cast Endurance. We are talking about very rare creatures, so summoning priority is at least double than buffing priority - for each TWO sky drakes, there is a 10% chance to cast an Endurance. So the real problem is, one out of 20 Sky Drakes will actually have the buff - the AI isn't human to cast buffs "on demand" when there is a new creature. Most buffs don't care about it because they can be used in combat if it's needed...but Endurance cannot.
The only advantage Endurance has is +1 movement, which I admit can be relevant, but there aren't many things in the game that can outrun a very rare creature - and on one out of 20, it's meaningless.

Holy armor should be in the top 5 because mono life should start with it.

Mono life will skip #2 Just Cause, so they get it at anyway if it's in the 6th position in the queue. Other realms won't, which is what you want - if it was at 5th, multi-realm wizards could still get it (if they pick exactly 6 life books).

Mono Life will research Just Cause - they have plenty of time to do so from turn 10 to turn 30 and don't need it earlier than that. (dropping the research cost might be a good idea though - 300 is a bit expensive)

I've only done the new research procedure up to uncommons, but Life will go in the order of (omitting starting spells) Bless-Just Cause-Astral Gate-Stream of Life-Holy Weapon-Star Fires-Guardian Spirit-Prayer-Divine Order-True Sight-etc

(why is holy weapon that good? It's just +25% damage, it won't help the unit survive at all...oh, I guess weapon immunity, I play Alchemy so often I forgot about that. I guess it might be good to move it between JC and AG then.)

[GermanJoey]

Endurance also gives +1 movement overland... and doesn't use any skill in battle. That's a major thing you're forgetting in your analysis. Not needing to cast Healing every turn allows you to cast other things - Prayer, Invulnerability, etc. Things that, when stacked with Endurance, allow you to absorb much more than just 5 damage per turn. Or you could... heal a different unit!

FWIW, that 1 damage per attack that Endurance will block actually does matter quite a bit for a Sky Drake, because the easiest way to take that unit out is to mass cheap ranged normal units and plink it to death. The +1 movement in battle is also useful as that lets you move twice plus attack twice - very useful against ranged units that are trying to kite you.

I don't think Healing is a very good spell for the AI either. For a human, it's fairly so-so except on Heroes and some top-end units, but the AI doesn't understand that like a human can. I agree with Nelph, it'd be usually much better off spamming buffs when restricted to Life.