Caster of Magic Release thread : latest version 6.06!

[Nelphine]

Ugh, why do transports work differently? Ah well doesn't matter (presumably something to do with air vs water). But yeah I don't care about getting the movement of both (and actually I wish you couldn't) so this should be fine. Unless. Say I have a windwalking hero and 8 warships (because for whatever reason I've decided warships are my best unit). Will the warships move with the windwalker after the transport movement is done?

[Seravy]

Ugh, why do transports work differently? Ah well doesn't matter (presumably something to do with air vs water). But yeah I don't care about getting the movement of both (and actually I wish you couldn't) so this should be fine. Unless. Say I have a windwalking hero and 8 warships (because for whatever reason I've decided warships are my best unit).  Will the warships move with the windwalker after the transport movement is done?

No, but you can't carry ships by other transports either so that's not different.

[Nelphine]

Fair enough.

[zitro1987]

With famous, I got 3 hero offers - Spyder, Shalla, and around 1407 the Paladin Hero!


Despite some earlier concerns of retort being underpowered, I consider this a strong effect.

[zitro1987]

And in regards to heroes, am I the only one who rarely researches/uses 'summon hero' ?

with 400 research cost, 'uncommon treasure', and 300 mana per hero (who may not be the kind you want), we'd be essentially spending 1000 power for 2 heroes.

I wonder if, just like 'enchant item', this should be a low research cost and 'common treasure' spell.

[Nelphine]

Jaer (the best hero in the game, bar none, including all champions and torin) is from summon hero. He's worth about 2500 mana or more. (If you are sorcery, and you don't fight dispelling wave, his value goes away.)

The MUCH reduced hero offers from current make the spell more relevant now. As per my life game with no hero offers before late 1415. It does feel high, but I think a more realistic description is not that summon hero is too expensive, its actually that summon champion is too cheap (it should be close to 1200 cost to cast).

[Catwalk]

I just got Summon Hero in my second sprite game (yes, I'm rushing through it as I was dared). I already got two nice items from lairs, looking forward to putting them on a decent hero.

[Nelphine]

For the list of current tasks, as a specific subsection to AI resources, can we also examine cost of globals?

Any global that costs 200 or less works fine; not because the cost is cheap, but the AI generally ignores them. Most of the globals that costore are starting to become effective enough that the AI knows to increase priority of disjunction and/or spell blast.

On extreme or impossible, by about, 1410, this means that an average AI (not talking about a crazy Uranus' blessing + aether binding sorcery wizard, who can achieve this result by about 1407) will have ~200+ casting skill, allowing them to cast a 400 (or 500) strength disjunction in 1 turn.

For the human player, for a 500 cost global, they have to spend ~3 turns casting it. That's a huge amount of time in the current state of the game where every turn matters. There will then be 2 or 3 AI (usually the others are too weak) who can cast disjunction as an instant; between them, you rarely get more than a few turns of your global lasting (assuming you are a specialist. Without specialist, I think the average would drop to 1 turn).

Adding to this is the sorcery crazy casting skill AI (who also get that dispel bonus, effectively negating specialist) - I've seen in 1412 sorcery ais who can disjunction multiple globals in one turn.

On top of that is spell blast - I understand why it needs to exist, but if the human is at war with an AI who can cast it, even though its a curse, and so its not crazy high priority* they will usually cast it at least once every 3 turns (and twice as often if they are sorcery crazy casting skill) - which is usually often enough to prevent the human from ever casting their best globals.


All of this combined simply means.. Humans don't get to use globals much, unless they have a massive lead on both research and casting skill, which just isn't likely on extreme or impossible - and even if it happens, in those games the human has probably already won, so the global doesn't actually matter.

Which I think sucks because I love globals. I think they're super awesomely cool, in all 5 realms, but only getting to actually use the same globals every game.. Is bleh. But I get to watch the AI use them a lot, because I can't do anything about it. -but- you say -if the AI get to use them a lot, doesn't that mean the other AI aren't disjunctioning or spell blasting them? Doesn't that mean you ate simply unlucky?- No. Unless specifically playing a diplomacy build, generally, the human player will always be in at least 1 war on these difficulties. Similarly, there are often multiple AI who are allied with each other. People at war like to smack your spells more often, and are less likely to go to war with another AI. The reverse for allies. This usually results in the human player taking the brunt of the disjunctions/spell blasts. Further, its often only the most powerful AI that actually gets to use globals - and by definition they're usually the one doing most of the disjunctioning.

Basically globals don't help unless you're already on top, and they just cement your hold - which is rarely the place for the human player

*I'm aware of my project to examine spell priority. While I think this topic could fall into the category of spell priority, I don't want to make that decision without a discussion and explanation of why I feel this is such a problem.

[jhsidi]

Unlimited mana and very high spell skill are the two reasons other than spell priority for ease of disjunction / spell blast, so it seems like you'd have to address one of those.

Personally I'm not a fan of unlimited mana on the part of enemy wizards -- by unlimited, I mean they've got 2k+ in the bank and an income higher than possible expenditure, which seems to always be the case in extreme+. The reason I don't like it is, specifically, that it cuts off the entire strategy of starving your opponent for mana. I've tried starving an opponent of mana by sending in weak units to provoke comba spells, but on those high levels it seems to make no difference, their income is just crazy.

Mana income for AI seems like a tough balancing problem to address, though, so it comes back to skill. Question is: does it make the "hardness" of the game more fun to face a wizard with high enough casting skill (and, of course, unlimited mana) to cast their best combat spells repeatedly? E.g. the wizard who can Flame Strike or Wave of Despair you 5 times in a row. What about casting a Great Drake in 2 turns? A disjunction in 1?

High skill also doesn't feel that good to me; I feel like the game wasn't designed or intended for the best / strongest spells to be spammed, whether it's me doing it or an enemy AI. The reason I love this mod is the spell / creature balancing and improvements to choice-making in AI, whereas these issues seem like side effects. But both of my opinions on mana supply / skill in AI also feel like just a personal opinion / taste, maybe you guys love the challenge of it aside from edge cases like disjunct/spell blast.

[Seravy]

Disjunction priority is hardcoded to a constant value for each spell and has no relation to casting cost.
Stronger spells simply have a higher priority to disjunction - if it's too high, feel free to suggest different numbers (they are on the first page of the AI thread)

(note that the reduction in summoning priority we recently did also means the AI has more chance to pick Disjunction)

Allies don't cast Disjuction against each other, so if multiple AI players are allies, their spell will last longer. If you are allies with the heavy sorcery AI, your spells are also safe - there is an exception though, if the AI starts casting Spell Binding (not Disjuction) against someone else's spell but yours is equal or better, it'll get stolen anyway.

Furthermore in the current system, the priority of casting disjunction is the sum of the priority for each enchantment that can be targeted (again allies don't count) so if there already are a lot of AI enchantment and you cast something that's "better" than those, it'll be dispelled. We might want to recode this part to only count the highest one. We might also want to change it so that war doubles (or nonwar halves) the priority. Maybe even define a separate priority value for each treaty status.

Mana supply cannot be drained through combat anyway - if the AI is too low (which is like 4-8 times their casting skill), they stop casting spells in combat and save the rest for overland/important battles. So this "hole" in the AI no longer exists. If you want them to run out of mana, the only way is to disable their income with Warp Node and Evil Presence on all their cities/nodes. Doable, but not easy and some sort of dispel resistance is highly recommended if you are behind in casting skill.

It's worth considering 2k is not high - as a human player I've had turns where I spent more than that in combat. For the AI who has more skill and cannot comprehend what "distance multiplier" means, this amount is even less - a high skill AI will aim at keeping several time as much in reserve. At 400 skill and a 3x penalty, 1200 mana only lasts a single battle. And it is fairly easy to prolong a battle long enough to make the AI use it all.

PS : Lower costs on globals mean the AI will be successful with their disjunction more often. It wouldn't really help.