Early game hero balancing

[Nelphine]

I'm a little confused. I play life only, without heroes, all the time. We have a slow ultra early game yes, no summons, but, if a human can do it, surely that's a matter of teaching the AI better? For instance, make them pick races that work well with buffs (similar to inquisitor being limited race choices.)

Or don't have the hero strategy as their primary strategy. If they get holy armor and endurance on a swordsmen of the right race super early, yes that's very strong. Since they're mono life they'll have all the hero options too even if they have to research them, so they'll be able to do that later anyway.

Alternatively, you could set your turn requirement based on fame (say 1 turn earlier per fame) - so a life AI that can cast just cause, could get a hero on turn 25.

[Seravy]

, surely that's a matter of teaching the AI better?

No, it's not. The AI uses a random chance based overland casting system, while buffing requires 100% focus on the desired goal over a longer period of time. This long term planning is something the AI will never be able to do.
Also, the human player is way too smart - buffing strategies won't work against them because they are hard to plan but easy to counter. (dispel magic for one, armor ignoring attacks and spells for another, targeting resistance instead of defense as third.)
Finally, the AI is not allowed to have strategies in diplomacy and overland unit use - the human player can say "I'm not attacking that Sorcery wizard because my buffed units are weak to sorcery magic", while the AI has no such choices - their wars are decided by game rules or human interaction, and their attacks are decided by raw numbers - they'll still send the buffed units even if the units are weak against that type of enemy. This is true for all units of course, but buffed units are the least cost-effective and all share the same weakness so unlike the partial success the AI can get in usual cases, they end up completely losing all battles this way.
Simply put, Life is a realm only the human player can play properly.

If they get holy armor and endurance on a swordsmen of the right race super early, yes that's very strong.

Sure, until it meets a Phantom Warrior...but it will probably stay in the capital anyway, doing nothing. The AI is busy building settlers and other required stuff, they won't have "Extra" swordsmen to send out of the capital, except on Extreme and Impossible where they get some for free. (even there they have those 3 or 5 units and that's all)

Going military first for them is an option, but that's way too risky - if they don't actually have to fight the human player, they wasted a lot of effort on nothing while everyone else was expanding.

[Nelphine]

I still think you underestimate buffing strategies in the hands of an AI. But, I'm OK with it staying as a human only tactic and life simply being a weak AI. I'd like to see you try it once, just to see how the AI does - I don't think it will be as bad as you expect. (For instance, I don't bother caring about sorcery wizards - aside from overland dispelling wave, they generally are one of the easier targets to deal with, because true sight is a thing. And I don't wait until I'm at war with a sorcery wizard to cast true sight - I always play on the assumption all 4 AI will declare war on me the next turn. And for the AI it only matters if the human is sorcery - against other AI, buffed units would dominate strategic combat and completely negate the sorcery magic and resistance based combat spells advantage.)

[Seravy]

because true sight is a thing

Certainly but we are talking about the very early game now - no uncommons.

Investing 10 turns of casting skill into that swordsmen with heroism, endurance, holy weapon and whatever else only to see it disappear to a single Psionic Blast of Phantom Warrior is definitely not the way to go and I don't think you'd attack a Sorcery wizard with that unit. The AI would.

But this shows very well what the problem is : The AI won't have true sight on their units. A few will have that buff, others won't. If they only lose 80% of their units before they are unprotected, they still lose the war.

I've had the "luck" to fight against buffed halfling swordsmen in this previous game. They are a pain to deal with. It takes forever to kill one if you are not Sorcery...but they are not a threat. Of course this is not universal, but horsebowmen can easily counter them by staying out of range. The problem is this - the human player will usually find something to counter the buffed unit. It might be crack's call, dispel magic, using a faster unit, simply not attacking the place with the buffed armies, using illusion damage, casting eldritch weapon, blood lust, etc.
Ultimately the human player is in control of the game - he decides which army fights what, and which unit in battle fights which unit 90% of the time. So buffed units work for the human and the human only, because the strategy relies on avoiding the wrong enemies.
(and ofc, buffs would have been a joke if it wasn't a HALFLING unit, and specifically a swordsmen with Large Shield against my bows. Most common summons are at least as good as a heroism swordsmen.)


So far, whenever I encountered a Life wizard in the early game, it went the same way : I was able to steal all their outposts, and they couldn't retaliate because they had no summons. Then I was able to deal with the units leaving their capital one at a time.
This doesn't work against other AI - they have stacks of 2-9 Nagas, Bears, Sprites, Ghouls etc running around - if I attack a city this early when I only have 2-3 units of my own, those will make me lose the game, or at least slow me down a lot, buying the wizard time to summon even more - but the Life wizard has no units outside their capital at all to attack me with on Hard.

[Nelphine]

Fair. But early heroes don't change that. And its true for the human mono life player (generally all human players that aren't going extreme rush strategies). The mono life AI having a hero early.. Means the hero guards their capital. It doesn't prevent you from taking all the outposts. So the hero change on turn limit (which is where this conversation started) wont change anything in terms of the problem you just described.

And in general, in terms of the human controlling the battle, down to the unit - of course they do. Guess what, if the AI attacks me with war bears, or nagas, or phantom warriors.. I control the battle and hard counter those too. Saying that the AI shouldn't use one strategy because the human will counter it means the AI should never do anything. Which us silly. The AI should still use strong strategies and force the human to counter it.

If one AI is sorcery and one AI is life, and they both declare war on the human, currently, as long as the human can beat the sorcery, he's OK - because life AI don't go for strong buffs. Imagine that same situation (human goes for armies with high resistance, and illusion immunity because of sorcery magic) and then runs into a whole whack of halberdiers with life commons. Of course he CAN counter them. But because he was trying to counter sorcery armies, he's not in a good position to do so easily.

[Seravy]

The mono life AI having a hero early.. Means the hero guards their capital. It doesn't prevent you from taking all the outposts. So the hero change on turn limit (which is where this conversation started) wont change anything in terms of the problem you just described.

I guess that's true for Hard unless the AI gets at least 2 heroes, as the minimal required units for the capital during these turns is 4. (or 3 on Normal)
So the problem is not related to heroes, which is good. (the heroes used to get pushed outside quite a lot, I guess that's not a thing anymore as the AI has different garrison priorities in the early game)

because life AI don't go for strong buffs.

It does. Why do you think it does not? The Life wizard has no summons, so all their overland casting power will be used on buffs - there is nothing else to cast, afterall.
I never said the AI is not doing it. I said it's not working out for them. Mainly because you don't have halberdiers in the first 30 turns which is what we were talking about.
(life AI is probably doing fine if it isn't found until it has a few stacks of halberdiers. The problem is the time before then. Those buffs on swordsmen are not enough, and the AI has no swordsmen anyway - they go settlers and fighter's guild first.)

If you feel the AI is not effective enough, we can start a topic about individual buff priorities - maybe the AI is just not using the right spells in the right proportions?

[Nelphine]

Yup if this is just early then I think we're on same page.

For later I find life summons too many angels instead of lionhearting troops. Angels aren't THAT good. I'd love a rule where angels can't be summoned if summoning circle already has an angel.

[Seravy]

For later I find life summons too many angels instead of lionhearting troops. Angels aren't THAT good. I'd love a rule where angels can't be summoned if summoning circle already has an angel.

So...the priority of casting a summoning spell VS the priority of casting a buff is the problem? (reducing angels inside the category would just make them summon a unicorn instead)

Is it a problem for ALL cases or just some?
What I want to say is, the priority to cast a buff, or a summoning spell has many global factors (and none of them specific to Life wizards).

In general the AI will
-Cast a lot more buffs if Perfectionist
-But cast more of other spell types for other personalities - each has it's own preference.
-Cast more summons if their military is behind the human player
-Cast more summons if a "Strong" summon is known - Angels are not one of these, but Arch Angels are.
-Each copy of Divine Order in play increases buff and reduces summon priority
-etc

If every wizard casts too few buffs then the base priority of buffing might be low. If only life wizards, then we might need a special case for "more buffs if only life books owned". If it's only specific life wizards, like Expansionists then  it's the personality.

Also, you specifically say "lionheart" but that's only one buff out of the 10 the AI knows. While it does have more priority than casting, say, holy armor, ultimately the AI will be casting everything, leaving the Lionheart as a choice for ~1/5 of the buffs cast while 4/5 are other buffs. So we might still want to go over these priorities and adjust them, although it probably won't help. Ultimately, buffing works for the human because they use the one-two spells that they need, while the AI has to cast all 20 it knows, even if not equally often.
(this is quite different from summoning where you don't need to use the weak ones when you have a stronger one. For buffs you need to use all of them because buffs are best if stacked, and they are functionally different.)

PS : Life wizards generally know a lot more and more expensive city buffs than other wizards. So those might be eating into casting capacity, not just summons. And that's a good thing as those city buffs are worth a lot for the AI if they are used.

[Seravy]

Meanwhile I'm done adjusting hero costs and fame requirements.

Fame :
Thief went up to require 5 as it's way better than 0 fame heroes - with Agility, Lucky and Charmed, it's as invincible as it gets, not even champions have a better set of defensive abilities.
Bard went down to require 0 - buffing the entire army is a powerful ability but it does require an army so it's less of a threat to early balance than I thought.

Cost :
100 : Dwarf, Barbarian and Orc Warrior - these are fairly weak heroes that have no defensive abilities despite being melee and have no abilities to produce resources. I don't think they are a threat to early balance after turn 30.
200 : Sage, Dervish, Beastmaster, Healer, Druid, Huntress, Warrior Mage, Bard, Rouge - these are better heroes that either produce resources or have some useful ability so they cost more.
300 : Thief - costs even more because it's that powerful.
400 : War Monk, Assassin, Ranger, Draconian, Golden One
500 : Wind Mage, Witch
600 : all fame 20 heroes
999 : all champions

We can easily change the numbers in the unit tables now - I managed to separate fame from gold cost.

[Seravy]

Posted the AI priorities for each buff here : http://www.realmsbeyond.net/forums/showt...p?tid=8724