Current 1.6 insights....

[Arcvasti]

Eh, if enemies are too difficult if they have good AI and super boosted stats/talents, then you can just nerf their talent scaling to compensate. I do agree that having enemies consistently run away will get old very quickly, especially enemies with good escapes.

[darkgod]

NPCs being smarter can never be a bad thing; I mean we don't want to play against target dummies do we?
If however being smarter exposes that they were too annoying/powerful/... but that it was unnoticed because they were stupid, the obvious solution is not to make them dumb but to make then less annoying/powerful/... That's more healthy in the long term

[Aesthete]

Yeah, but balancing that other stuff is hard. I guess I'll just content myself with being the devil's advocate for artificial stupidity.

[Micbran]

Eh, if enemies are too difficult if they have good AI and super boosted stats/talents, then you can just nerf their talent scaling to compensate. I do agree that having enemies consistently run away will get old very quickly, especially enemies with good escapes.

Which has already been done iirc, the way rares/bosses level has been completely redone. I’ll ask shibari to throw me the MR and I’ll add it to the OP

[Micbran]

I take that back, I'm not adding it to the OP because its apparently extremely tentative in the way it works. Anyways, here it is:
NPC Autoleveling Changes:
NPC's now respect actual game rules when auto leveling. This causes most rare+ and bosses to have more spread out skills.
As a great man named Shibari once said, the reason randbosses/rares are so powerful is because of the concentration of passive talents/sustains or particular hyper scaling talents. As a result of these changes, both of things would take a big hit as the AI would be less likely to have a bunch of points in one talent. This change as it is now, is actually a huge nerf to most randbosses, but like I said its extremely tentative and is being changed/messed around with.

[minmay]

NPCs being smarter can never be a bad thing; I mean we don't want to play against target dummies do we?
If however being smarter exposes that they were too annoying/powerful/... but that it was unnoticed because they were stupid, the obvious solution is not to make them dumb but to make then less annoying/powerful/... That's more healthy in the long term

In a single-player, asymmetrical game, the point of enemy AI is to create interesting/fun tactical situations, not to survive or kill the player as efficiently as possible. AI optimized for the latter is usually really annoying and can actually be less interesting to fight.
If you're writing an AI for chess or Go or Super Smash Bros or another symmetrical game where the AI is filling the role of another human player, then yeah, generally you want to make it as smart as possible. For PvE games, this is not the case. If the imps in Doom aimed their fireballs to lead your movements, it would be annoying. If centaurs in DCSS kited you it would be god-awful (and if you don't believe me you can try Crawl 4.1).

Even the current AI produces countless situations that are awful gameplay due to the "smart" things it does. Enemies that hang out of range of your attacks are just annoying. Enemies that dodge projectiles mean that you want to change the default target of every projectile so that it's aimed behind the enemy instead of directly at them - this behaviour does nothing but slow down the game. Enemies avoiding using talents you're resistant to means that e.g. having high lightning or fire resistance against multihued drakes is actually bad if you have stun immunity but not disarm/blind immunity, because it makes them more likely to use their breaths that cause disarm/blind. Having higher resistances should never be bad!
These AI "improvements" only make these problems worse, and I notice the AI has been going in this direction for quite a few years. I think you will ultimately find that this approach reduces the depth of the game rather than increasing it, and is unhealthy in the long term.

[Zeyphor]

+.3 to all talent masteries from a prodigy?
that seems a bit OP, at least for non-weapon classes; perhaps it should be toned down to .2

there is more that that could do than simply slightly increasing damage dealt:
it could increase the radius of catalepsy and cyst burst and make bone shield be better
it could increase the radius of circles and let corona hit more targets, and let hymn nocturnalist fire more and cheaper negative energy beams
it could let doomed get a 4th shadow without spending a category point or equipping umbraphage, and increase the range of hateful whisper's spreading, and cause creeping darkness to make more darkness
it could let doomed AND cursed increase the mastery level of cursed aura, allowing them to get more and stronger and less detrimental curses
it could increase the range AND radius of alchemist bombs
it'd make necromancer minions be better and make forgery of haze last longer, and increase the radius of chill of the tomb, and let liches get lvl 7 lichform without spending a CP to get it
it could make distortion bolt have a bigger radius, distortion wave be bigger and stun for longer, ravage last longer, and malestrom be bigger AND longer, and make sleeping effects last longer and such
it could make a summoner's summons last longer and reduce their cooldowns, and make frantic summoning last longer
it could enable seal fate to proc more times per turn
it could make the oozemancer's mucus AND mucus oozes last longer, and make moss be bigger and last longer, and cause mind parasite to put more talents on cd per activation
it could increase the talent level of chemistry and physics talents, enabling you to save on generic points when trying to get certain tinkers, such as crystal edge or unstoppable force salve
and much more!

theres alot of stuff that giving .3 to all known masteries can do; it really should be toned down to .2

and on top of all of that, it'd also increase the talent level of weapon mastery talents and light/heavy armor training

[Coldbringer]

I bet the huge nerfs to movement and heroism infusions will upset a lot of players. As in, "Well now what do we do?" The constant physical status applications opponents use is not going away, and now the best tool to prevent them is near useless for such. Vs groups of rare+ mobs on insane+ it feels like you're getting stunned/pinned/dazed every other turn by so many different abilities they use.

[minmay]

theres alot of stuff that giving .3 to all known masteries can do; it really should be toned down to .2

I can't speak for Shibari, but I personally tested Adept on all classes with several candidate mastery levels (0.3, 0.5, 0.7, and 1.0 which was Shibari's original suggestion in the in-game chat).

Briefly, the results are that the 0.3 and 0.5 versions are (in my opinion) worse than at least 2 other prodigies on every vanilla class except Possessor. Possessor is a special case because of Improved Form and Full Control.

As you alluded to, there are clearly superior prodigies for weapon classes. For caster classes, the thing is, direct damage talents almost always base their damage on the square root of talent level, so going from TL 6.5 to TL 8 only increases damage by about 11% - disregarding the constant part, and the constant is usually positive so the returns are actually worse than that. Obviously, Adept is better than just a 11% damage increase since you also have defense and utility talents and maybe talents that give you extra spellpower/damage/penetration/crit, but the point is, it's not as great as it might look at first glance.
When you take the diminishing returns on talent level into account, and keep in mind that the alternatives include AAD and Cauterize and Draconic Will, Adept doesn't seem (to me) like a #1 or #2 choice.

That said, I would love to see prodigies in general nerfed.

[Shibari]

The AI issue that immediately came up after those changes was randbosses edge of screen kiting melee endlessly as early as T1. This creates a situation where you either move back and LOS pull (correct play) or recklessly gap close to finish them off. Early game all forms of mobility are at a premium and it made melee starts infuriating.

This is temporarily solved by disabling AI escape behavior entirely on randbosses until mid game. I doubt there will be a solution that doesn't involve something similar, and am somewhat agnostic on how smart the AI should be about escaping mid game+. At the very least its clear that it should be worse at it than it currently is, something like decreasing Escape weighting with time might work.

i have to admit that a dozen of times being overwhelmed by a single rare+ on promising character in old forest has been the BEST experience i got in ToME, and thus this pointer is very cheesy for me. It makes you think ahead, like digging another way from the level change so that this unkillable monster can't just block the only one. Also, this game is not a "kill every single monster" challenge, so retreat (possibly till a few more level-ups) is ok and adds good challenge

also IMO the best insane thing ever is EOR approach, where no random rare+ spawn till you reach lvl 20; maybe in AoA it could be set to lvl 10, while saving all those precious overpowered monsters! kiting and dying is fun

General difficult spikes aren't the problem here. The majority of the arbitrary stat "capping" and similar is aimed at tier 1 situations that are incredibly irritating (even when not challenging) to certain builds, usually flat damage reduction or armor vs. multihit melee. On 1.5 you can spend an amazing amount of time either running from or taking 20 years to kill a tier 1 boss that can't hurt you but also takes 0 damage from you. This is not caused by innate scaling issues with the talents or randbosses, it's more about how few options players have then, *especially* with the T3 and drowning crutches removed. Mid/late game options open up a lot more and scenarios like this are not inherently bad.

NPC bases are just too varied to not adjust their stats when they get randbossed as a hard rule. Especially common NPC bases. Even with the damage reduction (the usual nerf) Midge Swarms and such are vastly stronger than other randbosss in the tier 2s, this just reduces the degree. Other examples include 500% attack speed horrors that will put out thousands of DPS if given a weapon class. I don't think were losing much by having this scenario only produce 200% IAS gods of death instead.

The EoR method does work and completely avoids the wide variety of ways randbosses can break the early game. That was actually me giving up on getting anything else to work well for making NM+ EoR playable. In general I think its better to do the extra work necessary to beat early game randbosses into working as its a lot less boring and removing them would have other side effects like reducing the loot higher difficulty players have entering tier 2.

I bet the huge nerfs to movement and heroism infusions will upset a lot of players. As in, "Well now what do we do?" The constant physical status applications opponents use is not going away, and now the best tool to prevent them is near useless for such. Vs groups of rare+ mobs on insane+ it feels like you're getting stunned/pinned/dazed every other turn by so many different abilities they use.

Stun only halves talent cooldown now and is directly hit by the randboss adjustments mentioned above. Specifically, any stun talent has its level capped through the early game which almost always translates to a duration reduction. This lets you deal with it early without an immunity or guaranteed cleanse, but mid game+ that won't be a sane approach.

The Movement crutch was horrible, unintuitive design left in because it addressed a specific problem. It got some slight numeric buffs along with this change and is still very good at what it is supposed to do.

Heroism is awkward. I don't like that a huge part of most builds HP is coming from a temporary buff in an inscription slot. I also don't want to remove a huge amount of HP at once, so Heroism loses its free offense but retains its main anti-burst role. Would be kinda nice to redesign this into something interesting someday but oh well.

[minmay]

Heroism is awkward. I don't like that a huge part of most builds HP is coming from a temporary buff in an inscription slot. I also don't want to remove a huge amount of HP at once

passive shield runes now plz, youll get free sound design in exchange for it

[Shibari]

Heroism is awkward. I don't like that a huge part of most builds HP is coming from a temporary buff in an inscription slot. I also don't want to remove a huge amount of HP at once

passive shield runes now plz, youll get free sound design in exchange for it

I can only offer another 0.001 passive mana regen on Manasurge Rune

[Zeyphor]

The AI issue that immediately came up after those changes was randbosses edge of screen kiting melee endlessly as early as T1. This creates a situation where you either move back and LOS pull (correct play) or recklessly gap close to finish them off. Early game all forms of mobility are at a premium and it made melee starts infuriating.

This is temporarily solved by disabling AI escape behavior entirely on randbosses until mid game. I doubt there will be a solution that doesn't involve something similar, and am somewhat agnostic on how smart the AI should be about escaping mid game+. At the very least its clear that it should be worse at it than it currently is, something like decreasing Escape weighting with time might work.

why disable it until mid-game if you can just disable it until T2 dungeons?
players should be able to beat those guys by the time they're in insane old forest and are at level 14+, right?
and if it is an issue for a few levels after 14, then you can do SWL where... you could disable the behaviour because not only would it cause enemies to get crushed to death by sand, but some enemies would get really hard to kill with their burrowing
or perhaps I don't understand the behaviour quite right

[Zizzo]

there is more that that could do than simply slightly increasing damage dealt:
[…]
it'd make necromancer minions be better and make forgery of haze last longer, and increase the radius of chill of the tomb, and let liches get lvl 7 lichform without spending a CP to get it

Ah, crap, that's going to further complicate the Celestial/Star-Fury-from-Lichform handling in the talent point planner, isn't it? *sigh* I guess I'm getting the alpha after all. I assume it'll be sufficient to check out a copy of git trunk and build from source?

[Davion Fuxa]

also IMO the best insane thing ever is EOR approach, where no random rare+ spawn till you reach lvl 20; maybe in AoA it could be set to lvl 10, while saving all those precious overpowered monsters! kiting and dying is fun

The EoR method does work and completely avoids the wide variety of ways randbosses can break the early game. That was actually me giving up on getting anything else to work well for making NM+ EoR playable. In general I think its better to do the extra work necessary to beat early game randbosses into working as its a lot less boring and removing them would have other side effects like reducing the loot higher difficulty players have entering tier 2.

Personally I think the EoR method was actually one of the things that made EoR better then AoA (of the few things that were better). I'd probably suggest making it that you can reach level 15 in the AoA campaign. Not only will this help with the irritating nature of randoms throwing a wild card that does a bit more then spice up the flavor of the game, but for harder starts a player won't have as many factors to worry about.

Really if it is a problem with loot, then there is a rather easy fix that could be pursued and likely one that should have the extra work directed at more then trying to make randoms appear early in the game:

The shops selling items should be a bit more then thematic eye candy, they should serve a function, make buying the equipment you need to play with a thing!