Crawl needs more HP percentile damages : EDITED

[RedGem]

We need more diverse HP percentile damage to prevent many interesting monsters from being obsolete too fast.

Diverse HP percentile damage can make game more interesting.
It can be both based on:
Max HP
Current HP

In player's perception, it'll make easier to understand the threat level of a monster.
In dungeon designer's perception, it'll make easier to control the threat level of a monster.
Currently we have
[ orc -> orc warrior -> orc knight -> orc warlord ]
[ centaur -> centaur warrior ]
This is very hard to balance and takes endless effort.

With HP percentile damage, all the interesting monsters can make very complex situation altogether.

Here are some examples:
1. Green imp([3~7% damage based on Max HP], melee) vs Red imp
A. Open ground. An enemy starts to summon imps. You ignore them and find a way to kill the summoner.
B. Open ground. An enemy starts to summon green imps. Being surrounded by green imps is extremely lethal. You retreat and find a better place to fight.
It's very hard to use red imps to make the game more interesting. They're nothing but annoying in almost every situation. Can be ignored easily.
On the other hand, green imps can be an actual threat in almost every situation. When many of them are in the LOS, players must avoid themselves from being surrounded.
In case B, positioning matters a lot.

2. Orc warpriest([7-17 damage or 14-34% of the current HP if it is over 50)], smite) vs Orc priest(7-17 damage, smite)
A. You're XL19 MiBe. You encounter a group of orcs at V:3 including a orc priest and 2 orc warriors. They're not dangerous at all. It doesn't matter much which one you kill first.
B. You're XL19 MiBe. You encounter a group of orcs at V:3 including a orc warpriest and 2 orc warriors. You immediately get out of the LOS of the orc warpriest.
In this case, Killing orc warpriest is the highst priority. Positioning and LOS matters.

3. Green imps and an orc warpriest and other monsters together
When they come together, you'll have to think harder to survive. You can't simply ignore these two. And they fit well in almost any phase of gameplay.

I think crawl should be more about strategy, and less about resource management. It's more fun and more challenging.
Diverse HP percentile damage is a good way to make the game that way.

----- Original Post -----
In crawl, when you have good defense, most enemies become meaningless.
Weak enemies are like hostile butterflies that follow you.

In late game, at 100% HP, almost nothing challenges you if there's no torment.
Torment matters too much.
I mean, torment is great. It makes risky, challenging situations, and makes the game more interesting, but it shouldn't be the only thing that matters.

I think HP percentile damage should be more diverse.
Think about enemies with HP percentile damage melee attack. It will make situation much more complex and interesting. Positioning and strategy will matter much more.
Or it can be ranged/magic. It can make things more interesting.

To make the least change to the damage calculation system,
Giving something like 1d35% damage instead of 1d35 seems worth consideration.

What do you think?

[dck]

I think is a good tool to convey a sense of danger and weakness to the player doing extended in a normal progression and that while utterly harmless it is also a good way of getting other damage sources available in the torment-heavy areas to kill players. The mix between fake and genuine danger there is very well achieved, I believe.
Off the top of my head flaying is the only other percentual damage source in the game and while all in all I like flaying because it is spectacular to witness, it doesn't do a good job at telling the player about what is actually going on and it is inflicted by one monster that is not common and often found in the company of much less varied types of threats.

[tedric]

In crawl, when you have good defense, most enemies become meaningless.
Weak enemies are like hostile butterflies that follow you.

In late game, at 100% HP, almost nothing challenges you if there's no torment.

Maybe I have a poor definition of "good defense," but it has not been my experience that "In late game, at 100% HP, almost nothing challenges you if there's no torment." Even characters who are ready to tackle Zot will often still find draconian packs, orb guardians, etc. rather challenging. (I'm assuming A. Liches and OOFs are the "almost nothing" that remains challenging for most characters in your view.)

Of course, there is a difference between "ready to tackle Zot" and "cleared every enemy from every floor in Dungeon, Lair, Orc, Elf, Swamp/Shoals, Spider/Snake, Slime, Vaults, Crypt, Depths and maybe even dipped into Abyss/Hell/Pan/Tomb already" -- if your games tend toward the latter, you'll have such a surplus of XP and loot that it's no wonder the late-game branches feel relatively easy. But the game is designed to be winnable, at a reasonable level of challenge, without completing every scrap of content and even while skipping huge chunks of content. In fact players are encouraged to skip tons of content since you can win with just 3 of 15 runes, which means visiting at minimum 7 (of, like, 20?) branches and maybe 45 (of 100+ and functionally infinite, because Pan) floors.

So the idea that more percentage-based damage is needed to make the game properly challenging is, I think, based on false premises.

I also agree with dck that percentage-based damage effects suffer from issues of transparency. Adding more of these is not the only way to make the game more challenging, if that were necessary, and I'm not sure it would be the best way either.

To make the least change to the damage calculation system,
Giving something like 1d35% damage instead of 1d35 seems worth consideration.

If you're suggesting that ALL damage become percentage-based, I think this is a terrible idea. Not even sure it would be that easy to change, since I think the current math on some attacks/spells can deal more than 100 damage and as a percentage that means one-hit-kills and then some. More importantly, it would turn Torment, which you describe as a "great" mechanic that produces "risky, challenging situations," into a boring mechanic that does exactly what all other attacks in the game do...just hits slightly harder. Percentage-based damage is currently interesting because it's different from the fixed damage of all other attacks, and Torment/Flaying adjusts the threat value of those attacks in real time.

[RedGem]

tedric, I think you misread the intention of the suggestion due to my poor expressions.
Crawl is very well designed game, and enemies such as OOFs or packs of draconian are valid, interesting challenge when you meet them in normal progress.
However, what I wanted to say is that many enemies become obsolete once you overcome them.
For instance, when you have grown strong enough, even packs of draconian don't scare you, they're not much different from packs of orc.

Also, I'm not suggesting that all damage should be percent based. That's ridiculous.
However, in crawl, there are many monsters with interesting gimmick,
which make situations more interesting, and therefore shouldn't be obsolete so fast.
Like jumping spider, boulder beetle, etc.
I think percent based damage can help them from being obsolete too fast.
For example, If boulder beetle's rolling attack always do 30% damage to your current HP, you can't ignore it even if you're at XL27.
That means that dungeon designers can deploy boulder beetles in any level of the dungeon, without considering whether it fits the level's power level.
It will be much easier to make situations complex.

[crate]

making torment more common in the parts of the game where it doesn't exist would be a good move imo! (i've suggested this in other topics here)

I don't really see a reason to have more percentage-based damage though, 50% does the job nicely.

[RedGem]

I think the current 50% damage torment is too lethal for early game.
To have more torments in earlier level of the game, it needs to be toned down a bit.
For example, orc priests deal [7-17] irresistible damage with smite.
What if the smite damage is [max(7-17, 7-17% current HP)] instead?
It will make them much more dangerous, but not too much.

[Leafsnail]

I think keeping it to 50% would probably be better, that's a lot easier to understand than having lots of arbitrary percentages. Torment early on would be a threat, but I don't think it would be unacceptably bad. Afterall, early in the game there's stuff that can hit for more than 50% of your HP, and also kill you.

e: The early game tormentor should probably only be able to torment and melee though, the standard smite/pain and torment combo would be too much.

For example, orc priests deal [7-17] irresistible damage with smite.
What if the smite damage is [max(7-17, 7-17% current HP)] instead?
It will make them much more dangerous, but not too much.

So they'd be very similar until you have 100HP, and then they'd become maybe a tiny bit more dangerous? Only it wouldn't matter too much because Orc Priests aren't much of a threat by that stage. The main problem I see with that proposal is that's it's really confusing.

[KoboldLord]

I've long thought that percentage damage would be a good effect for poison to have. Currently, a venomous monster is dangerous mostly because the most notable examples happen to hit like a truck, and even with relatively recent modifications the poison damage can mostly be ignored until combat is over. If poison dealt, say, 25% of your current hp every time it ticked, however, I dare say it would command your attention in battle even when the venomous monster causing it didn't have such a large physical component to its attack. You might even have situations where it is worth the bother to quaff curing to take care of poison during battle. Naturally, it would need to cause this damage a little less frequently than in the status quo version. Poison resistance could reduce the chance of getting poisoned, reduce the chance that poison damage is applied during any particular turn, or cut the damage from 25% to 12.5%, or some combination of these. Different monsters could also have venoms of different virulence, which could play with any or all of these three stats.

[RedGem]

I think keeping it to 50% would probably be better, that's a lot easier to understand than having lots of arbitrary percentages. Torment early on would be a threat, but I don't think it would be unacceptably bad. Afterall, early in the game there's stuff that can hit for more than 50% of your HP, and also kill you.

e: The early game tormentor should probably only be able to torment and melee though, the standard smite/pain and torment combo would be too much.

For example, orc priests deal [7-17] irresistible damage with smite.
What if the smite damage is [max(7-17, 7-17% current HP)] instead?
It will make them much more dangerous, but not too much.

So they'd be very similar until you have 100HP, and then they'd become maybe a tiny bit more dangerous? Only it wouldn't matter too much because Orc Priests aren't much of a threat by that stage. The main problem I see with that proposal is that's it's really confusing.

I admit that I made a bad example. However, I think it's confusing because the difference is too small.
The damage should be: [max(7-17, 14-34% current HP)]
So they'd be very similar until you have 50hp, and then they'd remain as significant threat throughout the whole game.
It doesn't have to be called 'torment'. The name 'smite' should remain as is. Players will think 'orc smite always hurt a lot'. I don't see much confusion from it.

[mikee]

A) You're on a wide open level and you see an orc priest at the edge of your LOS. As you walk towards it it smites you twice, and you lose approximately 60 hp. You leave, mash 5, and come back until this doesn't happen.

B) You're in a room with an orc priest and you have 1 hp. It can't possibly kill you.

C) You kill an orc priest, having taken over 100 damage in smites. You mash 5 for a while.

Changes like the ones being suggested would actually make the game easier to win and much more tedious. The reason that crawl becomes easier as the game goes on isn't that monsters don't do enough damage relative to your maximum health (there are in fact monsters in 3 rune games that can do more damage than some characters' maximum health in one turn). It's because the game is long enough that the player can accumulate enough resources to leave fights whenever he feels like it, and effectively doubling monster damage when you're at full health and reducing it when you're at low health exacerbates that situation.

[RedGem]

B) You're in a room with an orc priest and you have 1 hp. It can't possibly kill you.

This is wrong. The smite damage I suggested is [max(7-17, 14-34% current HP)].
This means that when you're under 50HP, the smite will work as it does now. 7 HP and another smite means you're dead for sure.
Let's assume that you have 200hp char, currently smite damage would go like:
[200 - 190 - 183 - 169 - ... 75 - 62 - 49 - 33 - 14 - 3 - 0]
It's expected to need about 15~16 turns to kill you.
With the system I suggested,
[200 - 144 - 110 - 95 - 68 - 49 - 33 - 14 - 3 - 0]
It'll take about nine turns.

In this case, two or three orc priests will always be a significant threat to any character.

Also, my point is, that we need to prevent monsters with interesting gimmicks from being obsolete too soon. Not that every monsters should have torment-like thingy.
Therefore, percentile damage based on max HP should also be considered.

[RedGem]

I'd like to summarize my suggestion. Maybe my thoughts were not well organized when I started writing the original post.

We need more diverse HP percentile damage to prevent many interesting monsters from being obsolete too fast.

Diverse HP percentile damage can make game more interesting.
It can be both based on:
Max HP
Current HP

In player's perception, it'll make easier to understand the threat level of a monster.
In dungeon designer's perception, it'll make easier to control the threat level of a monster.
Currently we have
[ orc -> orc warrior -> orc knight -> orc warlord ]
[ centaur -> centaur warrior ]
This is very hard to balance and takes endless effort.

With HP percentile damage, all the interesting monsters can make very complex situation altogether.

Here are some examples:
1. Green imp([3~7% damage based on Max HP], melee) vs Red imp
A. Open ground. An enemy starts to summon imps. You ignore them and find a way to kill the summoner.
B. Open ground. An enemy starts to summon green imps. Being surrounded by green imps is extremely lethal. You retreat and find a better place to fight.
It's very hard to use red imps to make the game more interesting. They're nothing but annoying in almost every situation. Can be ignored easily.
On the other hand, green imps can be an actual threat in almost every situation. When many of them are in the LOS, players must avoid themselves from being surrounded.
In case B, positioning matters a lot.

2. Orc warpriest([7-17 damage or 14-34% of the current HP if it is over 50)], smite) vs Orc priest(7-17 damage, smite)
A. You're XL19 MiBe. You encounter a group of orcs at V:3 including a orc priest and 2 orc warriors. They're not dangerous at all. It doesn't matter much which one you kill first.
B. You're XL19 MiBe. You encounter a group of orcs at V:3 including a orc warpriest and 2 orc warriors. You immediately get out of the LOS of the orc warpriest.
In this case, Killing orc warpriest is the highst priority. Positioning and LOS matters.

3. Green imps and an orc warpriest and other monsters together
When they come together, you'll have to think harder to survive. You can't simply ignore these two. And they fit well in almost any phase of gameplay.

I think crawl should be more about strategy, and less about resource management. It's more fun and more challenging.
Diverse HP percentile damage is a good way to make the game that way.

[varsovie]

Nice buff to the already overpowered magic users that don't rely on huge HP pool. Bye bye Trolls....

I'd like some more dangerous threats in extended (or even maybe late),but be it % based or not the problem as mikee said, unless you're drunk you nearly ALWAYS at that point have a way to avoid those encounters, and if you don't then you grab the orb and leave the dungeon.

Not counting the fact that it's kind of confusing if the 1st time you see an "priest" it takes a third of your health, and you think "nice it only does 10 dmg", next time you meet him you think "nice it will only tikles, I even got more Mr...." then getting rapped for 50Hp.

Torment at least has a special message and an effect grave enough to recognize easily and remembers it. I agree more alike should be thoroughly considered in the UI, we don't want cryptic stuff or more spam in the logs.

I think all in all scaling of the monsters are better done by increasing their types (orc, orc warrior, orc knight...), quantity (single orc, small pack, the whole fucking mines) or even different monsters (orcs, gnolls, draconians).

[RedGem]

I think HP percentile damage will actually make things less confusing. It'll make easier to understand the situation's threat level, once a player understands that 'some monsters deal HP percentile damage'.

It's actually very consistent. Players will learn by experience that they can't ignore some monsters even after they've grown much stronger.

Also, I think UI is not much of a problem for two reasons. First of all, when you think of torment, the main reason players start to think about it carefully is that it deals significant damage, not because of the message. I think 24% of current HP is good enough to trigger people to start thinking(24% is the average damage of the orc warpriest's smite I suggested). Also, it gives players enough chance to think about what's going on. They'll react and learn. Also, players who expirenced the HP percentile smite will be much less confused when they're tormented for the first time ever, and handle the situation more wisely.

Secondly, there are good ways to make percentile damage more recognizable. In the HP bar, the red bar is for the damage dealt in the last turn. We can divide it and give purple if the damage was dealt based on the Max Hp, Pink if it was based on the Current HP. Informing text can be something like 'It hits you from inside!!!' or 'It hits your soul directly!'.
This way players will understand what's going on easily. I think this should apply on current torment and flay too.

Once recognized properly, HP percentile damage is threat that's much easier to understand. Players can rely on their experience with more confidence. It's good way to diversify battle situations, and make the game more interesting easily.

[njvack]

If poison dealt, say, 25% of your current hp every time it ticked, however, I dare say it would command your attention in battle even when the venomous monster causing it didn't have such a large physical component to its attack.

The other nice thing about this suggestion is that poison would no longer directly kill you when you're otherwise out of danger.

[XuaXua]

So they'd be very similar until you have 50hp, and then they'd remain as significant threat throughout the whole game.
It doesn't have to be called 'torment'. The name 'smite' should remain as is. Players will think 'orc smite always hurt a lot'. I don't see much confusion from it.

I've long been an advocate for Orc Priest smiting to be %-based and never kill outright. Possibly scaled (but not eliminated) based on +rN to make +rN matter even more.

[Wahaha]

When monsters become too weak to be a threat, they also generally stop appearing in the areas that the player is in. Or at least are supposed to. Orc priests don't spawn in Depths (unless with a warlord? By the way why are warlords back to being nonexistent in Depths? They are a good monster unlike spriggans). So your premise that "x monster is too weak later in the game" is somewhat flawed because x monster is not encountered later in the game. Or shouldn't be. I think that right now all areas in Crawl present an appropriate challenge to the player. Individually some monsters might be weak for their area but some monsters being weaker than others is good. If a monster is TOO weak then yes it's a problem and it simply shouldn't spawn there.

An orc priest smites for 12 damage on average, that's easily 10% of a character's health in D and 10 to 8% in Vaults. So they're clearly dangerous enough for those areas without % scaling.

It's true that % scaling like the orc priest example makes the monster easier to balance and might be good, though like I said I think most monsters are fine already. So there's not that big of an advantage to making currently existing monsters scale with % damage. If new monsters are added that do % damage, then they need to be interesting monsters that do something interesting with % damage first of all.

The only advantage of % damage based on max hp (not current hp) that I see is easier balance. Assuming good balance, I don't think % damage based on max hp is any different from normal damage in terms of tactics.

Making orc priests do more damage to a level 15 troll than a level 5 troll is maybe ok. But why should a level 15 troll take more damage than a level 15 deep elf? Right now trolls take the same damage from orc priests as deep elves do. Why should that be changed? Note that this doesn't mean that both trolls and deep elves shouldn't take MORE damage after a certain point (although this by itself is very debatable), it just means that the "more damage" should still stay equal.

Possibly scaled (but not eliminated) based on +rN to make +rN matter even more.

Orc priests already have an attack that is resisted by rN.

[Siegurt]

While those small monsters stop being a threat, they also stop contributing meaningfully to your XP, and stop generating in the areas that you are currently 'exploring', orc priests for example are only going to generate as a random spawn while backtracking.

I dont' see why you would want serious "threats" to pop up while you are running back to that D:4 shop to buy an extra restore ability potions after finishing the Depths. I don't see how that meaningfully increases either the enojoyability or playability of the game.

Perhaps random spawns could be addressed in a better way (Either regular spawns could stop happening when the player is too far OOD for the level, or the spawns could start being OOD when you are significantly higher level than the dungeon level you're on) But honestly, the stuff you fought as a low level player *should* be not a threat at high level (If I skip killing sigmund on D:3 and return to blow him up at XL27, it doesn't contribute to the game in any way that he'd be just as likely to kill me.), and conversely high level threats should not do less damage just because you have lower max HP or are lower level.

Torment works well specifically because it's a specific threat that you have to deal with in specific situations, it would not work well as the "usual" case, that's just bad game design.

Trying to make low level critters a high level XP threat isn't a good idea at all, (And trying to make the XP that they are worth scale commensurate with that threat is an even worse idea.)

[RedGem]

I think all in all scaling of the monsters are better done by increasing their types (orc, orc warrior, orc knight...), quantity (single orc, small pack, the whole fucking mines) or even different monsters (orcs, gnolls, draconians).

Actually, it is not. The longer game goes, it's harder to make reasonable threat. It is harder to scale. And eventually, virtually almost every monster that you encounter become obsolete except for few. Scailing fails. That is the main reason why torment matters too much in extended, exacerbating problems that mikee pointed out. The longer you play the game, lesser the significance of the challenge. Eventually you end up in ziggurat, where the scaling methods you mentioned are in effect at extreme scale. But it's more of duration test, not much of strategy. Not much of a choice. Many strategic actions are forbidden there, making it much less complex and less interesting.

Percentile damage based on max HP can be a good solution to make this problem better. It is much easier to control.

[Siegurt]

Yes, torment (and monsters whose hit points and damage scale to the player) is a reasonable solution to a finite game which you can play infinitely. However the number of people who play a single game infinitely in crawl is vanishingly small and rearranging the game for those specific people isn't helpful to everyone else.

While you *can* play a single crawl game indefinitely, it's detrimental to both your chances of winning and your score to do so. Why would you also want to increase the "challenge" of doing so? That just makes something that's detrimental, grindy, and counter to the game design that much more attractive (to those who enjoy a 'challenge')

If nothing's a challenge any longer then you should've won a long long long time ago and started like 4 more games and won those too.

[RedGem]

@Siegurt
You are right. The main game is very good as is, and HP percentile damage is likely to do more harm than benefit for it.
Maybe it should be limited on the extended branches if it's to be applied. Or in regular end game too, as an introduction for extended.

[RedGem]

However, I'd like to point out one thing. Even with HP percentile damage, monsters can and will be less threatning as you grow. It will only slow the rate of them being obsolete. Not stopping it.

That's because of these three reasons.
1. As you grow, you'll have better ways to kill them.
2. You'll have better ways to avoid them.
3. HP percentile damage can be affected by AC or resistance or other means of defense. It doesn't have to always ignore your whole defense.
In this case monsters will do less damage if you have better defense.

So, basically, it's diversified means of dealing damage. Some attacks ignore AC. HP percentile damage attack based on Max HP ignore the size of Max HP.

[Tiber]

I very much disagree with the premise. The occasional monster that bypasses your defenses is interesting design. It works best on an enemy that is iconic, in the sense that torment is the central gimmick of a tormentor and hellfire is the gimmick of a hellion. In extended, however, torment is so common that it just feels like a cheap way to add difficulty, adds annoying resting, and basically makes offense the only form of defense.

That said, a torment-branded melee wouldn't be too bad, as "kill it as soon as possible" isn't the only way to deal with it. It doesn't help much with the melee/ranged disparity though.

[Galefury]

making torment more common in the parts of the game where it doesn't exist would be a good move imo! (i've suggested this in other topics here)

One thing about torment is that it damages all living things in los, and most monsters in the parts of the game where torment doesn't exist are alive, so torment would deal a lot of damage to the monster itself and its allies. But cutting everyone's HP in half might actually be a useful thing to do for, say, ogre mages. It might also be good on demon summoners like deep elf demonologists and Francis. How would you add torment to more places? Undead/demons who use it? Living casters? New monsters or additions to existing monster's spellbooks? How about agony instead (yes this would make MR even more important, but also it is blocked by a single level of rN)?

[njvack]

How about agony instead (yes this would make MR even more important, but also it is blocked by a single level of rN)?

Monster agony wouldn't need to work like player agony. If it was targeted torment, I think it'd make more sense than something MR-resistable and totally blocked by rN.

It might make sense as a shadow dragon attack.

[dck]

Monsters already have agony btw. It even has the extra power and everything.
Not sure if rN makes you immune.

[partial]

enemies that pierce armor are bad

[Patashu]

enemies that pierce armor are bad

Because...

[Sandman25]

enemies that pierce armor are bad

Because...

and basically makes offense the only form of defense.

[dck]

Corners are pretty offensive I suppose.

[KoboldLord]

Running away is the most devastating form of offense known to mankind.

[partial]

enemies that pierce armor are bad

Because...

Because they are unreasonably threatening to heavy armor melee characters. Obviously every character should have some ranged answer but that's not always the case, making offense the only defense.

Play pitsprint with a melee and you'll know what I'm talking about; a lone Phantasmal Warrior will almost certainly cost you a !heal or !rage or !agi (or kill you anyway).

[Sandman25]

Sorry, I didn't know 0.15 allows to train "CornerHiding" or "RunningAway" skills.
What's the point in training Dodging/Armour/Shields, hurting spell success by wearing heavy armour, getting elemental resistances if everything is ignored by Hellfire? What's the point in training Fighting if you still can die to 2-3 torments and hellfire anyway? It's possible to win 3-rune game with spear and no spells while extremely unstealthy, good luck doing it with 15 runer.

Would you love crawl if we had Hellions and Tormentors only? That would be very easy game (skill-wise) for Be - train Weapon/Fighting to increase damage dealt. And yes, only tactics matters, everything else is trash. Throwing a stone at every monster, hiding behind every corner, fighting dangerous monsters near teleport trap, you will love doing it hundreds of times in every game.

[tabstorm]

Sorry, I didn't know 0.15 allows to train "CornerHiding" or "RunningAway" skills.
What's the point in training Dodging/Armour/Shields, hurting spell success by wearing heavy armour, getting elemental resistances if everything is ignored by Hellfire? What's the point in training Fighting if you still can die to 2-3 torments and hellfire anyway? It's possible to win 3-rune game with spear and no spells while extremely unstealthy, good luck doing it with 15 runer.

Would you love crawl if we had Hellions and Tormentors only? That would be very easy game (skill-wise) for Be - train Weapon/Fighting to increase damage dealt. And yes, only tactics matters, everything else is trash. Throwing a stone at every monster, hiding behind every corner, fighting dangerous monsters near teleport trap, you will love doing it hundreds of times in every game.

By the time most melee characters hit extended they have like 30/30 defenses, +9 2h weapon, >200HP, etc., it's hard for a character with that much defense to die to normal attacks without really bad positional mistakes. That's why we have things like torment and hellfire. I think most damage your dude will take at this point will be magical or maybe ice fiend/archdemon melee.

IMO for extended you should try to take 10-15 stealth, it will make your life much easier since you can avoid tormentors or hellfire users some of the time and interact with them on your own terms. For this reason I favor enchanting moderate-weight dragon armours (up to storm) for my end-game armour, perhaps it is not optimal for a 3 rune game to absolutely maximize my win probability in Lair branches or whatever since I'm not enchanting plate, but if I know I am doing extended (i.e. I'm doing a speedrun) I want to make it easier on myself to get a reasonable amount of stealth.

If you worshipped trog at any point antimagic will neuter a lot of these enemies, it's one of the best uses of cleaving in my experience since you can block a few demons' magic at once. Another good thing to do is to take some translocations and train evocations for use of a ring of teleport, this way you can use spells like passage of golubria or evoke teleport to get out of potentially bad situations before your HP gets taken down to 40. Another useful fact is that Hellions and Tormentors don't see invis, so if you have any evokable invis it can allow you to just walk by them. Having good regen is also useful, since after a fight you can leave the area instead of stopping to heal, so that demons who come to investigate the noise won't find you. So, you don't need to do annoying stuff like throw stones and kite every dude, your tactics are just based more around having good mobility, good regeneration, good stealth, and maybe an antimagic weapon.

[and into]

Having some attacks ignore some (or all) of your AC is not necessarily unbalanced nor is it unreasonably dangerous for heavy armor characters. (And sprint should never form the basis for design decisions of the main game.)

The only thing I don't like about how damage is determined is that there are many exceptional or weird cases, some of which are not obvious. It is obvious that explosion-type attacks ignore EV, and that torment cuts your HP by a percentage, etc.; I have no problem with that. But when a single monster gets a melee attack that ignores your AC as its special little gimmick, and some (most?) electric damage is rolled against AC/2 rather than just your AC, and so on, it does end up adding a lot of opacity without (IMO) very much gain.