I'm tired of being one-shot

[Bananadine]

Man I'm playing on insane roguelike and I'm not even able to reach the point where an enemy can beat me in one turn, because I happen to have been playing antimagic characters, which (for some classes) seems to mean I either win access to Dream Walk from an escort quest or I eventually get caught by some super-tough monster's long-lasting pinning effect and then SLOWLY die. The universal solution to this used to be visiting the tool vendors in Zigur and Shatur. But now they no longer seem to sell torques of psychoportation, so antimagic characters had better load up on more wild infusions than I have been using I guess.

Hm I guess I could try using one of those gale force thingies to push away a super-strong plant boss that's got me pinned for twenty turns? I don't know whether that works. It seems like the plant might just resist the knockback though. It's no psychoportation... That ability was like the old version of Rush, in that it was so good that you'd probably always want it around. I don't know whether psychoportation is even gone?? But anyway it seems like it's gone from the early game at least. Try not to get stunlocked, Ziguranth!

what folks have been talking about in regards to insane play sounds to me like it's been every single time I've looked at it since the higher difficulties have been a thing -- lots of chaff and occasionally something just one-shots you and there's not much you can do about it besides stay well away from stuff that'll casually squish you (and playing to do that will probably be pretty unpleasant). Which is how insane's basically always came across,

Haha yeah that's ToME.

I think a crucial part of that has always been that this game hardly ever lets you get hurt in long-lasting ways. You can miss out on a quest here and there, but that's about it I guess. You can't take a severe injury that requires in-game months to heal (or that never fully heals). You can't have your equipment damaged by powerful attacks or bad cave-ins or corrosion from an acid-pool trap. You can't help yourself in the short term but hurt yourself in the long term by taking your time beating your immediate foe while the larger enemy army advances unimpeded behind the scenes, because the game doesn't usually simulate behind-the-scenes activity of that sort. Almost everything comes down to the immediate fight with just a few monsters, after which you can normally rest until you are in absolutely perfect health in all ways. So every fight has to cram in an entire game's worth of challenge; it's not good enough for a monster to just get in your way briefly or wear you down a little, because you have the ability to rest after every few fights, and resting is vastly overpowered.

I think that that is the root of this kind of problem in this particular game, and I think it will probably never change much. Adding better enemy AI could probably improve the situation; adding a new character class that relies heavily on a new resource that actually doesn't come back quickly or frequently and must be managed carefully over an entire run could maaaaybe solve the problem (albeit only for that class). But otherwise I think most of the game would just need to be reworked, for this problem to be properly solved.

I'm glad somebody is working on the enemy AI, even if the result isn't always ideal. I think that's the best hope there is for ameliorating this game's deepest, largest flaw.

Another potentially powerful and cost-efficient angle of attack might be to add one or more long-term time limits, driving the player to actually care HOW those easily beaten weak enemies are beaten - you'd be wanting to fight them efficiently, and not just safely, which could add a lot of interest and challenge. I think maybe that could help a lot. Time limits have great potential to be annoying or frustrating though... and they could possibly change the feeling of the game drastically.

[Snarvid]

1.5.10 did have similar problems, but they were the "devils you know" so you just put up with them as though they were normal.
1.6 has "devils you don't know", so they seem far scarier because you don't know what they are yet or how to deal with them.
Reverting the changes just freezes the development on all those areas that still do need improvements. As they can't sensibly test them while its sitting as a side branch. We all just saw that.

The Devs are still working hard, and I've seen AI discussion and changes going on for 1.6.6. There also appears to be improved tooltip code and something for making it easier to identify the cause of an insane death. There is also a lot of little bug fixes and tweaks going on. And this has been going on over a traditional Holiday period.

As of this post, the Vault contains 252 Insane 1.6.5 Winners. So the stupid one-shots aren't stopping people from winning. Perhaps the people that write guides are waiting for the dust to settle, or are busy doing something else at the moment?

To reuse a response I gave in another thread - there's no claim (by me, at any rate) that Insane is strictly unwinnable in its current state (since Mankeli posted some of their wins, I know it's possible). It's more that any given run might or might not turn out to be unwinnable based on ai omniscience and/or boss talent level/combinations RNG, making the final outcome something the player has no control over, cannot perceive from a world seed ahead of time, and cannot learn counterplay against. If you don't encounter these situations you can definitely win, but I want to push back against the "devil you know" narrative a bit if it's meant to imply a learning curve that, once mastered, would remedy the concerns in the the previous sentence. The impact of some devils (e.g. a whole Pride activating at once) aren't mitigated in the knowing.

I have no problem with people taking holiday breaks or time to search for and execute solutions, to be clear, and I don't doubt they're working hard. Much of the discussion and the patches thus far have mostly tended to focus on individual talent tweaks being the solution to 1.6x's one-shots, and I'm glad to hear that there is also movement on more systematic problems for an upcoming patch. My rollback suggestion was a hope for a Gordian knot-style fix to these, as thus far the changes to AI and boss power levels seem like a painful implementation of an idea for which I do not perceive the original demand (although my ability to perceive this is notably limited by not being on Discord or in-game chat, I mostly notice what is posted here).

[Snarvid]

I think a crucial part of that has always been that this game hardly ever lets you get hurt in long-lasting ways. You can miss out on a quest here and there, but that's about it I guess. You can't take a severe injury that requires in-game months to heal (or that never fully heals). You can't have your equipment damaged by powerful attacks or bad cave-ins or corrosion from an acid-pool trap. You can't help yourself in the short term but hurt yourself in the long term by taking your time beating your immediate foe while the larger enemy army advances unimpeded behind the scenes, because the game doesn't usually simulate behind-the-scenes activity of that sort. Almost everything comes down to the immediate fight with just a few monsters, after which you can normally rest until you are in absolutely perfect health in all ways. So every fight has to cram in an entire game's worth of challenge; it's not good enough for a monster to just get in your way briefly or wear you down a little, because you have the ability to rest after every few fights, and resting is vastly overpowered.

Can I interest you in Possessor? You get much bigger power peaks in exchange for being able to be hurt in long-lasting ways.

[whitelion]

Is there a reason besides sunk costs to not roll back to 1.59 and just add in new content (inscriptions, prodigy changes, classes, bug fixes for memory leaks and stack overflows) while reverting the AI, randboss talent system, and crit/damage formulas to the last version that didn't have these problems?

1.5.10 did have similar problems, but they were the "devils you know" so you just put up with them as though they were normal.
1.6 has "devils you don't know", so they seem far scarier because you don't know what they are yet or how to deal with them.
Reverting the changes just freezes the development on all those areas that still do need improvements. As they can't sensibly test them while its sitting as a side branch. We all just saw that.

The Devs are still working hard, and I've seen AI discussion and changes going on for 1.6.6. There also appears to be improved tooltip code and something for making it easier to identify the cause of an insane death. There is also a lot of little bug fixes and tweaks going on. And this has been going on over a traditional Holiday period.

As of this post, the Vault contains 252 Insane 1.6.5 Winners. So the stupid one-shots aren't stopping people from winning. Perhaps the people that write guides are waiting for the dust to settle, or are busy doing something else at the moment?

1.5 had its own problems yes, but I think there is significant agreement among many insane+ players who post here that the particular problem of one-shot deaths is noticeably worse in 1.6. Yes, this evidence is anecdotal, but as Mankeli said, we don't have access to statistics of that nature, and even if we did they may not tell the whole story either. And this is not to say that 1.6 hasn't improved on 1.5 in some areas. I think the reworked prodigies, races, and inscriptions were good things, for example.

And ultimately, in my and some other cases, the game simply feels less enjoyable in 1.6. We have been trying to identify specific changes from 1.5->1.6 that have contributed to this feeling both to verify this feeling is not simply "different = bad", and in the hopes that the devs will consider our experiences, which I am happy to say it sounds like they are. But it's also subjective and people play the game for different reasons. If you like 1.6 better than 1.5, I'm not saying you're wrong, rather you just have a different opinion. darkgod is right that he can't please everyone all the time, and ultimately it's his game so he can pursue the development vision that he has. And as Snarvid says, if after the dust settles the game ends up in a form we no longer enjoy, we will simply choose to do other things rather than discuss it indefinitely. And I certainly don't plan to harass or begrudge anyone who enjoys 1.6; good for them.

I, and I hope others, also do appreciate the work the devs are doing. And I understand the work takes time, and certainly don't begrudge anyone their holidays because of a game I like to play. I am happy to wait a while as these things are worked on, so long as things are being looked at, which it sounds like they are.

I think that is also a big reason for the lack of 1.6 guides so far. Since the release of 1.6, there have a lot of things in flux, as happens with a big new release, and so it seems understandable to wait for the dust to settle rather than write a guide that will be outdated a month later.

[HousePet]

Actually, I might be able to guess why some of the changes that have appeared in 1.6 that are bitchslapping Insane+ players happened.
There is a major problem with the Normal difficulty level. It becomes a cakewalk at level 30. (That is a weird phrase.) I could basically open Vaults whenever I found them, even with my sub-optimal build style. I moved to Nightmare to see if it would help, but it doesn't really.
Now, Nightmare makes all the enemies have higher stat and talent levels. Negligible early, significant late game. If better stats doesn't fix the problem, what other options are there?
Item or talent changes? Both players and enemies get those, so any changes will mostly cancel out.
So what remains to be improved? The rather bad AI.
However, what would happen if you give better AI to enemies that have much higher stats and talent levels than the player?
On Normal difficulty, enemies are now on par with the player because they have similar stats. On Insane, the Rare+ enemies can now roflstomp you.

[Snarvid]

We had a similar chat in “overtuned.”

The AI stuff I’ve seen and find objectionable is stuff like the AI activating and/or corner sniping without ever having LOS on the player - playing by a different rule set than previous versions to drop defacto Hunted! into Insane. I do agree that a smart enough AI can beat human players even when limited to similar APM and knowledge (DeepMind Starcraft AI), but the implication that this is what’s responsible for the losses we are concerned about (again, things like dropping 2k+ damage through capped resists before the player gets a turn) seems not likely to lead to relevant fixes. And good AI in a game is generally seen as praiseworthy, and I’d prefer not to valorize the 1HKOs ending 15 hour runs.

But we’ve a new version to look at with at least some Cursed fixes, so off to take a look before further critique.

[whitelion]

I had a chance to try the new version a bit yesterday, and my first impression is that the AI now does seem much less "Skynet" and generally doesn't come after you until the enemy has at least caught a glimpse of you, which seems more fair than monsters phase dooring in from who knows where. The AI also seems less good at finding you if you break LoS. It will look for you where you first ran away, but doesn't automatically know where you are at all times like before. This also seems reasonable to me, so overall I like these changes so far. I'll update once I have a chance to play more with further thoughts on things.

[Zizzo]

There is a major problem with the Normal difficulty level. It becomes a cakewalk at level 30.

*blink* I… can only conclude that you and I are playing completely different games. I'm still trying to figure out how to tone down the unrelenting stress pit that is Normal-mode late-game.

I could basically open Vaults whenever I found them, even with my sub-optimal build style.

Whereas I often run into trouble with Orc Pride vaults even after coming back to them several levels later.

So what remains to be improved? The rather bad AI.

See, I'm not sure that follows, though. I mean, as I interpret your comment, I gather you have no complaints with the game before level 30, so it seems you'd want a change that only (or primarily) affects the late game. Changing the AI, though, affects everything, more or less equally.

(On the other hand, I'm not entirely certain the new AI is the proximate cause of the particular problems I'm having, so I may well be talking out of my hat… )

[HousePet]

Improving AI doesn't affect early game as much, because nothing in early game has many tricks it can use.
Whereas by late game, everything has a bunch of things it can throw at you, but it was basically doing it in a stupid way.
This is a generalisation with many exceptions, but compare Wrathroot with Vor:
Wrathroot by default has Armour Training, Stun, Ice Storm, Tidal Wave, Freeze and a random Infusion. It barely matters how good its AI is, because it doesn't have many options about what it can do.
Vor has 4 times the number of different talents, 5 slots of runes and a random wand. With bad AI its just going to throwing random spells at you. With good AI its got a very large toolbox to make your life hell.

So yeah, we can give a Jelly the most advanced AI on the planet, but it still won't make it dangerous.
So how does changing the AI affect everything more or less equally?

[GlassGo]

I had a chance to try the new version a bit yesterday, and my first impression is that the AI now does seem much less "Skynet" and generally doesn't come after you until the enemy has at least caught a glimpse of you, which seems more fair than monsters phase dooring in from who knows where.

It seems AI doesn't come after you even if his friends does see you and that mob see his friends.
You can clearly see it in Trollmire with Bill - he almost never come after you, while all his frinds in his LoS come to you.
So it's actually bugged?

[whitelion]

I had a chance to try the new version a bit yesterday, and my first impression is that the AI now does seem much less "Skynet" and generally doesn't come after you until the enemy has at least caught a glimpse of you, which seems more fair than monsters phase dooring in from who knows where.

It seems AI doesn't come after you even if his friends does see you and that mob see his friends.
You can clearly see it in Trollmire with Bill - he almost never come after you, while all his frinds in his LoS come to you.
So it's actually bugged?

Yeah, it does seem like sometimes enemies who should act as groups don't for this reason. You can also notice it with those little groups of drakelings. Maybe messaging passing should be very effective only for enemies who are part of the same "unit" then? But either way, I still like this better than Skynet.

There is a major problem with the Normal difficulty level. It becomes a cakewalk at level 30.

*blink* I… can only conclude that you and I are playing completely different games. I'm still trying to figure out how to tone down the unrelenting stress pit that is Normal-mode late-game.

Anecdotally at least, I think normal is plenty hard for many players. It certainly was for me when I started playing a few years ago. I think that if you find normal too easy, you should play on higher difficulty. I don't think that normal difficulty needs to be made harder for everyone.

[HousePet]

Well I'm not sure that a more difficult late half of the game is the end goal.
Its been a long standing issue that The East can be a bit of a grind against tough enemies. And its not that its hard to beat them, its that there are lots of them and they take longer to go down.
If the enemies were a bit smarter then they could have their stats lowered, resulting in a similar level of 'difficulty', but less of a grinding feel.

[Tryble]

Here's another instakill with screenshots.

Log and mouseover of enemy. He dealt about 2400; I had 170 armor at the time, from thorny skin.
SS of attack tab.
SS of talents tab.

I'd been tearing through prides and all the rest of high peak with almost no threats to speak of, until this dude took me out. Sure, I could've survived I'd known he could do up to 2500 but I had no idea anything physical would hit that high, especially through so much armor+thorns. Logs say first was 374 damage after armor, 1090 after crit multiplier, 1307 after boosts. Ditto for same except it's not boosted. Crazy. I'd grown pretty lax after so much time without anything meaningfully threatening and didn't examine him that thoroughly other than to note the high melee damage.


Pastebin of the log file of that turn.

Edit: The burst from that prismatic strike should have dealt another 450 to me, I think, but I already died. Also, on examining that guy, he's got about 150 global and 120 attack speed so he might have doubled me that turn even if I'd survived, geez.

Edit2: I went at that guy again with storm up, and yeah he threw 3500 as an opener on me. thanks stormshield

[Delmuir]

I'm not terribly concerned about being one-shot as I just consider that part of the game but I do want to throw my vote out there for taking a look at Greater Weapon Focus. That thing seems to be the most common kill-skill against me, though primarily on Nightmare. I don't think I see it as much on Normal but... who knows?

I recently lost an Archmage on Nightmare in the Sandworm Lair through an 1100 damage shield, and about 450 life. One shot opener. I mean... that seems a little rough to me.

[ktg]

Grushnak just did 3307 against my Archmage on Nightmare with Greater Weapon Focus (1778) + Assault (1529). I also saw a log in Discord where one of the combat trainers did 4236 with a similar combo.