1.6 changes to saves - testing feedback

[supermini]

Physical/mental/spell saves have been changed in 1.6. As far as I can tell, the goal was to make saves always be useful. It used to be that when you were 20 power over a save, there was basically no chance of your effects getting resisted, and when you were 20 over with a save, there was basically no chance of being affected. Now the curve is much smoother, so there always seems to be a chance of a save reducing the duration/resisting the effect.

After testing 1.6 beta with multiple characters, I'd say that this change is a really, really bad idea. Lets take an example.

A reaver with 50 spell power is casting epidemic (base duration 6) from range on a training dummy with 7 spell save 100 times. It has 0% disease resistance (not that it would help it, epidemic ignores disease resistance).

(I accept that testing something 100 times isn't exactly statistically rigorous, but that's all the time and patience I have - and it matches the empirical observations).

FP - full power 56%
R1 - duration reduced by 1 8%
R2 - duration reduced by 2 12%
R3 - duration reduced by 3 7%
R4 - duration reduced by 4 2%
R5 - duration reduced by 5 0%
RS - resisted 15%

Okay, so what does that mean? It means that a midgame character fighting, say, a trollmire snake has about a 56% chance of his abilities actually working as advertised, and 15% of just nothing happening. And most enemies that you fight at your level have way more saves than 43 below your main power.

In practice, this leads to very frustrating and non-engaging gameplay where more often than not, your abilities simply don't work. I just don't see what purpose does it serve to have enemies with garbage saves resist anything.

For testing purposes, I got one character to the bottom of Dreadfell and another one to clear Tier 1 dungeons, and it was the most frustrating experience I ever had in ToME. I suggest a rollback to the previous save system, because this basically throws out the baby with the bathwater.

I invite everyone - especially the people working on 1.6 - to test it and draw their own conclusions.

[HousePet]

I agree that these number seem off.
100 tests is good enough for me, when you have a spread of effect strength like this. (Its not ideal, as the error is +/-10%, but halving that require 400 tests.)

[Doctornull]

Maybe rank should matter, so bosses and the PC can have 44% effect reduction -- but Trollmire snakes can suffer all full effects, because seriously f-----* those guys.

On the other hand, not being able to reliably stun a boss would be differently horrible.

Yeah I don't see value in this.

*) flambé

[Delmuir]

I don't really understand the issue with saves and resistances...

What would make sense to me is this:

1. Saves determine whether or not the effect lands.

2. Resistances determine the duration, i.e., 90% resistance reduces duration by 90%.

3. Cap resistances at some appropriate amount, likely between 70% and 95%.

Enemies that are immune to certain effects remain immune. Balance it such that if your saves are sufficiently high relative to the opponent's power, the effect will rarely land and vice-versa but even if it lands, you'll have resistances to back you up by mitigating the effect.

An ideal balance, to me, is that status effects should land relatively frequently. I'd say that if your power is equal to the enemy saves then the effect should land 75% of the time. If its double the enemy saves then it should land 100% of the time... or something like that.

Utilizing this approach, you could reasonably use saves to stave off most status effects but you'd wander into the occasional enemy that just overpowers them, and so you'd still want resistances to back you up and mitigate the duration. Getting hit with confuse is punitive but if its duration decreases to 1 or 2 from 8 because you have high resistances then that's manageable.

More so, your abilities would land the overwhelming majority of the time except against the occasional enemy with high saves. Then all the programmers need to do is ensure that enemies have reasonably high attack power and moderate to low saves to ensure that outcome because anything much lower than about a 90% of success radically reduces the utility of status effects such that they're not even worth using.

I'm going to post this in the ideas forum as well.

[supermini]

I don't really understand the issue with saves and resistances...

What would make sense to me is this:

1. Saves determine whether or not the effect lands.

2. Resistances determine the duration, i.e., 90% resistance reduces duration by 90%.

3. Cap resistances at some appropriate amount, likely between 70% and 95%.

The issue with this approach(and the 1.6 version of saves) is that not all effects are created equal. Duration reduction makes sense for damage over time abilities, while for most of the truly dangerous effects it doesn't matter as much. Having Inner Demons on you is in most cases as lethal when it lasts 6 turns as it is when it lasts 12 turns. You either clear it or you're in serious trouble. That's more or less the same for being blinded, silenced, stunned, confused, etc. It works like that in reverse as well - disabling an enemy for 4 turns is usually as good as disabling them for 6.

Another issue I have with your idea specifically is that the current way talents are balanced assumes that saves and resistances work, well, the way they work now. So it's not just that you have to change the complete system of saves and resistances, you have to go back and update all talents with a completely different system in mind. What you're proposing is a complete overhaul.

I'm not saying that a complete overhaul is a bad idea, but 1.6 is far enough along that I don't think it's realistic. At this point, it's more realistic to expect that a system that we already have will be tweaked so it works better than that someone will spend months on completely overhauling the game.

[Delmuir]

That's a very fair point on it requiring a complete overhaul... can't argue with that.

I will respond further to your criticism by noting that you could address that point by simply making infusions and runes that clear status effects have priority to those effects OR make it so that you can choose which effect is eliminated. Then keep the abilities that eliminate status effects the same (general) which would maintain the utility of runes and infusions.

But that's probably a moot point as your broader criticism is likely fatal to my idea.

Thanks!

[Zeyphor]

i think the changes to saves would cause more problems than they would solve; it isn't worth doing

[starsapphire]

Tried a few play on 1.6 beta and must say yes. It makes the effect of control (stun/daze/confuse) skills much more unpredictable than before, which was the basis of many classes. Those classes relies on DoT damage like poisons and Paradox Mage is also badly affected. The intention of this change may be proposed to deal randboss with high power in insane, but it will create more problem in normal and nightmare difficulty, while boss in insane+ still can land 10+ turn debuff on players because of their high talent level. And the most important, not being able to predict the outcome of most of actions will be subversive for previous tactical thinking, which will make people feel discomfort. Believe this change may need more examination.

[yutio888]

I found it really strange that in "mod/class/Actor.lua#on_set_temporary_effect", effect duration percentage is randomly determined by normal distribution with such great standard deviation(in 1.6b1 50%)， so all debuffs have ~16% chance to reduce their duration by at least half. And currently duration cannot be extended even if the percentage is above 100%, what is the point to set such hight standard deviation? Note that the mean percentage is (110-savedChance * 1.1), and "savedChance" is usually above 10%, then the chance for "normal" result is likely beyond 50%, and EVERY TIME you try to apply debuff, you need to pray a "normal" result.

[Nevuk]

According to the 1.6 beta 2 changelog all the save changes were reversed.

It does feel a little silly that we have two stats that do the exact same thing - status resistance and saves. It's somewhat redundant. The only real plus side is that the saves make it so that a class is more resistant to its own type of spells (ie, mages have high spell save while their freeze resist is entirely based on gear/skills, while bulwarks et al have crappy spell save). This could in theory allow a class to skimp on a certain resistance, but most effects are caused by multiple sources (ie, stun/freeze is both physical and magical. Same for daze. I think confusion may be mostly mindpower but I'm not really sure), so this doesn't really apply either.

In other, similar, games status resistance usually affects the duration of that effect (similar as in RPG type games, rather than roguelike. I know that in DotA and in Grim Dawn there are status effect duration resistances but they are capped in both games. In Grim Dawn the cap is 80 and the cap can be increased by probably ~10% max by gear. In DotA it's only a very few items and a couple of skills/heroes, where the average hero can't get past about 30%. It can also behave weirdly vs DoTs, where it can increase their DPS but shorten their duration, as most of them are attached to other debuffs like slows. Those two games came to mind as examples for being well known for their balancing efforts). However, these are real time rather than turnbased games, where a 31.357% reduction makes more sense.

I liked the idea in 1.6, but I do agree with criticisms in this thread. I think a wider spread that has an effect would be nice though, as +20/-20 are the only markers that currently seem to matter. IE, saves in the 20-30 range could reduce duration but not chance to be affected at all. The important thing is that it would need to be more predictable than the beta1's were, with a lower chance of drastically lowered duration.

A possibility hinted at earlier in this thread to make saves more important would be to allow for especially bad save values to increase duration. This would require a fair amount of rebalancing though, as some durations are already crazy high (20 turn hexes, would be awful on a bulwark with 10 mag and 0 spell save, etc.) so it wouldn't really be a good fit for 1.6.

[HousePet]

I don't find it odd to have generic save stats and status immunity percentage stats. I've seen this done many times before.

Saves aren't always about resisting status effects, and you would never create a immunity stat for every status effect.