So - physical/mental/magical saves. They're not well understood, and appear to be near-meaningless, unless you put a fair amount of effort into making them work properly. Currently, with a few possible niche exceptions of people who really understand the system and optimize around them, bonuses to saves are basically worthless - meaning that the skills that grant them grant nothing. As we are currently trying to work through fixing the various skills, I would propose that we institute a system of saves that actually works and makes relatively straightforward sense - both to rehabilitate those skills that provide saves (or at least get them to the point where their rehabilitation can be meaningfully discussed) and to make "and add a bit of this save or that save" a valid option for buffing certain skills out a little (or making them a bit riskier).
Given that, there's a few questions to start with.
- What's the point of saves? Why are they here at all? Do we want them in the game?
- What sorts of behavior are saves supposed to produce? Do we like having those behaviors around?
- As far as I can tell, the intended behavior is to make it so that low-level characters can't cripple high-level characters with debuffing, but I'm really not sure.
After that, I think we should try to think up a way to structure saves such that a bonus to saves is going to be useful to most people who do not specifically choose to ignore saves, and in a consistently predictable way with decent scaling. On the "consistently predictable" side, having a percentage chance popup on the character screen might be nice. "You currently have, on average, a 24% chance to save against physical effects inflicted by other creatures of your level. Adding 20 points of physical save will increase that to a 35% chance"
Actually, I think scaling may be one of the major issues. At least by my (very, very limited) understanding of the way saves work, scaling gets to be a problem. If you have a skill that grants you 15% stun resist, then that'll be 15% stun resist all the way up to level 50, stacking with any other stun resist you may come across. If you get +25 physical save from a skill, then you're going to be pretty well-protected from anything that requires physical saves at level 1, but by the time you get to level 30 or so, it's starting to look a bit underwhelming. As such, if I care more about long-term viability than the immediate benefits, I'll pretty much always choose a resist skill over a save skill - especially when there's just not that much saving to do early on.
Insight? Thoughts? Anyone?
fix to the saves system
Moderator: Moderator
Re: fix to the saves system
Some thoughts I had about this:
To make saves more relevant 'on average', what if there was a save for EACH possible turn of effect duration(or knockback square)?
Let's say you could be knocked back 7 squares. You roll 7 saves, only 3 fail, so you're knocked back 3 squares. Or you could be stunned for 3 turns, but you make 1 save and so are only stunned for 2.
Yes, this could nerf classes who rely on status effects significantly, but it would also help them against NPCs. Might need some rebalancing.
To make saves more relevant 'on average', what if there was a save for EACH possible turn of effect duration(or knockback square)?
Let's say you could be knocked back 7 squares. You roll 7 saves, only 3 fail, so you're knocked back 3 squares. Or you could be stunned for 3 turns, but you make 1 save and so are only stunned for 2.
Yes, this could nerf classes who rely on status effects significantly, but it would also help them against NPCs. Might need some rebalancing.
ToME online profile: http://te4.org/users/zonk
Addons (most likely obsolete): Wights, Trolls, Starting prodigy, Alternate save/resistance system
Addons (most likely obsolete): Wights, Trolls, Starting prodigy, Alternate save/resistance system
Re: fix to the saves system
Scaling is definitely the issue in beta18, and has actually been upgraded in beta19 so saves should be more important soon. Currently saves are ONLY checked before applying status effects, eg stun, confuse, freeze, etc. There may be some status effects that are not subject to saves, but that is probably unintended.
In terms of mechanics the saves use the same mechanics as hit/defense. More example stunning blow checks the attacker's "Attack" (in character sheet) number and uses the defender's "Physical Save" to see if the status effect hits. Following that there is a check for status resistance/immunity. Since saves were so much lower than the attack/spellpower/etc being used to attack with the saves were essentially meaningless. It may be useful to actually have a separate message for status resistance/immunity and a save to help players see if their saves are actually working...
In terms of mechanics the saves use the same mechanics as hit/defense. More example stunning blow checks the attacker's "Attack" (in character sheet) number and uses the defender's "Physical Save" to see if the status effect hits. Following that there is a check for status resistance/immunity. Since saves were so much lower than the attack/spellpower/etc being used to attack with the saves were essentially meaningless. It may be useful to actually have a separate message for status resistance/immunity and a save to help players see if their saves are actually working...
<DarkGod> lets say it's intended
-
Susramanian
- Spiderkin
- Posts: 454
- Joined: Sat May 15, 2010 3:09 am
Re: fix to the saves system
Weird timing. I was about to start a thread on the subject.
Saves as they work right now suffer from multiple problems. The first is that their effectiveness is entirely relative to unknown numbers. How good is a spell save of 100? Impossible to say without knowing hidden stats about what you're trying to save against. A spell save of 100 is perfect* when pitted against numbers below about 80. It's totally* useless when pitted against numbers above about 120. This makes it nearly impossible to give the player useful information about what their save numbers mean.
(* caps apply)
The second problem I see is that saves perform a function that is almost identical to that performed by the various effect resistances. The math looks much different, but from the user's point of view, they all feel the same. They make bad stuff less likely to happen to you.
To address both of these problems, I propose that we make saves a percent value that is used to reduce the duration of applicable effects. A physical save of 60 means that, for example, a bleed effect would have its duration reduced by 60%. This would require some rebalancing of save values granted by gear and stats. Caps might be a good idea, too.
The big problem I see with this is that saves would have a minor or unnoticeable effect for much of the game. Reducing the duration of a confuse effect by 12% in the early game wouldn't do much. What might be cool is to make a system in which save numbers are fairly easy to come by, even in the early game, but the caps shift upwards slowly as the game goes on. So maybe a berserker starts the game with a physical save of 40, but the cap isn't much above that at first.
There are quite a few ways to set up saves and save caps to achieve this. Maybe caps are determined by stats (strength, etc.), but the values are determined by gear and level. Or maybe the caps shift with level, and stats and gear improve your save values. Or maybe we can grant large, innate bonuses to save (and/or save caps) based on class and race. This would be a flavorful and noticeable bonus to differentiate races.
How about this: gear and stats increase your save values (as now), but the caps are determined by class and level. Increase these caps as the player levels up, with different classes obviously arriving at different caps at level 50. For example, maybe a berserker starts with caps of 50, 20, and 30 to physical, spell, and mental saves, respectively. By level 50, these are 80, 32, and 48 (if we want to keep everything proportional, which certainly isn't required.) Archmages would have very different caps.
To increase the value of saves, we could also introduce caps to the various effect resistances (stun, confusion, etc.). Once it becomes impossible to have total immunity to certain effects, it becomes much more desirable to find a way to reduce the duration of the effects when they inevitably hit.
Saves as they work right now suffer from multiple problems. The first is that their effectiveness is entirely relative to unknown numbers. How good is a spell save of 100? Impossible to say without knowing hidden stats about what you're trying to save against. A spell save of 100 is perfect* when pitted against numbers below about 80. It's totally* useless when pitted against numbers above about 120. This makes it nearly impossible to give the player useful information about what their save numbers mean.
(* caps apply)
The second problem I see is that saves perform a function that is almost identical to that performed by the various effect resistances. The math looks much different, but from the user's point of view, they all feel the same. They make bad stuff less likely to happen to you.
To address both of these problems, I propose that we make saves a percent value that is used to reduce the duration of applicable effects. A physical save of 60 means that, for example, a bleed effect would have its duration reduced by 60%. This would require some rebalancing of save values granted by gear and stats. Caps might be a good idea, too.
The big problem I see with this is that saves would have a minor or unnoticeable effect for much of the game. Reducing the duration of a confuse effect by 12% in the early game wouldn't do much. What might be cool is to make a system in which save numbers are fairly easy to come by, even in the early game, but the caps shift upwards slowly as the game goes on. So maybe a berserker starts the game with a physical save of 40, but the cap isn't much above that at first.
There are quite a few ways to set up saves and save caps to achieve this. Maybe caps are determined by stats (strength, etc.), but the values are determined by gear and level. Or maybe the caps shift with level, and stats and gear improve your save values. Or maybe we can grant large, innate bonuses to save (and/or save caps) based on class and race. This would be a flavorful and noticeable bonus to differentiate races.
How about this: gear and stats increase your save values (as now), but the caps are determined by class and level. Increase these caps as the player levels up, with different classes obviously arriving at different caps at level 50. For example, maybe a berserker starts with caps of 50, 20, and 30 to physical, spell, and mental saves, respectively. By level 50, these are 80, 32, and 48 (if we want to keep everything proportional, which certainly isn't required.) Archmages would have very different caps.
To increase the value of saves, we could also introduce caps to the various effect resistances (stun, confusion, etc.). Once it becomes impossible to have total immunity to certain effects, it becomes much more desirable to find a way to reduce the duration of the effects when they inevitably hit.
Re: fix to the saves system
I think whatever system is used needs:
1. An attack value versus a defense value that accounts for the attacker and defender's skills. (for example, spell power vs spell save)
2. A way of modifying chance and amounts (for example, 30% chance to stun OR stun for 4 turns)
A lot of this comes down to the math. Its hard to pick a system that works well for these cases: (especially #2)
1. A wide range between attacker and defender (for example, non-magic player vs magic boss)
2. Early levels that have small differences between attack and defense values versus late levels that have wide differences. (For example, an archmage's spellpower starts out close to most opponent's magic saves but quickly outpaces them)
Cursed's "Gloom" talent solved #2 by looking at the ratio of attack to defense for deciding who failed the save. At 2 to 1 it would use a maximum chance and at 1 to 2 it would use a minimum chance. Some quirky math was also added to get a better spread at low values. This works, but its a lot harder to explain and understand.
Here's a very simple percent based system that I kind of like:
value = value * (100% + attack% - defense%).
So if you have a physical talent like this: "The talent has a 30% chance of stunning enemies for 4 turns (* +15% physical power)"
Attacker: +15% physical power
Defender: +5% physical save
30% stun => 33% stun
4 turns => 4.4 turns
Defender: +40% physical save
30% stun => 22.5% stun
4 turns => 3 turns
Defender: +70% physical save
30% stun => 13.5% stun
4 turns => 1.8 turns
This handles issue #1 ok, but not #2. At low levels, this system means nothing. At high levels, it can vary wildly. It can be improved though by either improving all saves and attacks based on level (+1% per level for everyone) or by ignoring levels completely by keeping attack and defense pretty fixed throughout the game.
Anyway, I guess my point is that no explaination is complete without looking at the math and handling these spreads between attacker and defender.
1. An attack value versus a defense value that accounts for the attacker and defender's skills. (for example, spell power vs spell save)
2. A way of modifying chance and amounts (for example, 30% chance to stun OR stun for 4 turns)
A lot of this comes down to the math. Its hard to pick a system that works well for these cases: (especially #2)
1. A wide range between attacker and defender (for example, non-magic player vs magic boss)
2. Early levels that have small differences between attack and defense values versus late levels that have wide differences. (For example, an archmage's spellpower starts out close to most opponent's magic saves but quickly outpaces them)
Cursed's "Gloom" talent solved #2 by looking at the ratio of attack to defense for deciding who failed the save. At 2 to 1 it would use a maximum chance and at 1 to 2 it would use a minimum chance. Some quirky math was also added to get a better spread at low values. This works, but its a lot harder to explain and understand.
Here's a very simple percent based system that I kind of like:
value = value * (100% + attack% - defense%).
So if you have a physical talent like this: "The talent has a 30% chance of stunning enemies for 4 turns (* +15% physical power)"
Attacker: +15% physical power
Defender: +5% physical save
30% stun => 33% stun
4 turns => 4.4 turns
Defender: +40% physical save
30% stun => 22.5% stun
4 turns => 3 turns
Defender: +70% physical save
30% stun => 13.5% stun
4 turns => 1.8 turns
This handles issue #1 ok, but not #2. At low levels, this system means nothing. At high levels, it can vary wildly. It can be improved though by either improving all saves and attacks based on level (+1% per level for everyone) or by ignoring levels completely by keeping attack and defense pretty fixed throughout the game.
Anyway, I guess my point is that no explaination is complete without looking at the math and handling these spreads between attacker and defender.
Re: fix to the saves system
I think saves in general must be more reliable against a range of attack strengths. With a very high save you should expect only the strongest bosses and elites to overcome your save roll. With a low score you should still expect a few lucky saves. The current +/- 20 range is far far too small and needs to be stretched significantly, with particular bias towards high save values still being effective against even high attack values.
I'd say instead of the roll being based on the magical +/- 20 number it should consider the attack and defence numbers in proportion. If your save is half the attack you should expect about 25% of the rolls to save. If equal you should expect around 75% saves. If double you should expect to almost never lose.
Another solution is to remove the save system entirely and only use resistance %'s. This would be a shame, but shouldn't be ignored if the save system can't be made useable and transparent.
At the moment there aren't many ways to modify the save values. Certain classes have access to save-boosting talents which shoot them through the roof. The stats and equipment items seem to have a pitifully small effect. This needs some careful review.
I'd say instead of the roll being based on the magical +/- 20 number it should consider the attack and defence numbers in proportion. If your save is half the attack you should expect about 25% of the rolls to save. If equal you should expect around 75% saves. If double you should expect to almost never lose.
Another solution is to remove the save system entirely and only use resistance %'s. This would be a shame, but shouldn't be ignored if the save system can't be made useable and transparent.
At the moment there aren't many ways to modify the save values. Certain classes have access to save-boosting talents which shoot them through the roof. The stats and equipment items seem to have a pitifully small effect. This needs some careful review.