Survivability vs Sustainability (what's wrong with Guillotine)
DNA3000
Member, Guardian Guardian › Posts: 19,702 Guardian
Tl;DR: In one sense Guillotine 2.0's healing is just as good as before, but in a more important sense it is worse, and in a way that is invisible to Kabam's metrics.
Healing is actually, in game design terms, extremely complex. Far too complex to cover entirely here. But I wanted to highlight one aspect of healing that I believe is not commonly discussed or well understood and I think is at the heart of *one* thing wrong with Guillotine's update: what happened to her healing.
Kabam has already mentioned their desire to be cautious with Guillotine's new healing mechanics, but that's only half the story. The other half is (in my opinion) a perspective error that takes some time to explain. I've mentioned some of this elsewhere, but I thought I would articulate the entire story in one place. To start, healing is generally non-proportional. What does that mean?
When we think of conventional damage mitigation, we often thing in terms of stuff like armor or resistances. If you have 50% resistance, you take 50% less damage. What does this mean? Absent healing, if you take half the damage, you live twice as long. This is true everywhere, it is true in RoL, it is true in the Abyss. No matter what your health, and no matter how much damage you're taking, 50% resistance means half the damage taken, and you live twice as long. Proportional damage mitigation is relatively easy to understand for that reason. Healing doesn't generally work that way. There are proportional healing mechanisms in the game (think Mr Sinister's heal on crit, or Sorcerer Supreme's heal on block) but most healing is not proportional to incoming damage. When you see a heal, you cannot say something like "this will allow me to survive twice as long." If you're healing 100 h/s and you're taking 200 dps, you'll survive twice as long. If you're taking 150 dps you'll survive three times as long. And if you're taking 99 dps, you're immortal.
So how do we consider the value of healing? We need to construct models of damage and healing that let us say reasonably useful things about healing. One model is what I generally refer to as the survivability model. It goes like this: we model a fight as two dps cannons pointed at each other. My champ fires dps at the opponent and the opponent fires dps at me. Which ever side's health bar reaches zero first loses. In such a model, what is the effect of healing? Well, suppose I'm shooting 100 dps at the opponent and he's shooting 150 dps at me. Assuming our health bars are equal (for discussion purposes: of course this is rarely the case) I'm going to lose this fight. But if I have 60 h/s of healing, in a sense I'm reducing the opponent dps cannon by 60 dps. Instead of 150 dps I'm seeing 90 dps of net damage. Now I win. In this model, healing is a *rate* and it reduces the incoming damage *rate*. Higher rate is better. This is true no matter how fast or how slow the fight itself is.
This seems like a reasonable model for judging healing, but there's another model that is equally useful but generates different results. Suppose I'm doing multiple fights, one after the other. I might lose health in one of them, but still need to do more fights. I could ask the question: if I lose health in one fight, how much of that health can I regain with healing? In a sense, if the first model asks "will I win" the second model asks "do I need to spend consumables?" If I lose health and I can't get it back, given how much health I expect to lose later I might need to use potions. If I can get enough of it back, I might not need to. In this second model, healing is not a rate. Healing is an amount of health. This amount may vary based on conditions, but we're considering how much health we could regain in a fight to prepare for the next one. I generally refer to this as the sustainability model of health recovery.
Here's where the two models diverge. Suppose a champion is capable of healing for 1000 points of health. Maybe it is a one-time heal, like Iron Man has. As I mentioned above, we can roughly approximate Guillotine's heal in this way as well. The details are not important. In the survivability model how valuable that heal is depends on how long the fight lasts. If the fight lasts 100 seconds, that heal is worth 10 h/s. But if we can speed up the fight to 50 seconds, the heal is now worth 20 h/s. In effect, shortening the fight improves the value of the heal. In other words, if we buff the damage of the champion, according to the survivability model we also strengthen any heals they have, even though we didn't touch the heal at all.
But in the sustainability model, that's not true. In the sustainability model, 1000 points of heal is 1000 points of heal. It will let us erase a 1000 damage mistake. It acts like a free 1000 point health potion. Whether the fight lasts 100 seconds or 10 seconds, that heal is worth exactly the same. If we buff the damage of the champion, according to the sustainability model the heal gets no better and no worse. Its value remains the same.
Heals are just heals, right? One of these models has to be wrong? Actually, no: while it is true that the game just implements heals, how we treat heals as players can differ in material ways, because healing has situational value, and we often have control over those situations. We can choose to play more aggressively to end fights faster when we have high health, and more conservatively to prevent death when we have low health. How we view health can feed back into the game and change the way we benefit from healing.
Getting back to Guillotine. According to Kabam, Guillotine 2.0's healing is lower than OG Guillotine, but only by a tiny amount. Calculations say G2.0's healing should be substantially lower. Although the actual calculations are complex and factor in a lot of things, fundamentally Guillotine's heal is based on how much damage you deal, which is limited by the health bar of the opponent. You can't deal more damage than that (to a first order approximation) and thus you can't heal more than a percentage of that amount. In fact, in any fight you win, both OG Guillotine and Guillotine 2.0 have in a sense a fixed heal, that is spread out over the whole fight. That value is lower for Guillotine 2.0 than Guillotine 1.0. So G2.0's heal should be lower. So what is the cause of the discrepancy? It is that the Kabam metrics are looking at healing rate, not healing. Healing is expressed in h/s. So what's causing G2.0's heal to be almost equal to G1.0, and in effect stronger than calculations suggest? It is that G2.0's damage is higher. She's ending fights quicker, so while she is healing less, she's recovering that less health in shorter fights, which makes her healing per second higher.
So what? I just said there's two ways to look at healing, and they both have merits. G2.0 is almost as survivable as G1.0 by one model. Why isn't that good enough? Aren't both models equally valid? Out of context, yes, they are. Both offer valid ways to look at healing. But there's a larger question, and it is this: was OG Guillotine's heal *preserved* for *existing players?* I content it was not. OG Guillotine had a burst high magnitude heal. Everyone agrees it was not reliable. A bursty unreliable heal is not a good survivability tool. The survivability model presumes we can count on that heal to continuously offset incoming damage. If it doesn't for an extended period of time, you could be dead. More reasonably, the player will need to change the way they play to become more conservative to counter the fact that they will lose a lot more health.
But they will eventually get that health back. RNG will eventually average out. OG Guillotine's heal was not a good survivability tool, but it was a better sustainability tool. OG Guillotine's heal was like randomly having free potions fall from the sky. You had to play her carefully because you couldn't rely on her heal on a time scale of second by second fighting, but you could count on getting free potions erasing damaging mistakes constantly on a time scale of whole fights.
So preserving Guillotine's heal wasn't just about picking some random metric for healing and trying to hit it again. It was about preserving the use case for that healing: how players were seeing it and using it. And given how it worked, I believe most players (who played Guillotine at all) saw the healing as a sustainability tool. From that perspective, Guillotine 2.0's heal is vastly worse. As a survivability tool it is somewhat better, but not because the heal itself is better but rather because G2.0's damage is itself higher.
For some players, this makes no difference. She had a heal, she still has a heal, it works, and that's all that matters. I think that is how Kabam is looking at it as well. But in this case, I believe Kabam is looking at the wrong thing. The thing they are looking at is not a bad thing, it is just not a thing that is consistent with the champion they were buffing. And that's why a lot of players just don't see the value of the new heal, but some do. Some are looking at the heal as a survivability tool where it is about as good as it was before, while others are looking it as a sustainability tool and seeing it is far worse than it was (even if they wouldn't characterize their views in that way).
I believe if we're going to try to avoid unnecessarily disappointing players, we should be trying to avoid taking away use cases from existing champions whenever possible. I believe that was the lesson from the Hood incident. And that means we need to look beyond superficial metrics, and into how players play champions. Calculations help, but only when they calculate the right things. Guillotine's buff did not look at the right thing for healing, and while that's not the only problem with the Guillotine update, I believe this is why this one thing went wrong (at least, in my opinion).
I suspect for those that already believe Guillotine's heal was broken, this may provide an interesting framework for understanding why it might have been broken, but for those that think her heal is perfectly fine and the change was thus perfectly fine, this won't be convincing. For those people, I would ask to consider this hypothetical. Suppose the devs added a new Willpower-like mastery that would heal champions constantly. And then they deleted all the health potions from the game. I'm sure there exist players that would love this and benefit from this, and I'm sure there's a way to tweak the numbers so it averages to the same benefit overall. But the vast majority of players would hate this, and it isn't hard to see why. Healing is not just healing: *how* we receive this healing has tactical and strategic importance, and changing the way we get healing is a significant change, even if we hit a metric on a graph that says things are still the same.
For those that are interested and haven't fallen asleep yet, this is sort of part two of my Guillotine series. Part one was my original analysis of the buff which was posted in the thread below:
https://forums.playcontestofchampions.com/en/discussion/281706/guillotines-update-is-a-miss-heres-why/p1.
In that thread another poster (Dr. Zola) posed what I consider to be an important question: if these are the things I think went wrong with an update, what is my opinion on what we should do to get updates right. For those that don't want to read the whole thread, that post, which I consider to be part 1.5, is here: https://forums.playcontestofchampions.com/en/discussion/comment/1930944/#Comment_1930944
Healing is actually, in game design terms, extremely complex. Far too complex to cover entirely here. But I wanted to highlight one aspect of healing that I believe is not commonly discussed or well understood and I think is at the heart of *one* thing wrong with Guillotine's update: what happened to her healing.
Kabam has already mentioned their desire to be cautious with Guillotine's new healing mechanics, but that's only half the story. The other half is (in my opinion) a perspective error that takes some time to explain. I've mentioned some of this elsewhere, but I thought I would articulate the entire story in one place. To start, healing is generally non-proportional. What does that mean?
When we think of conventional damage mitigation, we often thing in terms of stuff like armor or resistances. If you have 50% resistance, you take 50% less damage. What does this mean? Absent healing, if you take half the damage, you live twice as long. This is true everywhere, it is true in RoL, it is true in the Abyss. No matter what your health, and no matter how much damage you're taking, 50% resistance means half the damage taken, and you live twice as long. Proportional damage mitigation is relatively easy to understand for that reason. Healing doesn't generally work that way. There are proportional healing mechanisms in the game (think Mr Sinister's heal on crit, or Sorcerer Supreme's heal on block) but most healing is not proportional to incoming damage. When you see a heal, you cannot say something like "this will allow me to survive twice as long." If you're healing 100 h/s and you're taking 200 dps, you'll survive twice as long. If you're taking 150 dps you'll survive three times as long. And if you're taking 99 dps, you're immortal.
So how do we consider the value of healing? We need to construct models of damage and healing that let us say reasonably useful things about healing. One model is what I generally refer to as the survivability model. It goes like this: we model a fight as two dps cannons pointed at each other. My champ fires dps at the opponent and the opponent fires dps at me. Which ever side's health bar reaches zero first loses. In such a model, what is the effect of healing? Well, suppose I'm shooting 100 dps at the opponent and he's shooting 150 dps at me. Assuming our health bars are equal (for discussion purposes: of course this is rarely the case) I'm going to lose this fight. But if I have 60 h/s of healing, in a sense I'm reducing the opponent dps cannon by 60 dps. Instead of 150 dps I'm seeing 90 dps of net damage. Now I win. In this model, healing is a *rate* and it reduces the incoming damage *rate*. Higher rate is better. This is true no matter how fast or how slow the fight itself is.
This seems like a reasonable model for judging healing, but there's another model that is equally useful but generates different results. Suppose I'm doing multiple fights, one after the other. I might lose health in one of them, but still need to do more fights. I could ask the question: if I lose health in one fight, how much of that health can I regain with healing? In a sense, if the first model asks "will I win" the second model asks "do I need to spend consumables?" If I lose health and I can't get it back, given how much health I expect to lose later I might need to use potions. If I can get enough of it back, I might not need to. In this second model, healing is not a rate. Healing is an amount of health. This amount may vary based on conditions, but we're considering how much health we could regain in a fight to prepare for the next one. I generally refer to this as the sustainability model of health recovery.
Here's where the two models diverge. Suppose a champion is capable of healing for 1000 points of health. Maybe it is a one-time heal, like Iron Man has. As I mentioned above, we can roughly approximate Guillotine's heal in this way as well. The details are not important. In the survivability model how valuable that heal is depends on how long the fight lasts. If the fight lasts 100 seconds, that heal is worth 10 h/s. But if we can speed up the fight to 50 seconds, the heal is now worth 20 h/s. In effect, shortening the fight improves the value of the heal. In other words, if we buff the damage of the champion, according to the survivability model we also strengthen any heals they have, even though we didn't touch the heal at all.
But in the sustainability model, that's not true. In the sustainability model, 1000 points of heal is 1000 points of heal. It will let us erase a 1000 damage mistake. It acts like a free 1000 point health potion. Whether the fight lasts 100 seconds or 10 seconds, that heal is worth exactly the same. If we buff the damage of the champion, according to the sustainability model the heal gets no better and no worse. Its value remains the same.
Heals are just heals, right? One of these models has to be wrong? Actually, no: while it is true that the game just implements heals, how we treat heals as players can differ in material ways, because healing has situational value, and we often have control over those situations. We can choose to play more aggressively to end fights faster when we have high health, and more conservatively to prevent death when we have low health. How we view health can feed back into the game and change the way we benefit from healing.
Getting back to Guillotine. According to Kabam, Guillotine 2.0's healing is lower than OG Guillotine, but only by a tiny amount. Calculations say G2.0's healing should be substantially lower. Although the actual calculations are complex and factor in a lot of things, fundamentally Guillotine's heal is based on how much damage you deal, which is limited by the health bar of the opponent. You can't deal more damage than that (to a first order approximation) and thus you can't heal more than a percentage of that amount. In fact, in any fight you win, both OG Guillotine and Guillotine 2.0 have in a sense a fixed heal, that is spread out over the whole fight. That value is lower for Guillotine 2.0 than Guillotine 1.0. So G2.0's heal should be lower. So what is the cause of the discrepancy? It is that the Kabam metrics are looking at healing rate, not healing. Healing is expressed in h/s. So what's causing G2.0's heal to be almost equal to G1.0, and in effect stronger than calculations suggest? It is that G2.0's damage is higher. She's ending fights quicker, so while she is healing less, she's recovering that less health in shorter fights, which makes her healing per second higher.
So what? I just said there's two ways to look at healing, and they both have merits. G2.0 is almost as survivable as G1.0 by one model. Why isn't that good enough? Aren't both models equally valid? Out of context, yes, they are. Both offer valid ways to look at healing. But there's a larger question, and it is this: was OG Guillotine's heal *preserved* for *existing players?* I content it was not. OG Guillotine had a burst high magnitude heal. Everyone agrees it was not reliable. A bursty unreliable heal is not a good survivability tool. The survivability model presumes we can count on that heal to continuously offset incoming damage. If it doesn't for an extended period of time, you could be dead. More reasonably, the player will need to change the way they play to become more conservative to counter the fact that they will lose a lot more health.
But they will eventually get that health back. RNG will eventually average out. OG Guillotine's heal was not a good survivability tool, but it was a better sustainability tool. OG Guillotine's heal was like randomly having free potions fall from the sky. You had to play her carefully because you couldn't rely on her heal on a time scale of second by second fighting, but you could count on getting free potions erasing damaging mistakes constantly on a time scale of whole fights.
So preserving Guillotine's heal wasn't just about picking some random metric for healing and trying to hit it again. It was about preserving the use case for that healing: how players were seeing it and using it. And given how it worked, I believe most players (who played Guillotine at all) saw the healing as a sustainability tool. From that perspective, Guillotine 2.0's heal is vastly worse. As a survivability tool it is somewhat better, but not because the heal itself is better but rather because G2.0's damage is itself higher.
For some players, this makes no difference. She had a heal, she still has a heal, it works, and that's all that matters. I think that is how Kabam is looking at it as well. But in this case, I believe Kabam is looking at the wrong thing. The thing they are looking at is not a bad thing, it is just not a thing that is consistent with the champion they were buffing. And that's why a lot of players just don't see the value of the new heal, but some do. Some are looking at the heal as a survivability tool where it is about as good as it was before, while others are looking it as a sustainability tool and seeing it is far worse than it was (even if they wouldn't characterize their views in that way).
I believe if we're going to try to avoid unnecessarily disappointing players, we should be trying to avoid taking away use cases from existing champions whenever possible. I believe that was the lesson from the Hood incident. And that means we need to look beyond superficial metrics, and into how players play champions. Calculations help, but only when they calculate the right things. Guillotine's buff did not look at the right thing for healing, and while that's not the only problem with the Guillotine update, I believe this is why this one thing went wrong (at least, in my opinion).
I suspect for those that already believe Guillotine's heal was broken, this may provide an interesting framework for understanding why it might have been broken, but for those that think her heal is perfectly fine and the change was thus perfectly fine, this won't be convincing. For those people, I would ask to consider this hypothetical. Suppose the devs added a new Willpower-like mastery that would heal champions constantly. And then they deleted all the health potions from the game. I'm sure there exist players that would love this and benefit from this, and I'm sure there's a way to tweak the numbers so it averages to the same benefit overall. But the vast majority of players would hate this, and it isn't hard to see why. Healing is not just healing: *how* we receive this healing has tactical and strategic importance, and changing the way we get healing is a significant change, even if we hit a metric on a graph that says things are still the same.
For those that are interested and haven't fallen asleep yet, this is sort of part two of my Guillotine series. Part one was my original analysis of the buff which was posted in the thread below:
https://forums.playcontestofchampions.com/en/discussion/281706/guillotines-update-is-a-miss-heres-why/p1.
In that thread another poster (Dr. Zola) posed what I consider to be an important question: if these are the things I think went wrong with an update, what is my opinion on what we should do to get updates right. For those that don't want to read the whole thread, that post, which I consider to be part 1.5, is here: https://forums.playcontestofchampions.com/en/discussion/comment/1930944/#Comment_1930944
31
Comments
You have outdone yourself with this one DNA 3000
Enjoyed reading every word
The reality is that with her it really was all the luck of the draw, especially if you had acquired enough Souls and a high signature level, she kept you alive in the fight like you said and it was exactly like getting free potions mid-fight but that doesn't mean that it made her a perfect champion in that regard.
Something I pointed out even when the initial Champion's Choice vote began was that a lot of what she was missing at least in terms of community feedback was some mystical utility that in retrospect could have easily been add it on through a moderate update, but again determining factors played a part into how she won and people decide to rebuild her from the ground up, so now that we understand the consequences of what an overhaul can do if things don't work out as intended, I still believe that there is some possibility of potential but in order to do so it's going to require some tweaks that can really compliment her damaged but at the same time add some stability even if it's a fixed rate with a fixed amount.
From what I understand G2.0 has close to the same healing rate as per the model described as G1.0 because she ends fights faster while healing the same amount i.e. she heals the same amount of health but would take less damage. Is this correct? If so, does this account for the fact that to end fights faster, the playstyle would have to change which could mean that more damage is being taken, for example the fastest way to end fights with G2.0 is parry heavy as much as possible to stack bleed curses and then lauch a special to stack bleeds, but this would mean higher block damage as well. On the other hand if one was to play normally it wouldn't end fights faster but there would still be damage taken unless she's played perfectly and no block damage is taken while attempting to intercept with MLLLM combos.
She's got roughly the same average health recovery per second.
But her average damage (apart from her SP3) is improved, meaning that fights are shorter, i.e. they've got fewer seconds in them.
So she's recovering health at the same rate per second; but for a smaller number of seconds.
Therefore less health is recovered in all.
One of DNA's points is: this healing visibly isn't the same: Guilly 2.0 is healing less. But if your metric for measuring regeneration is 'how much do you heal per second" it looks just as good. His post is partly about the fact that Kabam is using the wrong metric to measure success.
(Some companies use a strange metric called 'customer satisfaction', for example...)
I find it particularly odd that their response (the team are satisfied...) suggests that they're choosing to ignore 'customer satisfaction' in favour of 'designer satisfaction'.
I'm going to provide an analogy, to see if it helps Kabam understand what they've done to the players by changing Guilly's regen, and why people are unhappy with it. I'm purely focusing on the regen - I'm aware we all feel there are other issues!
So: people complain about RNG a lot. But the reality is that they love it. We all do. It's a basic human psychological response.
People like to gamble. They like to think they can beat RNG. They remember the times they win, and selectively forget the times they lose. When they win, their brains release dopamine and feel good.
Anyway, Guillotines original regeneration was always like a casino slot machine. Often nothing would happen. Often it wouldn't achieve much. But sometimes they'd have crazy luck, and see huge wins.
Now, the slot machine has been replaced with a treadmill that dispenses money. Keep walking, and money comes out. Run and it comes out faster. Stop, and nothing happens at all. But there's no highs and no lows. No RNG. That's not gambling - that's a job.
Mathematically, the players might recover almost identical amounts of health from 'new' and 'old' Guilly. But the experience they have doing it has changed substantially.
Guillotines high/low 'gambling-type' regeneration has been replaced with very dull 'just a grind' regeneration and that's a big part of why people don't like it. There's no risk, no reward, nothing interesting happening.
So that's my analogy; Kabam. I hope it helps
But seriously this was a good post
In this case I'm inclined to bet on me, and it isn't that Kabam's numbers are necessarily wrong (they could be, but they don't have to be) it is that they are looking at the wrong numbers. They think they are looking at the right numbers, obviously, and Math is Hard so my goal was to present a reasonable argument that most people could follow that justified my position that they are looking at the wrong thing, and also just how much margin for error there is when you look at the wrong thing. All sorts of complicated factors can influence what metrics to look at, affected by playstyle, by situational combat issues, etc.
Consider this: if you wanted to know if a champion had "better" healing than another champion, which statistic would you look at first? If you're tunnel visioned on math, you might look at h/s. You might even look at heal/fight, as I suggested. But the *first* metric you might look at, to give those numbers context, is how often Guillotine dies in combat. Because if G2.0 dies more often than G1.0 did, her h/s being similar *clearly* has an issue. The *second* metric you might look at is potions used to heal the champ. Because once again, if G2.0's healing looks similar to G1.0 in your metric, but players are pouring a lot more potions into G2.0, then once again you should distrust the metric you're looking at as measuring the right thing.
If I wanted to prove to you that G2.0's healing was better or approximately as good, that's how I would start. Because the factors you're talking about *can* affect things a lot, and my first goal would be to ensure that I was making as close to an apples to apples comparison as possible. That means making sure players are getting similar outcomes, not just similar stats.