D&D 5E The impact of overkill damage

[FrogReaver]

I agree with you. A spell like Hypnotic Pattern can end a encounter against an overwhelming group of monstrous HP with low Wisdom saves, and suck against low HP creatures like Derro. HP distribution is largely irrelevant, for this conversation

Hypnotic Pattern also does no DPS, no Overkill, but is potentially a high Target Removal threat....how does a focus Solely on DPS and Monster HP range account for Hypnotic Pattern:
it doesnt..that spell is invisible on those metrics.

Which is why we need to utilize many different metrics.

As I noted previously, the HP Distribution part of the discussion was in relation to a specific simulation. One that changing the hp distribution in would cause vastly different results.

I think it's worth making sure we don't mix points about it as general discussion points.

 

[Ovinomancer]

I think you are the one dying on a hill. I asked 1 yes or no question. It shouldn't have turned into whatever this is even if that information was clearly in one of your previous posts. It should have been a quick yes or no and move on. Why you stubbed up on that one question I will never know. The only reason I can assume is that you felt the question was in bad faith, but let me assure you - it was not.

Now, that said I agree there's a point when someone is making unreasonable demands on another and you don't need to engage with those demands - but a single yes or no question doesn't cross that line - especially when someone is telling you they want to understand your position better.

I think going around and around on this isn't beneficial for us or the forum and so let's drop it. I know nothing you have said is going to convince me that someone refusing to answer a single yes or no question like the one I asked was justified. I doubt I'm going to convince you that it wasn't. I doubt anyone else wants to hear anymore about it. So let's agree to disagree?

If it were one yes/no question, but you had, up to that point, asked me why I used f=6, which was clearly explained in the post you quoted and which I answered. You then asked if I accounted for the extra attack from GWM, which was clearly stated in the post that you quoted and which I also answered. When you then asked what builds I tested, which was also clearly stated in the post you had quoted, I still answered you question, only this time I pointed you to the post where all of the answers to all three of your questions were clearly explained. At this point, you accused me of not being willing to discuss, and, indeed, of literally being unwilling to answer any questions (oops, right?) because I pointed to my post, which you had just recently quoted, as the source of the answer. I even told you that I did this because you were asking many questions that were clearly stated in that post and I felt that discussion would be better if you re-read it and answered all of those questions yourself rather than having a back and forth while you continue to ask already answered questions.

Now we're here, where you state that I am acting in an unjustified manner, but that you'll be willing to move forward (presumably continuing to ask questions already answered) if I agree to drop the matter entirely after you get in the last word about how poorly I've behaved. I mean, really?

Sure, I'm willing to jump in and start discussing from this point

So there's 3 portions of GWM
1. -5/+10
2. Bonus action attack on a crit
3. Bonus action attack when you kill in an enemy.

Point 2 alone is about a 10% increase in DPR for a 2 attack fighter. Of note is that point 3 is ignored in nearly all GWM DPR discussions. Then there's the overkill part of the discussion.

(Note: I'm thinking through the problem here and you probably have already calculated some of these values)
So we need to establish what the DPR overkill effect is without GWM
We need need to establish what the DPR is with each portion of the feat both with and without overkill
The also need to factor in any diminishing returns or increasing returns when combining the parts of the feat together (also with overkill)

If I understand you correctly you are taking the -5/+10 part of the feat and applying all overkill damage towards that portion of the feat and then applying the bonus action attack on kill portion independently. Please correct me if I am wrong.

I am suggesting we apply overkill against the bonus action when you kill an enemy part of the feat. Overkill and that effect both happen at the same time and that effect more than offsets overkill and also provide some extra damage. In which case overkill doesn't impact GWM because it has a point never factored into DPR discussions that actually more than fully negates it.

I guess it would matter if your looking at a GWM fighter that uses -5/+10 and one that has the feat but doesn't. That's another part of the reason I was trying to establish what you were talking about.

This betrays a fundamental failure to understand the method I used. And, you reasked the same question about builds at the end. Honestly, just read my previous post and then come up with questions. If you can't be bothered to scroll up and read what I've already said so that you can critique it, the problem is yours, not mine for failing to link for you.

 

[Ovinomancer]

Not a red herring as it's specifically related to the simulation in question and it's important to that discussion.

But I agree that in terms of a mathematical model we just have to estimate the number of kills a PC will have on average in an encounter.

In a 4 round encounter with 4 enemies and 4 identical PC's, each PC would kill 1 enemy on average. The fighter type pc's we are looking at likely do more DPR and so will get a slightly larger proportion of those kills but the last kill also doesn't affect overkill. I think those factors will come close to canceling out. So to me I think 1 kill per encounter is a very fair picture but I'm open to alternative reasonings there.



So if you applied the extra attack on kill effect and overkill first and then the -5/+10 would that mean that the increase in damage was coming from -5/+10 and the bonus action attack did next to nothing? The point is whatever effect you apply last will appear to be the leading cause of the damage improvement. Thoughts?

The argument that the last kill can ignore overkill is flawed. DPR doesn't discount last kill, and overkill is a function of DPR calculations, so until you're going to discount last attack from DPR calculations, overkill is valid. What's happening here is a different set of criteria for evaluation is being used for overkill than what was used for DPR -- this is logically flawed because overkill only matters to DPR calculations. You can't use a different criteria to dismiss overkill and still keep DPR unchanged. Either you find a way to discount DPR for final kills and then ignore overkill on that (and I have no idea how you'd do that), or you don't but also have to keep overkill in the loop.

 

[FrogReaver]

Hypnotic Pattern also does no DPS, no Overkill, but is potentially a high Target Removal threat....how does a focus Solely on DPS and Monster HP range account for Hypnotic Pattern:
it doesnt..that spell is invisible on those metrics.

Which is why we need to utilize many different metrics.

I've done analysis on Hypnotic pattern. It's much better than nearly any amount of damage in nearly every multi enemy encounter (assuming no wake ups). I looked at it in terms of average rounds enemies lost. Which is a pretty powerful metric and can also be tied back into DPR.

I've also done analysis on AOE fireball. Fireball can be translated into single target damage roughly as follows .5+0.5*N where N is the number of enemies you are facing.

I've developed quite a few tools to aid in those kinds of discussions.

 

[Ovinomancer]

As I noted previously, the HP Distribution part of the discussion was in relation to a specific simulation. One that changing the hp distribution in would cause vastly different results.

I think it's worth making sure we don't mix points about it as general discussion points.

No, the CR distribution was coming sideways into frequency of overkill -- it was an attempt to find out how often overkill occurred by randomly scattering hps and then effectively counting killing blows. The red herring is thinking the CR distribution (and therefore hp distribution) matters. It doesn't -- only the frequency of killing blows functionally matters, and in a game with a party the actual hp of the target you're attacking rarely has much to do with it's CR because it's most likely already taken some damage. Which is why I say it only matters when the target hp is between 1 and average damage. This can be undamaged or after four rounds.

 

[FrogReaver]

The argument that the last kill can ignore overkill is flawed. DPR doesn't discount last kill, and overkill is a function of DPR calculations, so until you're going to discount last attack from DPR calculations, overkill is valid. What's happening here is a different set of criteria for evaluation is being used for overkill than what was used for DPR -- this is logically flawed because overkill only matters to DPR calculations. You can't use a different criteria to dismiss overkill and still keep DPR unchanged. Either you find a way to discount DPR for final kills and then ignore overkill on that (and I have no idea how you'd do that), or you don't but also have to keep overkill in the loop.

Huh? The only benefit of overkill is that it allows you to get an attack off against the next enemy. Without that it has no effect at all.

 

[FrogReaver]

No, the CR distribution was coming sideways into frequency of overkill -- it was an attempt to find out how often overkill occurred by randomly scattering hps and then effectively counting killing blows. The red herring is thinking the CR distribution (and therefore hp distribution) matters. It doesn't -- only the frequency of killing blows functionally matters, and in a game with a party the actual hp of the target you're attacking rarely has much to do with it's CR because it's most likely already taken some damage. Which is why I say it only matters when the target hp is between 1 and average damage. This can be undamaged or after four rounds.

The CR distribution was directly related to the sim every time I've brought it up and it matters a great deal there. Maybe you aren't thinking through why. It's because the more attacks you make without getting a kill the less effect overkill has on the average as most rounds you don't have to worry about overkill at all. Higher CR means higher hp, means more rounds to kill, means less effect of overkill on the average. If we aren't simming the rounds taken to kill (or trying to compute them, I am not for any purpose outside the sim) then they don't matter. But just to be clear, they matter a great deal for the sim as I explained here.

 

[Ovinomancer]

Huh? The only benefit of overkill is that it allows you to get an attack off against the next enemy. Without that it has no effect at all.

Nope. This is flawed. Overkill only matters in discussion of DPR. Usual DPR doesn't check to see how much damage is actually applied, it assumes all damage is fully applied. Overkill is the part where you look to see how true that assumption is and thereby modify DPR. Since DPR doesn't really care how fast you get to the next target, because all damage is assumed applied anyway, stating that overkill only matters when switching targets is a flawed analysis, using a criteria set different from how DPR is calculated.

What you're describing is an outcome of well matched overkill -- if you do very little overkill loss, you will maximize your DPR calculation. You do this by not blowing past target hp. But, overkill doesn't stop mattering or only mattering when you can switch targets -- it matters in all cases so long as DPR assuming all damage applies.

 

[Ovinomancer]

The CR distribution was directly related to the sim every time I've brought it up and it matters a great deal there. Maybe you aren't thinking through why. It's because the more attacks you make without getting a kill the less effect overkill has on the average as most rounds you don't have to worry about overkill at all. Higher CR means higher hp, means more rounds to kill, means less effect of overkill on the average. If we aren't simming the rounds taken to kill (or trying to compute them, I am not for any purpose outside the sim) then they don't matter. But just to be clear, they matter a great deal for the sim as I explained here.

Yes, it's because frequency of killing blows. Which I said, and which CR distribution is a red herring for, largely because the sim treats targets as bags of hitpoints the single PC chews through to arrive at the frequency of killing blows. The important consideration is this frequency of killing blows -- this is the only place overkill happens. Arguing over the which CR distribution is better is therefor a red herring -- it's an artificial method of generating frequency of killing blows that doesn't look like any real game but creates arguments as if it does matter.

What matters is making an assumption about how often killing blows occur and seeing what that does to overkill. This is why I used your suggestion of 1 kill per combat, paired with an assumption of average 3 rounds per combat and then looking to see how many attacks the PC builds do in 3 rounds, 1 of which is a killing blow. Viola, frequency of killing blows is 1 out of 6 attacks. The higher f is (the denominator), the better overkill looks for all builds.

 

[FrogReaver]

Nope. This is flawed. Overkill only matters in discussion of DPR. Usual DPR doesn't check to see how much damage is actually applied, it assumes all damage is fully applied. Overkill is the part where you look to see how true that assumption is and thereby modify DPR. Since DPR doesn't really care how fast you get to the next target, because all damage is assumed applied anyway, stating that overkill only matters when switching targets is a flawed analysis, using a criteria set different from how DPR is calculated.

What you're describing is an outcome of well matched overkill -- if you do very little overkill loss, you will maximize your DPR calculation. You do this by not blowing past target hp. But, overkill doesn't stop mattering or only mattering when you can switch targets -- it matters in all cases so long as DPR assuming all damage applies.

Now you've lost me. Let's say you there are 2 PC's with identical DPR. The first has 2 attacks and kills the enemy on his first attack and has nothing to swing at with his 2nd. The other has 1 attack and kills the enemy on his first attack. Why would you not be counting that 2nd unmade attack in any numbers? It matters just as much as the single attack character wasting more damage on his last attack. That is Attacks not made that could be made if more enemies were present need accounted for. Whether you count it as overkill or name it something different it has the same effect.