D&D 5E A benchmark for Encounter Deadliness

[GMMichael]

I once managed to pull a TPK against 7th level PCs with a single kobold (and a cut rope)... no system is perfect.

But being able to quickly judge if a fight could be deadly is helpful. Then you might know ahead of time, hey maybe these sea spawn really wouldn't fight to the death (unlikely since they are controlled by the aboleth, but you get the idea).

You're right on one hand: it's another tool. Use it if it's helpful. But on the other, there might not be a tool that can help a DM say, "hmm, I might kill my players with this encounter." Jgsugden killed an entire party with one kobold (level unmentioned). Which calculator could have predicted that?

It makes sense that, in a game where a lone halfling could easily dispatch a stone giant given enough levels and magic, DMs should have a little guidance in how to plot (prevent?) such things. So, the DMG, AngryGM, and Mike Shea provide systems for doing so. But to need such a system in the first place is to 1) assume combat must happen, and that someone has to die, and 2) ignore the introductory material of several books that states that GMs are in charge, and their job is to make sure everyone has fun. (I think rule 0 is in there somewhere.)

So "look in the mirror" points a DM back to the forest, if you will, if she's been staring at the trees. Which actually is helpful, if it prevents a DM from focusing on dice and ignoring the table, then looking up to realize that a bad session was had by all.

 

[dave2008]

You're right on one hand: it's another tool. Use it if it's helpful. But on the other, there might not be a tool that can help a DM say, "hmm, I might kill my players with this encounter." Jgsugden killed an entire party with one kobold (level unmentioned). Which calculator could have predicted that?

It makes sense that, in a game where a lone halfling could easily dispatch a stone giant given enough levels and magic, DMs should have a little guidance in how to plot (prevent?) such things. So, the DMG, AngryGM, and Mike Shea provide systems for doing so. But to need such a system in the first place is to 1) assume combat must happen, and that someone has to die, and 2) ignore the introductory material of several books that states that GMs are in charge, and their job is to make sure everyone has fun. (I think rule 0 is in there somewhere.)

So "look in the mirror" points a DM back to the forest, if you will, if she's been staring at the trees. Which actually is helpful, if it prevents a DM from focusing on dice and ignoring the table, then looking up to realize that a bad session was had by all.

In my history as a DM (30+ years, but many of those early years I was really bad at it) I have tried many different "styles" if you will. Of course I understand that the party can completely avoid a fight, or come up with great tactics or strategy to make it easy, or wade into it and make complete idiots of themselves. Any good DM doesn't need those things stated, those are a given. But I can't see how be able to quickly determine if a battle could be deadly if the dice fall randomly is a bad thing or why you feel the need to rail against.

Also, some DMs have fun by not knowing everything a head of time. Otherwise, why should the DM roll dice at all. Just declare hits and damage and success based on a fun story?

 

[clearstream]

I just read this article over at Mike Shea's website:
The Lazy Encounter Benchmark, a Simple Benchmark for D&D Encounter Deadliness

It's a nice method, and can be simplified further. One might observe that only two kinds of encounter really matter in 5e: attritional (hard or easier) and lethal (deadly+) due to character recuperative powers. Deadly encounters are those presumed to offer a palpable risk of character death. So with the caveat that the non-deadly encounters make sense for your game, the crucial question becomes - is my encounter deadly?

Crunching the numbers (finding the CRs that yield the XP for the encounter thresholds) you find that CL/CR >2 at every level (character level over CR for XP threshold) is always attritional (i.e. less than deadly).

Therefore Shea's guideline can be simplified to ignore the tier 1 adjustment. Simply CL/CR = ≤2 = potentially deadly.

Regards what it means to offer a chance off death, for my campaign I've defined deadly as a 1:12 chance of death in the encounter per character. That is consistent with my data over two years of campaign... but could differ wildly from yours! Attritional are then an order of magnitude lower - 1:120 - but parties go through a great many more such encounters. The two together yield decent expectations for deaths based on cumulative probability over expected number of encounters, given also my assumption that revival is available about half the time (so it takes on average two deaths to kill a character). Expectations for lethality scale and revival should scale together, of course.

 

[Cap'n Kobold]

It's a nice method, and can be simplified further. One might observe that only two kinds of encounter really matter in 5e: attritional (hard or easier) and lethal (deadly+) due to character recuperative powers. Deadly encounters are those presumed to offer a palpable risk of character death. So with the caveat that the non-deadly encounters make sense for your game, the crucial question becomes - is my encounter deadly?

Crunching the numbers (finding the CRs that yield the XP for the encounter thresholds) you find that CL/CR >2 at every level (character level over CR for XP threshold) is always attritional (i.e. less than deadly).

Therefore Shea's guideline can be simplified to ignore the tier 1 adjustment. Simply CL/CR = ≤2 = potentially deadly.

Regards what it means to offer a chance off death, for my campaign I've defined deadly as a 1:12 chance of death in the encounter per character. That is consistent with my data over two years of campaign... but could differ wildly from yours! Attritional are then an order of magnitude lower - 1:120 - but parties go through a great many more such encounters. The two together yield decent expectations for deaths based on cumulative probability over expected number of encounters, given also my assumption that revival is available about half the time (so it takes on average two deaths to kill a character). Expectations for lethality scale and revival should scale together, of course.

I'd say that confusing "deadly encounter" meaning "high attrition/difficulty" and it meaning "X% chance of character death may cause problems.
First encounter of the day and your players aren't expecting many more? Your "deadly" encounter is unlikely to have anyone die.
10th encounter of the day? Your deadly encounter is going to have a much higher chance of actual character death.

 

[clearstream]

I'd say that confusing "deadly encounter" meaning "high attrition/difficulty" and it meaning "X% chance of character death may cause problems.
First encounter of the day and your players aren't expecting many more? Your "deadly" encounter is unlikely to have anyone die.
10th encounter of the day? Your deadly encounter is going to have a much higher chance of actual character death.

I think this would be a concern, were the game played that way. I've found and heard reported that for most groups, number of encounters is low enough that you can - as a rough rule of thumb - treat each independently. Bearing in mind we are talking rules of thumb here, not strictly-applicable-in-every-case values.

Also the game is moreorless on easy mode out of the box. I use a longer rests system that pushes my players harder, and I still find the rule of thumb about right across most of my encounters.

 

[dave2008]

Crunching the numbers (finding the CRs that yield the XP for the encounter thresholds) you find that CL/CR >2 at every level (character level over CR for XP threshold) is always attritional (i.e. less than deadly).

Therefore Shea's guideline can be simplified to ignore the tier 1 adjustment. Simply CL/CR = ≤2 = potentially deadly.

Love it! However, can you clarify two things:

1. do you mean total character levels and total monster CRs?
2. What do you mean by CR for XP threshold

 

[Asensur]

This is the method I use.

(Total sum of CRs*4)/(total sum of levels of PCs)=difficulty

<=0.25 trivial
<=0.5 very easy
<=0.75 easy
<=1 medium
<=1.25 hard
<=1.5 very hard
Higher: avoid if possible or alternate win condition

 

[NotAYakk]

The funny thing is that all of the fancy XP math is very close to adding up CR, but not quite.

Most of the difference is in sub-level-5 CR, and that groups of monsters are a bit weaker than they would be from naive CR adding. (Two CR 5s are more like a CR 9 then a 10, 3 are more like a CR 13, etc)

The plan he put up also puts a higher threshold for "deadly" than the DMG deadly is. Which is legit, but should be made clear.

A medium encounter is CR matching average player levels with 4 PCs. A deadly encounter has 2x the XP of that, but that is only 1.4x the CR as CR -> XP is roughly quadratic. Mike's deadly is 2x CR of that.

It is possible that this was done intentionally. It is also possible Mike made a math error. I cannot tell, because how Mike worked out this rule of thumb is not told to us. Was it trial and error? Did he do math? Did he mix the two, and give us a guideline that is out of whack?

 

[UngeheuerLich]

I like the Idea of splitting the encounter. Actually that is what allows for 6-8 fights per day. Also sometimes it is your responsibility as a DM to split the encounter in two halves. Sometimes it is the party who might initiate the split.
Even if the sample encounter is split into two parts by luring half of them away and kill them making terrible noise, the remaining orcs will usually arrive to late to enter the fight in the same encounter. Now you will have two encounters back to back and so you will have a natural 2 encounter before short rest sequence.

 

[Matrix Sorcica]

Link?

Part 1
Part 2
Example 1
Example 2