Unearthed Arcana WotC's Mearls Presents A New XP System For 5E In August's Unearthed Arcana

I have to say...not terribly impressed with this month's article...

http://media.wizards.com/2017/dnd/downloads/UA-ThreePillarXP.pdf

I think my thread should be closed. So I repeat myself here:
Streamlined and probably very useful. I think level 1 and 2 should get an accelerator. Should be considered tier 0 probably.

 

[clearstream]

...discussion about encounters per Adventuring Day

So here is the number crunching. I simply divided encounter XP at each level and risk threshold into adventuring day XP. At the bottom are means, modes and medians. And the correlation of medium to hard encounters, which is significant and tight meaning that they are good predictors of each other (from a game design point of view, it suggests that the table is coherent).

Level Easy Medium Hard Deadly
1 12.0 6.0 4.0 3.0
2 12.0 6.0 4.0 3.0
3 16.0 8.0 5.3 3.0
4 13.6 6.8 4.5 3.4
5 14.0 7.0 4.7 3.2
6 13.3 6.7 4.4 2.9
7 14.3 6.7 4.5 2.9
8 13.3 6.7 4.3 2.9
9 13.6 6.8 4.7 3.1
10 15.0 7.5 4.7 3.2
11 13.1 6.6 4.4 2.9
12 11.5 5.8 3.8 2.6
13 12.3 6.1 4.0 2.6
14 12.0 6.0 3.9 2.6
15 12.9 6.4 4.2 2.8
16 12.5 6.3 4.2 2.8
17 12.5 6.4 4.2 2.8
18 12.9 6.4 4.3 2.8
19 12.5 6.1 4.1 2.8
20 14.3 7.0 4.7 3.1

mean 13.2 6.6 4.4 2.9
mode 12.0 6.0 4.0 3.0
median 13.0 6.5 4.3 2.9
correll 0.97091184

The real number of encounters per day using the adventuring day XP budgets is 4.4 to 6.6 i.e. about one and a half encounters fewer than advertised. I suspect this is a root cause of comments that 5e has an "easy" difficulty setting. For mass-market games it usually is correct to dial down the difficulty. Although ideally there should be a way supplied to dial it back up again. The problem might not lie in the adventuring day XP of course: it could lie in the monster CRs i.e. the XP may be paying for too few monsters. That could matter because if it is the case, increasing the XP serves to accelerate character advancement without appropriately increasing the risk.

What I'd like to do is figure out a simple method to provide a "hard" difficulty setting for 5e. We need to know if monster CRs need tweaking, which I think we can tell from the XP Thresholds table perhaps using Kobold Fight Club? Ideally, they won't, in which case all we'll need to do is up the Adventuring Day XP... perhaps by a quarter?

 

[FormerlyHemlock]

The real number of encounters per day using the adventuring day XP budgets is 4.4 to 6.6 i.e. about one and a half encounters fewer than advertised.

Excuse me, don't you mean 2.9 to 6.6? 4.4 to 6.6 is the Medium band only; 2.9 to 6.6 is the Medium/Hard band.

Remember, those numbers you're computing ratios for are floors, not medians.

What I'd like to do is figure out a simple method to provide a "hard" difficulty setting for 5e. We need to know if monster CRs need tweaking, which I think we can tell from the XP Thresholds table perhaps using Kobold Fight Club? Ideally, they won't, in which case all we'll need to do is up the Adventuring Day XP... perhaps by a quarter?

CR is lossy--this is immediately apparent as soon as you examine the DMG rules for computing CR. There is no way to derive a genuinely reliable difficulty metric from CR because it's already missing too much information. Furthermore, playstyle matters. There's an enormous difference between a T-Rex being run by a DM who roleplays a T-Rex's likely behavior given its goals and low Intelligence, vs. a DM who runs a T-Rex with the mind of a tactical battlecomputer focused on destroying the PCs at any cost. The first T-Rex can be satisfied if you just give it a horse to eat; the second T-Rex is going to leverage its high speed (50'), reach attacks, and grappling bite to strafe the PCs and pick off the weakest-looking member of the party. If it gets lucky on geography, that second T-Rex could potentially TPK a melee-oriented 7th level party.

But the things that would enable the T-Rex to do that aren't valued at all by the CR calculation. A T-Rex with 10' move and no grapple would have exactly the same CR by DMG rules.

I've tossed around potential metrics in the past, like say a Champion rating which says how many Nth level Champions you'd need on average to have a [better than 50%? essentially 100%? somewhere in between] chance of winning a computer simulation of a given fight/adventure. I haven't found a metric that has enough predictive value to make everyone happy with it, but I am at least pretty sure that it's possible to do much better than CR.

 

[Ristamar]

What I'd like to do is figure out a simple method to provide a "hard" difficulty setting for 5e. We need to know if monster CRs need tweaking, which I think we can tell from the XP Thresholds table perhaps using Kobold Fight Club? Ideally, they won't, in which case all we'll need to do is up the Adventuring Day XP... perhaps by a quarter?

I've leaned toward simply awarding less-than-standard XP for encounters in "short" adventuring days and bonus XP for pushing beyond the suggested range.

 

[clearstream]

Excuse me, don't you mean 2.9 to 6.6? 4.4 to 6.6 is the Medium band only; 2.9 to 6.6 is the Medium/Hard band.

The table formatting here might make it difficult to read. The minimum value in the Medium column is 5.8 and the Maximum value is 8.0. The minimum value in the Hard column is 3.8 and the maximum value is 5.3. The average for Medium is 6.6 and the average for Hard is 4.4.

CR is lossy--this is immediately apparent as soon as you examine the DMG rules for computing CR. There is no way to derive a genuinely reliable difficulty metric from CR because it's already missing too much information. Furthermore, playstyle matters. There's an enormous difference between a T-Rex being run by a DM who roleplays a T-Rex's likely behavior given its goals and low Intelligence, vs. a DM who runs a T-Rex with the mind of a tactical battlecomputer focused on destroying the PCs at any cost. The first T-Rex can be satisfied if you just give it a horse to eat; the second T-Rex is going to leverage its high speed (50'), reach attacks, and grappling bite to strafe the PCs and pick off the weakest-looking member of the party. If it gets lucky on geography, that second T-Rex could potentially TPK a melee-oriented 7th level party.

But the things that would enable the T-Rex to do that aren't valued at all by the CR calculation. A T-Rex with 10' move and no grapple would have exactly the same CR by DMG rules.

Agreed, it's very noticeable as soon as you go ahead and make up some creatures of your own using the rules and tables toward the end of the DMG.

I've tossed around potential metrics in the past, like say a Champion rating which says how many Nth level Champions you'd need on average to have a [better than 50%? essentially 100%? somewhere in between] chance of winning a computer simulation of a given fight/adventure. I haven't found a metric that has enough predictive value to make everyone happy with it, but I am at least pretty sure that it's possible to do much better than CR.

Ideally they'd just brute force it heuristically. My intuition is that it might be better to leave Adventuring Day XP as is and discount encounter values, e.g. we might build an encounter worth 550XP using the rules and then discount it by 20%*. Here the discount would impact the XP awarded, and it would be the discounted value that is paid out of the XP budget for the "day". This is perhaps the most conservative approach: it ramps difficulty without accelerating levelling. What do you think?

*20% is chosen even though the discrepancy is more like 25%, just for ease of calculation.

 

[FormerlyHemlock]

The table formatting here might make it difficult to read. The minimum value in the Medium column is 5.8 and the Maximum value is 8.0. The minimum value in the Hard column is 3.8 and the maximum value is 5.3. The average for Medium is 6.6 and the average for Hard is 4.4.

Oh, I understand now--I misunderstood the derivation of your table. You're already measuring from the midpoint of the Medium band, not from the threshold, so you don't need to double-correct. I stand corrected, no pun intended.

Agreed, it's very noticeable as soon as you go ahead and make up some creatures of your own using the rules and tables toward the end of the DMG.

Ideally they'd just brute force it heuristically. My intuition is that it might be better to leave Adventuring Day XP as is and discount encounter values, e.g. we might build an encounter worth 550XP using the rules and then discount it by 20%*. Here the discount would impact the XP awarded, and it would be the discounted value that is paid out of the XP budget for the "day". This is perhaps the most conservative approach: it ramps difficulty without accelerating levelling. What do you think?

*20% is chosen even though the discrepancy is more like 25%, just for ease of calculation.

Hmmm. I'm not sure I'm following what problem your proposal is trying to solve. If you want to slow down levelling, the most straightforward way to do that is to just change the shape of the XP table. I've run games where all XP requirements are x10 (but you get bonus XP for spending treasure offscreen on your campaign goals, essentially converting gold to XP on a 1 gp:1 XP basis). Likewise, Mike Mearls' most recent UA actually linearizes the XP table and changes the way XP get awarded. In any case, I don't see that the adventuring day necessarily has to be involved.

If on the other hand you're just trying to make the game more interesting, you can simply crank up the length of the adventuring day and the difficulty of individual encounters and leave the level advancement XP table unchanged. My very first 5E campaign did precisely this. PCs die occasionally, but they also advance really fast, so a replacement PC that starts off at 1st or 2nd level takes only a few sessions before it's a fully-functioning 8th level member of the party. It's not like 5E's advancement rate is delicately-tuned in the first place, or milestone levelling wouldn't even be mentioned in the DMG.

In any case, I don't quite follow what the goal is of your proposal, so I'm not sure how to give feedback. Could you be more explicit please?

 

[clearstream]

Hmmm. I'm not sure I'm following what problem your proposal is trying to solve. If you want to slow down levelling, the most straightforward way to do that is to just change the shape of the XP table. I've run games where all XP requirements are x10 (but you get bonus XP for spending treasure offscreen on your campaign goals, essentially converting gold to XP on a 1 gp:1 XP basis). Likewise, Mike Mearls' most recent UA actually linearizes the XP table and changes the way XP get awarded. In any case, I don't see that the adventuring day necessarily has to be involved.

If on the other hand you're just trying to make the game more interesting, you can simply crank up the length of the adventuring day and the difficulty of individual encounters and leave the level advancement XP table unchanged. My very first 5E campaign did precisely this. PCs die occasionally, but they also advance really fast, so a replacement PC that starts off at 1st or 2nd level takes only a few sessions before it's a fully-functioning 8th level member of the party. It's not like 5E's advancement rate is delicately-tuned in the first place, or milestone levelling wouldn't even be mentioned in the DMG.

In any case, I don't quite follow what the goal is of your proposal, so I'm not sure how to give feedback. Could you be more explicit please?

The goal isn't to slow down levelling, it is to increase the game difficulty without accelerating levelling.

As you noticed in your campaign - if you keep the XP as it is in the DMG and increase the encounter CRs, then encounters become more dangerous and characters advance faster. So that increases both the challenge and the pace of the game. The proposal here attempts to achieve the former without the latter. My motive is that I like the overall thinking and design represented in the DMG. Building encounters to an "adventuring day" budget works well from several angles. However, the values the designers tuned the system to are far too low for experienced gamers.

Another approach could be to increase the adventuring day XP budgets and the XP costs per level by the same amount. That would have the same effect as discounting the encounter XP values. The change needs to be applied in two places, but it's done once instead of done on every encounter. However, it's also more visible to players and could weaken suspension of disbelief.

 

[FormerlyHemlock]

The goal isn't to slow down levelling, it is to increase the game difficulty without accelerating levelling.

As you noticed in your campaign - if you keep the XP as it is in the DMG and increase the encounter CRs, then encounters become more dangerous and characters advance faster. So that increases both the challenge and the pace of the game. The proposal here attempts to achieve the former without the latter. My motive is that I like the overall thinking and design represented in the DMG. Building encounters to an "adventuring day" budget works well from several angles. However, the values the designers tuned the system to are far too low for experienced gamers.

Another approach could be to increase the adventuring day XP budgets and the XP costs per level by the same amount. That would have the same effect as discounting the encounter XP values. The change needs to be applied in two places, but it's done once instead of done on every encounter. However, it's also more visible to players and could weaken suspension of disbelief.

Sure. Equivalently, you could simply reduce the XP value of monsters. Equivalently, you could simply have fewer, larger encounters. 10 quaggoths attacking at once, led by a drow mage, is much harder than five groups of 2 quaggoths whom you get to ambush, plus a drow mage in a tower. But they both give the same XP.

I wouldn't expect increasing XP advancement costs to have any effect on player suspension of disbelief. What would that even look like?

"Sorry, DM, but my first level fighter has killed three orcs today and he still has only 12 HP. I just can't believe in so preposterous a world--it's really straining my suspension of disbelief not to have 20 HP and Action Surge."

Trying to imagine those words coming out of a player's mouth strains my suspension of disbelief. Instead, just say, "I want this campaign to be harder, but I don't want to speed up advancement rates, so in this campaign you need twice as much XP as the PHB says for each level. Don't worry, I'll be throwing twice as many monsters at you too."

My players found 10x XP requirements more believable, not less. (After a while though they got tired of it and wanted to go back to the regular advancement pace, so we switched campaigns.)

 

[FormerlyHemlock]

So here is the number crunching. I simply divided encounter XP at each level and risk threshold into adventuring day XP. At the bottom are means, modes and medians. And the correlation of medium to hard encounters, which is significant and tight meaning that they are good predictors of each other (from a game design point of view, it suggests that the table is coherent).

Level Easy Medium Hard Deadly
1 12.0 6.0 4.0 3.0
2 12.0 6.0 4.0 3.0
3 16.0 8.0 5.3 3.0
4 13.6 6.8 4.5 3.4
5 14.0 7.0 4.7 3.2
6 13.3 6.7 4.4 2.9
7 14.3 6.7 4.5 2.9
8 13.3 6.7 4.3 2.9
9 13.6 6.8 4.7 3.1
10 15.0 7.5 4.7 3.2
11 13.1 6.6 4.4 2.9
12 11.5 5.8 3.8 2.6
13 12.3 6.1 4.0 2.6
14 12.0 6.0 3.9 2.6
15 12.9 6.4 4.2 2.8
16 12.5 6.3 4.2 2.8
17 12.5 6.4 4.2 2.8
18 12.9 6.4 4.3 2.8
19 12.5 6.1 4.1 2.8
20 14.3 7.0 4.7 3.1

The table formatting here might make it difficult to read. The minimum value in the Medium column is 5.8 and the Maximum value is 8.0. The minimum value in the Hard column is 3.8 and the maximum value is 5.3. The average for Medium is 6.6 and the average for Hard is 4.4.

Oh, I understand now--I misunderstood the derivation of your table. You're already measuring from the midpoint of the Medium band, not from the threshold, so you don't need to double-correct. I stand corrected, no pun intended.

Actually, vonklaude, I did some further spot-checking of your math and now I think I have to retract my agreement here. I think my initial understanding of your method was correct and you are high-balling the numbers. Take first level:

The typical (median) Medium encounter will be worth (50 + 74)/2 = 62 XP. Since the adventuring day budget is 300 XP, that means you can fit 300/62 = 4.83 Medium encounters in the adventuring day budget. Or you can fit 300/((75 + 99)/2) = 3.49 Hard encounters in the budget. The easiest possible Medium encounter is however 50 XP, and you could fit in 6 of those before running out of budget--and 6 encounters is what your table reports. In this case your table is clearly measuring from the thresholds, not the medians, and is therefore highballing the actual number of typical encounters that you can fit.

Likewise at third level, your table reports 8.0 Medium encounters, corresponding to the easiest-possible Medium encounter, but the median Medium encounter will have 187 XP, and you can only fit in 6.4 of those in the typical adventuring day. You could fit 3.84 median Hard encounters.

I'm a bit confused because I spot-checked a couple of your numbers previously and they came up in the right ballpark; but now I think I actually misread the columns when I was spot-checking. (I think what happened is that I computed the medians for 3rd level, came up with 6.8, looked at your table and accidentally read the 4th level column, which reads 6.4, and confused myself into thinking I'd seen the same number in both places and that you were using the median method after all.)

Sorry for the churn here, but my point is: the correct way to read your chart is as a bunch of thesholds, since that's what it's derived from. If you want to know the approximate range for medium/hard encounters, start at the Medium number, and read up to but not including the Deadly number. If we take the average across levels as meaningful, that means that since you got:

mean: 13.2 6.6 4.4 2.9

we can crudely infer that 2.9 encounters per day or more means they are either Deadly or over budget; 3.0 through 4.4 encounters per day means they are Hard (or over budget); 4.5 through 6.6 encounters per day means they are Medium; 6.7 through 13.2 encounters per day means they are Easy; and more than 13.3 means they are not even easy.

 

[clearstream]

Sure. Equivalently, you could simply reduce the XP value of monsters. Equivalently, you could simply have fewer, larger encounters. 10 quaggoths attacking at once, led by a drow mage, is much harder than five groups of 2 quaggoths whom you get to ambush, plus a drow mage in a tower. But they both give the same XP.

I wouldn't expect increasing XP advancement costs to have any effect on player suspension of disbelief. What would that even look like?

Okay, so if you change the levelling up XP amounts then that pulls it into the foreground. It undermines the validity of the XP per level because those costs starts to appear arbitrary. In general it pays not to tinker with the mechanical pillars too frequently. Here's an extreme example - change all the XP/level thresholds. Then change them back a few sessions later. Then change them again to something different. You'll very likely notice players reacting. Those reactions are hard to predict. Human behaviour is often indirect.

 

[clearstream]

The typical (median) Medium encounter will be worth (50 + 74)/2 = 62 XP. Since the adventuring day budget is 300 XP, that means you can fit 300/62 = 4.83 Medium encounters in the adventuring day budget. Or you can fit 300/((75 + 99)/2) = 3.49 Hard encounters in the budget. The easiest possible Medium encounter is however 50 XP, and you could fit in 6 of those before running out of budget--and 6 encounters is what your table reports. In this case your table is clearly measuring from the thresholds, not the medians, and is therefore highballing the actual number of typical encounters that you can fit.

In my defense, I'm not high-balling or low-balling them. I'm using the book values exactly as I said I did. Our goal is only to understand the game difficulty. The table is a means to that ends. What you are suggesting sounds valid although we should admit that we don't know the empirical average and that could be higher or lower than the assumed average. A secondary problem is that as Deadly has nothing to average with, it isn't clear what values you use. A tertiary problem is that the greater our attempts to be precise, the shakier our reliance on magic ranges like 6-8 will start to look. Here is the table adjusted for your averaging principle -

Level Easy Medium Hard Deadly
1 8.1 4.8 3.4 2.8
2 8.1 4.8 3.4 2.8
3 10.7 6.4 3.8 3.3
4 9.1 5.4 3.9 3.2
5 9.3 5.6 3.8 3.1
6 8.9 5.3 3.5 2.9
7 9.1 5.4 3.6 3.1
8 8.9 5.2 3.4 2.9
9 9.1 5.6 3.8 3.1
10 10.0 5.8 3.8 3.3
11 8.8 5.3 3.5 2.9
12 7.7 4.6 3.1 2.6
13 8.2 4.8 3.2 2.7
14 8.0 4.8 3.2 2.6
15 8.6 5.1 3.4 2.8
16 8.3 5.0 3.3 2.8
17 8.5 5.1 3.4 2.8
18 8.6 5.1 3.4 2.8
19 8.2 4.9 3.3 2.8
20 9.4 5.6 3.8 3.2

I noticed Medium are about 2x Easy, and Hard are about 3x Easy, and Deadly are about 4x Easy, so I assumed a super-deadly of about 5x Easy. Sorry the layout makes it hard to read.

Level Easy Medium Hard Deadly
mean 8.8 5.2 3.5 2.9
median 8.7 5.2 3.4 2.9
min 7.7 4.6 3.1 2.6
max 10.7 6.4 3.9 3.3
correl 0.87471915

So now we see call it 5.2 Medium and 3.5 Hard per "day". Medium is less well correlated to Hard because of the way we're shifting all the numbers. Pointing a finger at the arbitrary nature of our choices. It's still significantly correlated. If we like this table and have faith that an expected 6-8 medium-hard encounters represents good game balance, then we might say that about 50% more encounters per day are needed. We could factor this without accelerating levelling rates, by discounting our encounters about 30% e.g. 4 orcs are worth 140XP not 200, so we add 2 orcs to bring it roughly to where we want it.

Thus I believe I would suggest at this point that a good simple way to ramp game difficulty from Easy to Medium without accelerating levelling is to discount encounters either 20% or 30%, depending on whether you more often use values close to the book value, or above that. Does that seem right?