D&D 5E (2024) Mike Mearls explains why your boss monsters die too easily

[pemerton]

Flat math is here to stay.

<snip>

Flatter math is easier to design around, easier to balance and often easier to teach to players. It's also just more resilient to a variety of situations.

The only game that really breaks the trend that comes to mind is Pathfinder Second Edition, but that is a little misleading in that it's fairly flat within the scope of play since almost all the gains in accuracy are level based and you are expected to keep play within a -4 to +4 level band of the characters.

This sounds similar to 4e D&D, where the maths relative to a given level of play is pretty flat. The rationale for level-based escalation in 4e D&D is to ensure a type of "story progression" as the game goes on, from Kobolds up to Orcus. It doesn't affect the difficulty at any given level. The actual escalation over the course of play is mostly a result of more intricate player-side ability suites and more severe conditions (on both sides).

I don't know, it seems to be working fine in Draw Steel. The only attrition there is to your recoveries, while your offensive power actually increases over the course of the day (because every previous Victory gives you one extra starting resource to work with, which can often be the difference between opening with one of your moderately powerful signature abilities or one of your strong 3-resource abilities or, with a few victories available, even a 5-resource ability).

This sounds like an amped-up version of 4e's milestones.

 

[Zardnaar]

I doubt that.

If 4e had open playtesting, it would look more like 4e Essentials.

That's at best though.

No risk of open playtest revealing that the whole system/concept was a bad idea?

Essentials didnt save 4E if they don't like the basic idea to begin with. I highly doubt a more essentials 4E on release would have helped.

It doesn't fixed the two main problems.

1. It wasn't a fixed 3.5/3.5 wasn't organically dead.
2. Playstyle was still essentially 4E.

 

[M_Natas]

You don't need statistical testing to tell you that at-will paralysis on any attack, in a game where save success is in the neighbourhood of 50/50, will likely be broken. You can work that out from eyeballing!

Yeah - like, you don't need a statistician. Like, for new DMs a good advice is, to play out encounters they plan in advance to see how it could go and to see how they could use the creature abilities to their advantage - it also helps run encounters more smoothly. And then you would also see if you could TPK your party easily.
How you run most monsters also is like half of the difficulty.

If you take 6 goblins against a Level 1 party (a moderate challenge) - they could be either wiped out in 2 rounds or the could TPK the party, depending completely on how the goblins are run.

Like, are the just in an empty room and having to fight the characters in melee? Or are we in a forest, where the goblins will ambush the party at 80 feet distance with their shortbows, then hide behind trees and run, staying out of melee range for at least 2 rounds and forcing the characters to dash to close the distance?
When you plan an adventure, you playtest that. When you create monsters, you should playtest that. Like, 90% of Goblin Power is in the nimble escape ability. Hiding allows them to attack with advantage. Disengage allows them to flee quickly.

The same with Ghouls, their power comes from the paralysis. If 3 ghouls ambush your party and get to the Low-AC-Low-Con Characters, its game over. If they got stuck on the front liners with High-AC and Con-Saves, there will be likely no paralysis in a typical 3-Round Battle.

I would expect them to test most monsters that has anything more than a simple attack or multiattack option to be playtested that way.

 

[Minigiant]

That's at best though.

No risk of open playtest revealing that the whole system/concept was a bad idea?

Essentials didnt save 4E if they don't like the basic idea to begin with. I highly doubt a more essentials 4E on release would have helped.

Nope.

The whole system wasn't a bad idea. It was just to drastic a shift.

If 4e Essentials were the original idea, it would have been loved. The issue was that Essentials came too late. The people who like 4e didn't need it. And the people who would have liked Essential had already abandoned 4e.

A Playtest would have fixed the 3 of the 4 main criticisms. Every class would not be AEDU. Roles would be less hard locks or there would be more colored classes to fill class role fantasies .And the math would have shrunk for faster fights.

But 4e more or less solved the problems stated by this thread: 4e assumed you went into every fight with all your encounters powers and some mixed of dailies, item powers, and action points. So the offensive capabilities of a given party could be accurately gauged and bosses could be calculated to handle full nova.

while...

Tables could still enjoy the gameplay of attrition and resource management because...

1. Automatic Encounter shutdown effects were limited, shifted to rituals or DM tokens, or non-existent
2. You regained lost magic item charges and action points as you adventure so you stay with the ability to ramp impact even if you nova earlier but you can't hoard them.
3. Gold had a core use

Again the #1 issue with the problem Mearls mentioned is that characters have too many powerful abilities after a long rest and for some classes those abilities are VERY powerful. You enter the dungeon with too many grenades and DMs are then forced to push you to toss some early.

 

[Gimby]

It is worth noting that, consistently, statistics from video games indicate that, for most games, players overwhelmingly (like more than 4:1) favor "good" options over "evil" ones when they're put to the choice. Good example, there's a dog in the Dragonfall campaign of Shadowrun Returns, and he belonged to the PC's friend who perished in the opening. He sometimes comes to you for comfort, as a dog missing his human would do. You can choose to be mean to him or nice to him.

For about 2/3 of the game, that's all it is, the occasional scene where you can be nice to a dog. Then, when your home base is under attack....it turns out he's half-hellhound and can kick some butt. If you were more nice to him than mean to him, he'll help fight. He's not particularly strong and can't wear most equipment, being both magical (which makes augmentation difficult/harmful) and not humanoid. But he is helpful. The efficient option, for min-maxers, is to be mean to him so you can fight him and get more XP. Something like ten times as many people have the achievement for recruiting him than those who have the achievement for killing him, even though a single game can easily accommodate doing both things without restarting.

That's far from the only example, it's just one where I've seen the data with my own eyes. Every time data of this kind gets reported, the pattern remains strong.

Yes, some people see the freedom of the TTRPG space as a chance to "cut loose" and do all the hedonistic, violent, selfish, horrible things they can't do IRL. But the vast majority, backed up by actual statistics and analysis, genuinely strive to be good people, perhaps even better people than they are/have been IRL.

"People" is a distribution, there will always be some in the extreme tails, and the bigger the population, the more extreme outlier examples you expect to see. But it turns out, video games actually reveal that most people want to do good by others, even when the rewards actually do favor being evil.

I know you are a strong proponent of the idea that players can be trained into certain behaviours by their experiences and I wonder if this is reflected in this - at least in part. In many games, "being good" just gives better rewards - it was a bit of a trope in BG1/2 for example that turning down a monetary quest reward got you both reputation points and magic items worth more than the reward.

Even in BG3 the "good" routes are often better fleshed out and the rewards are pretty similar (getting Minthara as a companion would be counter example, but the general case it a bit uncommon).

I don't disagree with your overall thesis, but there might be a bit of a feedback loop here - "good" routes historically provide better rewards so players choose them more, Devs spend more time on content players are likely to see, players choose the routes with more/better content and so the cycle repeats.

The question is kind of - if the "good" routes required significant content/gameplay sacrifices, would people still choose them to the same extent?

 

[Minigiant]

Another thing to remember

5e was mostly playtested off a single adventure. IIRC The bosses were at the end of dungeons or solo partykillers.

 

[Parmandur]

I know you are a strong proponent of the idea that players can be trained into certain behaviours by their experiences and I wonder if this is reflected in this - at least in part. In many games, "being good" just gives better rewards - it was a bit of a trope in BG1/2 for example that turning down a monetary quest reward got you both reputation points and magic items worth more than the reward.

Even in BG3 the "good" routes are often better fleshed out and the rewards are pretty similar (getting Minthara as a companion would be counter example, but the general case it a bit uncommon).

I don't disagree with your overall thesis, but there might be a bit of a feedback loop here - "good" routes historically provide better rewards so players choose them more, Devs spend more time on content players are likely to see, players choose the routes with more/better content and so the cycle repeats.

The question is kind of - if the "good" routes required significant content/gameplay sacrifices, would people still choose them to the same extent?

Actually, yes. Bioware found with Mass Effect, where they pit wqual energy into Paragon and Renegade paths and rewards, almost everyone went Paragon anyways.

People simply gravitate towards playing the good guy in video games, it's an established discovery of looking at the numbers.

 

[EzekielRaiden]

So you dont think an open playtest would have revealed what people thought of it?

People literally cheered when WotC told them Vancian casting was being removed.

"What people think of it" was wildly variable. People got HUGELY HUGELY mad about

Or thst the math didnt actually work and they had to rewrite the MM in effect?

Such a playtest would have revealed such a problem, yes, so then it wouldn't be one. That's...literally the point of real playtesting, as opposed to performative playtesting. It's literally about testing things to make sure they work, and if they don't, fix it.

Can't see any of those possibilities cropping up?

The first one is irrelevant, and the second is literally the point of the exercise, so....I mean the answer is too complicated to be a one-word "yes" or "no".

But if I must answer as you've specifically presented it? No, I cannot see those possibilities in the way you described. Instead, I see them as follows:

1. During the playtest, nothing meaningful would be learned about presentation, because players expect pretty basic presentation at this point. No one complained that the "D&D Next" docs were black-and-white word docs in PDF form. So that part's just outright gone. Second, it is--or should be--expected that ideas will grow and evolve, and that feedback will be factored in. Thus, the only real element that people could have "revealed what [they] thought of it" would be...pretty much the exact things we already saw. In other words, I don't see how anything meaningful would change, other than the designers getting more direct/specific/usable feedback because of better survey design. That's....all good, as far as I can tell!
2. Fixing math errors like the thing you describe is one of the greatest benefits of doing what I've described. Most of the math errors in 5e, for example, come from not testing things. Not being rigorous. Not having well-constructed surveys, nor conducting robust simulations, nor proper statistical analysis. And the exact same thing is true of 4e. The math errors you love to crow about so much would never have happened, because this testing process would catch them and fix them before publication. So....yes I foresee that happening and it is a directly good thing.

Hence, the things you describe are either utterly irrelevant.....or precisely part of the plan.

 

[tetrasodium]

So you dont think an open playtest would have revealed what people thought of it?

Or thst the math didnt actually work and they had to rewrite the MM in effect?

Can't see any of those possibilities cropping up?

It's complicated and I think blizzard has done a good job of both succinctly summarizing why as well as growing from a verbal faux pas with the well known blizzcon interaction
"You don't want that. You think you do but you don't. Remember when you had to spam cities 'need a tank, need a tank, need a tank' during TBC days? You don't remember that because you now push a button to go to the dungeon. You don't want to do that". Wanting "[this]" in regards to specific mechanics and design ethos doesn't always mean that the specific element in question is solely responsible for accomplishing the goal absent supporting elements that are removed or modified for the [this] that someone else wanted

Spoiler

Yo think you do but you don't interaction
Evolution of it as a gag up to "you think you want this and I think maybe that you do"

A lot of 5e's biggest problems being discussed over and over in this thread and others like it over the last decade or so can often be traced back to that design by committee in isolation process where too often the results were shaped by "management said give the players what they want and the players wanted these at the time". Following that up with "now they want more of those plus these" is where the complicated hairline cracks start really breaking down because of two reasons.

Firstly ttrpgs are a 3 legged stool (players/cohesive ruleset/gm). When the design ethosvputs too much weight on using awful polling practices to "give the players what they want" it shots out that third leg from the process and completely strips away the cohesive bit for all but the loudest most numerous subset of players. When there is a critical minority at nearly every table who is not part of "the players", when it comes to needs and wants whil their presence is required for the game to work at all (the gm) polling needs to recognize up front they because GM's are going to be an absolute minority in polling & have an internal gm advocate on the team with significant pull when it comes to saying no while overriding what the players wanted.

Even if the self selected survey nature of the polling leaves them somewhat overrepresented, that whole cohesive ruleset leg of the stool is critical to the GM's and not particularly important to players who can just rely on the gm to kinda fix it while blaming said gm instead of cold unfeeling rules for the results of trying to fix areas where rules cohesion fell down. The 6-8 encounter per long rest target juxtaposed with giving players near guaranteed successful resting under almost any circumstances under the harsh light of what 13.33 encounters once represented is a great example of how "give the players what they want" directly undermines the other two legs of the stool and blames the gm both for running too few encounters per rest to allow the PC's to roflstomp everything as well as throwing blame for the boredom of running so many sessions of nothing but combat should they actually try to railroad in so many sessions of nothing but combat.


Secondly is where the complicated really comes into play. There needs to be serious consideration for "why [that] worked" between "the players say they want [this]" and the gm advocate needs to have extreme authority in almost unilaterally deciding both the supporting reasons and which conflicting wants/reasons the players said they wanted and simply antithetical to each other a cohesive ruleset and GM needs/wants. The 5.0024 play test saw some of this where Crawford told us they saw people vote high overall on a class but low on individual features§ao they decided respondents forgot and chose to help by assigning the higher value to both. They choice was the kind of thing a GM/rules advocate SHOULD be empowered to do, but it was done backwards to further empower the majority (players) in service of ignoring anyone who was voting for rules cohesion or GM needs/wants.

Tlr: design needs to consider that the game is a three legged stool & "Give the players what they want" makes a mess when there is noone advocating for the other two legs of the stool or that person actively exacerbates it by advocating for the players if even the players start noticing problems they don't want in large enough numbers to be statistically noted.

§ or vice versa, I don't recall
Edit: Blizzard relevance for those who don't get the blizzard/wow reference, they got asked to provide servers with world state as it was in the day and after saying no while scoffing at the idea they gave it much thought and implemented servers that pretty much give that with some changes to support the why behind why what was wanted worked.

 

[Paul Farquhar]

The whole system wasn't a bad idea. It was just to drastic a shift.

A “drastic shift” is a bad idea. When people buy a game called D&D they expect to be able to play D&D, not some other game, no matter how good it is.