Larian Studios Forums Baldur's Gate III Baldur's Gate III - Build Discussion Hold Person vs Drow Poison, a Mathhammer Bonanza

So this is in response to discussion in another thread, the one on balance, where the argument is being made that Hold Person is *less* effective than Drow Poison and Crawler Mucus. The link to it and the specific quote I'm responding to are below, I just felt it was worth splitting this off so as to not derail the entire conversation into an argument over which form of CC is superior. I do think it's relevant to the balance discussion since oddly some folks seem to be arguing damage is better than CC as a form of saying the game is imbalanced as a result, but I plan on getting deep enough in the weeds I think it'll be better this way. Anyway, buckle up ladies and gentlemen, this is going to be a super nerdy math hammer bonanza, I'd bail now if the sight of expected damage calculations, probabilities, and endless numbers in general make your eyes bleed. Consider yourself warned. Here's the link and quote:

https://forums.larian.com/ubbthreads.php?ubb=showflat&Number=898331#Post898331


So first, let's talk some general principles why I disagree before we dig into the numbers.
1) The most egregious thing by far, trying to say that a 2nd level spell slot is limited, while saying Crawler Mucus is spammable. Crawler Mucus is not so common that you can just toss it out there willy nilly in every encounter. 2nd level spell slots on the other hand, *are*. So we're not going to compare Hold Person (able to be cast literally every encounter if you want) to Crawler Mucus, they're not comparable. Instead, we'll compare it to Drow Poison, and then after, compare Crawler Mucus to Hold Monster; it's still far more limited in use from a rarity perspective but it's at least a more fair comparison.
2) Poisons target Con saves, while CC spells target Wis (or Int or Cha depending on the spell), generally speaking. Enemy bosses, those worth CCing, generally have higher Con than Wis / Int / Cha. It's not just a pattern, it's the overwhelming majority.
3) You can improve your save DC with spells you cast and pump it up, you can't do so with the DC from poisons.
4) It's a false comparison, as they're not using equivalent resources. It's not Drow Poison vs Hold Person, you can use both (whomever goes first go for their CC, if it fails use the other person's CC, if it succeeds dive into damage). The actual comparisons *should* be hold person vs Fireball and Drow Poison vs say Simple Toxin or Wyvern Venom in isolation. However we'll ignore that and go ahead with the faulty half comparisons for the sake of expediency.
5) Almost as egregious as the first point, when an enemy saves against a poison, they become "inoculated" against it, which means they can't be further affected by it over the next 2 turns. This means they have a significant chance of the poison simply becoming ineffectual, while you can't be inoculated against hold person. This is actually even worse with legendary resistances,

But let's set that aside and instead of speaking in generalities get into the nitty gritty. First, we'll use a boss from Act II. We'll assume a party level of 7, and we'll choose one where their Wis and Con are actually the same, as otherwise the comparisons are too unfavorable to Drow Poison to be interesting. Just note that if you were level 3 or 4 and didn't have a second attack on your Martial characters, or if you were facing a target with a higher Con than Wis, it would be even worse. We'll use the same party comp I recommended to someone recently just because it's easy and balanced, with a ranged damage dealer, Pact of Blade Warlock / Paladin, Wiz/Cleric/Sorc, and Druid. Our target has an AC of 17, 165 HP, a Wis of 14 and a Con of 14 (and a Dex of 16), with 3 attacks per round doing 8.5 damage each and a bonus to hit of +7. We're just ignoring positioning and turn order altogether here, and we'll track across 2 rounds of combat, looking at average damage, best case damage, and failure rate. We won't worry about crits unless they're auto-crits, for simplicity. Trust me it doesn't make that big of a difference.

[Drow Poison Team]

Round 1
(The Druid)
Apparently they're not on CC, it's better for them to be blasting I hear, so to be as favorable to this party as possible they'll cast Haste, which is best spent on the Face.

(The Face)
They've got a Str of 16 at this point, 2 attacks per round, and Great Weapon Master to dive into damage when warranted. However they likely *don't* want to use it unless she's already CC'd. So first round is 2 attacks at +7 (+3 Prod, +3 Str, +1 Magic Weapon) dealing (2d6 average 7 + 3 + 1 = 11 average damage on a hit, 18 on an auto-crit. Their first attack in that combo has a 45% chance of missing outright, a (.55*.5 = .275) 27.5% chance to apply Sleep, and a 27.5% chance to result in them being inoculated. Their second has a 55% hit chance if they missed the first time (with the same net chance to either apply sleep or inoculate), a 55% hit chance if inoculated (with a 100% chance of inoculation of course in that case and thus no sleep chance), and a 75% hit chance if they're asleep, which will wake them and trigger neither sleep nor inoculation. In all cases they'll have GWM off of course, so the average damage total for the two attacks is ((11*.55=6) + ((18*1 = 18) * .275) + ((11*.55 = 6) * .725) = 15.3 damage

Their second attack action then starts with a 40% chance they're awake and inoculated (.275 from the 1st hit, and .45*.275 from a miss then save), a 12.5% chance they're asleep (a miss then a hit and a failed save), and a 47.5% chance they're awake and not inoculated. The 47.5% where they start awake and not inoculated pans out exactly the same as the first attack. The 40% where they're inoculated is super simple, just 11*.55*2 expected damage (12.1). The 12.5% chance they're asleep they would turn on Great Weapon Master to leverage the auto-crit, for 28 average damage on the 1st hit and a 30% chance of hitting on the 2nd for 21 average damage (6.3 average damage total, 15% chance of inoculation). Total average damage (15.3 * .475 = 7.26) + (12.1 * .4 = 4.84) + (34.3 * .125 = 4.28) = 16.38; Total chance they're inoculated (.4 from the 1st set + (.475 * .4 = .19 if starting awake) + (.125 * .15 = .02 if starting asleep) = .61) 61%, total chance they're asleep (0 + (.125 * .475) + (.125 * .15) = .08) 8%, 31% chance they're awake and not inoculated.

(The Scout)
They've got a Dex of 18, gloves of archery, +1 weapons, and the Ranger Archery fighting style by now, so they have +10 to hit. They also have sharpshooter, and use it here because it provides a mathematical advantage to their damage output. They apply their drow poison while stealthed, and then get their massive initial turn's worth of attacks - 2 attacks + 2 bonus action attacks, 4 total. If the target is autocrit they deal 24 damage (4 Dex + 2 Archery + 1 Weapons + 2d6 weapon crit damage + 10 Sharpshooter); if they get autocrit sneak attack damage it's 4d6 (14) otherwise it's 7, non-autocrit average damage is 21, just missing 1d6 weapon damage.

If the target is asleep (8%) their first attack is an autocrit sneak attack. 2nd attack is a 45% hit chance (.45 * 21) 9.45 average damage (.225 they're asleep, .225 they're inoculated). 3rd attack is only different if they're asleep (.225 * 24 = 5.4 average damage if asleep, .775 * 9.45 = 7.32 damage if not, total average damage 12.72); after the 3rd attack there's a (.225 + .55*.45*.5 = 35%) chance they're inoculated, (.55*.45*.5 = 12.5%) chance they're asleep, 4th attack then is ((.125 * 24 = 3) + (.875*9.45 = 8.25) = 11.25 average damage, (.35 + (.525 * .45 * .5 = .12) = .47) 47% chance they're inoculated, and a 12% chance they're asleep.
If the target starts awake but not inoculated (31%) it's a 45% hit chance, (.45 * 21) 9.45 average damage (.225 they're asleep, .225 they're inoculated). They're basically just an attack ahead of the path above, so the 4th attack this time is basically like a 5th attack above, meaning (.12 * 24 = 2.88) + (9.45 * .88 = 8.31) for 11.19 average damage, with a (.41 * .45 * .5 = 9%) chance they're asleep or newly inoculated or asleep at the end of it, for a total 56% chance of being inoculated
If the target starts inoculated (61%) the first attack (and all subsequent attacks) are just 9.45 average damage.
Coming out of it we've got:
Average damage- (9.45 * 4 = 44.8 average damage if inoculated (61%)) = 27.3 + (24 + 9.45 + 12.72 + 11.25 = 57.45 average damage if they start asleep (8%) = 4.6 and (9.45+12.72+11.25+11.19 = 44.64 + (14 * (.225+.12+.09 = .435) + 7 * (1 - .435 = .565) = 10 from sneak attack crit / non crit), 54.64 total if starting awake and not inoculated (31%) = 16.9; total 48.83 average damage.
Chance they're asleep (.09 chance to come out asleep if starting awake and not inoculated * .31 chance of that happening = .03) + (.12 chance to come out asleep if they start asleep * .08 chance of that happening = .01); 4% total
Chance they're inoculated (100% if they started inoculated = .61) + (.47 if they started asleep * .08 of that happening = .04) + (.56 if they started awake * .31 = .17) total of .82


(The Caster Cleric)
They're casting lightning bolt, because apparently that's better than hold person. It deals 8d6, average 28, but they'll gladly spend their channel divinity to maximize that to 48, and their save DC is 8+3+3 or 14. Against Dex that means a 50% chance of saving, unless they start asleep (.09 * .31 = .03 + (.12 * .08 = .01); 4% chance of that) in which case it's an auto-fail. So 48% chance to save overall, total average damage ((.52*48 = 24.96) + (.48*24 = 11.52) = 36.5 average damage.

Round 2 -
(Druid) - They'll cast Create Water to help the Cleric out.

(The Face) - Same as the 1st round but there's a 82% chance they're inoculated going in, in which case it's (11*.55*4 = 24.2) total average damage for the round (82%), 31.7 otherwise (39%) (.82 * 24.2 = 19.8) + (.18 * 31.7 = 5.7) = 25.5 total average damage

(The Scout) - Same as the 1st round, but there's a (.82 + .18 * .62 = 93% chance they're inoculated going in, and only a (.18 * .08) 2% chance they're asleep. That means it's 44.8 average damage if inoculated (93% chance) (41.7 total) + 57.45 if asleep (2% chance) (1.5 total) + 54.64 if awake and not inoculated (5% chance) (2.7 total) = 45.9 total average damage for the round

(The Caster Cleric) - Same as last round, but doubled, since now the target is wet, 73 average damage total

Conclusion
Total average damage among everyone was 117 in the first round, 71.4 in the second before the Caster Cleric sealed the deal, more than enough at that point already but the Caster had the nail in their coffin ready if needed. So they won in 2 rounds, they didn't need any CC spells or messing about, they did awesome! Surely this is all that is man, there is no reason to even *try* CC, right?

[The Hold Person Team]
Round 1
Druid - Casts Hold Person, save DC (3+3+1+8 = 15), against Wis that's a 40% save rate, 60% fail rate

Cleric - if Hold Person worked, they can just chill and worry about other targets if there are any, otherwise they cast Hold Person (45% save rate, 55% fail rate),

Face - No Haste this time. Nor Drow Venom, they'll instead just coat with a candle for an additional 1d4 fire damage. In the event the target is Held (.60 + .55 * .4 = .82% of the time) they auto-crit on 2 attacks, hitting for the same 28 auto-crit damage from the first team, plus 5 damage from the fire (2d4), for 33 damage each, or 66 total. In the event they're not held, we're, well, you guessed it, we're casting hold person (55% fail rate).

Scout - In the event the target is held (.82 + .55*.18 = .91) we just auto-crit with our Sharpshooter 4 attacks each dealing the same 24 damage + 5 fire damage since we're using that, so 29*4 or 116 damage, + 14 damage from the sneak attack for 130 damage total. In the event they are not, we just attack 4x at a .45 hit rate and don't get the crit on the sneak attack, for 52.2 average damage

Round 2
As you may have noticed, if Hold Person hit, the target is already dead. The only exception being if it hit just before the Scout went (happened 9% of the time), and in that case the scout finishes them off, regardless of whether or not Hold Person dropped on their turn. If none of them hit (happened 9% of the time), then the Face tries again, and if they hit, the Scout finishes off the target (.55 chance of that) or else the Druid casts Create Water, the Cleric burns a Channel Divinity charge for 96 damage, 48 if they save, and then the scout finishes them off.

[The Conclusion]
In both scenarios, the target is comfortably dead after 2 rounds. The difference being, on the Hold Person team, they burned at most 2 spell slots on the Warlock (short rest guy), a 1st and 2nd on the Druid, and a 2nd and 3rd on the Cleric along with a Channel Divinity Charge. But that Channel Divinity Charge and 3rd level spell slot only got used in 4% of scenarios, in the rest of them you burned 1 maybe 2 2nd level spell slots, and took 0 damage from your opponent. It was over *quick* (she was dead in the 1st round) in 82% of cases. Only in 18% of cases did she survive the first round, and only in 13% of cases did she get a turn off.

In the Drow Poison scenario, she was inoculated on the 1st hit 27.5% of the time and thus was never CC'd at all. She guaranteed had a turn at the end of round 1 regardless, and guaranteed was still alive. 2 3rds and a Channel Divinity charge were guaranteed to be used, and if she managed to say knock the scout off a cliff on her turn, another 3rd and a 2nd Channel Divinity had to be used to take her out. The whole thing was just like reading about it or calculating it - a struggle. Did it work, and did they win? Of course they did, this party is absolute ballers, even if piloted suboptimally they'll clean house. But was it the superior option? No, not even close. In fact, no part of it was superior, by any measure. Nor was it, at any point, something you could really plan around, you just kinda hoped you got lucky, and not just a little lucky, like *very* lucky or else you were just pumping out mediocre damage (albeit at high volumes, again, the team is super strong).

The Crawler Mucus vs Hold Monster will be a later post, this one took forever. Spoiler alert it's going to be on the Dragon, and while the Dragon needs a 17 to save vs Hold Monster it becomes inoculated against the Crawler Mucus on a 5 or higher, so it's dramatically more lopsided than this one even before you consider legendary resistance. I know that I'm proving something anyone who has actually tried the spell already knows, that it's absurdly powerful and instantly ends battles the moment it lands. The idea that Drow Poison is somehow comparable to it is absolutely incorrect. I also know that if someone can't be bothered to try something, but judges it as trash literally without having tried it, they are unlikely to be convinced by math and logic. This was just for fun for me, to show with cold hard numbers, just how large the disparity in power level is between them. To that end, let's do final average damage tallies!

Drow Poison Team: 117 in Round 1, 261 average damage total, they almost killed her twice over
Hold Person Team:
The Cleric will burn resources in the 1st round if Hold Person hit just to make resource expenditure even. They do 36*.6 + 96*.75 + 73*.25 = 114.8 average damage
The Scout deals (130*.91 + 52.2*.09 + 130 * .75 + 52.2 *.25) = 118.3+4.7+97.5+13 or 233 average damage
The Face deals (66*.82 + 66*.5) = 54.1 + 33 = 87.1 average damage
The Druid didn't deal any damage
Total: 434.9 average damage, they killed her nearly 3x over, dealing a clean 2/3 or 66.6% *more* average damage than the other team.

There’s a lot here to process, but I believe something key is missing from the calculus. The boss gets to save against Hold Person at the beginning of their own turn; for any PC going after the boss, the odds of the boss being held are more or less halved.

Moreover, if the boss goes right after the caster, Hold Person does nothing at all if they succeed either save. (Odds of failing twice are 0.6^2 = 0.36 ; odds of not failing twice are 1-0.6^2 = 0.64). My point here is I believe that the computed odds for when the boss is held are overestimated by quite some margin.

That could be the case for the poison as well, but I don’t know if or when poison triggers additional saves.

If the initial save versus poison is made then I believe there is a follow up immunity to that type of poison. So you can't just keep hitting an enemy with the same poison type hoping they will eventually fail the save.

I'm not certain how long the immunity lasts.

Yep that is not missing, I promise. I didn't do great later on in the post of explicitly stating everything but it's absolutely part of the calculus. For the sake of whiteroom analysis we assume we're starting with all 4 of our heroes able to freely rearrange their turns and the boss at the end. It honestly doesn't matter the relative initiatives, if the boss is the only enemy all players will always go between the boss and the next time the boss has a turn, and players adjacent in initiative can always choose their turn order. So while one could argue over the realism of the setup which is primarily done for convenience, based on the game mechanics it's actually not too far off from a standard scenario, based on how turn mechanics work in the game.

That save is why it's for example 13% of cases where the boss gets a turn - 9% chance of them having passed all saves so far, plus of the 9% where the scout was the only one who went after they got held, in 45% of those cases the boss makes their save at the beginning of their turn, thus 13%. For the poison it's absolutely the case as well, for continual effect poisons like the Drow poison or the Crawler Mucus, they get a save at the beginning of their turn, and worse, if they succeed, they're inoculated. That'll come into play heavily when it comes time for crawler mucus vs hold monster, but regardless, not missed, and the odds are not overstated.

They may be inapplicable to a given scenario which will have its own odds calculations, but you can envision basically any setup you want, the difference in efficacy between hold person and the drow poison is simply too severe for it to practically matter, the conclusion will always be the same - Hold Person is a far more reliable and effective form of CC than Drow Poison, and being Paralyzed while a Martial with GWM or Sharpshooter has a turn to exploit that paralysis is an outright death sentence. It's basically a 2nd level "Save or Die" effect. Even in the situation you describe (which is worst case scenario), that's 36% of the time where the boss does nothing, stays paralyzed, then dies immediately after. Ridiculously good.

Edit: Also of note, and imho just as important and arguably far more important, once they fail the boss *does not* get any saves *outside* of the start of their turn. So if they fail that one, that's it, they're done for. If you get them at any point prior to their turn, you're guaranteed to have them until their turn occurs. That's why the numbers are how they are, but imho your perspective is a bit flipped, it's more that they *only* get a save at the start of their turn, not when damage occurs or under any other condition.

@JandK it's in the top, the immunity is referred to as "inoculated", it lasts 2 turns and is poison type specific, you can be inoculated against multiple types of poisons simultaneously though.

It's down to the poison staying on your weapon for a specific time. Some are cumulative.

Yes… one could argue… Don’t threaten me with a good time ^^

The thing is, I don’t freakin’ know. If I were to argue, I’d say that having 16 strength on your level 7 great weapons master is a misallocation of ability scores, that initiative swapping isn’t a reliable assumption given how many mobs usually surround bosses (in my limited experience), that Hold Person requires maintaining concentration (which isn’t necessarily a problem, provided initiative order lines up), that facing a boss seems like the perfect time to use rare consumables or that poison has low opportunity cost and the benefit of immediate payoff, reducing the odds that something goes wrong between a failed save and whatever horrific fate you have planned.

All that said, you kinda sold me on Hold Person. The fact that, under favorable initiative conditions (I can’t really let that go - but it speaks more of me than you) you’re 60% to land 6 crits with your fighter seems deece.

I still wish Larian would change the saving throw trigger to be at the end of turn rather than at the beginning. But so many threads are claiming BG3 is too easy, I don’t know if boosting a spell that clearly already has a following is a good idea.

In the end, I’m glad you love this spell, OP. I’m sure Larian are, too. What’s life for if not strong feelings?

They should change it to be at the end of its turn because even if a spell like hold person at least allows you to get SOME benefit if you pass one roll but fail another, so long as you have other characters with turns in between, a spell like confuse gets literally NO benefit, and the bug essentially means "All creatures roll with advantage against this spell." This is on top of OTHER bugs that hurt the power of CC spells.

As for too easy, let's split it into 3 sections:

Low levels (1-4) feel not that bad; some fights are even relatively challenging on tactician.
Mid levels (5-10) start getting easier, because you start to get equipment that is too powerful, and because haste gives you a full action instead of +1 attack (which means that martials, who double their number of attacks, start being able to do a good deal more damage than they should), and because around this time, if you explore everything, you will consistently be overleveled for the content you are facing.
High levels (10+) you become stupendously overpowered because of the absurd items available to you, haste making martials even MORE powerful, and (if you bother with them) OP tadpole powers.

I really, REALLY don't think we should build in what seems to me an OBVIOUS bug in CC spells, to make up for the fact that the game is too easy due to a bunch of other things.

How did the caster cleric get lightning bolt?

@Mordant One level of Wizard.

@Flooter it’s because your Cha in this build is 17 so you can pump it to 22 with the in game stat boosts, since you’re replacing Str anyway when using an elixir and eventually when wearing gloves. I agree regarding consumables and optimally would be using something like Wyvern venom, but didn’t want to distract by having a different consumable and candles are free. I’d be down with them triggering at the end instead of the beginning, certainly, but that would make an already powerful option even more powerful, so I agree it isn’t really needed.

To be fair.. They "balanced" poisons in patch #1.
Used to be, even though the save DC is low, they're able to be applied multiple times (esp as a dual wielding rogue). But now it makes them immune when it fails, aand these poisons will likely fail the first time, because poisons are con save (DC 13), and npcs are good at con save. When they inevitably fail, you don't get a second try until round 3, because it's a universal immunity.
Compare that to spells, which natively enjoys a scalable, higher save DC, and incurs no penalty/immunity on failure, which means we get to try multiple times on a single round, if we have multiple casters.

Why would you use something that is: 1. cost a bonus action, 2. has lower save DC, 3. incurs penalty on failure, 4. doesn't recover during full rest. ?
It could easily be the case, that before patch, poisons are more efficient and reliable. But they nerfed it (too much), making poisons pretty useless mid to late game.

Or you could look at it as one costs a spell slot or a scroll and the other can be made for free or at least cheaply if you buy the ingredients and one only works on humanoids (guessing) and the other has a broader scope. One only works during combat while the other can be applied at any time and is 'fire and forget'.

And one is a concentration spell whereas the other can be prepped pre-battle. Not that killing an enemy twice or three times over means anything. All that matters is you kill it, which happens in 1-2 rounds anyways. With a Hex into magic missiles with glove clowns it's 200+ damage or a chain lightning with the legendary staff amp doing 130+ damage forking to multiple targets and can be twin cast if you want to be even more ridiculous and play sorc. Most spells are garbage, but the few damage spells that work with supporting items are way more worthwhile than a hold person, especially for casters.

Because you can get a paladin and get the ring that lets you cast a spell as a bonus action after a melee attack and you can just have your martial party member apply hold person just as well, making martials yet again a more effective CC applier while being tankier and doing way more single target damage anyways.

It makes perfect sense to prepare it before battle, and treat it as a potential boost to critical chance.
I refrain from doing pre-battle preparations though, so it always cost something to me.