Larian Studios Forums Baldur's Gate III Baldur's Gate III - Suggestions & Feedback Several dicerolls for one outcome is bad

Yeah, and sorry to get all 'Team 5e' on you, but having random rather than guaranteed results would be fine in a better written game. Being more likely to succeed rather than guaranteed to succeed is fine because failure in TTRPG gaming is supposed to lead to more interesting situations and rarely to outright fail-states. It should give the game master a chance to have some fun by humanizing your character and making you react to more out-of-control situations, but all the 'GM' of BG3 wants to do is randomly kill children or force you to fight. It extra sucks because 'it's fun to fail forward!' was supposedly a big talking point around this game when it was being introduced through streams and marketing.

...and I'm like 99% sure that this is actually true.

But there were too many red flags in the whole dice roll mechanic for me to enjoy it. You named two of them which I absolutely - ABSOLUTELY - hate.

I remember how much I enjoyed being able to solve situations in D:OS2 in different ways (example: houndmaster, dogs and red ball).

Here, currently, I feel like the dice decides for me. And thats not what I expect from an RPG. Because this is MY role, not the role of the dice.

You seemed to solve Nettie a much different way than I did.
Go along with everything, when she says "Nope, die" beat her up, and use knockout attacks, so you can still feel like a good person.
Side note, was this the first room to use crafting for everyone?

Combat isnt determined with one die roll. It seems weird to want social encounters resolved the same way. They should be resolved with X successes before Y failures though, so its not a one failure ends your chances like it is currently.

So ive spend the better part of the last hour trying to win that er..... Conversation? And 3 guesses how many times I managed to pass it? xD

Yeah you guys are right about the implementation of it. Its horrid. Even with guidance which gives a D4 bonus on each skill check I dident get past the 3rd skill a single time.

What I noticed is that the skill checks are just there to be there. You cant try to convince her by forming a narrative or by making counter points to evnentually pursuade her. You just have to keep hitting your head against the brick wall until you either fail and have to attack her or by some miracle will eventually get through to her; which I havent been able to do once while trying to do for an hour straight of quikloading after I failed. Even if the rolls needed to pursuade her are low, eventually you will fail and thats that.

The concept is fine. The way its included needs some work.

To me this reads like you just want to click the usually blatantly obvious "right" choice so you can dump CHA, social spells and skills. You don't get to pick the "kill them" option and have that automatically work either in most cases.

Honestly, if they wanted to have multiple skill checks to swing the conversation back and forth and put some suspense into it before resolving based on your aggregate, that would be an interesting approach. It would make conversation feel a bit more like combat, and even though it breaks from DnD it would still make sense and be satisfying to play out.

The approach they got now though is just frustrating and stupid. What were they even thinking?

Table top players realize a single die roll can change a good/bad situation instantly. When a DM has multiple rolls it being used to create tension like the Nettie situation. If the tension isn't there, multiple rolls are being used because a DM wanted a different outcome but the player either rolled too good or too poorly. Never let something really important be decided by the dice alone. Meaning, a DM shouldn't let something occur in which only the dice can get a player out of. The player needs to make several mistakes on their own and perhaps that one last die roll determines their fate.

I think the idea is fine but the implementation should be adjusted.

Maybe have each test be you alleviating one of Nessie's worries so she eventually concedes, rather then you using things like intimidation to try and pursuade her. Which in this situation would probably do the exact opposite of help your case here. Youre basicly confirming to her that you are a threat rather then debunk it.

It would make you think but still give you ways to pursuade her; some might be more effective then others. For example if she mentions the former druid making an exact point it might be a good start to drop some of the quests as you are making the same point her master did. Some points might not need a test while others (like pursuading her that you arent a threat) might be. Then they could make that one hard to be convincing like it should be but make the rest irrelevant or even easier. That said easy tests can still fail and with RNG arent optimal. I think personally that 1 or 2 tests there or 1 if she notices you make the same point that the missing druid did would eleviate her worries. Or at the very least ease the conversation enough that outright tests arent needed to make your point. Sometimes less is more and I think that applies here.

Now its just a stack of test that you have to complete and what options you pick doesent seem to matter. Which basicly means that 999.999 out of 1.000.000 times you end up having to attack her.

In this case the player did make multiple mistakes though?

They casually told a healer that they have a illithid tadpole in their body.
They followed said healer to a secluded room.
They let the healer slash them with the *cough* antidote.

Prior to the last point the players could back out and escape the 'were fucked' situation and walk off.

The implementation needs work but the situation here is something that does warrant multiple tests. It just needs to be tweaked to make sense and be more forgiving. Its insane in its current implementation...

+1

i would stress the point that repetative checks for the same outcome results in repetitive lines of dialogue. Nettie case is awful, somehow however tadpole case is not, but that's good wrinting covering bad design i guess

Here's how I would design the encounter for a tabletop game:

Introduction
Medicine check succes: Hm, this druid can't even heal a bird fully. This is probably a waste of time.
Option to leave.

Walking towards backroom
Insight check succes: Her mood has darkened considerably and she doesn't meet our eyes. Something's afoot.
Option to leave.

When entering backroom
Nature check succes: That vial she picked up was wyvern venom.

Nettie speaks again
1+ succes on previous checks: Call Nettie out. Persuasion check to pinky swear to poison self, or murder hobo option
0 previous succeses: Get poisoned. Deception/Intimidate/Persuasion/murder hobo to get antidote.

This I feel will give several character builds a non-lethal way out while building tension, and it won't serve four match points against you.

What if the skill checks were to make the final make-or-break one easier/harder? So with 3 skillchecks, if you pass the first 2, the third gets a +2 for your roll, or something?

Lowering the DC of the final check with successes would work, but you'd have to allow the conversation to proceed, success or fail, instead of one failure and you're out.

I was lucky and passed all rolls first time I talked to her. But yeah those multiple rolls for one purpose seem like bad design to me. Set it up differently or give me one roll which reflects the same low chance of passing 4 rolls in 1. If I see that I need to roll 18 I know that this was meant to be really hard outright.

This kind of thing starts as soon as the wounded mindflayer on the ground by the way.

I would be fine if they implemented that scene in that manner. When I first encountered that scene I was a bit annoyed that my Ranger character didn't have the option to identify the plant she was holding. Up until she tells you she is trying to kill you there is little to say that doing as she says isn't a bad idea. Then you have to go through the multiple skill checks and frankly, if I'm successful at a "give me the antidote or I'll kill you" type intimidation I expect the result to be a cowed and compliant Nettie not an indifferent one that I have to then intimidate again. Logically, multiple threats of "give it to me or die" should be less threatening not more.

I think the entire multiple successive skill checks for a single action and encouraging a murderhobo approach via mechanics is a terrible idea and needs to be rethought. The only reason I can think of for them doing this is that they wanted to make it a difficult check, but because of the way they have implemented skill rolling they need to ensure a final target number between 1 and 20.

I did eventually manage to succeed at the conversation once, which required a persuasion (DC 9), deception (DC 11, no persuasion option) and another persuasion (DC 9) roll. I assume this means a base DC of 10 for each check, my character had proficiency in persuasion but not deception and despite having a 10 charisma seemed to be suffering from a -2 charisma penalty (from the poison?). On straight rolls this results in an 18% chance of success, with disadvantage (which I assume was also applied due to poison) this drops to a 3% chance of success. For a single skill check this would require a base DC of 18 for persuasion, a non-proficient skill check with an 8 charisma would therefore result in a DC 20 check.

Careful, you might confuse Larian with character agency based on our skills and the ability to use them

The difficulty and multiple-die-roll questions should be disaggregated. If there was only one check in the Nettle convo, it would have a hellacious DC. It's not like the devs didn't know what the probabilities were.

Definitely agreed on this. It feels like I'm being tricked by the game because it intends to have me fail instead of being given an actual chance to succeed.

+1.

And having 2 relatively "big" encounters where this happens in the first 2 hours of gameplay is "a bit" too much to be considered a coincidence.

Good vs. bad outcomes should be 50:50.