Larian Studios Forums Baldur's Gate III Baldur's Gate III - Suggestions & Feedback please the truth, dice result really random

If you look at this webpage it gives the likelihood of a result for straight d20 rolls, with Disadvantage, and with Advantage. If you look in the box for Advantage on a 5+ you see the probability is 96%.

Advantage/Disadvantage Probability

Yes, all the data in that post is for weighted rolls. If weighted dice only function to reduce streaks, then it should become an even distribution faster than Larian's true random.

Here is the chart from patch 3 (unweighted).

If I said I had a seed for your goodness of fit, would you salt my validation? Here's the catch! I did the same thing you're doing with a real deck of cards. And I was really actually unlucky for a whole night. Everyone else was using the same deck of cards. And how could they know where I was going to sit?

@DragonSnooz ,

I thought about this, and I'm going to help you. If you think the dice rolls are non-random, it could be something personal to you. And if that's the case, you're the center of attention for the magical effect. And therefore your response to it matters for all people interested in the product of that effect.

P.S.
Things Math Professors Never Say!

I'm confused, what makes you think I'm in the opinion that dice rolls weren't random? The weighted dice setting is a form of pseudo-RNG. But it still has randomness to it.

There is no room for opinion when the correct answer is that true randomness does not exist. The more I learn about how randomness is a presumption, the more capable I am in knowing what will happen if I do one thing or the other, and the more confused the rest of people will be about why the dice rolls seem non-random. All randomness is stupid. You're replying to Wofenring. I'm Gristly-Knuckle. Well met.

Ah. I was of the mindset that weighted dice made it more likely to roll a middling number, not reduce streaks.

Thanks for the unweighted chart. It's Chi^2 is 23.5 which is still below the 95% confidence value of 30.14.
The most extreme values in that chart are 6 (4 times) and 17 (16 times):
The chance of rolling at least one number only 4 times is 3.5%.
The chance of rolling at least one number 16 times is 0.8%.
Neither of these are too incredibly unlikely.

BUT! If we combine your unweighted chart with @Saberem's unweighted data, we get a Chi^2 of 32.26 which is GREATER than the 95% confidence value of 31.14. Thus, we can conclude that Larian's unweighted dice rng is NOT random.
The most extreme values were 6 (7 times), with a ~1.8% chance of happening.
and 17 (rolled 28 times), with a ~0.02% chance of happening. This is pretty unlikely.

Math

I still have no idea what point you're trying to make.
I think we're all in the understanding that RNG is not true random. It's just easier to refer to Larian's RNG as such relative to weighted dice. Larian stated they were testing a form of true RNG in early access to see the players' response. Currently we have a weighted dice option and we're are reviewing how random either are. Sure nomenclature was juggled when comparing different forms of RNG, but I'm still confused by your posts.


I definitely prefer weighted dice as RNG, I was getting tired of seeing 5, 10, or 17 xD

The point I'm trying to make is that I have my own dice. Larian should just let us type in our own die results. If you thought I was trying to make any other point, then you are wrong. That is the right answer.

Alright, you do you.

Hotfix #10 "From now on, loaded dice will only bend RNG in the rolling character's favour. That means you will not be made to miss to make up for a lucky streak of hits."

So my understanding that means the target roll affects the outcome of our rolls. I was curious to see how this change affected our roll distribution.

Here is my comparison against the same low & high AC target:

Hotfix #10 Weighted Dice Target 4 (AC 9 - 5 Proficiency 85% chance to hit)

Number of rolls 3 or below: 7
Number of rolls 4 or above: 93
Longest Streak 3 or below: 1 roll

Number of rolls 11 or below: 45
Number of rolls 12 or above: 55
Longest Streak 11 or below: 7 rolls

Hotfix #10 Weighted Dice Target 12 (AC 16 - 4 Proficiency 45% chance to hit)

Number of rolls 3 or below: 8
Number of rolls 4 or above: 92
Longest Streak 3 or below: 2 rolls

Number of rolls 11 or below: 42
Number of rolls 12 or above: 58
Longest Streak 11 or below: 2 rolls

Comparison Chart: https://imgur.com/a/BXTd7NI

I could have been extremely lucky when I tested on the higher AC target but the overall distribution was similar to the low AC target, except my rolls below 12 streak was reduced from 7 to 2 which means I never missed more than twice in a row even against the higher AC target.

The chart shows that against low AC there is a more even spread of rolls above 4 because any roll above 4 should be unaffected. Against higher AC, the distribution is more above 12 which seems to confirm that the change in Hotfix #10 is doing its job and rolling more hits than misses.

Some of your replies actually made me chuckle in combination with your portrait, it's quite funny imagining the figure in your portrait having eloquent meta-magical philosophical monologues on statistical theory.

Anyhow, just wanted to add a +1 to the option to add your own dice results. I plan to play BG with the gf and it would be such a nice and social little extra to be able to roll our own dice in the sofa in front of our screen. Def needs to be optional though, no need to needlessly anger any part of the fanbase who might find this sacrilege.

Originally the weighted dice just had added logic to reduce streaks for all outcomes [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20].
The recent hotfix changed that logic to ignore streaks above a threshold, (let's assume it's 10). So [11-20] are allowed to streak but other outputs are not, this will undeniably bias dice rolls in the player's favor. Which your chart shows.

Original Post