Are dice weighed? Is that average D20 roll in combat "fair"?

[KocLobster]

Over the past few weeks, I have noticed that all of my companions seem to roll natural 1's...kind of a lot. I've tried my best to be objective, to truly observe the scope of all rolls and not get tunnel vision when rolling 1's just because it is more significant/memorable & sticks out.

But it really does seem like rolling a natural 1 is too common. Without the "Always a Chance" mythic ability, even having insanely high attacks rolls are irrelevant when rolling a 1. And it just feels stupidly common.

On attack rolls, it does not feel like each side of a D20 is weighed equally. It appears that rolling a 1 is much more common than the 5% chance it should be. Is this on purpose? Does anyone else experience this?

[pckirk]

yes, multiple people report this same thing every week or so..

[Orion Invictus]

Confirmation bias. Yes, sometimes you get clusters of nat 1s, but you also get clusters of every other number. You only notice the extremes because they're the only ones that are noticeable due to the game's mechanics.

[Stormsong]

Players and mod makers have done pretty extensive testing on the die rolls. They are as fair and "random" as they're going to get being computer generated. IIRC, the roll doesn't even know what entity is doing the rolling (like it doesn't differentiate between friendly and hostile). You can find more specific information if you flag down one of the mechanics modders on the discord. Well, if they reply anyway lol. Busy folks with the EE around the corner.

[KocLobster]

Originally posted by Orion Invictus:

Confirmation bias. Yes, sometimes you get clusters of nat 1s, but you also get clusters of every other number. You only notice the extremes because they're the only ones that are noticeable due to the game's mechanics.

I thought I was being objective but you definitely could be right, maybe the nat 1s just stick out so much more that they seem more common.

Originally posted by Stormsong the Fallen:

Players and mod makers have done pretty extensive testing on the die rolls. They are as fair and "random" as they're going to get being computer generated. IIRC, the roll doesn't even know what entity is doing the rolling (like it doesn't differentiate between friendly and hostile). You can find more specific information if you flag down one of the mechanics modders on the discord. Well, if they reply anyway lol. Busy folks with the EE around the corner.

Thanks for the info! I will continue to just grumble when I get nat 1s then lol

[Maple]

5% of the time is actually pretty common. For every 100 rolls you make- which can be 1 or 2 large encounters, especially as the game goes on and more attacks are made a turn- 5 of them are going to be critical misses.

And of course, statistic abnormalities happen. You've got to roll the dice not just hundreds, not thousands, but millions of times to wrinkle out the statistic abnormalities.

It feels bad because most video games *do* weight the odds, in the player's favor. It is well documented that video game players by and large do not have a very good sense of probability, so they weigh the dice rolls in the player's favor to make them feel fair. This game very clearly does not.

[Gregorovitch]

The RNG in this game was extensively tested a year or so ago by the player @mread2 who wrote a program to extract rolls and analyse them from real combat log files submitted by a number of other players. There is a long thread detailing his work and the results on this board.

The long and short of it is he proved beyond reasonable doubt that the RNG in this game is totally fair.

[Uzkin]

How many times one would have to roll to "find out statistic abnormalities" depends entirely on how severe those abnormalities are.

If, for example, the d20 die was completely biased and only gave 1s then you'd only need about 4 rolls! First you'd declare that now you start to collect data, then you'd roll 4 times, record 4 x 1, and you'd already have large enough sample size to conclude that the die is flawed beyond any reasonable doubt.

[Maple]

Uh. You absolutely need to roll more than 4 times to come to that conclusion. I regularly have stretches where my real, physical dice roll multiple 1s in a row. It happens, startlingly often. Even when we're not talking statistical certainty but just "beyond reasonable doubt", you gotta roll more than 4 times, lol.

[Uzkin]

Nope. You are thinking of a different scenario where you just keep rolling and rolling, and when a string of three 1s (or so) occurs you pay attention to it.

Statistics can be counter-intuitive but it can go both ways, as my example shows.

[Maple]

No, like, if you want to determine the dice is flawed and only gives 1s, you gotta do more tests than that. That's not enough tests to determine if the dice only gives 1s or if it's just a bad string of luck right at the start. You absolutely need more data to come to that specific conclusion.

[Orion Invictus]

Originally posted by pkdragon:

No, like, if you want to determine the dice is flawed and only gives 1s, you gotta do more tests than that. That's not enough tests to determine if the dice only gives 1s or if it's just a bad string of luck right at the start. You absolutely need more data to come to that specific conclusion.

The odds of getting four 1s in a row on a d20 are 0.000625%, or 1 in 160000. If you got four 1s on your first four rolls with a specific d20, you could (and should) suspect foul play.

On the other hand, getting four 1s in a row on a d20 in, say, half a million rolls is not only likely, it's expected.

[Maple]

I'm sorry, even 1 in 160000 odds seems pretty high when we're looking at the amount a dice gets rolled over it's lifetime. Yes, I know you're not looking at that because you're framing this in a very limited, weird way, but that's what we've gotta look at to determine if the dice is actually busted or not.

Like, that it happened when you were specifically recording it is a surprise, but I just can't imagine saying "yep, dice is broken, it only rolls 1s" without rolling it at least 10 times. And even that is statistically way too low a sample size but that's where I'd feel comfortable making the call.

[Orion Invictus]

Originally posted by pkdragon:

I'm sorry, even 1 in 160000 odds seems pretty high when we're looking at the amount a dice gets rolled over it's lifetime. Yes, I know you're not looking at that because you're framing this in a very limited, weird way, but that's what we've gotta look at to determine if the dice is actually busted or not.

Like, that it happened when you were specifically recording it is a surprise, but I just can't imagine saying "yep, dice is broken, it only rolls 1s" without rolling it at least 10 times. And even that is statistically way too low a sample size but that's where I'd feel comfortable making the call.

Statistics are not intuitive, as that other person said.

Rolling four 1s in a row in a vacuum is extremely unlikely, to the point you should suspect something's up.

Rolling four 1s as part of a much larger (known) set of dice rolls is fine.

You cannot just assume that the dice has been rolled a bunch of times in the first scenario because you don't know, nor do you know the results. At the same time, past dice rolls do not influence the probability of the next dice rolls, so even if it has been rolled a million times without ever landing on 1, that doesn't somehow increase the probability of the next roll being a 1.

[Immortal Reaver]

If we are talking about real dice you can put them in water if they float and push them around to see what they roll, if they always roll over to same place, it is not properly balanced.

Agree with pkdragon, 1 in 160000 is very high chance. You definetly can roll more dice in 1 playthrough, so it should happen few times.

Also do you know what chances are to roll 4,15,17,6? Yes 1 in 160000. But that "feels" like right like randomly rolled. Right?

[ScrapMettle]

Check the statistics page for each character and you can get stats on how many critical misses, hits and whatnot. I felt my own character was getting an insane amount of critical misses... but looking at the stats it wasn't as bad as I thought.

Sometimes you just have a string of bad luck. Playing D&D recently myself I got 3 1's in a row. So did another player I was playing with in the same session.

Honestly without the physical element of dice somehow being flawed or weighed slightly wrong I think digital dice should give closer to "True random" value.