Low Die Rolls?

[malvok]

Has anyone else had issues with the Fantasy Grounds dice often rolling very low numbers? In my gaming group we roll low so often that there is a running joke.

Player 1 rolls dice with a very low result.
Player 2: "Didn't you buy the dice expansion yet?
Everyone groans.

Once we did some rolls to see who, out of a group of 5, would make it to a door before it closed. A result of 1 on a d6 meant that PC or NPC didn't make it in time. We rolled 5 1's in a row.

It is not unheard of for a player to roll nothing higher than a 7 on 2d6 all session long.

It could just be very bad luck, but we've noticed that the dice in Fantasy Grounds seem to just like rolling low numbers. If they were physical dice I'd throw mine out.

This isn't meant to be a rip on Fantasy Grounds, I love the program. I'm just wondering if anyone else has had similar issues?

These sessions were conducted with both the 3.5e ruleset and a modified version of the 3.5e ruleset. No alterations in the modified ruleset changed the default dice.

[JohnD]

Funny the players in my campaign were complaining about all the high rolls I was getting last time....

Can't say as I've noticed anything myself though.

[Trenloe]

There have been a few threads over the years mentioning this, the most recent one here: https://www.fantasygrounds.com/forums/showthread.php?t=18042

In the end most of it all boils down to sometimes there are good dice rolling sessions and sometimes there aren't. This is just like my experience with face-to-face game dice rolling, so I like it. Personally, I wouldn't want rolls in a game to be statistically average all the time, that wouldn't be as entertaining as good and bad streaks.

[grapper]

This is most likely a clustering illusion (https://en.wikipedia.org/wiki/Clustering_illusion). Streaks in random results are pretty common because a proper random number is not influenced by the previous results. In fact, the absence of streaks is often an indicator of someone falsifying data.

[Mellock]

Joshuha once made a script that rolled the dice 100000 times. The results are on https://www.fantasygrounds.com/forums/showthread.php?t=7632&page=3 around post 22. It had a pretty good distribution.

[Velocinox]

What we need is for Smiteworks to license Gamescience dice from Lou Zocchi instead of those tumble polished dice...

Dice School Part One (https://www.youtube.com/watch?v=bR2fxoNHIuU)

Dice School Part Two (https://www.youtube.com/watch?v=PxmkWrDbn34)

And if you're wondering if this guy is for real about being around since 1974...

Check out this joke post (https://www.fantasygrounds.com/forums/showthread.php?t=18594) I made about old D&D ads... look at the second one and check out the second to last distributor in the far right column... That's him!

[Mellock]

I have 2 sets of gamescience dice, and they're very pretty. Unfortunately they also cost me 40 bucks in shipping, and another 10 for customs. Still, they're my favourite dice.

[grapper]

Joshuha once made a script that rolled the dice 100000 times. The results are on https://www.fantasygrounds.com/forums/showthread.php?t=7632&page=3 around post 22. It had a pretty good distribution.


Let me preface by saying that the discussion of the linked article is off topic from the OP, which was about streakiness.

Second disclaimer: The linked thread is from 2007, things may have changed in the program since then.

So, I'm going off topic to discuss this, but wow. The die results posted in the above thread certainly are skewed. Not by a lot, but the bias is consistent. Granted, the bias probably isn't any worse than a real 20-sided die.

The posted data all show: 1, 2, 5, 13, 14, 18, 19 and 20 all come up more frequently than expected, while 3, 4, 7, 10, 11, 15, 16 and 17 come out less frequently than expected. The worst was 11 which consistently appeared only about 4.7% of the time. (6, 8, 9 and 12 were all close to the expected 5% chance of occurring)

As a sanity check, I generated 2 million values using Matlab and looked at the percentages of occurrences. As you can see in the plot, the rand function results cluster much closer to the 5% value than the FGII die rolls.

Out of curiosity, I know that that the die roller is done using a physics simulation, but there must be some kind of randomization in the initial conditions, especially for script generated rolls.

I certainly found it very interesting, but I'm not sure if it is something worth worrying about.

https://home.comcast.net/~grapper/d20probability.jpg

[Velocinox]

So you called me out for being OT to post some data that you're...


not sure if it is something worth worrying about.

lol, don't ever let it be said math geeks don't have a sense of humor, because you certainly made me laugh... well done, sir! :D

[grapper]

LOL

It wasn't meant as a call out, but more of an apology for geeking out on the data.

My suspicion about the data is that the random number generator used to initialize the die roll may not be as robust as we would hope. For anyone interested take a look at Chapter 7 in Numerical Recipes (https://www.nr.com/), especially the history section.

[Velocinox]

Hell, don't apologize, you were right, I was OT. Especially when you can back up your post with charts and graphs. I thought the thread was over, I didn't know you were waiting to drop the probability bomb on the thread...

[Moon Wizard]

Yeah, the alignment (i.e. rotational position) and the 2D throw vector are randomized for each die rolled at the beginning of the physics simulation.

Regards,
JPG