Dice broken tonight?

Our group has rolled an inordinate number of 1s and 20s tonight. It's almost like flipping a coin. I'd say out of 100 rolls, we've had 15 20s and 10 1s. It's really crazy. Anyone else noticing any issues with rolling on the prod server tonight?

Sounds like a run of luck (good or bad). 100 rolls is not a large enough sample size to indicate a problem.

I'd always encourage anyone questioning the randomness of the dice engine to do a little reading on the subject: <a href="https://app.roll20.net/forum/post/63487/from-the-s" rel="nofollow">https://app.roll20.net/forum/post/63487/from-the-s</a>... And know that part of the Data Delve update is going to involve really publicizing the rolls in a way that will make it even more apparent that the issue is perceived. There were 800 games played last night in Roll20, which means there's a lot more rolls happening across the spectrum....

our HeroQuest GM said: I kept rolling ALL WHITE SHIELDS. 10 IN A ROW all told. (proof in pictures.) I don't even know what the odds are of this happening, but I'm sure it's astronomically small. I would equate the improbability is in the neighborhood of whales and potted plants falling simultaneously from the sky. (For the curious, the dice - a d6 - has 3 Skull faces, 2 White Shield faces, and 1 Black Shield face. Someone do the math on that.) I even double checked my table weights, to make sure something wasn't screwy. Nope, 50, 33, 17, all weighted correctly. Nothing was wrong but it was a rarity to be sure. He even took screenshots to prove it.

The odds of rolling 10 white shields in a row (1 in 3 on a d6) is 1 in 59,049. With your GM's setup the odds are a bit different (33% rather than 33.3333...%), 1 in 65,292.

I've now been in multiple games where we've rolled an improbably number of 1's in a row. I had 5 in a row one game and another player experienced the same in another game. Plus I've seen this in other games. Statistically, almost impossible. The dice engine is broken. Hopefully they have a fix soon for it.

Bruce, could you define "statistically impossible"? Even if you rolled 5 1's on a d20 in a row that is not statistically impossible. In fact, it is nearly guaranteed to happen to someone, sometime. The odds of 5 1's in a row on a d20 are 1 in 3,200,000. Better odds than winning the lottery and we know people do win that. The dice roller went through extensive testing, the link has been posted by Nolan (above).

That's why I said "almost" impossible. In the last several games I saw exaggerated examples of what is being brought up here over and over again. It was amusing to here the players complaining that they've never thrown so many 1's in a row. Other people used to seeing that as of late in Roll20 just simply lamented with them and said that this would be fixed hopefully in an upcoming release. I've seen this to a lesser extent with 20's. With 1's, they seem to statistically show up waaaaaaaaaaay too often. I forgot how many 1's got thrown (by a different player) the last time I played, but it was enough to get noticed and commented on the near impossibility of the situation if throwing a real die. I really like Roll20, but something is likely very wrong with the random die throwing algorithm. As a software developer myself, I don't believe in statistical coincidences like this. Something is amiss.

It's not broken. It's not wrong. There is absolutely nothing amiss. There are hundreds, if not thousands of people rolling dice all at the same time as you.

Just a note regarding throwing a real die. Most real dice are flawed. For example, Chessex dice are usually slightly oval shaped with the long axis on the 1-20 axis and the short axis on the 10-11 axis. This means you will roll fewer 1's and 20's than you should. You can test this with a micrometer. Additionally, some dice are weighted. You can test that by creating a salt water solution with a high enough salt content to float your dice. Then spin the die with the spin axis being on the plane of water. If, near the end of the spin, it changes spin direction then it is weighted. The faster the change the more it is weighted. Even precision dice are flawed although they are less flawed than Chessex dice. Regarding statistically showing up way to often, do you have at least a thousand rolls worth of data? A million rolls? Until you have a sufficient sample size any pattern you see is just human nature.