RNG

I have read the previous threads about Roll20's random number generator, and about QuantomRoll.  It all sounds fine, and it's typically okay, but our group have encountered situations that are statistically unlikely.  I had a group of 5 who were in stealth and ambushed the party all roll a 1 on the attack in a row.  One of my players rolls a 1 nearly every round.  Then I have another player who rolls pretty consistent 20s.  It just seems like we're seeing rolls in the extremes far more often than we did when playing PnP.   Does anybody else have similar experiences to this?  Is this normal, and is there a fix if so?  Should we use a different RNG for game play?  Thanks for any replies.

Everyone will usually have similar experiences to this.  Its normal.  There is a chart with the randomness of the rolls of roll20 and the distribution looks fine (I dont remember where it is exactly). If you are concerned, just have your players roll real dice and announce their results if you want?

I was considering that, and I did see the chart you mention.  But the chart seems to represent an extremely large number of rolls.  Again, I've never seen that many 1s in a row while playing at the table.   But thank you for your advice.  Maybe I'll give it another session before I look into switching to real dice.

You might like to read this to set your minds to rest about Roll20's random numbers:  Introduction to Randomness and Random Numbers Runs of 1s and 20s are entirely to be expected. Runs of any numbers are to be expected. In the long run, I think you and your players will find the convenience of rolling on Roll20 and automating your game using Roll Templates or PowerCards much nicer than rolling your own dice and doing your own arithmetic. My players and me certainly do in my 13th Age games.  However, you might like to suggest a per user record of dice rolls and a page that shows the bar charts just like the system wide one. This would set people's minds at rest in the long run. Being able to see how all your D20 rolls even out over time would be comforting, I think. 

Flannigan , you might want to check out the QuantumRoll page of the Roll20 Wiki. QuantumRoll is actually more random than physical dice . Further discussion and readings on this topic: <a href="https://app.roll20.net/forum/post/63487/from-the-source-roll20-rng-testing" rel="nofollow">https://app.roll20.net/forum/post/63487/from-the-source-roll20-rng-testing</a> <a href="http://cocosci.berkeley.edu/tom/papers/hard.pdf" rel="nofollow">http://cocosci.berkeley.edu/tom/papers/hard.pdf</a>

Random numbers optimally should not produce patterns , each time a generation occurs there is an equal opportunity for all of the possible numbers to appear. You could make an absolutely perfect cube with the numbers 1 through 6 on the sides. Rolling that cube ( you could call it a die ) will give any of the sides with equal probability on each roll. Rolling 1,000,000 times with no 6's rolled has no effect on the next roll, there is exactly a 16.666...% chance for a 6 to come up. In order to prove a system random, you need thousands of iterations, so years of game play. People tend to notice what appear to be exceptional results, not the huge number of average results. Nobody finishes a game remembering how the dice grouped around the averages, they remember when they rolled three critical hits in a row, or conversely, when they have a run of bad luck.