Dice roll randomizer

I was wondering what the equation for this was or how exactly it was working? I found a post from a while back saying they had increased the spread of what should be rolled, and I personally have not had a problem... but my players have. I get it is a game of chance in a sense but my players have been complaining so I figured I would ask if anyone else had terrible luck with D20 rolls specifically. I play open ended D20 where 20 is of course crit or success and 1 is auto fail. My players even counted back in the log last night, something in a few thousand rolls only warranted about 60 twenties in total, but over double the amount of that in ones. I also have seen them roll 1, 1, 8, 1, 11, 1, 3 back to back to back. So for their sanity and sake I told them I would check to see if there have been any issues with this or how it was running, Thanks. Had a lot of player stress and bad luck going from actual dice to Roll20 rolling.

It was one of my players that counted it up, when we first started they made some really big cluster rolls as well so he pulled from that pool too I believe. I think it was like 7,000+ or so rolls and in total only something like 60 something 20's between  all of them. At the same time they rolled about double or more of that in 1's. It isn't so much the problem they're having rolling more 1's and/or 20's, I think it is the fact they have seen so many more 1's than 20's if you compare the two. I was just wanting to check for them and see how it was working to prove that it was in fact random. (player aggro) I assumed it was random and explained it was a randomizer, even found the post stating where you had done tests and expanded on the spread you should see from the rolls. Was just wanting to check for their sake and see if I could get any information to prove it. Thanks! My players just have bad luck as I assumed. :P Will link them to this though for clear proof so they don't just trust my word of mouth alone.

The global stats would be good I think. Or perhaps a "dice tester" where you can just enter the die roll you want to test, the tester runs it an acceptable number of times (10k would probably lag the user a bit, but give a fair deal of spread) and show a graph. Anything that would show the distribution in a clear and easily interpreted way (for those human thingies that didn't enjoy taking statistics & probabilities as a minor) Seriously though, while the dice here aren't 100% random, no dice are. These are far better at being "fair" than any real dice you will ever see. Anywhere. (cause physics are mean)

Yeah I get ya. I don't think they so much wouldn't believe me when I told them it was random, they were just in a bad spot to have a streak of ones and were raging about it. Linking them to this thread and article should do it for them though, just wanted some solid proof that I could show them should it ever come up again. On that same note I have yet to have terrible luck with the system personally, rolling in my favor that is. Anyways, thanks again for the help. I would like to state on that note that I personally love Roll20 and as the GM it is great. Maybe the fact I have had a hot streak the last few plays and they have had a poor string of bad luck, and not just one or two of them but most of the party would get in a 1 rolling rut. But eh, it happens, fate of the dice! 

I think this falls under the same situation that you may run into IRL.  You roll a die a few times and you end up getting alot of low rolls, so you switch to a new die to see if it hits better for you.  Roll20 unfortunately doesnt have confirmation bias functionality built into the system.

Is roll20 using a PRNG or a TRNG to power its underlying rolls?  The simplest way to prove that Roll20's rolls are truly random would be to use a service like random.org to make those rolls, and just be open that you are using it.  All of my online dice rollers use random.org, specifically, because its API is incredibly fast (I can make a request for 10,000 d20s in under a tenth of a second).  If you make the request as your client's IP, you won't run into ratelimiting, either.