Roll20 uses cookies to improve your experience on our site. Cookies enable you to enjoy certain features, social sharing functionality, and tailor message and display ads to your interests on our site and others. They also help us understand how our site is being used. By continuing to use our site, you consent to our use of cookies. Update your cookie preferences .
×
Create a free account

RNG Errors

During the last month or so when rolling D10's and D100's our group keeps encountering repeating (as in getting the same roll value multiple times) numbers as well and Max and Minimum values following each other in sequence (In other words getting a 1 followed a by a roll of 10 or 100 on 1d10/1d100 respectively.) Because 1 and 100's on percentile system can either be critical successes or Critical failures their regular occurrence makes roll20 occasionally frustrating (as we have to keep re-rolling until it returns to normal) or almost unusable as it will sometimes repeat up the same roll up to six times greatly slowing down game play and especially needlessly inflating combat encounter times. This has happened in multiple campaigns, my own and two others I participate in, and though I love Roll20 its beginning to frustrate the other players greatly. The following composite screen shot is from two campaigns the first two from one, the third example from another. I would have added more but I couldn't get the chat archives to load so I just loaded the individual games and scrolled up for the most recent rolls. Is this something on our end that might be tweaking the RNG unintentionally? Is lag a or the GM's connection a factor on the rest of the groups rolls? My stats- OS: Windows 7 (64-bit) Browser: Firefox 15.0 Flash: 14.4.402 JavaScript; Enabled Worth noting: Other members of our party are using chrome, though I don't know which version.
Do you know when these rolls were performed? Was it before or after the August 29th update?
The first two examples were today at approximately 5pm EST, the third was a week or two ago so before the update. That said we have had similar repeats happen within the last 48 hours I just couldn't get the chat log to load for me when I was trying to get screen caps. I can list the campaigns Ive seen it happen in if that helps you.
Also, how far apart are the rolls in the first two examples? Are they done at the same time or a few minutes (or more) apart?
Generally within a minute of each other, usually when the GM calls for more than one of us to roll.
Are any of your players using the 3d dice?
Most of us actually, we've enjoyed it quite a bit despite the occasional error.
Okay, so here's what we're thinking on this. First off, the far-right screenshot is irrelevant at this point, because we completely redid the RNG stuff in the August 29th update, due to exactly that problem (the old RNG didn't have good distribution of numbers, which could lead to streaks like that). So that leaves the other two. The two d10's rolling min/max right in a row is likely coincidence, especially since those were rolled with (I'm assuming) 3D Dice. Small sample size and all that. Let us know if you keep seeing that problem specifically with the 3D Dice, but since it's more of a physics simulation than a instant RNG operation, it may be a bit more "streaky" than the normal dice roller, although over time the results should have a suitably random distribution. The thing that's concerning to us is the two players getting the same dice roll result in the same order. The way that PRNGs work is that they are predictable -- if you start with the same seed, you always get the same numbers back in the same order. In *theory* we're randomizing the seed so that it should be nigh on impossible for two players to get the same one (and therefore it should be nearly impossible for two players to get the same results in order like that). So we're wondering if we've somehow missed something with our seed generation. Again, the problem is no longer one of bad distribution of rolls (getting a 63, then 59, then 100 is fine from a random-number perspective), the concern is that two players got the exact same results in order like that. On the other hand, again this is a small sample size, so it could just be a coincidence. But we're going to assume for now it's not :-) I just pushed out a new update where we tried a slightly different approach to that problem, so let us know if that fixes things for you in your next game. I would also recommend disabling the 3D Dice for now (I know, sorry!) if you're really concerned about this, since there's a lot of new code there, and even though the d100's aren't actually rolled in 3D yet, it's possible that just going through that new code could be causing an issue. We'd like to determine if this is truly a problem with the new PRNG, or if it's something to do with the 3D Dice code. Thanks for answering our questions so promptly :-)
Thanks so much Riley! (and Eric too!) I wanted to provide a larger sample size but my connection right now is not behaving so I pulled the best examples i could at the time, I agree the 1 and 10 may be just coincidence and I appreciate you looking into the issue. I'll see if i cant get some of my players to run some tests tomorrow night with 3d dice off and see if we still get the same errors, and if so I'll provide a larger sample size. we will miss the 3d dice but we will survive :P
Well it should only a be a temporary loss :-)