Not so new topic of bad dice rolls, oh yes this old chestnut is here for roasting again and I’m not sorry for posting it. ;)

Roll20 random Dice gen.  This is going to be fun.  As soon as someone writes Player/DM/GM/All suffer with bad dice the posts ends with a Mod stating it’s not Roll20 it’s got to be something else. As a player or DM/GM that’s not very helpful and in fact a bit on the rude side of things. Then they carry’s on and says look at the number of rolls created by our wonderful dice god of a system compared to the amount of post and comments that people actually complain about the problem, so it can’t be our system, like just because most people don’t complain there’s the proof the system is great. Or our system is better than really life randomness you only have to look at the Monte Casino effect or my most favourite the Will Wheatley effect. Any of these problem can be explained away with either a false understanding/logic/maths/bad coding or a good solicitor. :) But still every now and then a new posts raises out of the dead of the last one and it all starts again. Let’s try and stop it from happening again, where the player, DM get a review of the system and sensible answers to this problem as it is a problem.  Roll20 is here to help us have fun at our past time not to make it stressful and possibly drive people away from an amazing hobby. Being told by people it must be me or its something I’ve done/not done is wrong. Roll20 needs to deliver a better solution not nerf the player by allowing Mods and other members of the community to make the person making the complaint feel like it’s their fault and it’s just negativity not a real world problem. Maybe their more random than a    real world dice system needs to be nerfed to deliver the results people want in their games. Just like Apple did to the iPod random music player.  I have never moaned to anyone outside of the group of fine people I game with about my bad luck with the Roll20 system. I never posted about it anywhere before, I can imagine there are many more people like me that just go along with it and then enjoy the banter at the table. That was until today after reading some threads about this on Roll20 and I finally got annoyed enough to finally stop shaking my fist at the screen and denouncing the small gods of Dice and all things random and finally, finally put fingers to keyboard. Now this cannot just be linked to me and what I’ve done to my character sheet or not done, as others in my group have similar fortune as I and not in just one gaming system or even in one campaign, we play D&D red box and Star Frontiers. I am guaranteed to have the lowest rolls to succeed or pass saving throws (and I’m playing a 6 level Dwarf) in D&D and the highest in Star Frontiers, the system also are different in dice D20 and D100. I don’t type anything in on the prompt just use the dice roller icon all the time.    When I check the roll with mouse over there are no unusual characters missing or inserted that includes white space. I also us different computers and system. i.e PC or Mac book Pro and different Win OS. So there is no problem with either keyboard maps and different charters (as can happen using a US keyboard mapped to UK English), I’ve even gone as far as using different browsers during play sessions.  One of my great lifetime friends and the DM/GM (all so a long time IT professional and like most of us that have been in IT since the 80/90’s covered many areas of the subject and that includes Dev) has put forward an idea.         Every account essentially has an ID. If the RNG(random number generator) when asked to generate a number does more than what it should I.e. Generate a random number between x and y. Then an additional bit of information is being sent or returned that could weight of the result. If the ID of a user is sent to the RNG as part of any logging or return reference this could when the code is broken down from the program code to machine code cause some sort of affinity between the ID of the user and certain numbers. Essentially making it less than a random number.    In essence your id acts as a magnet to certain number ranges shifting the random number bell curve…  Let’s try this time to think outside the normal troops and finger pointing to work out why some people suffer the issues I am having.  Also am not hexed or cursed as I have been to the temples in f may gods to get cleansed, I’ve also been de-Geist-ed just for good measure.  🤣🤣🙃🙃

OK, let's look at this idea. First assumption: the Roll20 die roller is about as random as can be reasonably engineered. You don't seem refute that, so I think we can take that as a given. Second assumption: Most people claim they want truly random results. Third assumption: People are unhappy with a string of bad results. Proposal 1: Your weighting system above. Pros: Fewer bad results Cons: Opens doors to complaints about removal of randomness, and accusations of favoritism. The majority of users will complain, since the whole point of using dice is for randomness. Proposal 2: House rule--Allow rerolls for players who have had a string of bad rolls. Or a card deck of a certain number of guaranteed successes per game. Or anything. Pros: Fewer bad results. Only affects game of people who are dissatisfied with randomness. Cons: Slightly more initial set-up or play testing time.

I would say the best way to find out if rolls for a player over time are truly averaging out to be random is to track them and compare. There’s a script  that can do just that! (Roll20 Pro subscription required). 

I don't think the question is over testing for randomness. My understanding is that the suggestion is a method of making the rolls less random. I.e. skewing rolls so they "feel" more random to the player. This might work at a table where all agree (hence the second suggestion above), but it would likely be unacceptable to enforce across the site.

Actual dice are not particularly random, unless they are quite expensive and cut and weighed to precise measurements. Casinos are constantly changing dice, knowing that one set of dice is going to lean slightly in some direction, but by mixing them up you approach true randomness. When I went to school in the dark ages we used books of random numbers when we needed random numbers for studies. I have no idea how those were made, but they guaranteed there would be no patterns within certain parameters. At some point you have to assume that faux random numbers are good enough for the purpose at hand, playing a game.

I used to test my physical dice for fun, they are anything but random.  Take a micrometer to your dice sometime, measure each axis. I guarantee they won't be even.  Float them in a high saline solution, roll them (not spin like a top)...watch them change direction as the momentum of the roll winds down. The faster the direction change, the more weighted they are.  IIRC (its been a long time) some dice manufacturers even admitted to their dice being designed to roll on a flattened bell curve, where 10-11 axis is designed to be the highest probability. I guess the logic is that if one axis is going to be rolled most often, it is best if it is the middle axis.  Now, Roll20 has reportedly done a lot of work to make their RNG as random as they could make it. Problem is? People cannot perceive random. Any grouping of numbers is perceived as a pattern. But in any random system groupings will happen. That is part of being random.  To be random in the way humans would perceive it the RNG would have to be NOT random. It would have to remember previous rolls, and then avoid those rolls. The problem is human perception. 

Just don't use the Roll20 dice roller if you don't like it... I have groups where some use it and others roll physical dice. Some groups I've been part of use discord bots, google/alexa or custom app die rollers. Find what works for you and your group for the right feel of 'random'... heck, put pieces of paper in a hat and draw from that. I truly feel your pain but if the shoes aren't fitting, change your cobbler.

Ziechael said: Just don't use the Roll20 dice roller if you don't like it... I have groups where some use it and others roll physical dice. Some groups I've been part of use discord bots, google/alexa or custom app die rollers. Find what works for you and your group for the right feel of 'random'... heck, put pieces of paper in a hat and draw from that. I truly feel your pain but if the shoes aren't fitting, change your cobbler. Unfortunately, for me, it doesn't matter where I roll (R20, real dice, something else), my rolls just suck. LOL 

Finderski said: Unfortunately, for me, it doesn't matter where I roll (R20, real dice, something else), my rolls just suck. LOL  Same here, I guess we are just resigned to it! I would recommend the 'numbers in a hat' option though... I never said which numbers you had to put in the hat... natural 20s every time :D

Corret me if I'm wrong, but Roll20's "RNG" is using some sort of not-so-random scripts... Most games are plague with those nowadays, and it makes me really sad... I would rather have RNG based on pure randomness. Note: RNG generated by a computer used to based on the CPU's clock. Can't say if it's still the same these days, I've been out of the loop for a couple of years. The way to compensate for that organically was to burry RNG under layers of RNG. Making the randomness exponentially more random with each layer, just used more resources -.-

Neo said: Corret me if I'm wrong, but Roll20's "RNG" is using some sort of not-so-random scripts... Quantum Roll : QuantumRoll is the Roll20 dice engine which was enabled by default in the Data Delve update. Rather than relying on client-side pseudo-random number generation to perform dice rolls, Roll20 now utilizes a "true random" source of entropy, based on the fluctuations in the power of a beam of light. All rolling is done via our server. This ensures that dice rolls on Roll20 are as random as possible.

Sounds like it is related to if not exactly based on this technique .

Hmmm, interesting... From the data they provide, that looks legit. o.O Is there still a way to tell if the roll has been quantum signed when it's done inline? Also doesn't explain why sometime a whole table has their rolls always come up as the same 2-3 numbers for over an hour -.-

Inline rolls are also signed. Substrings of similar numbers occur in any randomly generated number strings. If you've had long strings you are likely either on the far end of a probability curve, or experiencing confirmation bias. (sorry, I know that sounds harsh—it's not meant to.) The script Jarren linked upthread can track and give you true statistical data to work with.

Question is; does Roll20 randomly generates numbers? Or are they generating strings of numbers? Because in no system generating strings of numbers does RNG ever feels like RNG. I do NOT use API.

keithcurtis said: Inline rolls are also signed. Thanks, I'll keep an eye out for those now.

As has already been mentioned, you need to track dice over a very large number of rolls to try to make a case that the random number generator is not truly random / not doing what it is supposed to do.  I.E., I've tracked a series of ~50 d20 rolls and I was averaging less than 7 on those 50.  That obviously makes for a session that is not very fun but other sessions I have averaged closer to 14 on 50 rolls. In my experience, some people truly do have bad luck.  But more often than not, this becomes a self-perpetuating mindset.  I.E., had bad luck some times so now any bad rolls are what the player focuses on / remembers over a period of time. As for anyone being not helpful based on their response, the only way to help you is offer advice.  If you find someone's advice not helpful, just skip it - it might be helpful to someone else reading the thread. -Adam

Now that I think of it, the OP did bring something worth talking about: At every single table I've been on since I got on Roll20, there's been this weird convo at one point or another that Roll20's RNG was weighted toward Pro users, compared to Free users. Now let's assume for a sec the RNG they uses pre-generate a string of numbers instead of generating numbers as they are being requested. Then let's assume the player's ID requesting the roll is factored in the actual value to be returned. The overall RNG curve for the generator itself wouldn't be affected. But some players would consistently get worse RNG than others. Not saying that's what's happening, but it's not something I'd rule out as a possibility. All in all, food for thoughts. o.O In Excell, back in the days, I used to randomly generate a number between 0 and 1, then multiply it by a "ceiling" number (rounded up), to generate a semi-random sample. The same logic could be used to return a generic random number, and apply it to any die roll.

Neo said: Question is; does Roll20 randomly generates numbers? Or are they generating strings of numbers? Because in no system generating strings of numbers does RNG ever feels like RNG. I do NOT use API. I'm not sure that makes a difference. Sequential generations of numbers is no different from generating a strong of numbers. Also, it's very unlikely that any random number generator is generating 1 digit integers. And "feel like" is not really a good metric for judging randomness. "7,7,7,7,7,7,7" is no more or less statistically likely than "1,2,3,4,5,6,7" or "2,4,3,3,5,6". The Quantum Roll engine has been proven over and over and over to be random and without bias toward any result. You can run a statistical analysis of your own by rolling an astronomical number of dice and putting the results into a spreadsheet. This will not capture rolls made in play, but it's the easiest solution without a script.

And back to the original point, since the discussion of the Quantum Roll generator has been discussed ad nauseum in other threads: Simon D, were you able to get anything useful from this thread? It sounded like you were after a system that felt more random.

Neo said: At every single table I've been on since I got on Roll20, there's been this weird convo at one point or another that Roll20's RNG was weighted toward Pro users, compared to Free users Hmm, I don't think Quantum Roll has gotten the memo about my subscription....  I mean, sometimes it seems to when I run games.  It certainly isn't me trying to kill the characters of my players, but definitely Quantum Roll.  It seems to give me plenty of simply mediocre when I play however.

       Every account essentially has an ID. If the RNG(random number generator) when asked to generate a number does more than what it should I.e. Generate a random number between x and y. Then an additional bit of information is being sent or returned that could weight of the result. If the ID of a user is sent to the RNG as part of any logging or return reference this could when the code is broken down from the program code to machine code cause some sort of affinity between the ID of the user and certain numbers. Essentially making it less than a random number.    In essence your id acts as a magnet to certain number ranges shifting the random number bell curve…  Let’s try this time to think outside the normal troops and finger pointing to work out why some people suffer the issues I am having. I don't believe the OP is looking for "a system that seems more random."  It sounds like he and some other users feel they are prone to getting rolls that are either in the same range, less that average, or higher than average, more often than they should.  The quoted paragraph is speculation that there may be something inherent in the way the data is generated or handled during the process that could be influencing the outcome of the "rolls".  I think he is asking that the devs look at the way the request for the random number is handled, if there is any possibility that anything in the process could be skewing the results (more often.)

Neo said: At every single table I've been on since I got on Roll20, there's been this weird convo at one point or another that Roll20's RNG was weighted toward Pro users, compared to Free users. I've got a pro account. My co-GM has a Pro account. Our dozen or so players over the many years we've played on Roll20 have had unpaid accounts. Our non-scientific tracking of who is luckiest has me (Pro!) as unluckiest and one unpaid account with far better rolls than anyone else. It's as if the anecdotal evidence is all contradictory.

On topic : You cannot truly measure how random the rolls are until you take a large result set. And by large I'm talking in the multiples of tens or hundreds of thousands or even millions really. Mostly off topic : Anecdotally, Roll20 seems fine to me. My table all have our share of 1's and 20's; certainly no one's ever complained and it's caused quite a few laughs when some poor PC gets a run of bad luck. If you feel some sort of pressure that you must roll high every time, you're missing the point of the game or perhaps are suffering from some poor table etiquette? Are players mocking the low die rolls and you end up taking it personally ? Remember that the system rolled badly, not you, and that translates into your character performing in an unexpected/poor way. Ideally, you should be coming up with a fun or interesting reason WHY that happened, in other words start roleplaying . Unless you're playing a wargame there should always be some opportunity for fun whether you are rolling high are low. Both high and low die rolls create the story . As a DM, if I've got PCs needing to make skill checks in a game, and I'm not just talking about DnD here, I allow them to come up with ways why they should get a bonus on their die rolls. This helps with both the roleplaying as well as the outcome, so you are less at the mercy of a low die roll and encourage to roleplay your character as well as start thinking creatively. We've got memes at our table now based on some of the really flimsy reasoning put forward by some players as to why they should get a bonus to their skill check "as an Agitator ... " :-D Ultimately, it's just a game so try to find a way you can have fun at the table that does not rely on rolling great all the time. You can also encourage your table to take the approach of "yes, but" for die rolls. For that I mean, stop making your die rolls binary yes/no results. You can introduce a system where, for example, if you fail a skill check by less than 5 you still get the result you wanted but maybe something goes slightly wrong or you don't quite get the outcome you expected etc. You can even do this in combat: a glancing blow could be the result of missing by less than 5. I'm actually introducing glancing blows into my Waterdeep: Dragon Heist game as the PCs have had some terrible die rolls - those lower levels can be tough.

I've been a notoriously bad die roller since I or anyone I know can remember. Whether it's D&D, Squad Leader, Admiralty, Parchisi, Craps or even Yahtzee! It doesn't stop me from having fun; I simply modify whatever circumstances that I can to compensate, i.e., I don't even try to play craps, and you'll never get an intelligent monster as a random encounter in my games because I take plenty of time to plan their strategy before they meet the PCs.