RNG Issues Nothing but ones.

[Zilimar]

Week after week I have been experiencing a crazy amount of ones rolled on my dice, to the point where everyone in the games are starting to count and take note.

So far in tonight's game I have rolled 4 ones out of 6 attempts.
Last week I stopped counting after 5.
The week before I had 5 ones in a row on a d20 with the 6th being negated by advantage. This is no exaggeration, the others in the game can attest to this.
Quite often my first roll of the game is a one, for initiative of course. Good to get it out of the way ....or so one would hope. What follows? Much more of the same....


So I started to test the system; when I grab a bunch of dice it seems to be random however rolls done one at a time the die seems to get stuck. I can't explain it but I have seen for example a d20 roll a 9 four times in a row. Etc. But it seems to like the number 1 more often then not. On excess of 60 % per my test.

Five 1s in a row??????????
And why not five 20s?
In fact I can't recall the last time I rolled a 20 during a live game. Crazy.

So I ask, what gives?
Its getting frustrating.

[LordEntrails]

The chat window creates a log. I suggest you grab that and post it here so that a dev can look at it.

[LordEntrails]

Oh, the file is <FG data>/campaigns/<campaign>/chat.html

[Moon Wizard]

There is no random number generator. The dice are a physics simulation of a perfectly weighted dice dropped from a fixed height onto a flat surface. The dice is given a random rotation based on the computer RNG every time it is picked up or rolled. Also, the lateral direction and power of the die roll is determined by the speed of the mouse prior to the die release. If the speed is too slow, the speed is increased to a minimum. If you double click instead of rolling with the mouse, then a random direction is chosen with a random power based on the computer RNG.

cheers,
JPG

[Nickademus]

It should also be mentioned that the sides of the chat window act as walls that the physics simulation interacts with. There is a lot more randomness going on here than a simple RNG.

[damned]

Hi Moon Wizard there have been two (at least) other posts with similar experiences recently with dice seemingly getting stuck on a particular (usually low ball) number... Is it possible that something is glitching and a close and re-open of FG might clear?

[Moon Wizard]

We get a couple posts from time to time about this. However, nothing has changed for the dice in years, so I would think it would come up more often if it was a systemic issue.

There was even a community dev that wrote an extension to measure d20 roll distribution. Since it takes quite a while to run 100k of more rolls, they only did a couple runs. The distributions did not show a particular weighting.

I think this is just the nature of randomness. I've had similar sessions at a live table myself.

If you want to increase the randomness, pick up and throw the dice outside the chat box from time to time.

Regards,
JPG

[lostsanityreturned]

that is how randomness works. It is possible there is a person who will never roll anything but ones. Stupidly unlikely, but it is possible. That is chance.

[Nylanfs]

Try changing your dice colour?

[Zilimar]

Thanks for all the responses, however I think I need to clarify my concern here.

On the side of it being a RNG issue or not, I have no idea the development that went into trying to get the dice results as perfect as possible. Physics, chat window walls...etc. I used RNG to help describe what I am experiencing. In one response above I'm told RNG has nothing to do with it and in another I'm told its all about randomness. What am I missing here?

Regardless

Guys, gals, the point is: something is either low-balling the dice or maxing the results. How the physics work and all the development process while I predicate it all and such from a customers stand point I just want it to work. I'm sorry but I should not have to change the color of my dice nor toss them around randomly outside the chat window to force a randomness.

Take this picture from last nights game: One person maxed his dice on three consecutive rolls. This is a common result. Unfortunately for me I get the low rolls.

18441

Anyways, I like the product and all but I must repeat this has gotten beyond frustrating. Last nights gaming experience once again just sucked. No one is THAT unlucky.
I should note that I am using a MAC and have recently moved to this platform however I have also experienced this on my Win-based machine as well. Just not as noticeable.
One other thing I should note is that I do tend to "double-click" on the die to roll them. Maybe that is a lead. Not sure.

Thanks again for looking into this.

[JohnD]

Speaking as a DM, everything sounds OK. ☺

[LordEntrails]

Zilimar, please post the chat log. With that we can actually look at what is being rolled when. Without it, your story is just an anecdote and there is no reasonable way to troubleshoot an anecdote.

[Zilimar]

I'm trying to find them, I'm on a mac (not my comfort zone) using Wineskin (even worse) but I'll manage. Later tonight I'll get you what you want.
I can also bring in 5 other members of the game if you want backing to my short and amusing story if we really need to go down that path.

[Zilimar]

LOL. Couldn't agree more :-). Here is a sad but somewhat amusing note; The DM in last nights game actually awarded me inspiration for hitting a creature. It was the only successful roll all night and it was after 3+ solid hours of game play. Sad but funny all the same. Hey.... keep plugging away right?



Speaking as a DM, everything sounds OK. ☺

[Nickademus]

Use that inspiration to hit another creature; maybe you'll start a domino effect.

[Zilimar]

I'm sorry, can you help me out here please? I'm really not good with Macs. I found the correct directory that holds the campaign data and log files however I have no files to the games that are not hosted on my machine. Just the ones that I host. Do you need these from the DMs machine? Are these in a different place?

Thanks in advance.



Zilimar, please post the chat log. With that we can actually look at what is being rolled when. Without it, your story is just an anecdote and there is no reasonable way to troubleshoot an anecdote.

[Zacchaeus]

The chatlog is held on the DMs computer, so if you weren't the DM you won't see a log.

[Zilimar]

The chatlog is held on the DMs computer, so if you weren't the DM you won't see a log.

Thanks kind sir.

[SirGraystone]

Sometime the dice just want to kill you :-D

[ddavison]

When this happens to me, I normally melt one of the offending dice while the others watch in horror. They normally start acting better after that... for a while.

[LordEntrails]

When this happens to me, I normally melt one of the offending dice while the others watch in horror. They normally start acting better after that... for a while.
With a magnifying glass in the backyard? Or do you save that special "horror" for little green army men?

[Nickademus]

One of my friends used to use the microwave. He said the hum was therapeutic.

[ddavison]

I reserve all horrors for dice only -- and players when I am the GM.

[keyroll]

I am willing to testify against Zilmmar dice, so they can get the death penalty. I have been present at the table the treason was commited, and recomen a virtual grinder for this offending dice!

Hope he can provide you with the log, is really weird for his rolls to come up that badly in the last few sessions.

[MentalChillness]

Here's the chat log for the last session:

[lostsanityreturned]

Here's the chat log for the last session:

Pending approval.

Probably quicker/easier to dump the text into a pastebin.

[Moon Wizard]

I did a quick manual accumulation of all the rolls for the session in the chat log, and attached the text with all the rolls from the session.

From that chat log, I have the following d20 rolls from Zilimar: 14, 6, 6, 1, 1, 8, 14, 6, 1, 14, 7, 8, 3, 19, 3, 7, 12, 12(A)

Dropping the d20 roll with advantage (since I don't know the result of both rolls), I ran this through an online chi square test for randomness page that gave a P-value of 0.18024. I'm not a statistical expert, but according to what I can figure out, this is still considered random. Of course, also from what I read, a test should use a much larger sample of rolls if possible (1000+).

Any statistics experts around to run the calculations?

Regards,
JPG

[damned]

Its a little low on average value too - around an 8 vs 10.5 but it is on too small a sample to make anything remotely resembling a call either way.

[Zilimar]

Regarding the logs; its a capture in time that did not include any of the sessions of what was discussed. However I must point out that it was at the begging of this session that I started to make my post on the forums after rolling numerous ones yet again at the beginning of yet another game on yet another day. These logs were reset in the middle of our game as well. We needed to do a server reboot so you again are not seeing what is entirely experienced.

That being said; 1) if we knew the log files were going to be inconclusive because they are reset and 2) knew we would need more samples (1000s) why did we ask for them? Was this just to prove or disprove a story? Seems like an entirely was of everyone's time.

Sorry, I don't get it. I'm just a customer trying to say that maybe something is wrong. Much like others of stated mind you. I'm not trying to poke fun or call your product an ugly baby or anything like that. I like the platform. It has great potential.

However I no longer trust its "randomness" because at times its not.

Go back to that picture I posted earlier. That's all that should matter. Why? Because that is what is happening daily in our games. Either maxed rolls or all 1s in succession. What are those odds?
Again, What are the odds of having lets just say a max number roll on any dice three times in a row? Now lets say that scenario happens 2, 3 or 4 times in a typical 2 hour (not always rolling dice) game?
That's what's happening. Its not a funny little story. Its not make believe. And frankly, its not enjoyable when it happens.

I will continue to post pictures as they come up.

Anyways, thanks for listening and I hope that one day this mystery gets solved.

[damned]

Regarding the logs; its a capture in time that did not include any of the sessions of what was discussed. However I must point out that it was at the begging of this session that I started to make my post on the forums after rolling numerous ones yet again at the beginning of yet another game on yet another day. These logs were reset in the middle of our game as well. We needed to do a server reboot so you again are not seeing what is entirely experienced.

That being said; 1) if we knew the log files were going to be inconclusive because they are reset and 2) knew we would need more samples (1000s) why did we ask for them? Was this just to prove or disprove a story? Seems like an entirely was of everyone's time.

Sorry, I don't get it. I'm just a customer trying to say that maybe something is wrong. Much like others of stated mind you. I'm not trying to poke fun or call your product an ugly baby or anything like that. I like the platform. It has great potential.

However I no longer trust its "randomness" because at times its not.

Go back to that picture I posted earlier. That's all that should matter. Why? Because that is what is happening daily in our games. Either maxed rolls or all 1s in succession. What are those odds?
Again, What are the odds of having lets just say a max number roll on any dice three times in a row? Now lets say that scenario happens 2, 3 or 4 times in a typical 2 hour (not always rolling dice) game?
That's what's happening. Its not a funny little story. Its not make believe. And frankly, its not enjoyable when it happens.

I will continue to post pictures as they come up.

Anyways, thanks for listening and I hope that one day this mystery gets solved.

The comment on the log size is relevant because the number of rolls is still small.
There was one instance of 2 successive ones, one instance of 2 successive 6s and one instance of two successive 12s.
The GM has a chatlog that doesnt not get autoreset for the whole campaign - it might show more info.
A small sample size can be relevant when the results are wildly similar but these ones are not and nor are the ones that your fellow players are throwing...

[Moon Wizard]

I understand what you are saying, and I'm definitely not trying to downplay your report.

However, I only have data and statistical tests in order to measure the issue you are reporting. Based on the data that I have seen so far, the statistical tests are saying that the results are still random. And, the code that I reviewed when this was brought up seems to create a very random scenario.

At this point, it appears to be a lower than average set of rolls reported so far, but they're within the expected behavior of a random d20.

We can continue to keep an eye on the rolls, and watch for any issues.

Regards,
JPG

[Zilimar]

And the picture that was posted as an example, we are supposed to take that and those cases that happen like it on an all too frequently basis as random?
Just want clarification on your response here.
Do you guys call that random?






I understand what you are saying, and I'm definitely not trying to downplay your report.

However, I only have data and statistical tests in order to measure the issue you are reporting. Based on the data that I have seen so far, the statistical tests are saying that the results are still random. And, the code that I reviewed when this was brought up seems to create a very random scenario.

At this point, it appears to be a lower than average set of rolls reported so far, but they're within the expected behavior of a random d20.

We can continue to keep an eye on the rolls, and watch for any issues.

Regards,
JPG

[Callum]

It is random, it just doesn't appear to be random. Here's a blog post that explains this idea in more detail:

https://www.analyticbridge.com/profiles/blogs/7-traps-to-avoid-being-fooled-by-statistical-randomness

[damned]

And the picture that was posted as an example, we are supposed to take that and those cases that happen like it on an all too frequently basis as random?
Just want clarification on your response here.
Do you guys call that random?

Is there a different picture?
The posted picture shows no 1s but does have one player roll max on 3 successive rolls with 2 different dice type. Interspersed with a GM that rolls 3 completely different dice for 3 completely different results.
Grab the GMs full chat log - hopefully it will show more examples?

[ddavison]

The GM has a chatlog that doesnt not get autoreset for the whole campaign - it might show more info.


This is the file I am requesting because it should have a much longer history of rolls.

[LordEntrails]

Note, I had mentioned posting the chat log early on thinking you would be able to post the GM log so that it would have a large history and number of rolls.

Hopefully you don't feel anyone is trying to be dismissive. Everyone really is interested and it's important to all of us that the rolls are random. But right now we really don't have any other way to run this down (as Moon Wizard mentions) other than through looking at the statistics.

[ddavison]

So this is just the beginning of these sorts of things. I've been meaning to build some statistical analysis tools based on chatlogs for a while. I think it would be fun to spit these out and allow people to easily post them up to web pages for online campaign management.

The uploaded attachment is a zip file containing an excel document.

The short summary - looking only at all dice (not yet by player)

Your game had only 135 rolls to look at (from the chat log text)
6.75 Average # of occurrences per die
10.71111111 Average Roll

Roll Types
normal 113
advantage 21
disadvantage 1

https://www.fantasygrounds.com/forums/attachment.php?attachmentid=18499&d=1491836977


I ran the same analysis on a version I use for testing over a slightly longer period of time and I got these results:
Total rolls reviewed 292
14.6 Average # of occurrences per die
11.21575342 Average Roll

Roll Types
normal 286
advantage 5
disadvantage 1

https://www.fantasygrounds.com/forums/attachment.php?attachmentid=18498&d=1491836971

[Zilimar]

Sorry guys, but I still think your missing it.
Collecting large samples and looking at those results are only going to mess up what I believe to be the larger issue is at hand.
Its blindsiding my point.

If I take the results of five 1s in a row and mix them into the results of 995 other rolls you will never see the point that i just rolled five 1s in a row.
Someone will stay say it falls within the specs of randomness.

It does not. Not even close. Its off. By a lot.
Hence I don't care statistically what the overall average is of all the dice rolls.
I'm sorry.
Its meaningless to what I have been trying to say.

My belief based on past experience and the others in the game agree, can and have attest to it: Something is locking the results into 1) a given number 2) the lowest number of a die or 3) the highest value.
When it happens you will see them them repeat for an unknown amount of time.

Why? I don't know. How many times does it happen in a row? I have no idea. ...its random. Yep, I said it.

Go back and look at that picture again You don't need any other information.
That is an example of the dice locked in a HIGH state.

For me, those numbers have been unlucky, they have been 1s, for my fellow gamer above they were 20s and 10s (damage dice) locked high.

If I asked for anything it would be for you to answer why that happened in the picture. Please, just focus on that picture. That is NOT random.
Its not even tied to any given player because the final die locked high was system generated (player rolled to hit, crit, rolled damage, got max, system capped the added die).
It happens all the time!

The DM has already come forward as has a couple others. I really have nothing else to add here.
Good luck with this. I just hope that one day may gaming experience within FG will improve.

[Callum]

You really should read that blog post I linked to earlier.

[ddavison]

So the problem with randomness and probability from my recollection of various college math classes (of which I have a Math minor), the value you roll "next" is not impacted at all by any of your previous rolls because they are classified as Independent Events. It is just as statistically valid for you to roll 5 1's in a row as it is for you to roll 5 other dice values in a row. It is the same reason why you can roll 19 times and not get a 20 and that doesn't make your 20th roll any more likely to roll a 20 than any other value. This is something that almost everyone introduced to the concept of probability struggled with.

By looking at more results, you can see if there is something occurring more or less than expected. If you have five players rolling dice and someone rolls a 1, then next player rolls a 1 and then a 3rd player rolls a 1 then that may seem strange. Similarly, if you roll a 1, but the other players didn't and then you roll 1s again for the next two rolls you make, that again looks strange to you because you decided to only look at a select group -- either all players in the first set of rolls or just your rolls spread over a time but viewed as a single event. How you personally are choosing to view the "grouping" is the entire reason it seems unfair or broken. Looking at the charts above, you can see that in your case rolling a 1 was slightly higher in representation than the average (7 vs 6.75) and less in my sample set (12 vs 14.6). The #4 was actually lowest on both. Both datasets show that 4 was underrepresented in the sample.

Where there could legitimately be a problem that isn't exposed is advantage or disadvantage rolls producing the same results. Our chatlog doesn't currently feature that, but I could envision a scenario where they essentially give the same force, angle and rotation to both die and end up with the same results. Moon Wizard will probably be able to look that up and confirm or deny that aspect. We are not able to statistically determine if that is happening since the chat log is not keeping the thrown away die values.

[Zilimar]

You really should read that blog post I linked to earlier.

I did. This is not random. It's a forced result. Unwanted I might add.

[Zacchaeus]

Sorry, Zilimar, but if you don't accept it is possible to randomly roll 5 one's or 20's or whatever in a row then no-one is going to be able to convince you otherwise.

To enable this issue to be investigated then as the devs have said they need some raw data with which to work. Your picture doesn't say anything other than some dice were rolled and happened to be the maximum on those dice; unusual certainly - but it doesn't in itself prove that there is a problem with the way FG handles the generation of the dice roll. As has been said in order to determine if there is a problem then a much larger sample of data is needed. If, for example you roll a d20 1000 times then theoretically every number should come up 50 times. It won't happen of course exactly like that but each number should come out at more or less 5% of the number of rolls made. Now if we looked at a large sample and saw that 1's or 20's were way off the expected 5% then we'd know there was a problem.

At present apart from some sporadic posts here and there like yours where unusual results have been noted there's no hard evidence of a problem.

[Moon Wizard]

I'm going to go ahead and close this thread, since we're just rehashing at this point.

Zilimar, if you'd like us to investigate any other chat log histories to analyze them or have any other questions, please send them directly to us at [email protected]

Regards,
JPG