Forum adverts like this one are shown to any user who is not logged in. Join us by filling out a tiny 3 field form and you will get your own, free, dakka user account which gives a good range of benefits to you: No adverts like this in the forums anymore. Times and dates in your local timezone. Full tracking of what you have read so you can skip to your first unread post, easily see what has changed since you last logged in, and easily see what is new at a glance. Email notifications for threads you want to watch closely. Being a part of the oldest wargaming community on the net. If you are already a member then feel free to login now.

Im not willing to pay the crazy prices to get PRECISION, and CUSTOM dice foir the marginal diff it might make. I took a couple hours today to investigate the alternative companies and after quite a bit of searching have decidedthat this level of precision in CUSTOM dice is just too unwieldy a price tag.

 Zagman wrote:
And as cool as that data is, it isn't very applicable to what we were going to see on the gaming table. If nearly 30% of roll came up 1s we'd know. That is not the case on average on the gaming table.

There is much we don't know about that study. How was each dice held, what orientation? Were the dice rolled the same way by hand? What was the coefficient of restitution in relation to the table?

All we know, is that when each die was rolled 1000 times by a particular individual on a particular substance we see such a distribution and can reasonably believe that it is indeed the case.

But, how does the data differ on different surface, especially uneven and softer surfaces commonly used for gaming. How does rolling large handfuls of dice on these surfaces affect results? As soon as variables are added such as different surfaces, uneven surfaces especially those with small bumps, and obstacles such as terrain and other dice are involved result will normalize to a significant degree or at the least invalidate the precision of such studies.

My best guess is that under those varying circumstances we'll end up with a much more normalized distribution, just as forcing someone to roll against a wall on a craps table does. The supplied theories are indeed interesting, but rely on clinical precision to be valid and the reasons for the dice behaving in such a manor are likely to be affected to a high degree by uneven surfaces or during normal gaming. Any anything that differs from clinical precision is going to reduce any predictability in the results.

This is true, I don't think they well defined HOW they were rolling the dice, you'd have to do it on a gaming table and in batches similar to what you see when rolling for things in games.

Definitely could be more rigorous testing, I agree.

My point was primarily that, yes, a gamer can roll a die 1000 times and if the result is off by more than a few percent, the chances of the dice (or the rolling technique) being biased is quite high.

And the most important part, everyone I've encounter during Wargaming are using the same kinds of dice from the same manufacturers and therefore any effect the dice may be having is irrelevant. Sure, there was one guy using precision dice, but he isn't beating me nor rolling any less polar to a significant enough degree to be noticed.

I don't agree on this point. If the dice are biased it will affect different armies in different ways. If a set of dice is biased toward 1's and 2's, that will have a far larger effect on a Space Marine army that hits, saves and quite often wounds on a 3+. It will a lesser effect on an Ork army that doesn't have a lot of "3+ is a success" rolls, they're typically 5+ or 6+.

For poops and giggles, I just did my own test.

I rolled 53 Games Workshop dice (I have more GW dice than anything else). They were bought over an 18 year period, mostly gotten from GW boxed sets. I rolled them on soft carpet, 20 times, from ~4", gave them a decent shake in my hands and then a decent roll. Because it's such a large batch of dice, they do hit each other a lot as they roll. I picked the dice up pretty randomly (53 dice only just fit in my hand, so the effect of how I hold them is negligible IMO). All of this, I think, would reduce any bias in the dice themselves. The carpet is far softer than any gaming table I've ever played on which makes them roll more, I think this would make them more random.

The results?

side / how many times it came up / %
1 / 168 / 15.8%
2 / 173 / 16.3%
3 / 172 / 16.2%
4 / 172 / 16.2%
5 / 186 / 17.5%
6 / 189 / 17.8%

So, a pretty even spread. Of course, I could roll the dice again tomorrow and get 29% 1's. But if there were a 29% chance of rolling a 1, the chance of me rolling less than 170 1's out of 1060 rolls (I rolled 168) is only 0.00000000000000000006%.

Soooo, I can say pretty safely say my dice when rolled the way I did it are NOT biased by 29% to 1, either that or I got super duper massively lucky (the equivalent of rolling 27 6's in a row sort of lucky).

Now... I did not roll these on a gaming table... I do not have a gaming table at home anymore and I'm not driving to the FLGS to do it But if someone else did in the realm of 1000 rolls on a gaming table I would definitely look at their results as statistically significant.

If someone has a day to waste, it would be worthwhile picking a few styles of rolling (single roll vs 2D6 leadership test vs big batch) on a couple of different surfaces (carpet, hard surface, gaming table) and roll 1000-2000 dice in that form and see the results. What I described would probably take you an afternoon to do, if any of the results are majorly off 17% per side, do the test again to check, so lets allow a full day of testing.

I do not have a day to waste doing that though, so yeah

AllSeeingSkink wrote:

 Zagman wrote:
And as cool as that data is, it isn't very applicable to what we were going to see on the gaming table. If nearly 30% of roll came up 1s we'd know. That is not the case on average on the gaming table.

There is much we don't know about that study. How was each dice held, what orientation? Were the dice rolled the same way by hand? What was the coefficient of restitution in relation to the table?

All we know, is that when each die was rolled 1000 times by a particular individual on a particular substance we see such a distribution and can reasonably believe that it is indeed the case.

But, how does the data differ on different surface, especially uneven and softer surfaces commonly used for gaming. How does rolling large handfuls of dice on these surfaces affect results? As soon as variables are added such as different surfaces, uneven surfaces especially those with small bumps, and obstacles such as terrain and other dice are involved result will normalize to a significant degree or at the least invalidate the precision of such studies.

My best guess is that under those varying circumstances we'll end up with a much more normalized distribution, just as forcing someone to roll against a wall on a craps table does. The supplied theories are indeed interesting, but rely on clinical precision to be valid and the reasons for the dice behaving in such a manor are likely to be affected to a high degree by uneven surfaces or during normal gaming. Any anything that differs from clinical precision is going to reduce any predictability in the results.

This is true, I don't think they well defined HOW they were rolling the dice, you'd have to do it on a gaming table and in batches similar to what you see when rolling for things in games.

Definitely could be more rigorous testing, I agree.

My point was primarily that, yes, a gamer can roll a die 1000 times and if the result is off by more than a few percent, the chances of the dice (or the rolling technique) being biased is quite high.

And the most important part, everyone I've encounter during Wargaming are using the same kinds of dice from the same manufacturers and therefore any effect the dice may be having is irrelevant. Sure, there was one guy using precision dice, but he isn't beating me nor rolling any less polar to a significant enough degree to be noticed.

I don't agree on this point. If the dice are biased it will affect different armies in different ways. If a set of dice is biased toward 1's and 2's, that will have a far larger effect on a Space Marine army that hits, saves and quite often wounds on a 3+. It will a lesser effect on an Ork army that doesn't have a lot of "3+ is a success" rolls, they're typically 5+ or 6+.

For poops and giggles, I just did my own test.

I rolled 53 Games Workshop dice (I have more GW dice than anything else). They were bought over an 18 year period, mostly gotten from GW boxed sets. I rolled them on soft carpet, 20 times, from ~4", gave them a decent shake in my hands and then a decent roll. Because it's such a large batch of dice, they do hit each other a lot as they roll. I picked the dice up pretty randomly (53 dice only just fit in my hand, so the effect of how I hold them is negligible IMO). All of this, I think, would reduce any bias in the dice themselves. The carpet is far softer than any gaming table I've ever played on which makes them roll more, I think this would make them more random.

The results?

side / how many times it came up / %
1 / 168 / 15.8%
2 / 173 / 16.3%
3 / 172 / 16.2%
4 / 172 / 16.2%
5 / 186 / 17.5%
6 / 189 / 17.8%

So, a pretty even spread. Of course, I could roll the dice again tomorrow and get 29% 1's. But if there were a 29% chance of rolling a 1, the chance of me rolling less than 170 1's out of 1060 rolls (I rolled 168) is only 0.00000000000000000006%.

Soooo, I can say pretty safely say my dice when rolled the way I did it are NOT biased by 29% to 1, either that or I got super duper massively lucky (the equivalent of rolling 27 6's in a row sort of lucky).

Now... I did not roll these on a gaming table... I do not have a gaming table at home anymore and I'm not driving to the FLGS to do it But if someone else did in the realm of 1000 rolls on a gaming table I would definitely look at their results as statistically significant.

If someone has a day to waste, it would be worthwhile picking a few styles of rolling (single roll vs 2D6 leadership test vs big batch) on a couple of different surfaces (carpet, hard surface, gaming table) and roll 1000-2000 dice in that form and see the results. What I described would probably take you an afternoon to do, if any of the results are majorly off 17% per side, do the test again to check, so lets allow a full day of testing.

I do not have a day to waste doing that though, so yeah

I appreciate the sentiment and echo your message, but I want to point out, 1060 is hardly a good sample size. The original article (I do not remember the exact but I believe it to be in the tens of thousands) did a much larger sample size and got similar results. I appreciate what you did, but 1000 is way too small of a pool.

 jreilly89 wrote:
I appreciate the sentiment and echo your message, but I want to point out, 1060 is hardly a good sample size. The original article (I do not remember the exact but I believe it to be in the tens of thousands) did a much larger sample size and got similar results. I appreciate what you did, but 1000 is way too small of a pool.

My response to that is to highlight one of my sentences in that post...


Of course, I could roll the dice again tomorrow and get 29% 1's. But if there were a 29% chance of rolling a 1, the chance of me rolling less than 170 1's out of 1060 rolls (I rolled 168) is only 0.00000000000000000006%.

That's less than the chance of rolling 27 6's in a row.

So, yeah, I'm pretty confident that my dice set is no more than a few % biased one way or another.

The idea that you have to roll millions of dice to get close to the average is a fallacy. It's a binomial distribution, you just need to roll enough dice that the % chance of being far from the average is small. If there was really a 29% bias toward 1's (rather than the expected 17%), the chance of me not seeing it in 1000 rolls is extremely small. If the bias were only, say, 20%, yeah, I agree, I might miss it in only 1000 rolls (in fact I can tell you the chance of me rolling what I did if the bias were 20% is less than 0.03%).

I actually do experimental aerodynamic research for my job, part of that is knowing where you draw the line of "ok, this is enough data to be confident enough to make a statistically significant statement". IMO, 1000 rolls is enough to say something statistically significant within a few % of the expected mean. If you were more than a few % away from the expected mean, I'd totally say repeat the experiment, given the expected mean is 16.7% and my largest variation from that was 17.8%... I am satisfied with that.

EDIT: Note, when I say 1000 samples is enough for a statistically significant result, that's partly because the chance of getting any 1 side is 17%, if we were talking about a D20 where the chance is only 5%... yes, you'd have to roll it more to get a significant result).

pizzaguardian wrote:
Wait wait so trying to get dice that gets averagely 16.6% on every result of d6 is cheating?

And lets elaborate , if i get some precision dice to a game and tell you that " hey these dice are precision dice lets use these" make me a cheater?

I am really having a hard time understanding you that somehow trying to get average results in cheating?

And also what if my opponent is incistent or using their dice even after i tell them their dice might not put average results will i become a cheater then?

Somehow i am a cheater in your eyes for trying to get average results, or even worse not playing with dice that doesn't give you average results is cheating as well.

Nobody here is trying to get "roll better" in this thread as far as i can see. But trying to defend bad dice, wherever its from; just because somehow getting average results in d6 (which i hope the game is designed for the average result and not bad dice, which doesnt mean rolling just ones it means rolling not averagely on every result) makes somebody a cheater and we should all be playing with skewed results is beyond reason.

jreilly89 wrote:

Seriously?....wait, no, seriously? So, in a game that uses dice EXPLICITLY, me getting FAIR dice is cheating? That's stupid. If my opponent wanted to go out, blow money on "precision dice" and use them for our game, I would gladly let him. It's just a game, with, for me, no money being at stake.

As much as 40k is a game of chance, using precision dice will nott ultimately swing the game in his favor. Guess what? If you have precision dice and a crappy list, you'll still get beat, no matter how fairly your dice roll. Come on, dude.

Ou both missed my point, by a massive margin. Knowingly increasing your odds by rolling better than your opponent, even if it is only bringing your dice to average by knowingly pursuing perfectly balanced dice, if garnering an unfair advantage.

If you know Chessex dice roll below average, but consistently above average, and that almost every opponent is using these substandard dice, then pursuing precision dice to increase your odds to average is trying to gain an unfair advantage over your opponents. That is cheating in its simplistic form.

Now, I've postulated that these Cheap dice won't perform nearly as poorly in non clinical settings, but hey, I'm working off the assumption the provided data is accurate. If it is accurate, then it's knowingly creating an unfair advantage over your opponent.

AllSeeingSkink wrote:

 Zagman wrote:
And as cool as that data is, it isn't very applicable to what we were going to see on the gaming table. If nearly 30% of roll came up 1s we'd know. That is not the case on average on the gaming table.

There is much we don't know about that study. How was each dice held, what orientation? Were the dice rolled the same way by hand? What was the coefficient of restitution in relation to the table?

All we know, is that when each die was rolled 1000 times by a particular individual on a particular substance we see such a distribution and can reasonably believe that it is indeed the case.

But, how does the data differ on different surface, especially uneven and softer surfaces commonly used for gaming. How does rolling large handfuls of dice on these surfaces affect results? As soon as variables are added such as different surfaces, uneven surfaces especially those with small bumps, and obstacles such as terrain and other dice are involved result will normalize to a significant degree or at the least invalidate the precision of such studies.

My best guess is that under those varying circumstances we'll end up with a much more normalized distribution, just as forcing someone to roll against a wall on a craps table does. The supplied theories are indeed interesting, but rely on clinical precision to be valid and the reasons for the dice behaving in such a manor are likely to be affected to a high degree by uneven surfaces or during normal gaming. Any anything that differs from clinical precision is going to reduce any predictability in the results.

This is true, I don't think they well defined HOW they were rolling the dice, you'd have to do it on a gaming table and in batches similar to what you see when rolling for things in games.

Definitely could be more rigorous testing, I agree.

My point was primarily that, yes, a gamer can roll a die 1000 times and if the result is off by more than a few percent, the chances of the dice (or the rolling technique) being biased is quite high.

And the most important part, everyone I've encounter during Wargaming are using the same kinds of dice from the same manufacturers and therefore any effect the dice may be having is irrelevant. Sure, there was one guy using precision dice, but he isn't beating me nor rolling any less polar to a significant enough degree to be noticed.

I don't agree on this point. If the dice are biased it will affect different armies in different ways. If a set of dice is biased toward 1's and 2's, that will have a far larger effect on a Space Marine army that hits, saves and quite often wounds on a 3+. It will a lesser effect on an Ork army that doesn't have a lot of "3+ is a success" rolls, they're typically 5+ or 6+.

For poops and giggles, I just did my own test.

I rolled 53 Games Workshop dice (I have more GW dice than anything else). They were bought over an 18 year period, mostly gotten from GW boxed sets. I rolled them on soft carpet, 20 times, from ~4", gave them a decent shake in my hands and then a decent roll. Because it's such a large batch of dice, they do hit each other a lot as they roll. I picked the dice up pretty randomly (53 dice only just fit in my hand, so the effect of how I hold them is negligible IMO). All of this, I think, would reduce any bias in the dice themselves. The carpet is far softer than any gaming table I've ever played on which makes them roll more, I think this would make them more random.

The results?

side / how many times it came up / %
1 / 168 / 15.8%
2 / 173 / 16.3%
3 / 172 / 16.2%
4 / 172 / 16.2%
5 / 186 / 17.5%
6 / 189 / 17.8%

So, a pretty even spread. Of course, I could roll the dice again tomorrow and get 29% 1's. But if there were a 29% chance of rolling a 1, the chance of me rolling less than 170 1's out of 1060 rolls (I rolled 168) is only 0.00000000000000000006%.

Soooo, I can say pretty safely say my dice when rolled the way I did it are NOT biased by 29% to 1, either that or I got super duper massively lucky (the equivalent of rolling 27 6's in a row sort of lucky).

Now... I did not roll these on a gaming table... I do not have a gaming table at home anymore and I'm not driving to the FLGS to do it But if someone else did in the realm of 1000 rolls on a gaming table I would definitely look at their results as statistically significant.

If someone has a day to waste, it would be worthwhile picking a few styles of rolling (single roll vs 2D6 leadership test vs big batch) on a couple of different surfaces (carpet, hard surface, gaming table) and roll 1000-2000 dice in that form and see the results. What I described would probably take you an afternoon to do, if any of the results are majorly off 17% per side, do the test again to check, so lets allow a full day of testing.

I do not have a day to waste doing that though, so yeah

Outstanding, thank you for doing this. Granted, it's only 1/50 the number of rolls each die in the Dakka study was subjected to, it strongly hints that the surface used is of paramount importance to the end result. According your your data, using a soft surface allowed the substandard dice to roll better than average which is the reverse of the Dakka study showed using a perfect hard surface. Gaming surfaces are going to be closer to the soft surface than the hard surface making any of the Dakka conclusions of little relevance if the data could be expanded and validated in some way.

More data could support this theory of mine, but your evidence was of great use.

Thank you for doing this, I was considering pulling out my five Chexxex cubes and doing a few dozen rolls on varying surfaces myself.

Wow, people can be really anal about the whole dice side of 40k (not saying it's a bad thing). I guess it matters if your're playing in a GT with prizes, but I just wanted to know if I could use Chessex in my FLGS without ruining my games.

 Random Dude wrote:
Wow, people can be really anal about the whole dice side of 40k (not saying it's a bad thing). I guess it matters if your're playing in a GT with prizes, but I just wanted to know if I could use Chessex in my FLGS without ruining my games.

The answer to that is up to how much your opponents know and care


Automatically Appended Next Post:

 Zagman wrote:
Outstanding, thank you for doing this. Granted, it's only 1/50 the number of rolls each die in the Dakka study was subjected to, it strongly hints that the surface used is of paramount importance to the end result. According your your data, using a soft surface allowed the substandard dice to roll better than average which is the reverse of the Dakka study showed using a perfect hard surface. Gaming surfaces are going to be closer to the soft surface than the hard surface making any of the Dakka conclusions of little relevance if the data could be expanded and validated in some way.

More data could support this theory of mine, but your evidence was of great use.

Thank you for doing this, I was considering pulling out my five Chexxex cubes and doing a few dozen rolls on varying surfaces myself.

Actually it wasn't 1/50th, I did 1/36 the number of rolls they did.

But yeah, when you read this...

"The 6 is too light to stop the momentum of the die, the rounded corners cannot prevent the die from turning due to the weight... I then proceeded to buy more GW dice and we filled in the corners of the very same dice that we used, carefully melting the new plastic on to the old dice and filing in the corners to the right size and leveling them to .001 for accuracy. The dice then rolled more accurately but still 19% rolled ones."

...in the Dakka study, it does set off alarm bells that the surface they rolled them on must be important. As they are assuming the bias comes from the different amount of balancing force created by the rounded edges and varying weight, that factor is very different depending on the surface you roll on and you're right in thinking that on a soft surface that difference is minimized.

There's definitely ways of rolling that will accentuate the weight bias more than others, so after rolling my own dice I can't help but think either....

a) They were rolling them in a way that made the bias worse (though I admit the way I rolled them probably reduced any bias... so I'd like to see a test on a typical gaming surface).

b) Not all GW/Chessex dice bias toward 1, they just got unlucky in that their sets were biased toward 1.

c) They are right in that there is a 29% bias toward 1 and I just got extremely lucky that my 53 dice don't have as much bias and/or I managed to pull off the 0.00000000000000000006% chance of rolling less than 168 ones on dice that are biased to roll ~307 ones.

Given how small the chance of c) happening is, I'm leaning toward a) and maybe a little bit of b).

There's also the chance that the dice quality from Chessex/GW have degraded over the years. Most my dice are 15+ years old, that article was written 6 years ago, maybe there's a difference.

 Random Dude wrote:
Wow, people can be really anal about the whole dice side of 40k (not saying it's a bad thing). I guess it matters if your're playing in a GT with prizes, but I just wanted to know if I could use Chessex in my FLGS without ruining my games.

This is of primarily academic interest since almost ...almost...NO ONE has a full set of 30 precision dice and over the course of games maybe 10% will be willing to pay THIS much for the moral high ground... and those seeking it seem well intentioned but ultimately...Im not gonna pay that much.

 Jancoran wrote:

 Random Dude wrote:
Wow, people can be really anal about the whole dice side of 40k (not saying it's a bad thing). I guess it matters if your're playing in a GT with prizes, but I just wanted to know if I could use Chessex in my FLGS without ruining my games.

This is of primarily academic interest since almost ...almost...NO ONE has a full set of 30 precision dice and over the course of games maybe 10% will be willing to pay THIS much for the moral high ground... and those seeking it seem well intentioned but ultimately...Im not gonna pay that much.

It is mostly academic, but depending on the variation it can make a difference, though really I don't think most people would care.

Though if the Dakka study is right and GW/Chessex dice are 29% biased toward 1's, that does impact the game depending on what army you play. That's saying on average it takes only 17 wounds to kill a Terminator instead of 30


Automatically Appended Next Post:
For those interested, I just rolled the 53 dice another 20 times, so another 1060 trials.

1 / 333 / 15.7%
2 / 361 / 17.0%
3 / 356 / 16.8%
4 / 345 / 16.3%
5 / 347 / 16.4%
6 / 378 / 17.8%

So that's 2120 trials, compared to the Dakka study that was 36,000 trials, but really, the chance of it swinging to 29% 1's like the Dakka trial is so incredibly small.

I'm actually erring on the side of 6's being dominant.

For the people who like statistics, if you want to check out the convergence:



As you can see, after a couple of hundred trials it was well on its way to converging, from about 600 trials to 2000 trials the trend is pretty well established. Series 1 is chance of rolling a 1, Series 2 chance of rolling a 2, etc.

 Zagman wrote:

pizzaguardian wrote:
Wait wait so trying to get dice that gets averagely 16.6% on every result of d6 is cheating?

And lets elaborate , if i get some precision dice to a game and tell you that " hey these dice are precision dice lets use these" make me a cheater?

I am really having a hard time understanding you that somehow trying to get average results in cheating?

And also what if my opponent is incistent or using their dice even after i tell them their dice might not put average results will i become a cheater then?

Somehow i am a cheater in your eyes for trying to get average results, or even worse not playing with dice that doesn't give you average results is cheating as well.

Nobody here is trying to get "roll better" in this thread as far as i can see. But trying to defend bad dice, wherever its from; just because somehow getting average results in d6 (which i hope the game is designed for the average result and not bad dice, which doesnt mean rolling just ones it means rolling not averagely on every result) makes somebody a cheater and we should all be playing with skewed results is beyond reason.

jreilly89 wrote:

Seriously?....wait, no, seriously? So, in a game that uses dice EXPLICITLY, me getting FAIR dice is cheating? That's stupid. If my opponent wanted to go out, blow money on "precision dice" and use them for our game, I would gladly let him. It's just a game, with, for me, no money being at stake.

As much as 40k is a game of chance, using precision dice will nott ultimately swing the game in his favor. Guess what? If you have precision dice and a crappy list, you'll still get beat, no matter how fairly your dice roll. Come on, dude.

Ou both missed my point, by a massive margin. Knowingly increasing your odds by rolling better than your opponent, even if it is only bringing your dice to average by knowingly pursuing perfectly balanced dice, if garnering an unfair advantage.

If you know Chessex dice roll below average, but consistently above average, and that almost every opponent is using these substandard dice, then pursuing precision dice to increase your odds to average is trying to gain an unfair advantage over your opponents. That is cheating in its simplistic form.

Now, I've postulated that these Cheap dice won't perform nearly as poorly in non clinical settings, but hey, I'm working off the assumption the provided data is accurate. If it is accurate, then it's knowingly creating an unfair advantage over your opponent.

You think that somehow all the bad dice are biased towards 1, what if dice is biased towards 5s and 6s which doesn't matter at all.

I would not be cheating my opponents with rolling an average outcome dice , they would be cheating me with rolling a non-average outcome dice.

You are wrong that trying to get average resutls is cheating but anything else is not.

Frankly I never trusted those two studies about chessex dice favoring 1s by 30%. Namely because that percentage is absurdly high (almost double what it should be). But also because the way chessex dice are set up they should weighed towards the 1s side (because least amount of pips drilled in) so it should be more likely to land 1s side down resulting in 6s being the more likely result (or at the very least 1s being the least likely result)

I personally use 16mm Chessex and in my experience they are skewed but towards the higher numbers (which coincides with how chessex make their dice)

 CrownAxe wrote:
Frankly I never trusted those two studies about chessex dice favoring 1s by 30%.

One thing I found odd about the dakka study was this statement:

"We removed any statistical anomalies and came up with 29%."

What statistical anomalies are they coming across by rolling dice? Even if you roll 36 out of 36 6's, you should still count it. You shouldn't be removing outliers unless you have some reason to think they're outliers for some abnormal reason, which would cast aspersions on your testing technique.

The other thing I found odd... they say they tracked each individual die, yet they only offer very broad statements instead of giving any rigorous analysis of the results. I sure as hell didn't track each individual die as I think it was unnecessary, but I still have more statistics to offer than what they gave.

I feel it's worth noting that just because dice statistically roll ones slightly more often they're not automatically bad. You're not always trying to roll 5's and 6's. Sometimes you need a 1, like for scatter or morale.

 pizzaguardian wrote:

 Zagman wrote:

pizzaguardian wrote:
Wait wait so trying to get dice that gets averagely 16.6% on every result of d6 is cheating?

And lets elaborate , if i get some precision dice to a game and tell you that " hey these dice are precision dice lets use these" make me a cheater?

I am really having a hard time understanding you that somehow trying to get average results in cheating?

And also what if my opponent is incistent or using their dice even after i tell them their dice might not put average results will i become a cheater then?

Somehow i am a cheater in your eyes for trying to get average results, or even worse not playing with dice that doesn't give you average results is cheating as well.

Nobody here is trying to get "roll better" in this thread as far as i can see. But trying to defend bad dice, wherever its from; just because somehow getting average results in d6 (which i hope the game is designed for the average result and not bad dice, which doesnt mean rolling just ones it means rolling not averagely on every result) makes somebody a cheater and we should all be playing with skewed results is beyond reason.

jreilly89 wrote:

Seriously?....wait, no, seriously? So, in a game that uses dice EXPLICITLY, me getting FAIR dice is cheating? That's stupid. If my opponent wanted to go out, blow money on "precision dice" and use them for our game, I would gladly let him. It's just a game, with, for me, no money being at stake.

As much as 40k is a game of chance, using precision dice will nott ultimately swing the game in his favor. Guess what? If you have precision dice and a crappy list, you'll still get beat, no matter how fairly your dice roll. Come on, dude.

Ou both missed my point, by a massive margin. Knowingly increasing your odds by rolling better than your opponent, even if it is only bringing your dice to average by knowingly pursuing perfectly balanced dice, if garnering an unfair advantage.

If you know Chessex dice roll below average, but consistently above average, and that almost every opponent is using these substandard dice, then pursuing precision dice to increase your odds to average is trying to gain an unfair advantage over your opponents. That is cheating in its simplistic form.

Now, I've postulated that these Cheap dice won't perform nearly as poorly in non clinical settings, but hey, I'm working off the assumption the provided data is accurate. If it is accurate, then it's knowingly creating an unfair advantage over your opponent.

You think that somehow all the bad dice are biased towards 1, what if dice is biased towards 5s and 6s which doesn't matter at all.

I would not be cheating my opponents with rolling an average outcome dice , they would be cheating me with rolling a non-average outcome dice.

You are wrong that trying to get average resutls is cheating but anything else is not.

No, according to the supplied data Chessex style dice roll below average. Knowingly trying to gain an advantage over your opponents buy buying dice that are known to roll better is cheating,just as using a cube weighted to roll better would be.

Level playing field is fair, by trying to garner an advantage over the vast majority of opponents through dice is wrong.

Call me crazy, but I'm happy playing with the same, likely imaginary, handicap as everyone else. Plus, I've already postulated that this difference is almost nonexistent in a non clinical setting which has been backed up by AllSeeingSkinks data of over one thousand rolls.

9 out of 10 people use Chessex. Possibly more. Don't worry about "fairness". Even if Chessex rolled 1's 40% everyone would be affected and playing on the same field.

All our tournament participants have to use the dice provided by us. All of them are Chessex. So...np

Plus, they got badass Necrons-ish dice (Bright Green Vortex).

AllSeeingSkink wrote:

 jreilly89 wrote:
I appreciate the sentiment and echo your message, but I want to point out, 1060 is hardly a good sample size. The original article (I do not remember the exact but I believe it to be in the tens of thousands) did a much larger sample size and got similar results. I appreciate what you did, but 1000 is way too small of a pool.

My response to that is to highlight one of my sentences in that post...


Of course, I could roll the dice again tomorrow and get 29% 1's. But if there were a 29% chance of rolling a 1, the chance of me rolling less than 170 1's out of 1060 rolls (I rolled 168) is only 0.00000000000000000006%.

That's less than the chance of rolling 27 6's in a row.

So, yeah, I'm pretty confident that my dice set is no more than a few % biased one way or another.

The idea that you have to roll millions of dice to get close to the average is a fallacy. It's a binomial distribution, you just need to roll enough dice that the % chance of being far from the average is small. If there was really a 29% bias toward 1's (rather than the expected 17%), the chance of me not seeing it in 1000 rolls is extremely small. If the bias were only, say, 20%, yeah, I agree, I might miss it in only 1000 rolls (in fact I can tell you the chance of me rolling what I did if the bias were 20% is less than 0.03%).

I actually do experimental aerodynamic research for my job, part of that is knowing where you draw the line of "ok, this is enough data to be confident enough to make a statistically significant statement". IMO, 1000 rolls is enough to say something statistically significant within a few % of the expected mean. If you were more than a few % away from the expected mean, I'd totally say repeat the experiment, given the expected mean is 16.7% and my largest variation from that was 17.8%... I am satisfied with that.

EDIT: Note, when I say 1000 samples is enough for a statistically significant result, that's partly because the chance of getting any 1 side is 17%, if we were talking about a D20 where the chance is only 5%... yes, you'd have to roll it more to get a significant result).

I understand that and I agree, my point was just that I wouldn't say 1000 rolls is enough to publish a report or support a theory in a scientific setting. On Dakka sure, that's large enough, but if you wanted to use it as supporting evidence, I'd want a larger sample pool.

Ah, probability and dice. The thing is, it's a probability. To develop or truly see the probability curve of just one die, you have to roll it the exact same way, the exact same distance, with the exact same force, onto the exact same surface, and it has to stop on the exact same spot, or your results will be skewed. Oh yeah, and you'll have to do this thousands of times. The only other way is to invest in one of those dice checkers they use in casino's, and who knows how much they cost and if you can even get one.

Unlike other systems (such as Privateer Press' Warmachine), 40K requires a lot of dice to be rolled once. In Warmachine, you at best will ever need 5 dice. Bad dice in Warmachine is much more noticeable as you throw the same dice over, and over, and over again. In 40K, there are fewer dice rolls for those dice, but you are usually required to roll more dice... 10, 20, 30, heck, I've seen an Ork player roll 70 dice, and I'm sure you could come up with more dice to roll all at once. Most 40K players will have at least a "brick" of 36 dice. Making a lot of assumptions here, and it all depends on your army, but let's say a Space Marine army of 40 models, plus some tanks/flyers/whatever. You'll probably roll each of those 36 dice five times? Maybe six during a game? 5x36=180 dice rolls, so, yeah, that feels about right for me in that army. I'd say a typical game of Warmachine at a similar level would have maybe 60 dice rolls? The point here is that with such few dice rolls, any die that isn't "perfectly balanced" will skew all the results, but with the technique used to throw them and the surface they roll on will have a greater impact on the results... we could really chase the rabbit down the rabbit hole with this discussion.

Of interest to me is that people really get locked into this dice argument, but no one ever questions measuring sticks/rulers/tape measures or "widgets". How many times has your opponent measured the distance for one model, then just moved the other 9 models in his squad? I've had a game where my opponent was supposedly able to shoot at one of my models on the bottom of turn one, despite the fact that I showed him on paper that it was mathematically impossible with the movement, deployment, and range of his weapon. And yet, when we both measured it with our tape measures, he had inches to spare. So who cheated their movement there?

Have you ever measured the blast markers from GW? Just for kicks, measure the opaque clear, the green ones, and those fancy ones with the blue acrylic lightning in them. Notice a difference? Another fun exercise- measure the red sticks that come in Dark Vengence.

People are squabbling over the probability of 0.001 vs. 0.002 on a six sided die when you are using measuring devices that are up to an inch or more off, not to mention what I like to call "The Emperor's Inch" that every player seems to add to their movement.

It's a purely internet "thing", same as most YMDC threads. It belongs to the discussion on internet forums and its surrounding culture, but it has zero bearings on actual playing.

This is fine and not supposed to sound like "haha what you say is stupid". Far from it. Just saying that in actual play, this isn't an issue.

It's interesting to read and just to drop in - those Dakka "findings" are highly debatable. 29%? Not buying it

As odd as it sounds, try looking up a few "Dice Etiquette" videos. There ways to roll dice normally that have a more random result than simply throwing dice from a larger height. The video's I've seen has shown that the paint on the 6 side, tends to make that side slightly weighted down. Increasing its chance of ending face down with the 1 up. The way you roll your dice has a very large effect on this. If you roll your dice from a higher height you will have the top heavy portion of your number 6 landing face down more than your would otherwise

Aside from this purely theoretical notion. Dice are dice to me, but I don't shake them, I stir

 Zagman wrote:


No, according to the supplied data Chessex style dice roll below average. Knowingly trying to gain an advantage over your opponents buy buying dice that are known to roll better is cheating,just as using a cube weighted to roll better would be.

Level playing field is fair, by trying to garner an advantage over the vast majority of opponents through dice is wrong.

Call me crazy, but I'm happy playing with the same, likely imaginary, handicap as everyone else. Plus, I've already postulated that this difference is almost nonexistent in a non clinical setting which has been backed up by AllSeeingSkinks data of over one thousand rolls.

It is fun when you take the data that supplies your opinion to be true and others as needs more study. And i am not accepting that rolling lower then average results is fair play for anybody in a game where stuff should be designed on average outcome. We should not be playing with non average outcoming dice because it's "fair". We should all be playing average outcoming dice, period.

 jreilly89 wrote:
I understand that and I agree, my point was just that I wouldn't say 1000 rolls is enough to publish a report or support a theory in a scientific setting. On Dakka sure, that's large enough, but if you wanted to use it as supporting evidence, I'd want a larger sample pool.

If I was writing an article to be published in a scientific setting... I'd be getting paid to do it so I'd be willing to spend more than half an hour (or about an hour after I expanded to 2000 rolls) doing it

Actually, in a scientific setting, I'd be drawing a convergence graph like I did in my other post, I'd calculate according to the binomial distribution how many trials are needed for my desired level of accuracy and then post my convergence graph as evidence.

Honestly, if I were actually posting in an engineering/science journal, I don't think I'd need anywhere near 36,000 trials. You could do it once to prove that you DON'T need 36,000 trials... but actually in a scientific setting, you only need to show that you've done sufficient trials and a calculation of the standard deviation is enough to show that (if there's less than 1 in 1,000,000 chance that you are more than a couple of % away from the expected average, I think that would be more than enough for a peer reviewed scientific journal... at least any scientific journal intended for practical application).


Automatically Appended Next Post:

 pizzaguardian wrote:
It is fun when you take the data that supplies your opinion to be true and others as needs more study.

Actually his hypothesis (which I'm tending to agree with) is that different surfaces/rolling techniques/environmental conditions will produce different biases. The evidence for that is there's 2 different studies with very different results.

You can either call one study crap and the other true, or you can say there's circumstances beyond what the studies have analysed that produced the different results.

I encourage any gamers who are actually interested to test it themselves and post the results. It took me about an hour to do 2000 rolls, I think 1000 is probably enough to detect any large bias (more than a few %).

And i am not accepting that rolling lower then average results is fair play for anybody in a game where stuff should be designed on average outcome. We should not be playing with non average outcoming dice because it's "fair". We should all be playing average outcoming dice, period.

I agree with you on this, many of us use statistics to value things and while you might say things balance out if both players are using the same dice (even that's not really true, some armies are much more sensitive to rolling 1's than others). If there is a 29% bias to 1's and you KNOW you have a 29% bias toward 1 and your opponent doesn't, you can alter your list to take units that are less sensitive to rolling 1's and put yourself at an advantage.

Of course, after my own testing, I'm not convinced there is anywhere close to a 29% bias toward 1, my rolling is suggesting a much more manageable 1% bias toward 6.

Again... the alternative for the ACTUA:L:Y precision custom dice is cost prohibitive. No sane person should pay that much. If as a Christmas gift or something, very cool Its the kind of thing you ask for as a gamer geek.

But I'll take the Chessex dice and run happily.

 Jancoran wrote:
No sane person should pay that much.

Are we still talking about dice or GW's latest release?

 Ghaz wrote:

 Jancoran wrote:
No sane person should pay that much.

Are we still talking about dice or GW's latest release?

LOL

 Jancoran wrote:
Again... the alternative for the ACTUA:L:Y precision custom dice is cost prohibitive. No sane person should pay that much. If as a Christmas gift or something, very cool Its the kind of thing you ask for as a gamer geek.

But I'll take the Chessex dice and run happily.

Except if you can avoid bias with Chessex simply by making sure you aren't rolling them a certain way then you don't have to buy the expensive precision dice

 pizzaguardian wrote:

 Zagman wrote:


No, according to the supplied data Chessex style dice roll below average. Knowingly trying to gain an advantage over your opponents buy buying dice that are known to roll better is cheating,just as using a cube weighted to roll better would be.

Level playing field is fair, by trying to garner an advantage over the vast majority of opponents through dice is wrong.

Call me crazy, but I'm happy playing with the same, likely imaginary, handicap as everyone else. Plus, I've already postulated that this difference is almost nonexistent in a non clinical setting which has been backed up by AllSeeingSkinks data of over one thousand rolls.

It is fun when you take the data that supplies your opinion to be true and others as needs more study. And i am not accepting that rolling lower then average results is fair play for anybody in a game where stuff should be designed on average outcome. We should not be playing with non average outcoming dice because it's "fair". We should all be playing average outcoming dice, period.

Wow, please reread my posts with understanding, either you have misread much of what I have posted or have simply failed to grasp it.

My hypothesis, which now has some support, is that the surface and dice rolling technique has a large impact on the outcome of the roll. Rolled in a clinical setting it has been demonstrated that cheaper dice such as Chessex can show an extreme bias towards 1. But, that is only in a clinical setting. I have postulated, and AllSeeingSkink supported it with 2000 rolls, is that much more expected results can be achieved when large numbers of dice are rolled together on softer less precise surfaces.

The fact that precision dice are not required aside, if there truly existed such a difference, it would be garnering an unfair advantage over opponents to play with precision dice while they used Chessex and would be akin to cheating, especially if they were unaware and you were aware of the difference in dice and its outcome on the average die rolls.

Of course it doesn't matter, because cheap awesome looking Chessex dice roll extremely close to average under normal gaming conditions which differ wild from the clinical Dakka study rendering it a moot poitn.

Couple of questions...

How do people feel about custom dice? I've got custom Chessex for both my Sisters and Emperor's Children, with symbols and words instead of 6s. I always get stick for them from my gaming group, with people suggesting they are unfairly weighted after a couple of decent rolls in a row. They always seem to balance out overall though - some good rolls, some bad, very occasionally awesome and very occasionally terrible.

Secondly, I always give my dice a really good shake before I throw them to increase the randomness [dunno if this works but I'm applying the same principle as shuffling cards really well]. A friend of mine lets them kind of drop out of his hand, which always annoys me a little [he is a pretty lucky with his dice too]. Is there a certain amount of time people should shake?

I use Chessex dice for RPGs, and they appear to be pretty even.

Does a dice-tower make a difference, to even out quirks?

Taking the physical side out of the problem, are any dice-roller apps confirmed to be accurate?
I have FFG's Star Wars app, which has normal dice in it, too.