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.

 CrownAxe wrote:
I'm not trying to prove the article is lying. In fact I think those 72 dice are in fact unbalanced.

I'm saying that single article is not enough evidence to claim that Chessex and GW dice are on the whole unbalanced. A basic understanding on manufacturing and producing goods is enough to prove that it's sample size is too small to be conclusive.

In science you can't prove any hypothesis with a single study, you have to have multiple people repeat that experiment to verify those findings. That has not be done with this article.

Okay. How many Chessex and GW dice do you think are unbalanced then? Because this article seems to suggest the answer is "at least 72" not "0."

And it seems to have proved that at least 72 dice are unbalanced by testing a total of 72 dice.

Also, given what the study required to be available to the person doing it, expecting anyone to ever even be able to attempt to replicate it is unreasonable.

 Pouncey wrote:

 CrownAxe wrote:
I'm not trying to prove the article is lying. In fact I think those 72 dice are in fact unbalanced.

I'm saying that single article is not enough evidence to claim that Chessex and GW dice are on the whole unbalanced. A basic understanding on manufacturing and producing goods is enough to prove that it's sample size is too small to be conclusive.

In science you can't prove any hypothesis with a single study, you have to have multiple people repeat that experiment to verify those findings. That has not be done with this article.

Okay. How many Chessex and GW dice do you think are unbalanced then? Because this article seems to suggest the answer is "at least 72" not "0."

I never said 0 and doesn't need to be 0. If 99% of all dice were balanced that would be excellent and statistically indistinguishable from 100% balanced dice. But 99% of 1 million dice still means you have 10000 dice that are unbalanced. That's why a person testing only 72 dice from a single pack is not statistically significant because that is a small number when compared to 10000 unbalanced dice and that number is insignificantly small when compared. to the 990000 balanced dice


Automatically Appended Next Post:
How is it unreasonable to expect someone to replicate an experiment someone has already done? If one person can do it another person can repeat it. We aren't splitting the atom here.

 CrownAxe wrote:

 Pouncey wrote:

 CrownAxe wrote:
I'm not trying to prove the article is lying. In fact I think those 72 dice are in fact unbalanced.

I'm saying that single article is not enough evidence to claim that Chessex and GW dice are on the whole unbalanced. A basic understanding on manufacturing and producing goods is enough to prove that it's sample size is too small to be conclusive.

In science you can't prove any hypothesis with a single study, you have to have multiple people repeat that experiment to verify those findings. That has not be done with this article.

Okay. How many Chessex and GW dice do you think are unbalanced then? Because this article seems to suggest the answer is "at least 72" not "0."

I never said 0 and doesn't need to be 0. If 99% of all dice were balanced that would be excellent and statistically indistinguishable from 100% balanced dice. But 99% of 1 million dice still means you have 10000 dice that are unbalanced. That's why a person testing only 72 dice from a single pack is not statistically significant because that is a small number when compared to 10000 unbalanced dice and that number is insignificantly small when compared. to the 990000 balanced dice

You remember earlier in this thread when I talked about how my white GW dice seemed to roll way more 1s than they should so I invented the idea of ballast dice?

Is it possible that the 10 or so white dice in my collection of at least 200 or so GW/Chessex dice, that roll that many more 1s than normal are actually so poorly manufactured that I could probably say I have dice in my collection that are balanced even worse than the worst offenders in the study, in my personal collection? Even though the rest of my collection of dice seem to roll totally fine all the time?

Because the fact that 5% of my GW/Chessex dice are unbalanced severely is messing up my gameplay badly.

To which I must ask.

Does it even matter that most GW/Chessex dice are fine when the ones you are personally using are badly imbalanced and messing up your personal gameplay?


Automatically Appended Next Post:

 CrownAxe wrote:
Automatically Appended Next Post:
How is it unreasonable to expect someone to replicate an experiment someone has already done? If one person can do it another person can repeat it. We aren't splitting the atom here.

Why exactly haven't you personally replicated it yet to prove it wrong then? Just include more dice than 72.

 Pouncey wrote:

But I'm dealing with you guys who are saying things like 29% of 72,000 rolls are 1s is equally as unlikely as 100% of 1,000,000 rolls being 6s, and even I know that's completely wrong just from the numbers you're comparing, while you guys seem to think it's perfectly okay to say and defend as a valid comparison!

I already explained what I meant by that. You obviously didn't understand. I am not going to explain it again. Go back anf read again my last post on the topic.

"I don't believe you, so you're lying."

That is your argument to ignore 72,000 dice rolls that show GW and Chessex dice are poorly-made.

This article has been discussed numerous times over the last ten years. I read every page of every discussion on this article I could find. In these numerous threads there were a lot of people who ran their own experiments. None of them replicated the results of this study. None of them even hinted that ones would be significantly more probable. And I ran my own tests, which also did go against the results of the study. So no, I do not believe the results of this study. Whether due some colossal feth up or dishonesty this study's results are highly erroneous.

 Pouncey wrote:

 CrownAxe wrote:

 Pouncey wrote:

 CrownAxe wrote:
I'm not trying to prove the article is lying. In fact I think those 72 dice are in fact unbalanced.

I'm saying that single article is not enough evidence to claim that Chessex and GW dice are on the whole unbalanced. A basic understanding on manufacturing and producing goods is enough to prove that it's sample size is too small to be conclusive.

In science you can't prove any hypothesis with a single study, you have to have multiple people repeat that experiment to verify those findings. That has not be done with this article.

Okay. How many Chessex and GW dice do you think are unbalanced then? Because this article seems to suggest the answer is "at least 72" not "0."

I never said 0 and doesn't need to be 0. If 99% of all dice were balanced that would be excellent and statistically indistinguishable from 100% balanced dice. But 99% of 1 million dice still means you have 10000 dice that are unbalanced. That's why a person testing only 72 dice from a single pack is not statistically significant because that is a small number when compared to 10000 unbalanced dice and that number is insignificantly small when compared. to the 990000 balanced dice

You remember earlier in this thread when I talked about how my white GW dice seemed to roll way more 1s than they should so I invented the idea of ballast dice?

Is it possible that the 10 or so white dice in my collection of at least 200 or so GW/Chessex dice, that roll that many more 1s than normal are actually so poorly manufactured that I could probably say I have dice in my collection that are balanced even worse than the worst offenders in the study, in my personal collection? Even though the rest of my collection of dice seem to roll totally fine all the time?

Because the fact that 5% of my GW/Chessex dice are unbalanced severely is messing up my gameplay badly.

To which I must ask.

Does it even matter that most GW/Chessex dice are fine when the ones you are personally using are badly imbalanced and messing up your personal gameplay?

To the individual no. But if you are only looking at an individual's personal dice then it doesn't matter what any article or study says anyway. All that actually matters is if those specific dice are unbalanced and you can only prove that by testing them yourself

Automatically Appended Next Post:

 CrownAxe wrote:
Automatically Appended Next Post:
How is it unreasonable to expect someone to replicate an experiment someone has already done? If one person can do it another person can repeat it. We aren't splitting the atom here.

Why exactly haven't you personally replicated it yet to prove it wrong then? Just include more dice than 72.

I because I don't care about proving that cheesex dice are balanced. I'm just pointing out that the little evidence we do have isn't sufficient enough to prove that they aren't balanced.

 Pouncey wrote:

You remember earlier in this thread when I talked about how my white GW dice seemed to roll way more 1s than they should so I invented the idea of ballast dice?

Take those dice and throw them thousand or at least couple of hundred times and write down the results. Then you have pretty good idea whether they actually roll more ones or whether you're just imagining it.

May as well toss this into the mix; I've noticed that my Munitorum Issue dice roll 1's and 2's far more often than I'd consider normal within the terms of random probability. However, I know these dice aren't exactly the greatest - some have indentations, others chips on the corners.

With every other dice pack I've used over the years, I've never really noticed any consistent pattern. Even if I believed that those dice were not random enough, I'd throw them into a dice pool with my opponent like most people have suggested here to solve the issue.

G.A

 Crimson wrote:

 Pouncey wrote:

You remember earlier in this thread when I talked about how my white GW dice seemed to roll way more 1s than they should so I invented the idea of ballast dice?

Take those dice and throw them thousand or at least couple of hundred times and write down the results. Then you have pretty good idea whether they actually roll more ones or whether you're just imagining it.

Or, I could not do that, and instead stop using those dice and use a different color dice for melta rolls, which would solve the problem I am imagining to myself instead of just proving there is a problem and then doing so anyways.

I mean, you're saying I should prove something about a problem I can just solve by using different dice.

Whether it's true or not, if I believe those other dice are giving me random results, why should I prove it instead of just doing what I'd do after proving it - use different dice from the 95% of my collection no one believes is unbalanced?


Automatically Appended Next Post:
New question, since you're people of science:

Why is the response to:

"I believe that 5% of my personal dice are heavily unbalanced."

To ask me to prove it?

Instead of saying:

"So throw those dice in the garbage."

If you don't care about the truth, then you don't care. Not my problem if you throw your dice away.

Why would you throw away something you spent good money on just because they might possibly be unbalanced?

 CrownAxe wrote:
Why would you throw away something you spent good money on just because they might possibly be unbalanced?

Because I have 190 other dice that seem to work just fine and the amount of money I spent on 10 GW dice is less than the cost of a single Battle Sister.

Also, I don't even have to throw them out. I could make really good use of them by always using them for Leadership rolls and nothing else.

The study says that me doing so would make me a cheater.

The study even said you should keep using GW dice after everything it said. Leadership tests and other times where rolling more 1s than normal is beneficial already exist in WH40k. And the guy who wrote the article thought that over an entire game, the fact you used GW dice or casino dice would make no difference to the outcome of the game.

Then he said that people who use GW dice for leadership tests, and casino dice for everything else, are people you should call out for cheating now that you had read a scientific analysis of Chessex, GW, board game and casino dice that proved they were cheating.

I mean, you DID read the part where he said that rolling as many extra 1s as they did was a problem that would solve itself in a normal game of 40k, right?


Automatically Appended Next Post:

 Crimson wrote:
If you don't care about the truth, then you don't care. Not my problem if you throw your dice away.

What does your doctor tell you when you tell them your arm hurts when you make a weird motion with it you won't ever make unless you want to be making it?

"Stop doing it."

Do they attempt to prove anything about why you were feeling that pain, or do they just tell you to stop doing the thing you will never be doing unless you want to be doing it?

Does it matter to you personally WHY you felt that pain you will never feel again in your life by not doing that thing anymore? Do you insist on tests to prove it?

Or do you just stop doing the thing you don't have to be doing, that is causing you pain, and leave it at that?

I'm only here to point out that that article isn't significant evidence of anything.

Or are you going to keep ignoring my last post on the matter?

 CrownAxe wrote:
I'm only here to point out that that article isn't significant evidence of anything.

Or are you going to keep ignoring my last post on the matter?

You know, GW could've known it already and written the rules around the fact their dice roll more 1s than normal. The article says that may have happened.

WH40k is already making use of rules that require a non-even distribution of results, literally every time it has you roll 2d6.

"1d6 results in 29% 1s" isn't anything you couldn't design rules around more than the following:

https://www.google.ca/?gws_rd=ssl#q=distribution+of+2d6

Why are you even so obsessed with proving or disproving something that will never be used to affect anything more important to the world than a tabletop game being played for fun?

It's not. That. Important.

Anything where it is important to use perfectly fair dice isn't using GW or Chessex dice, it's using casino dice.

If it's not that important Pouncey, stop harping on it, and accept that other people's results and opinions are different.

The fact that a 29% 1's is dependent on flawed manufacturing, and the % is liable to change over time, would make planning a game around it flawed practise.
And really, GW sat down and made a statistical analysis of their dice before writing rules?
2D6 is predictable and regular, and is a much better model then the one GW uses.

Otherwise, I ran my own test with my Chesex.

30 dice, tossed in two lots of 20, so 1200 rolls total.

First test, in a box on carpet (so shock absorbent) averaged 13% 1's, with a Stderr of 1.1%. 600 rolls.

Second, on a table, so a more 'realistic' test, averaged 17% 1's, with a Stderr of 0.9%.

Together, averaging 15% with Stderr 0.7%.

Much closer to average, much more inline with the majority of results reported. I'm going to add that I think the 'big study' is probably the outlier.

And lo, the Dreadnought has been awoken from slumber and doesn't have coffee (pepsi is a poor, but caffeine rich substitute).

Ok, seeing more back and forth here than at Wimbledon I decided to do a test of my own to see if there's any inherent bias.

Using 4 different sets of dice: Blue GW dice cube (BGW), White GW dice from numerous starter sets (WGW), 'Lotus' coloured Chessex dice (LtsChx) and Green w/ bronze pip Chessex dice (GrnChx) I made 599 rolls total, due in part from rolling individually to start but getting tired and moving to whole sets. I did throw 4x9 twice to spice up the random. Shouldn't have done that but meh, I'm a medical scientist not a statistician (which is someone better qualified than a scientist to conduct this experiment).

population of each set is as follows:
n=25, BGW
n=42, WGW
n= 36, both LtsChx and GrnChx




Now I'll admit this isn't a properly performed experiment, fraught with inconsistent methods of rolling the dice (singly, batches of whole populations, 9 from each set rolled together) but the percentages are roughly equal. I am on holidays and really can't be stuffed to do this scientifically, so there.

The results won't be significant or anything but damn are all result percentages close to some kind of bell curve barring the green Chessex dice, which didn't want to roll a 3 all that often.

What was interesting to note was that there were some pretty statistically anomalous dice rolls. In one set of Lotus Chessex rolls (dice rolled = 36), 13 were 5's. Another example of statistically unlikeliness was the disproportionately low number of 3's rolled with green Chessex dice. Undoubtedly if n=10,000 or higher, the results would be closer to 16% across the board, with slight bias one way of the other. Point is that with the percent of 2's rolled being >30% doesn't indicate my dice are weighted to give 2's more often than not. Blue GW dice seem to have a nicer distribution, almost some kind of bell curve.

Take home message is that the dice gods hate you and will always hate you and that Mork (or possibly Gork) is the god to pray to. The End.

PS: Please report any unregistered Psyker activity to the Inquisition. Such activity may include, but is not limited to: Telepathy, telekinesis, biomancy, love of disco, and failure to praise the Emperor.

I tried the 'if it floats it's a witch!' test just out of curiosity, but I'm not going to get a joke out of it (or have to burn any of my dice at the stake) since all the dice I tested (GW dice, PP starter set dice, FFG RPG dice, Godslayer dice, and a few MTG spindowns just in case) behaved exactly the same. So if you're having trouble rolling any of those buying a new set isn't going to help unless it's going to exorcise demons of some sort, the quality control is probably working.

Also spent a bit of time adding to SpinCycle's data set. Rolled a batch of fifteen blue GW dice (on the logic that large batches is close to 'normal game conditions') twenty times and came up with:

1: 38 (12.6%)
2: 51 (17%)
3: 46 (15.3%)
4: 49 (16.3%)
5: 57 (19%)
6: 59 (19.6%)

Adding that to SpinCycle's data set we're actually pretty close to what we'd expect out of fair dice (16.67% chance of any given result), except that we've gotten many fewer 1s than we'd expect (12.5% of rolls).

If you've got a spare ten minutes and some GW dice you could help add to the data set, see if any patterns emerge as we go.

(The explanation I usually hear is that GW dice have indented pips so the '1' face is actually the heaviest, so they're slightly more likely to roll high than low)

 Farseer Anath'lan wrote:
If it's not that important Pouncey, stop harping on it, and accept that other people's results and opinions are different.

The fact that a 29% 1's is dependent on flawed manufacturing, and the % is liable to change over time, would make planning a game around it flawed practise.
And really, GW sat down and made a statistical analysis of their dice before writing rules?
2D6 is predictable and regular, and is a much better model then the one GW uses.

Otherwise, I ran my own test with my Chesex.

30 dice, tossed in two lots of 20, so 1200 rolls total.

First test, in a box on carpet (so shock absorbent) averaged 13% 1's, with a Stderr of 1.1%. 600 rolls.

Second, on a table, so a more 'realistic' test, averaged 17% 1's, with a Stderr of 0.9%.

Together, averaging 15% with Stderr 0.7%.

Much closer to average, much more inline with the majority of results reported. I'm going to add that I think the 'big study' is probably the outlier.

I didn't even want to discuss this! My OP wasn't ABOUT this article at all! It's about what the guy at the store said!

Also, why the hell does it matter if you roll twice as many 1s?

Here's the difference between 6 dice rolls:

16.7% 1s:
1,2,3,4,5,6

33% 1s:
1,1,3,4,5,6

You probably would just chalk it up to random chance making you a bit unlucky!

No one cares what some guy in the store said, and it matters massively if ones are twice as likely as they should. Think about it, if you're playing a terminator heavy GK army, you're failing armour saves twice as often as you should. That is a huge difference.

 Pouncey wrote:
...Also, why the hell does it matter if you roll twice as many 1s?

Here's the difference between 6 dice rolls:

16.7% 1s:
1,2,3,4,5,6

33% 1s:
1,1,3,4,5,6

You probably would just chalk it up to random chance making you a bit unlucky!

You do know this is why people have been trying to make a point by posting the results of 300 or 1200 rolls, not the result of 6, right?

 Crimson wrote:
No one cares what some guy in the store said, and it matters massively if ones are twice as likely as they should. Think about it, if you're playing a terminator heavy GK army, you're failing armour saves twice as often as you should. That is a huge difference.

I cared, so I thought I'd share the story, like we share stories around here for fun.

Yeah, it is. But it already happens that you take 5 lasgun wounds on 5 Terminators and roll 5 1s.

Over a thousand dice rolls it's a big difference, yeah. But when the hell have you EVER rolled a thousand dice at once?

Usually the most dice I roll at once is about 20 when firing a squad of 10 Battle Sisters' bolters. Rolling twice as many 1s as normal just means that I roll six 1s instead of 3 on average. And that's with the absolute worst offender in the article. The least was only at 23%, which is a third extra 1s, which would mean I'd roll four 1s instead of three.

The scale on which it matters is bigger than what you'll see in a game. You're not rolling enough dice at once to notice.


Automatically Appended Next Post:

 AnomanderRake wrote:

 Pouncey wrote:
...Also, why the hell does it matter if you roll twice as many 1s?

Here's the difference between 6 dice rolls:

16.7% 1s:
1,2,3,4,5,6

33% 1s:
1,1,3,4,5,6

You probably would just chalk it up to random chance making you a bit unlucky!

You do know this is why people have been trying to make a point by posting the results of 300 or 1200 rolls, not the result of 6, right?

When was the last time you rolled 300 dice at once in a game, much less 1200? Do you make a habit of fielding extremely large Green Tide formations or something? How did you even get that many models in range of your target?

What the hell does it matter if the dice are rolled at once? You roll many dice over the games and even more over several games.

If the imbalance as high as suggested in the article was real, the dice being used would probably be the biggest single factor deciding the outcome of the game.

 Crimson wrote:
What the hell does it matter if the dice are rolled at once? You roll many dice over the games and even more over several games.

If the imbalance as high as suggested in the article was real, the dice being used would probably be the biggest single factor deciding the outcome of the game.

You know what dice rolls don't matter whatsoever to the outcome of a game? The dice rolls you made in your previous game.

Also, rolling 1s is often good, you know. Rolling more 1s than normal helps you pass Leadership tests. Leadership tests can be hugely important, and if the dice are imbalanced, you're also passing more of those. Characteristic tests can often be a matter of "roll low or die instantly."

So it balances out since rolling 1s isn't always a bad thing. Sometimes you WANT to roll 1s.

There are way more situations where rolling low is bad than good. For starters space marine armies don't much worry about LD tests. Sure, they may matter a tiny bit, but it's not a big deal.

 Pouncey wrote:
Automatically Appended Next Post:

 AnomanderRake wrote:

 Pouncey wrote:
...Also, why the hell does it matter if you roll twice as many 1s?

Here's the difference between 6 dice rolls:

16.7% 1s:
1,2,3,4,5,6

33% 1s:
1,1,3,4,5,6

You probably would just chalk it up to random chance making you a bit unlucky!

You do know this is why people have been trying to make a point by posting the results of 300 or 1200 rolls, not the result of 6, right?

When was the last time you rolled 300 dice at once in a game, much less 1200? Do you make a habit of fielding extremely large Green Tide formations or something? How did you even get that many models in range of your target?

You're the one insisting slight imperfections in the dice have something to do with the 'ultimate condemnation' of the game. No game cares about a half a percent variance, no matter how well-made it is, just because we'd all get bored and go home long before we rolled enough dice to notice.

(80 is probably the soft limit for the number of dice you're getting out of a single attack; you can get more, but if you are it's pobably because you're doing something silly just to set the record rather than because you actually need to. For reference the two units off the top of my head that will put out eighty shots are Legion Tactical Squads (in which case that's eighty bolter shots) or Secutarii Peltasts (in which case they're S3/AP-/Shred).)

 Crimson wrote:
There are way more situations where rolling low is bad than good. For starters space marine armies don't much worry about LD tests. Sure, they may matter a tiny bit, but it's not a big deal.

Yeah, uh, a lot of people don't play Space Marines.

And your squad passing a morale test to avoid running off the table seems pretty important.


Automatically Appended Next Post:

 AnomanderRake wrote:


You're the one insisting slight imperfections in the dice have something to do with the 'ultimate condemnation' of the game. No game cares about a half a percent variance, no matter how well-made it is, just because we'd all get bored and go home long before we rolled enough dice to notice.

(80 is probably the soft limit for the number of dice you're getting out of a single attack; you can get more, but if you are it's pobably because you're doing something silly just to set the record rather than because you actually need to. For reference the two units off the top of my head that will put out eighty shots are Legion Tactical Squads (in which case that's eighty bolter shots) or Secutarii Peltasts (in which case they're S3/AP-/Shred).)

If you're gonna argue with me what I meant when I wrote the OP, you should probably read more of it than the first sentence. Also you should read the full title instead of all of it except the last word.

If I were given a choice between balanced dice and ones that roll ones almost twice as often as they should in order to play a game of 40K, I would choose the balanced dice every time. And if my opponent was then using the imbalanced dice, I would have a huge advantage. Thinking that this wouldn't be the case shows a shocking lack of understanding regarding both 40K rules and probability.

 Crimson wrote:
If I were given a choice between balanced dice and ones that roll ones almost twice as often as they should in order to play a game of 40K, I would choose the balanced dice every time. And if my opponent was then using the imbalanced dice, I would have a huge advantage. Thinking that this wouldn't be the case shows a shocking lack of understanding regarding both 40K rules and probability.

The article disagrees with you. The article thinks it doesn't matter if you use casino dice or unbalanced GW dice, so long as you use the same dice for everything. The article even says that if you use GW dice and your opponent uses casino dice, it doesn't matter.

You're making a bigger deal out of this than the article did, you know.

The article is bollocks and you're silly for believing it. Thinking that such a massive increase in failure rate wouldn't matter is completely absurd.

 Crimson wrote:
The article is bollocks and you're silly for believing it. Thinking that such a massive increase in failure rate wouldn't matter is completely absurd.

Okay.

So then ignore it and play with your GW dice like the article suggested you should do even if you believed the article.

What exactly would you do differently if you believed the article and took its suggestion of playing with GW dice anyways?

 Pouncey wrote:

What exactly would you do differently if you believed the article and took its suggestion of playing with GW dice anyways?

Well obviously I would never play with such dice. The dice that imbalanced would have a huge effect on how the game plays. The areas most obviously affected would be units with good armour saves and get's hot weapons which would be massively worse than they should. Weapons relying on scatter would be somewhat better than they should. And of course if only one side was using such dice, that side would be completely fethed unless they built their entire army around scatter-based weapons (and probably still, because they would have significant disadvantage while wounding and making armour pen rolls.)