- Gnome Stew - https://gnomestew.com -

Things I Learned while “Inking” 14 GameScience Dice with a Crayon

In preparation for the release of the upcoming DCC RPG [1] from Goodman Games, I recently acquired a set of GameScience [2] dice.

If you’re not familiar with GameScience, Col. Lou Zocchi, or why these dice are different from the dice you’re probably used to, these two GenCon 2008 video sales pitches by the colonel himself are a good place to start: part 1 [3], part 2 [4].

Here’s the short version: GameScience is most famous for making precision-edge dice — razor-sharp, very lightweight, and with flat faces. The license to produce these dice currently resides with Gamestation.net [5], and they make and carry the full line in all sorts of colors. Significantly, GameScience dice are by default not inked and not tumbled; tumbling is the process that allows other dice manufacturers (like Chessex [6], to cite my favorite) to ink the numbers and smooth the edges of their dice. (GS does make a few non-precision dice, notably the d7 and d100.)

Non-tumbled dice are more likely to roll truly random results because, as I understand it, tumbling is never as even as you’d think. That’s why casino dice aren’t tumbled: They have sharp edges, totally flat faces, and their pips are filled in with plastic that’s the same density as the plastic of the die itself. They also cost a shitload of money and, unsurprisingly, don’t come in any shape but six-sided.

Sooooo, history lesson aside, why am I writing about GameScience dice? Because I thought they sucked, I was wrong, and I learned some things in the process of discovering that I was wrong. I couldn’t find too much info about them online, particularly not from the perspective of someone who initially didn’t like them, so I figured I’d help fill in that gap.

Why I thought they sucked

GameScience dice are relatively expensive, they always include a blemish from where they were broken off the mold sprue — usually either a small chip or a protruding piece of plastic (supposedly neither affects the results) — and you either have to pay more to have them inked or ink them yourself.

For what GameScience dice cost, you can get much fancier, pre-inked dice with no imperfections. Depending on the die type, these are in the Q-Workshop [7] price range, and well above Chessex or other big manufacturers. I’ve used precision dice before, way back in 1987, and I hated them.

Why I bought them

Because I’m weird about dice… [8]

But apart from that, because the DCC RPG uses some decidedly non-standard polyhedrals: d3, d5, d7, d14, d16, d24, and d30. (It also uses d4, d6, d8, d10, d%, d12, and d20.) I want to play it, and I don’t want to substitute other dice or use control dice or anything like that.

On top of that, I’ve read accounts online written by folks who own them, love them, and even came to love the stranger aspects of using them — like inking them yourself. I started to think it was worth biting the bullet and giving these dice a try.


I read a few things about how to ink GameScience dice before settling on my approach, notably Jeff Rients’ post about doing it with a Sharpie [9] and MJ Harnish’s review of the dice themselves [10].

I dug out a couple I had in my collection, grabbed a white crayon and a fine point Sharpie, and gave it a try. The Sharpie worked pretty well, but it smeared easily and I could see that if I wasn’t more careful than my fat fingers allow, I’d probably wind up with smudged dice. Bleh.

But the crayon was actually kind of fun to use, the results looked good, and it really popped in white on gem dice. I decided to spring for a set of ruby red gem dice [11], which I rounded out with a couple extra d6s (gotta have three), a hideous d7 [12], and a precision d30 from another vendor [13] (which was hard to find).

(I didn’t try a paint pen because I don’t own one, and ditto for grease pencil; I didn’t try actual paint because that sounded like way too much work. Apparently those are all good methods as well.)

How I inked them

I stole a white Crayola crayon from one of my daughter’s sets, put down a paper towel to catch stray crayon rubbings, and set to work. Here’s some uninked dice as a reference:

I rubbed every face of the die with the crayon, making sure to focus on going across the numbers, not with the grooves. I rubbed until every number looked perfect, then wiped the excess wax away from the face with a paper towel. If a number wasn’t totally filled in when I was done, I added a bit more wax before moving on.

Here’s my partially “inked” d16:

An example of how these won’t be perfect — note the mold protrusion on the right, which has snagged some crayon wax I can’t remove, and the fine crack across the 13, which has also gotten inked:

Most of the set, fully crayoned in:


What I learned

Like I said, I expected to dislike these dice — I didn’t even want to buy them. But I learned a few things after I went over to the precision-edge side [18]:

Will my GameScience dice be displacing my bag full of Chessex and Q-Workshop dice? I don’t think so. But I’ve been wrong before…

In any case, I hope you enjoyed seeing some photos of weird dice, reading about coloring things with crayons, and are either fondly reminiscing about your own GameScience dice or considering buying a set of your own. They’re weird, but they’re kind of fun, too.

39 Comments (Open | Close)

39 Comments To "Things I Learned while “Inking” 14 GameScience Dice with a Crayon"

#1 Comment By theeo123 On March 28, 2012 @ 1:55 am

I’ve done the “dice stacking” thing personally, I play white wold a lot so i had a veritable bucket of D10’s and I have to say I was impressed.

one of the things that i sort of both like and dislike about my game science dice, how LIGHT they are. this feels kind of nice in my hand ,don’t know why, just does, sadly on occasions when I’m using a dice tower/boot they have a tendency to bounce out.

I took some joy in inking mine, but that’s because I like to pant mini’s in the first place, I bought the emerald green set, and inked with Gold, looks great.

My opinion, everyone should buy at least once set, just to try it.

#2 Comment By Kristian Serrano On March 28, 2012 @ 5:37 am

Be sure to let us know whether they do end up replacing your other dice. I’m tempted to get a ruby red set, too, but I’m in the place you were before you bought a set.

#3 Comment By danroth On March 28, 2012 @ 7:20 am

I watched the Colonel’s videos a few months ago, and was pretty set on wanting to get a set. (I haven’t gotten a set yet, though.)

I had figured I would just get hobby paint, but I really like your crayon idea. I think I’ll have to try that when I end up getting a set.

#4 Comment By Roxysteve On March 28, 2012 @ 8:08 am

The problem with these dice (and I’ve had many of the little buggers over the years since at one time they were the only game (aid) in town) is that the plastic is a super-smooth one with a “soapy” feel to it that doesn’t hold ink well (as you discovered with the Sharpie experiment). Using paint is problematical since the lines are shallow and smooth edged allowing paint to bleed onto the surface easily (I used a toothpick as a brush some years ago with less meh results).

If your coloring agent of choice isn’t bone-dry when you handle the dice you’ll smudge them too, often to the point of decoloring the number itself.

Over the years I’ve tried paint, ink, paint pen, pilot pen, pilot pen dipped in paint, whiteout, cat-o-nine-tails and blasphemous rites calling upon the Elder Gods.

I think on the whole that crayon is the best idea with Gamescience dice.

#5 Comment By Genevieve On March 28, 2012 @ 8:39 am

I *just* re-inked one of my GameScience sets the other night! My first set I did with crayon and it turned out great, my second set… not so much. There were too many fine scratches on them, so I ended up using an extra-fine point Sharpie paint pen, a toothpick and a paper towel. It didn’t turn out perfect, but I can actually read the numbers now.

#6 Comment By Lee Hanna On March 28, 2012 @ 8:43 am

I bought a set 2 or 3 Origins ago, mostly on the strength of that video. I can testify that the PC I use them for no longer sucks quite so bad, and he’s spent a lot less time unconscious/dying. He’s actually hit a few things, too.

Not the most powerful of testimony, but there it is.

I did have fun coloring in the numbers with my sons, (they have sets now, too) even the non-game player participated.

#7 Comment By Alan De Smet On March 28, 2012 @ 10:46 am

I inked my current set with paint. I eventually gave up on trying to paint inside the numbers and just globbed a bunch of paint off, squeegied the excess off with my finger, then cleaned by finger with a paper towel. Came out pretty well, but now that I’ve seen how crayon filled dice look, I’m going to try that next.

For those who are curious, here is a relatively high resolution photo of Zocchi’s old diagram/chart/photo of dice stacks. [19]

My wife and I were suspicious, since his photo is of some _very_ old dice, so we bought a bunch of dice from multiple companies. Measurements with a digital ruler accurate to 0.001″ show that Gamescience dice are quite good, and the competition ranges from okay to embarrassingly bad. I hope to write up a longer post soon, but I really want to recreate Zocchi’s diagram first. Unfortunately I discovered that creating a 10-die tall stack of d20s is insanely hard. We’ll try again soon, next time with some sort of frame to help.

#8 Comment By Kurt “Telas” Schneider On March 28, 2012 @ 11:18 am

“When I roll them, I feel like a kid again.”

‘Nuff said!

I wish GS made larger sizes, but thye’d probably chew the heck out of my dice tray.

#9 Comment By Roxysteve On March 28, 2012 @ 11:33 am

Heh heh. My original Gamescience d20s had only two sets of 0-9 on them. You had to ink them in different colours to get a d20, otherwise you just have a fancy d10.

I recently brought them to a game. Two full sets, one yellow, one blue, d20s inked in black and white. The inking was not identical on both dice, nor were the dice made to the current “opposite faces sum to 21” scheme (on mine they sum to 19 and one orientation sums to 21 I think). Every legal d20 value was there, and there only once per die. I figure if they are properly tumbled it doesn’t matter where each colour or number lies in respect to the other possibilities. YMMV.You should have seen the twenty-something players crossing their eyes trying to understand the “Black is high” d20 system. When I then rolled 2d20 for a percentile result with a casual “Yellow is high” there was a small riot.Tch! Kids!

#10 Comment By Patrick Benson On March 28, 2012 @ 12:54 pm

I own 6 sets of Game Science dice. I use crayon to fill in the numbers on all of them. I would highly recommend washing them with dish soap first, then patting them dry, and then blasting them with a hair dryer for a few seconds to make sure that there is no moisture left on the surface of the dice. It makes the crayon “inking” process easier, and the crayon tends to last longer (you will have to apply it again from time-to-time just to touch up).

I do like how the dice feel, and I enjoy the nostalgia of using such simple dice since they remind me of the dice that came in the boxed sets that I started gaming with. That is why I buy them – personal taste.

But I think the claims made by the Colonel are BS. The dice are not superior to other brands like Chessex. Let me put it like this – air bubbles. All of my GS sets have them, and some of the bubbles are freaking huge (well for dice I guess).

I wish that there was a company that made casino quality polyhedral dice, but the cost would probably be outrageous. The requirements for casino dice are heavily regulated by gaming commissions. That is one reason why they are so expensive, but there are other reasons like the quality of the material itself that the dice are made of. Unlike most dice we gamers use the plastic from which casino dice are made of settles much better during the casting process and is of a purer quality. Add the pips being of the same density, the completely smooth surfaces, and the measuring of all details to ensure quality (surface area, weight, angles, etc.) and that is why they cost so much.

My only reason for bringing this up is because a friend of mine said “GS dice are like casino dice!”

Far from it. I really like GS dice, but if you buy even used casino dice (only so many hours/rolls before the casino has to remove them and discard them from play forever)you will get better dice. Of course, you’ll only get D6s. 🙂

#11 Comment By Patrick Benson On March 28, 2012 @ 1:05 pm

[20] – I was able to recreate the stacks by creating a sleeve that I stacked the dice within and then removed. It was a tedious process, but it worked. My Chessex dice stacked just fine side-by-side meaning that they were stacks of the same height.

I agree that the photo is far from being proof that the dice from GS are better. For all we know they might have acquired rejected dice for that photo. I’m not saying they did, but just that we should be skeptics and ask for verifiable evidence.

#12 Comment By randite On March 28, 2012 @ 1:22 pm

I absolutely adore my Gamescience dice. The precise edges feel good in my meaty paw; kind of like the hard, unforgiving edge of a calculated risk…
All of mine are opaque and I had no trouble inking them with an xtra-fine tipped sharpie. Even that place where I inked outside of the 5 spot, but that’s okay. It gives that d6 character… no really, it does… or so I try to convince myself.
My wife got a jewel set that she still hasn’t inked, and seeing the way that crayon just popped; I’m definitely gonna recommend that method to her.
I’m a huge proponent of Gamescience’s precision manufacturing and enjoy Zocchi’s well-earned pride in his product. While Chessex may look snazzier, nothing matches visceral pleasure of rolling a Gamescience die.

#13 Comment By Roxysteve On March 28, 2012 @ 2:00 pm

[21] – I’d just like to say that after seeing the demo I repeated the stacking experiment and reproduced similar results to those shown in the video. The oblate-ness of the D20s was not predictable, but it could be found to be there on pretty much any selection I tried.

This did not make me change to using only Gamescience dice, since I love the special FX Chessex can provide. I do, however, rotate my D20s over a working set of never less than four and in a busy game of Delta Green vs Cultists more than thirty in order to randomize any unrandomness.

As for the BS factor, the only way to prove one way or the other that the claims for GS dice over tumbled brands would be a Mythbusters-style roll test.

#14 Comment By Kurt “Telas” Schneider On March 28, 2012 @ 2:06 pm

I don’t have a link, but I’m pretty sure Col. Zocchi took a pair of calipers and tested dozens upon dozens of dice to get those uneven columns of dice.

I also haven’t seen any evidence that the ‘out of round’ dice significantly affect the randomness of them. Only a thousands-of-rolls test would tell for sure.

#15 Comment By Kristian Serrano On March 28, 2012 @ 2:14 pm

[22] – I’ve also noticed that those stacks of dice tend to have the non-GS dice leaning against on another, one column supporting another, which can create uneven columns, yet the GS stacks don’t lean at all.

I also don’t buy his claims of his dice being more random because they don’t roll as much. There is a countless number of environmental and conditional factors that influence a die roll. How the dice are held, how they are released, the friction on a surface, the deflection off of other objects, how fast the die spins in the air, which side was facing up in the cup or hand, how long it takes for the die to fall, and so much more. Yet at the same time, he doesn’t acknowledge whether the break point from the molding influences the die roll.

I’ve yet to see any adverse effects to my games from using Chessex, Koplow, or Q-Workshop dice. My only reason for considering GS dice is simply because of how the translucent versions look. 😛

#16 Comment By Patrick Benson On March 28, 2012 @ 2:22 pm

[23] – I’m one of those atheist skeptic types. We’re a pesky bunch. So what you are saying is true, that only a roll of thousands of dice taken from random samples would prove or disprove the validity of the Colonel’s claims.

But the burden of proof belongs to the person making the claims. So unless the Colonel has actually conducted an experiment that was found to be sound when peer reviewed and verified he should not be making such claims.

Plus, I’m not trying to sell anyone dice. He is. He might be a nice guy. His product might be superior. But if you are going to make claims about a product to your customers have the evidence to back it up, and be ready to have it scrutinized by others.

And with my limited and personal testing of GS versus other brands of dice they are not anymore or less likely to roll an even distribution from what I could tell.

I encourage people to be just as skeptical of my claims as I am asking them to be of the Colonel’s. Don’t take my word for it. Challenge my claims and his by doing your own research and testing. 🙂

#17 Comment By Patrick Benson On March 28, 2012 @ 2:27 pm

[24] – Same here. I buy the GS dice because of the look, feel, and memories that they bring back. Not because of the claims of superior quality.

And to anyone who thinks I am bashing these dice I’m not. I’m bashing the claims that are made about the dice. I really do like the dice, but for different reasons. Why else would I buy them?

#18 Comment By Martin Ralya On March 28, 2012 @ 7:57 pm

I think it’s kind of neat that dice in general, and GameScience dice in particular, tend to invoke memories, make gamers feel nostalgic or happy or something else, and generally play a bigger role than one might think.

I remember the GameScience ads with the stacks of shoddy competitor dice, and I never bought that argument even as a kid. My personal experience with other dice was that they were just fine.

I guess I’ll find out for sure when these particular dice hit the table. If they feel more random than my other dice, I’ll be pleasantly surprised.

#19 Comment By Roxysteve On March 29, 2012 @ 8:15 am

[22] – Your weasel word there is “significantly”.

Sausage-shaped d20s will tend to show scores clustered on the length of the sausage rather than those the ends. No one score in that set will be favored, but ten of the available scores will come up less than the other ten and will do so repeatably, and the tendency will be exacerbated by the rounded edges.

But the point about voids in the GS dice that Patrick made is a valid one not addressed by Zocchi in his videos. Full disclosure would have been more honest.

And I agree that the Zocchi videos are first and foremost a sales pitch.

#20 Comment By BryanB On March 29, 2012 @ 9:45 am

I bought a set of GameScience dice at GenCon 2010 just to have the experience of owning “precision crafted gaming dice.” Sure, they are overpriced, but I’d read about them in Dragon magazine for ages.

The salesman was so arrogant, condescending, and rude that I was barely able to get myself to complete the sale. I kind of regretted buying them afterward. It’s nice that they think they have the greatest gaming dice in the world, but they don’t need to beat people’s heads into the ground with a smug sales pitch and swarmy attitude.

So I own a set of their dice. I considered it part of my GenCon 2010 trek to buy a set of GamesScience dice. Mission accomplished. No need to do so again. Chessex for the win. 🙂

#21 Comment By Alan De Smet On March 29, 2012 @ 10:06 am

[25] – I’m guessing you were dealing with Col Zocchi himself. I found him endearing with his mania for his dice, but I could see him rubbing people wrong. These days someone else is selling GameScience at Gen Con, and they’re just as businesslike as any of the other dice sellers.

#22 Comment By Matthew J. Neagley On March 29, 2012 @ 11:34 am

I saw several people mention that to check your dice would require “thousands of rolls” but in actuality due to the relatively small variance of the uniform distribution, it’s not all that bad. To get a .5 margin of error in a 90% confidence interval, you’d only need to roll a d4 14 times, a d6 32 times, a d8 57 times, a d10 90 times, a d12 129 times or a d20 360 times, which isn’t so bad, at least compared to “thousands”.

So your die’s “average roll” is (with 90% confidence) between (the average roll on the number of rolls given above for that die type)-1/2 and the average roll+1/2.

In case you want to find the required n for a non-standard die, or different confidence levels or margin of errors, it’s n=Z^2*(1/E)^2*(b+1)*(b-1) where b is the highest number on your die, E is your desired margin of error and Z is the Z-score associated with your confidence level. Common Zs are 1.645 for a 90%CI, 1.96 for a 95%CI and 2.576 for a 99% CI

#23 Comment By Alan De Smet On March 29, 2012 @ 12:47 pm

[26] – THANK YOU. I’ve been looking for exactly this sort information for the past few days. I can find lots of handwavy answers, but very little concrete and backed up with numbers and equations. A few questions if you don’t mind:

If I’m interested in looking further into this, can you recommend a book/web page/whatever with more information. I’m hoping for something more focused than “get an BS in stats”. 🙂

Also, where is this confidence applicable? I see you’re discussing it in the context of the average roll. Is it, or something similar, meaningful in the context of doing, say, a chi-squared test?

#24 Comment By Matthew J. Neagley On March 29, 2012 @ 2:19 pm

[27] – For most questions, a BS in stat would be overkill. A 100 level class or two would be more than sufficient. :p

I’m sure there are web sites that go over stats, but I unfortunately don’t know of any. However, many intro stat courses have easily understood books, back editions of which go relatively cheaply. My school uses a version of Elementary Statistics by Mario Triola and I snagged an old edition because I thought it was particularly easy to read and had lots of real-world examples and useful tests. You can get one on amazon for as low as a few bucks: [28]
Alternately if you know the right professors you may be able to get one free. They get comp copies of many major books in an attempt to woo them into using them so they always have stacks of old textbooks that have never been nor are likely to be opened.

The confidence interval comes from the context of the normal distribution (the same bell curve that you hear so much about). Using something called the central limit theorem, it assumes that the average result of die rolls is normally distributed even if the individual die rolls aren’t. Then it uses a known property of these distributions where no matter the specifics, moving away from the mean by a certain number of standard deviations (which aren’t but can be thought of as the average amount a given result varies from the mean) includes a specific amount of data. So a 90% confidence interval essentially (kinda) says that “If what we assume about this die’s average roll and variance is true and we did this test 100 times, 90 out of 100 times, we’d get a result between these two numbers.”

I hope that helps some. It’s tough to discuss Stat without getting bogged down in jargon and mumbo-jumbo which I know is of not use to most people.

You may also want to check out my two prior articles on probability here on the stew, which may or may not be something that interests you.

#25 Comment By Martin Ralya On March 29, 2012 @ 2:30 pm

[29] – I personally would be interested in seeing an article about quickly testing whether your dice are random that uses as few technical terms as possible (i.e., I didn’t understand a lot of what you said in this thread, even though I’m sure it’s very clear to less-dense people).

Like a little guide: Getting screwed by your dice? Follow these X steps and you’ll have a pretty good idea if they’re biased or the universe just hates you.

#26 Comment By Kurt “Telas” Schneider On March 29, 2012 @ 2:41 pm

[30]Significantly is not a weasel-word; it’s the crux of the argument.

Yes, an out-of-round die (or, more likely, one with inconsistently rounded edges) will produce a less-than-perfectly-random array of numbers. But the question is, how much less than ‘perfectly random’ will it be? Would it matter than your chance of rolling a natural 20 with a ‘flat die’ is 4.999% instead of 5.000%? Or is it more like 4.000%? Without testing, we honestly don’t know.

My guess at ‘thousands of rolls’ would be to determine if the average GS die is more consistent than the average non-GS die. Matt may still show that it would take less, but I was assuming a significant sample size of each. Otherwise, we’re just talking about two individual dice. 🙂

#27 Comment By MonsterMike On March 29, 2012 @ 3:28 pm

I have seven complete sets of Chessex dice. After yet another horrible night of rolling single digits all night, I decided to test them (the d20’s anyway).

I did 1000 rolls on a vinyl chessboard on a hard table. I rolled all 7 d20’s at once to save time and tallied the results in excel. Tracked results separately for each die, since they are distinguishable by color.

(I didn’t do this all at once. I did about 200 rolls a day for several days. 200 rolls took maybe 30 minutes to roll and record.)

Some dice did appear to favor some numbers over others, but a chi-squared (statistical) test said the amount of variability was not enough to disprove the hypothesis of a true flat random distribution (i.e. each number equally likely). It doesn’t prove the dice are fair (this can’t be done) but fails to prove that they are unfair within an acceptable margin of error.

If I can’t find a bias in the dice after 1000 rolls, any bias that would take 10,000 rolls to discover is small enough to not bother me. How many times do you roll a d20 in a night? Two or three dozen?

In the end, I decided that my dice are close enough to truly random, and that my d20’s are simply controlled by the Elder Gods whenever I play a d20 game.

And an aside: I started gaming in 1980 and left the hobby in 1986. Clear polyhedral dice with sharp edges were all you could get back then and you had to ink them yourself (I used blue and red crayon.) When I started gaming again in 2010, my fellow (and much younger) gamers reacted with stunned silence when those old-school gems hit the table. (Whisper: “My God! It’s like he’s from the stone age or something!”)

#28 Comment By BryanB On March 29, 2012 @ 3:56 pm

[31] – Actually it was a much younger guy not the guy with the dice hat. Don’t remember his name but he must have gone to the school of poor customer service and shameless marketing practices. 🙂

#29 Comment By BryanB On March 29, 2012 @ 3:57 pm

[32] – It is indeed the Elder Gods of d20. I too have suffered their wrath and have cried myself to sleep wondering what I had done to earn such a horrible stretch of luck. 😀

#30 Comment By tenkar On March 29, 2012 @ 8:25 pm

I got my set a month or so ago at one of the Games That Can Not Be Named game nights – I actually one a set. Chose a dark gray / black set.

I was offered the use of a white inking / paint pen of sorts. What a PITA to use, and very easy to over paint. I had to do some major buffing / rubbing to get off the excess.

I really should have gone the White Crayon Method. It’s what I had always used in the past for my un-inked dice. Besides, crayon is definitely the Old School way to color in your dice numbers 🙂

#31 Comment By Patrick Benson On March 30, 2012 @ 11:07 pm

[26] – While you are correct that 360 rolls would test a single die, it isn’t the right test. We aren’t looking to test one or two dice, we’re looking to test the quality of a product line. So you have to test a certain percentage of dice from both brands. Even a sample of much less than 1% from each manufacturer would probably be large enough that when you were done rolling each die 360 times you would definitely be in the thousands for the actual number of rolls made for a study of any real merit comparing various manufacturers.

I mean, I have 7 sets of GS dice and I have no idea how many Chessex dice. If I just rolled 7 d20s from each manufacturer 360 times that would result in 5,040 rolls and it would not be a large enough sample to actually speak to the quality of either manufacturer.

#32 Comment By Matthew J. Neagley On March 31, 2012 @ 10:04 am

[33] – Needing to test a percentage of a whole is a common misconception about sampling. In fact, as a population grows, the sample required grows, but quickly reaches an asymptotic limit, so it can’t be expressed as a “percentage of the population”. Rather, the variance of the population is more important, and since dice are (assumed) uniform distributions, they have relatively small variances (b+1)(b-1)/12 and so require relatively small sample. I’m not 100% sure that’s what you were saying, but it’s a common enough misconception that I wanted to point it out.

As for what we’re testing, I thought we were discussing testing a single die, but I agree 100% that if you’re trying to ascertain the quality of a line of dice that it would require more than a few hundred rolls. You wouldn’t use the same kind of tests though. Due to the assumed symmetrical nature of the error of dice, you couldn’t use a basic confidence interval, as the error from the low rolling die would cancel that from the high rollers. We’d probably be more interested in a test of the variance of the average rolls of dice. I’d say the main problem with testing a LINE of dice is familiarity with the industry and production processes and sampling error. For example, if Chessex has three production facilities, and we just walk into our FLGS and grab 20 of their dice, we’re probably only getting a sample from one of the three and we have no reason to believe them to be identical (other than they fact they’re trying to be identical, but they’re trying to be perfectly uniform too, and if we don’t believe that, why would we believe they’re identical across facilities?)

To sum up: yeah, I assumed we were talking about your favorite d20. testing all your dice or a brand of dice may well require 1000s of rolls.

#33 Comment By Patrick Benson On March 31, 2012 @ 12:24 pm

[34] – My point, which you agreed to, is that you do need thousands of rolls of dice in order to test a product line with based on how many dice each manufacturer has produced. I was just providing an example of how even a test of a sample that could not be considered conclusive would result in thousands of rolls.

#34 Comment By Roxysteve On April 2, 2012 @ 10:11 am

[35] – The good colonel’s point,( the crux of his video argument as it were) was not that the rounded corners were the problem causing non random results, but that the abrasion-tumbling process would result in oblate dice almost as a certainty (once it starts, and it starts almost immediately, the dice will tend to take up the same orientation and so exaggerate the rolling-motion in the “sand” over a tumbling one), and those oblate dice will favor ten of the scores over the the rest.Sharp edges work to mitigate lack of randomicity in out-of-balance dice (such as those that have voids in them), which will tend to come to rest with the center of gravity at the lowest position. In an homogenous oblate die they don’t have any effect other than to continue the rolling motion on the table, often until the die is on the floor. 8o)

#35 Comment By zegerman On August 22, 2012 @ 7:42 am

Quick, somebody get this guy some Game Science and Chessex dice! http://gamesbyemail.com/News/DiceOMatic

1.3 million rolls a day should do nicely.

#36 Comment By Blysse On March 3, 2013 @ 5:50 am

For those interested in the stats/deviation side: [36]
My only remaining questions about gs dice:
1. pre-ink- does the different amount of plastic gouged out/missing of the face cause the dice to become weighted- ie: the 20 face is missing more plastic than the 1 face due to having approx. 3x the grooves in the face surface to create the number 20 than the number 1.
2. If there’s any weight distribution difference due to the amount of groove caused by the numbering, does it affect the randomness of the rolls?
3. If there is an effect from the weight distribution from numbering grooves, what ink/crayon/paint best mimics the density/weight of the material the dice are made up of to counter this?

#37 Comment By Matthew J. Neagley On March 3, 2013 @ 8:45 am

Regarding the material missing from the numbers, this article may be helpful to you:

They say that due to some physics stuff that I’m not really up on that less material makes that side LESS likely to roll not more and that filling them in with plastic of a different color is the way to go.

#38 Comment By Blysse On March 3, 2013 @ 9:54 pm

Holy crap those findings are completely counterintuitive- the heaviest sides roll more- so really I need to fill in gamescience die with something heavier/denser than the plastic they’re made of- which I suppose crayon is since the wax is definitely dense.
This “should” result in more 10+ rolls on a d20 :p

#39 Comment By Matthew J. Neagley On March 4, 2013 @ 11:32 am

What you can do is mass a crayon and mass your dice and then find their volume (by displacement of water in a graduated cylinder is likely easiest) then density = mass/volume. So you can ideally choose the crayons with density closest to your dice. (or the heaviest you can find you big cheater!)