PDA

View Full Version : I never knew that about polyhedral dice . . .



LurkerInPlayground
2009-02-15, 12:39 PM
Louis Zocchi explains dice-manufacturing and explains why your polyhedral 20 might be loaded against you.
http://www.youtube.com/watch?v=bR2fxoNHIuU&eurl=http://grognardia.blogspot.com/2008_11_01_archive.html

Zaggab
2009-02-15, 01:50 PM
Huh. Interesting. Might explain some dice-related superstition.

LurkerInPlayground
2009-02-15, 02:20 PM
Pretty much. I actually took out my dice and examined them.

You see, I have this glow-in-the-dark d20. I looked at one of the facets and realized that it wasn't a perfect isosceles triangle. I didn't have to squint or anything, the that sucker was not radially symmetrical.

tcrudisi
2009-02-15, 02:27 PM
Thanks Lurker. Those were two very interesting videos that I definitely think every gamer should watch.

Jack_Simth
2009-02-15, 02:34 PM
Yep. Some dice are "lucky" - because they were accidentally loaded by the factory to a state where they'll roll high numbers more often that low numbers. Others are "unlucky" - because they were accidentally loaded by the factory to a state where they'll roll low numbers more often than high numbers (this is why casinos pay a LOT for their dice - they want them as random ("fair") as possible - a casino will, no joke, pay something like $20 for a pair of d6's - and they'll retire them after a night or two (sometimes they'll retire multiple sets at the same table over the course of a night), because the repeated impacts subtly change the shape of the dice).

If you're willing to put in the work for it, you can distinguish the two in a simple (but nontrivial) manner:
Roll the die a thousand times (using a cup), and track the results. If you get a disproportionate number of one or more numbers (with 1000 rolls, on a d20, each side should come up about fifty times, give or take a few - it won't be exactly 50, even on a fair die), chances are the die isn't fair. If you're evil, you can use that to your advantage. If you're good, you filter out the ones that are notably off average in either direction.

snoopy13a
2009-02-15, 03:03 PM
Also, crafting a fair D20 has to be significantly more difficult than crafting a fair D6.

I suppose the best way to keep things fair (other then the mind-numbing rolling the die a thousand times :smalltongue: ) is to use many different d20s and rotate through them throughout the game. In theory, this should balance out loaded dice. Additionally, if a player has "lucky dice" then there is a good chance that they are loaded towards higher numbers. Oddly, what was previously considered a superstition may actually have some merit to it.

Superglucose
2009-02-15, 03:34 PM
and they'll retire them after a night or two (sometimes they'll retire multiple sets at the same table over the course of a night), because the repeated impacts subtly change the shape of the dice).


I'm pretty sure dice are switched out every few hours, actually.

Wafflecart
2009-02-15, 04:28 PM
I'm pretty sure dice are switched out every few hours, actually.

You're correct, they are in most places...those videos make me want Casino Dice for trying out the Burning Wheel system, which I understand is entirely d-6 oriented.

Jack_Simth
2009-02-15, 04:38 PM
I suppose the best way to keep things fair (other then the mind-numbing rolling the die a thousand times :smalltongue: )You can actually do that while you're gaming - just keep a notepad of graph paper and track each individual roll on that d20. When you hit a thousand rolls, count them up and judge the die.

Additionally, if your dice are easily differentiated, you can test multiple dice simultaneously by rolling them simultaneously.

You're correct, they are in most places...those videos make me want Casino Dice for trying out the Burning Wheel system, which I understand is entirely d-6 oriented.
Ah, cool. Every few hours, not every night - I suppose that tells you how often I hang out at casinos....

Neithan
2009-02-15, 04:39 PM
So, if you#re invited to a shadowrun game, you know where you'll get your dice. :D

Quincunx
2009-02-15, 05:22 PM
At the end of their gaming lifespan, discarded casino dice are drilled through to prevent re-use (on the 1/6 axis, if memory serves). This makes them somewhat useless as fair d6s, but very cool trinkets. My dad had worked in casinos and I think all he had to do was ask for the discarded dice, which is a marvel given how everything else around there needs to be accounted for in triplicate.

Nohwl
2009-02-15, 05:27 PM
so is the only fair way to roll dice is by using a computer program?

AslanCross
2009-02-15, 06:02 PM
so is the only fair way to roll dice is by using a computer program?

Theoretically, but I've seen some soft die rollers roll in a very bizarre fashion. I once used a computer die roller for some games I held in my classes, and they'd roll the same number over and over again. I think once I rolled five straight 1s, then six straight 13s, and some other weird way. Even with computer die rollers, not all are created equal.

Also programs should be executed in sight of everyone. I've had one of the guys I game with (who is also my student) attempt to cheat in another DM's game. :P In my upcoming game I've told him that he MUST roll with my physical dice and in my presence.

Greg
2009-02-15, 06:27 PM
"Fresh oats cost more than used oats even though the used oats only passed through the horse one time."

This guy is pretty cool. He basically tells you how to fix your dice if you're a horrible cheater.

BizzaroStormy
2009-02-15, 06:45 PM
All of the sudden I feel like stopping all my projects in my machining class and making a set of dice. Even with the High School's low funding I can make them accurate to within .00001" :)

LurkerInPlayground
2009-02-15, 08:04 PM
"Fresh oats cost more than used oats even though the used oats only passed through the horse one time."

This guy is pretty cool. He basically tells you how to fix your dice if you're a horrible cheater.
Shave off the edges of the high numbers of your d20 while retaining crisp edges on the low numbers?

Prometheus
2009-02-15, 09:22 PM
Most of the peculiarities that gamers claim about are not the inaccuracies that the dice have. You have to roll a lot of times to have dice actually deviate from what they should be to a meaningful degree, even if they are defective. By the time you roll the dice that many times, you will have already forgotten it's true inaccuracies and confirmation bias will actually dictate what you think is wrong the dice.

If don't believe me, flip a coin fifty times and see if you get 25 and 25 (you probably won't, although you know it is even). But if the coin was inaccurate, you can see why this kind of phenomenon would make it hard to cover up. The effect is even less noticeable in d20, because they have more faces over which the probability is split, even though they are harder to make.

Caeldrim
2009-02-15, 09:49 PM
If I understood Lou right, he's saying that dice that haven't been tampered with (shaved) and that seem to roll a lot of 20s, are actually just as likely to roll a lot of 1s, due to their being 'thinner' on the 20-1 axis.

This would explain why i had a d20 that always seemed to roll 17s and 3s. I cannot believe i didn't think of this earlier.

we play on a pool table, and i habitually rub my dice on the table with the 20 side up. I now realise that if anything, this is probably decreasing my chances of getting a 20 on any given die that's been polished in this manner.

DOH.

LurkerInPlayground
2009-02-15, 10:10 PM
If I understood Lou right, he's saying that dice that haven't been tampered with (shaved) and that seem to roll a lot of 20s, are actually just as likely to roll a lot of 1s, due to their being 'thinner' on the 20-1 axis.

This would explain why i had a d20 that always seemed to roll 17s and 3s. I cannot believe i didn't think of this earlier.

we play on a pool table, and i habitually rub my dice on the table with the 20 side up. I now realise that if anything, this is probably decreasing my chances of getting a 20 on any given die that's been polished in this manner.

DOH.


Keep in mind that even though the "egg" is flatter with respect to the 20-1 axis, the "egg" favors coming to rest on the one of two facets on that axis with the more surface and sharpness.

So while both the "20" and the "1" are more likely than the other probable outcomes due to being the "privileged" axis, the "20" could still come ahead with a higher likelihood of probability than the "1."

Also, this doesn't account for the fact that there are also multiple axis with their own damage, which also might make these other axis "runner ups."

Reinboom
2009-02-15, 10:58 PM
Theoretically, but I've seen some soft die rollers roll in a very bizarre fashion. I once used a computer die roller for some games I held in my classes, and they'd roll the same number over and over again. I think once I rolled five straight 1s, then six straight 13s, and some other weird way. Even with computer die rollers, not all are created equal.

Also programs should be executed in sight of everyone. I've had one of the guys I game with (who is also my student) attempt to cheat in another DM's game. :P In my upcoming game I've told him that he MUST roll with my physical dice and in my presence.

That's because it's pulling from a slow moving/nonmoving/repetitive point on the machine, such as a slower moving piece of ram.
It's generally a better idea to have programs that understand to change their random seed between higher ram references and to keep adjusting even that number with the system clock (to add human 'random' to it).

Dyllan
2009-02-15, 11:39 PM
That's because it's pulling from a slow moving/nonmoving/repetitive point on the machine, such as a slower moving piece of ram.
It's generally a better idea to have programs that understand to change their random seed between higher ram references and to keep adjusting even that number with the system clock (to add human 'random' to it).

Yes, but how many programmers would bother to make software anywhere close to that intelligent, in days when I've seen print drivers that require .NET?

herrhauptmann
2009-02-16, 12:31 AM
I've heard that for the dice software programs, they (or some at least) are programmed to achieve the mean value. Say mostly between 5 and 16 on a d20, with each number past that being more and more rare, so 20s and 1s are rarer than 2s and 19s. 2s and 19s rarer than 3s and 18s.
I tested it on a dice roller I downloaded somewhere for about 30 minutes, and I definately seemed to have a clustering in the middle of the range. But that was also about 8 years ago, they might have more fair ones these days.

Halaster
2009-02-16, 05:57 AM
At least it should be easy to test any given software die roller. Just run it a few hundred times, copy the results into Excel and run some functions over them (counting out each number, calculating the average, that kind of stuff). If the results roughly match statistical averages, you're fine. Otherwise, find another tool.

For your dice, I like Lou's method. Take them all in one hand, roll them and see which ones stop first. They have the crispest edges. His stacking method is nice too, but few players actually have this number of dice from the same manufacturer.

I have actually seen people use casino dice in tabletop wargames. However, since the gaming tables are rarely smooth, the dice probably lose some of the randomness there.

CU,
Halaster

arguskos
2009-02-16, 06:30 AM
Actually, I recently picked up some dice from Lou's company, and after running some tests against my other dice, I have to say, his do tend to be more random. I basically rolled sets of d20's (one was from Lou, the other 5 were mine) maybe 1500 times or so (took me a few hours) and voice recorded the results. As far as I can tell, his dice are in fact closer to proper distribution. Not completely sure, of course, since I'm just a layman, but I figured I'd toss that out there.

I don't understand the finer points of die-rolling theory, but the man seems to know his dice. :smallcool:

Starshade
2009-02-16, 06:43 AM
That's because it's pulling from a slow moving/nonmoving/repetitive point on the machine, such as a slower moving piece of ram.
It's generally a better idea to have programs that understand to change their random seed between higher ram references and to keep adjusting even that number with the system clock (to add human 'random' to it).

grab random junk from higher ram and use system clock to randomize it? Sounds good, but ive seen PC games with less sofisticated, or real "Random" random number effects.

Great idea though. I might knock together such a program if i feel the need, instead of using my tumbled and polished edge dice. :smallamused:

Rasilak
2009-02-16, 07:30 AM
For the software faction: Actually, one of the best ways to get random data out of not perfectly random input, is to split the source in to streams of bits, A and B. Then you check if A XOR B is true, and in this case take A and drop B (or vice versa, just do it consistently), else you just drop A and B.
Or, if you have a LINUX machine, just use dev/random. It's pretty good (IIRC it uses the above method on calculation errors by the CPU).

PinkysBrain
2009-02-16, 07:35 AM
Just run Mersenne Twister with date-time as a seed ... not cryptographically secure, but more than good enough.

Jack_Simth
2009-02-16, 08:13 AM
If you have access to an older computer, you can just use Assembler to grab the lowest bit off the processor clock. The books on the language note that the low-end bits are not reliable - which means they're effectively random. Combine with the XOR trick, and you should be good to go.

DigoDragon
2009-02-16, 10:36 AM
At the end of their gaming lifespan, discarded casino dice are drilled through to prevent re-use...

Really? That's odd since my brother gave me four casino d6s from his trip to Vegas a few years ago and they're not drilled. Now I wonder how my brother came into these dice. Heh heh... :smallsmile: Interesting!

memnarch
2009-02-16, 10:51 AM
A good source for randomness is Random.org (http://random.org/).

LurkerInPlayground
2009-02-16, 11:16 AM
Really? That's odd since my brother gave me four casino d6s from his trip to Vegas a few years ago and they're not drilled. Now I wonder how my brother came into these dice. Heh heh... :smallsmile: Interesting!
Not really. It wouldn't surprise me if they had some way of inventorying the dice they give away to players.

Prometheus
2009-02-16, 04:54 PM
Yes, but how many programmers would bother to make software anywhere close to that intelligent, in days when I've seen print drivers that require .NET?
It's actually not that hard to write. I thought that was standard practice when generating random numbers. Of course, it depends on how precise the system clock the language interface has access to is and how frequently a random number is required. Perhaps browser based ones aren't as effective, I dunno.

My favorite random number generator is digits of Pi. Of course, calculating digits of pi is actually pretty difficult so a computer would just reference a table. From there you can always use that as the seed or as a logical operator with the random output. Scramble enough and you are a-okay.

You'd think the real trump card is the quantum mechanical random number generators which work off comparing atomic decay to it's "known" decay rate. But the error would be proportional to the difference between the true decay rate to the known decay rate, with "rate" defined by the accuracy of the measurement system. Don't get me wrong, its good, but it can't be perfect.

PinkysBrain
2009-02-17, 10:19 AM
All the error would introduce is bias, de-biasing a binary random number generators is pretty easy.