Dice Testing

JimTullis · 11-17-2020, 10:18 AM

After some unfortunate rolls in X-Wing I decided to check my dice for hidden air bubbles.

I took a 20 oz wide-mouth canning jar and filled it with water, then I microwaved it and stirred in Epsom Salt until a die would float in the jar.

I dropped each die in the jar and used a grease pencil to mark the side that it surfaced on. I did that 10 times per die and looked for dice that had trends. I also rolled each die 10 times using the same process.

2 of my defense dice surfaced on the same blank side every drop and almost never landed on that side . My guess is that they had an air bubble close to that side.

Another defense die and 2 attack dice tended to surface on adjacent sides. They rolled a bit more randomly than the 2 suspect defense dice. These also seemed to have something off kilter.

None of the rest seemed to favor any particular side.

I sequestered all the suspect dice and felt that my dice rolls improved.

I realize that this was not a truly scientific test, but I felt that it was good enough for finding the really bad dice.

I plan on doing the same thing with the Car Wars dice when they come in.

I had a couple of people who thought testing the dice was silly, and a couple who questioned the ethics of "only rolling the good dice". I'm interested in what Car Wars players, MIBS, and the powers that be think of this.

43Supporter · 11-17-2020, 02:53 PM

Quote:

Originally Posted by JimTullis

I had a couple of people who thought testing the dice was silly, and a couple who questioned the ethics of "only rolling the good dice". I'm interested in what Car Wars players, MIBS, and the powers that be think of this.

Well, I've seen the dice-testing protocols for Vegas; while this isn't that level, it's still useful to know if a die is badly biased, and needs to be done away with.

woodchuck · 11-17-2020, 03:41 PM

Testing your dice to see if they're biased is only cheating if you're exclusively using dice that roll in your favor or lending dice that roll unfavorably to your opponents.

baakyocalder · 11-17-2020, 05:15 PM

Given that dice manufacturing is imperfect, testing your dice is a reasonable proposition.

10 attempts per die is going to give you a better test than a few rolls, but I'd probably do more to be scientific--20 tests per die starts getting you into a pretty good confidence level. If the method of testing is consistent, with more trials you get closer and closer to ruling out non-chance factors in how the dice roll.

As long as one is transparent about their dice testing, using those dice in games should not be a problem. Gamers can be suspicious and superstitious--in my current online game using Discord the players often make 'sacrifices' to the dicebot. For physical dice, well if they are fair and roll true, I'd rather play with those than dice I'm not sure are fair.

Anthony · 11-17-2020, 05:35 PM

On 10 rolls, you aren't really statistically significant (>95%) if you get at least 6 duplicates (the odds of 5 of any given value are only 1.3%, but there odds of five of any value at all are 7.8%). It sounds like you were over statistically significant for the dice you segregated, though.

JimTullis · 11-17-2020, 08:31 PM

Quote:

Originally Posted by woodchuck

Testing your dice to see if they're biased is only cheating if you're exclusively using dice that roll in your favor or lending dice that roll unfavorably to your opponents.

The dice I sequestered are stored separately from the rest and I always inform my opponents. Opponents are welcome to borrow the dice I'm using, but I think Karma would get me if I loaned out the bad ones.

Quote:

Originally Posted by Anthony

On 10 rolls, you aren't really statistically significant (>95%) if you get at least 6 duplicates

X-Wing dice are 8 sided so that changes the statistics as well. BUT, the 10 rolls were in addition to the Salt Water Testing. Salt Water Testing is not nearly as random as rolling. It's still not perfectly scientific, but all of the dice I sequestered either floated to the same side or a cluster of the same adjacent sides every time. In my mind that was enough to make them suspect.

If I had one of the dice ballancing calipers I'd try that. I suspect that the 2 defense dice would still fail.

11-17-2020, 10:18 AM	#1
JimTullis Join Date: Feb 2013	Dice Testing After some unfortunate rolls in X-Wing I decided to check my dice for hidden air bubbles. I took a 20 oz wide-mouth canning jar and filled it with water, then I microwaved it and stirred in Epsom Salt until a die would float in the jar. I dropped each die in the jar and used a grease pencil to mark the side that it surfaced on. I did that 10 times per die and looked for dice that had trends. I also rolled each die 10 times using the same process. 2 of my defense dice surfaced on the same blank side every drop and almost never landed on that side . My guess is that they had an air bubble close to that side. Another defense die and 2 attack dice tended to surface on adjacent sides. They rolled a bit more randomly than the 2 suspect defense dice. These also seemed to have something off kilter. None of the rest seemed to favor any particular side. I sequestered all the suspect dice and felt that my dice rolls improved. I realize that this was not a truly scientific test, but I felt that it was good enough for finding the really bad dice. I plan on doing the same thing with the Car Wars dice when they come in. I had a couple of people who thought testing the dice was silly, and a couple who questioned the ethics of "only rolling the good dice". I'm interested in what Car Wars players, MIBS, and the powers that be think of this.

11-17-2020, 03:41 PM	#3
woodchuck Join Date: Dec 2007	Re: Dice Testing Testing your dice to see if they're biased is only cheating if you're exclusively using dice that roll in your favor or lending dice that roll unfavorably to your opponents. __________________ Sapor similis pullo.

11-17-2020, 05:15 PM	#4
baakyocalder Join Date: Jan 2009 Location: Sacramento metro, California	Re: Dice Testing Given that dice manufacturing is imperfect, testing your dice is a reasonable proposition. 10 attempts per die is going to give you a better test than a few rolls, but I'd probably do more to be scientific--20 tests per die starts getting you into a pretty good confidence level. If the method of testing is consistent, with more trials you get closer and closer to ruling out non-chance factors in how the dice roll. As long as one is transparent about their dice testing, using those dice in games should not be a problem. Gamers can be suspicious and superstitious--in my current online game using Discord the players often make 'sacrifices' to the dicebot. For physical dice, well if they are fair and roll true, I'd rather play with those than dice I'm not sure are fair. __________________ Currently Running: Without Number family games which use a lot of GURPS material for details when the players start asking(online, sporadically) Waiting For: Schedule Sanity to Play Car Wars and my Fnordcon special alt Car Wars cards!

11-17-2020, 05:35 PM	#5
Anthony Join Date: Feb 2005 Location: Berkeley, CA	Re: Dice Testing On 10 rolls, you aren't really statistically significant (>95%) if you get at least 6 duplicates (the odds of 5 of any given value are only 1.3%, but there odds of five of any value at all are 7.8%). It sounds like you were over statistically significant for the dice you segregated, though. __________________ My GURPS site and Blog.