Eh? Did you see those highs and lows? The statistical LOWS were akin to 1 in 4 (on a six-sided die). That's not merely a little off the expected, but significantly so. Also ... how in the world is the plastic from hollow pips and rounded corners on two dice going to be enough to actually make a third die? Those must be some pretty deep pips!
Still, it's pretty interesting. If the deviation is THAT outrageous, then it shouldn't be too hard to perform a test of my own (that is, roll dice for a predetermined large number of times, record the results, and see how it pans out).
True, that does seem like a pretty hefty claim regards getting enough plastic by shaving the dice by a third apiece!
What it makes me wonder though is, aren't most dice done the same way, i.e. gouged pips and rounded edges? If so, even if these particular dice are unusually deeply carved, most dice should exhibit the same flaws.
Clearly it's time to use the MUCK roll function for your tabletop gaming!
(though the MUCK pseudorandom number generator does have some issues of its own)
The ones that rolled the strangest turned out to have voids inside when cut open (with a hydrogen-cooled laboratory saw, to avoid melting the plastic while doing this). I seriously doubt the pips contributed significantly.
I'm a bit leery of the "we removed statistical anomalies" bit. They also didn't compute confidence intervals for their result, but it's been long enough since I've done a chi-squared calculation that I'm having trouble doing it too.
This is still a good argument for building a miniature x-ray tomography widget for testing dice, though };>.
It's probably more a case of equipment that *absolutely never* produces voids being expensive. Any entrained gas from melting/mixing the plastic will do it. At least, that's my best guess.
Still plotting kytteny tomography };>. The cheapest way to do it is probably to discharge a known amount of charge across a 10-20 kV spark gap (with a separation of a few mm or less, so that you strike a stable arc quickly), and put as big a CMOS image sensor as you can on the other side of the die. A CCD sensor will give you lower noise, but would be more expensive. Both will pick up x-ray photons just fine, and your gap is close to a point source if you use a needle as a cathode and have a rounded anode (so there's a tiny spot that's closest, rather than a wide area). The anode has to be fairly thin (1 mm or less if you're doing tomography of 1 cm dice). Add a not-too-imprecise way to reposition the die, and you're there.
![[identity profile]](https://www.dreamwidth.org/img/silk/identity/openid.png){kind=small}
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
timeasmymeasure
no subject
Date: 2010-06-21 06:52 pm (UTC)Still, it's pretty interesting. If the deviation is THAT outrageous, then it shouldn't be too hard to perform a test of my own (that is, roll dice for a predetermined large number of times, record the results, and see how it pans out).
no subject
Date: 2010-06-21 07:00 pm (UTC)What it makes me wonder though is, aren't most dice done the same way, i.e. gouged pips and rounded edges? If so, even if these particular dice are unusually deeply carved, most dice should exhibit the same flaws.
Clearly it's time to use the MUCK roll function for your tabletop gaming!
(though the MUCK pseudorandom number generator does have some issues of its own)
no subject
Date: 2010-06-22 03:05 am (UTC)I'm a bit leery of the "we removed statistical anomalies" bit. They also didn't compute confidence intervals for their result, but it's been long enough since I've done a chi-squared calculation that I'm having trouble doing it too.
This is still a good argument for building a miniature x-ray tomography widget for testing dice, though };>.
no subject
Date: 2010-06-22 06:22 am (UTC)no subject
Date: 2010-06-22 06:32 am (UTC)Still plotting kytteny tomography };>. The cheapest way to do it is probably to discharge a known amount of charge across a 10-20 kV spark gap (with a separation of a few mm or less, so that you strike a stable arc quickly), and put as big a CMOS image sensor as you can on the other side of the die. A CCD sensor will give you lower noise, but would be more expensive. Both will pick up x-ray photons just fine, and your gap is close to a point source if you use a needle as a cathode and have a rounded anode (so there's a tiny spot that's closest, rather than a wide area). The anode has to be fairly thin (1 mm or less if you're doing tomography of 1 cm dice). Add a not-too-imprecise way to reposition the die, and you're there.
no subject
Date: 2010-06-22 07:08 am (UTC)Actually, this gets me wondering about the quality control casinos impose on their dice suppliers? I understand they're pretty strict about cards...
no subject
Date: 2010-06-22 07:22 am (UTC)