craigmaloney, 1 year ago

@futurebird The examples I have of D100s suggest that would be rather difficult. They're essentially golf-balls that roll forever without a braking mechanism (the Zocchihedron has ball-bearings inside the die to help with this.)

craigmaloney, 1 year ago (edited 1 year ago)

@futurebird Though it appears you could put the middle numbers around the middle of the die and have them have a higher probability of them appearing.

This was apparently a problem with earlier Zoccihedron dice.

https://en.wikipedia.org/wiki/Zocchihedron

DoesntExist, 1 year ago

@futurebird

11d10 - 10

catselbow, 1 year ago

@futurebird
Rolling 12d6 gives you a pretty good normal distribution with a mean of 42 and a s.d. of 6.

For quick-n-dirty simulations I use the sum of 12 random numbers between 0 and 1, and subtract 6. That gives an approximately normal distribution centered at zero and with an s.d. of 1. It's easy to shift, or scale up the s.d.

KanaMauna, 1 year ago

@futurebird A lot of answers in the comments. But what application are we using it for?

alakest, 1 year ago

@futurebird There's a 120-sided die sold.

You could have every number represented and 20 to play with/add.

More effective might be to roll 2 100-siders and always choose the number closest to 50.5.

You might want to color the dies so one of them is decisive in cases where a 50 and a 51 are rolled.

riley, 1 year ago

@futurebird: I suspect it couldn't be done as a static mechanism, but if one were allowed to sneak a little bit of electronics and a couple of reaction wheels in there, it should be doable as a B.Sc. thesis project.

That having been said, consider the shape of a 5d20 distribution. It's not a perfect Gaussian, but it may be close enough for most games.

derwinmcgeary, 1 year ago

@futurebird hang on, would that be almost the same as 2D50?

futurebird, 1 year ago

@derwinmcgeary

I need to think about this. You can get a normal distribution (roughly) by rolling more dice depending on what the most common sum would be.

Would that be 25D4?

derwinmcgeary, 1 year ago

@futurebird I didn't think too much through the details of the curve - trivially you can't roll a 1 on 25D4, so maybe you need to offset it

futurebird, 1 year ago

@derwinmcgeary

There are D10s with 0 on them so maybe 10D10 is the best bet for a roughly normal distribution at 50... using "normal" dice

neia, 1 year ago

@futurebird @derwinmcgeary 11d10 gets you a range of 0-99 inclusive centered at 49.5. You need an odd number of possible values to center it at 50.

There aren't any common odd-sided dice, but you could do a d6 numbered 1-3, 1-3 to make a d3. Pretty sure I've seen that before.

7d3+4d20 gets you a range of 101 distinct values, but I'm not awake enough to check the distribution.