8-sided Dice as an Educational Device at IBM

The bill was addressed to R. W. Bemer, c/o I.B.M. Corp., 112 East Post Road, White Plains, New York (no ZipCode). That was where we were working on a new 8-bit character set for various IBM computers, and it looked at that time a lot like ASCII does now.

I had had two sets made a couple of years earlier, just to promote the idea of 1) octal numbering when IBM computers used a 6-bit character and 2) the fact that numbering for computers had to start with 0 -- a concept unpopular with the public at the time. Each die was a true octahedron, and each face was engraved with both a decimal number 0 to 7 and its corresponding 3-bit value. I confess that I could not figure how to get a 16-sided die for the 8-bit characters without breaking the laws of nature. Hexadecimal was out.

At that time I figured the odds as though used in a game like craps. For me, at least, that was much more difficult than the computation for 6-sided, so I abandoned thoughts of remaking Las Vegas.

Moreover, an octahedron does not roll as well as a cube, so in my 1960 order I specified:

"I understand that the actual die will not be an octahedron like the sample but rather a sphere of approximately 2/3" diameter with eight flattened faces."

I ask for history because all of these seemed to have to do with RPG (Role Playing Games), like "Dungeons and Dragons". And I know they did not show until after PCs did. I'm talking 1958-1960!

And who has those original 12 sets, I wonder? I still have my original set of the plain octahedron type.

On Feb 14 of 2002 I finally got meaningful comment to the above from a John Bwah . He has permitted his verbiage as an appendage to this piece.

===================================================================

I was reading one of your stories (immensely entertaining to those who understand the significance of your contributions), and came upon the bit about dice, Dungeons and Dragons, et cetera.

A sixteen-sided die, and in fact any even face-count die under something around thirty faces, can be made as follows:

1. Let N equal the target number of faces
2. Create a cone.
3. Slice pieces off of the cone, such that it becomes a regular period whose bottom face has N/2 faces (that is, if you want a 10-sider, make a pyramid with a pentagon bottom).
4. Create a second, identical cone.
5. Connect the two cones on their bottoms.

Improvements can be made by staggering the two cones by rotation, such that the midpoint of a line on the bottom face of Cone A is lined up with the endpoint of a line on cone B. A bit of tapering needs to be done; rather than triangles, the faces become extremely distorted diamonds (rhomboid schmomboid). Etc.

It doesn't work past thirty unless you make your die unusually large, as it has a nasty tendency to roll away, and it becomes difficult to get the thing to pick a random cone. But for reasonable sizes, such as the 16, it works well.

Effectively, by the way, the D&D style 8-sider popularized by 2nd ed. basic D&D follows such a pattern. Note that it does not have the rotation described above, but those are two regular pyramids with square bases, fused.

Also, if you want a seemingly impossible die size - like a five or a seven - the thing to do is to make a 10 (or a 14) and have two 1s, two 2s, and so forth.

Your new challenge, should you choose to think about dice ever again, is to find a more traditional format for 30-sided dice - that is to say, not using the above trick. It's possible, and I own one (they're a buck fifty at my local gaming store). If you want a photo, I'll mail ya one.

(A note: there is another way: you can bisect the faces of an eight sider, and bend the bisector line, such that it's a convex but shallow arc; then, round the remaining faces, so that it must pick one or another; lastly, fill the form, a clear plastic, with an opaque or dark liquid, and introduce an air bubble, so you know which face on the far side is used. But that's uglier than the Windows API.)