I read that link and it's true, you could easily reduce precision, and yes, you COULD reduce the randomness, but they are mutually exclusive. If I wanted to turn a d6 with a particular level of randomness into a coin flip (d2) by doing modular arithmetic, you don't automatically increase the predictability of that d2 merely because of the math you're performing.

I got so furious with nerd rage, I didn't even read the last line of your post! :-)

It seems there is a fair amount of waste by the buckets not being full. When the machine started up, the buckets were all packed with dice. Only after they started cycling did you see the gaps. This tells me you could put many more dice in the hopper to ensure a full bucket each trip. If you were okay with the current rate of rolls, you could slow the track down, thereby extending the life of your dice. So long as each bucket was full, you would get the same number of total rolls per day.

This is silly. If you have truly random d6s, applying math against their outcome doesn't make them any less random unless you divide the roll by itself.

Please don't use this method. You are not creating an evenly distributed d8. From all 36 possible outcomes of 2d6, your "mod" d8 comes out like this: 1 - 5 times; 2 thru 6 - 4 times each; 7 - 5 times; 8 - 6 times. You COULD use some kind of conversion table, but even then you would need to throw away 4 of the 36 combinations to make it work.

ME: PISS OFF!

ELIZA: Would it make you happy for me to PISS OFF!?

ME: No.

ELIZA: Don't be so negative.