Binary Dice; d2, d4, d8 and d16 from coins

[Bohandas]

I imagine a lot of people use a coin toss for d2; that's what I've always assumed a d2 was meant to be.

Considering this, and considering how the d% is constructed I was wondering if anyone had ever thought to use coins for higher order dice. Personally the idea just hit me about a week ago but I think it's a good one. You have a penny be 1 or 2 and then have nickel be 2 or 0 and you've got a d4 for cheaper than a d6 (6 cents as opposed to 12.5 cents for a d6 if you buy them in a pack of 8 at Dollar Tree). Add a further dime (4 or 0) and you have a d8, and finally add a quarter and you have a d16 that's more accessible than any of the traditional die sizes.

I think using a bunch of coins as dice would also be good thematically in a fight with a dragon.

Anybody have any thoughts?

[Cluedrew]

Main issue is that flipping a bunch of coins takes longer than rolling some dice. Or it least the way I know to flip coins. As an extreme case there is a place in my system where I roll dice and then (effectively) map them to coin tosses. Because it is faster that way.

[Grod_The_Giant]

I feel like you'd get some really weird probability distributions that way.

[exelsisxax]

I imagine a lot of people use a coin toss for d2; that's what I've always assumed a d2 was meant to be.

Considering this, and considering how the d% is constructed I was wondering if anyone had ever thought to use coins for higher order dice. Personally the idea just hit me about a week ago but I think it's a good one. You have a penny be 1 or 2 and then have nickel be 2 or 0 and you've got a d4 for cheaper than a d6 (6 cents as opposed to 12.5 cents for a d6 if you buy them in a pack of 8 at Dollar Tree). Add a further dime (4 or 0) and you have a d8, and finally add a quarter and you have a d16 that's more accessible than any of the traditional die sizes.

I think using a bunch of coins as dice would also be good thematically in a fight with a dragon.

Anybody have any thoughts?

None of these replicate dice because the distributions are super weird.

[Jormengand]

None of these replicate dice because the distributions are super weird.
I feel like you'd get some really weird probability distributions that way.

You don't get weird distributions because the way he's suggesting you do it is analogous to the way you ordinarily do d%, not 2d10. You get the same distribution as a standard d4, d8 or d16.

The real thing, though, is that it's entirely pointless to do this.

EDIT: To be clear, this (http://anydice.com/program/fab6) is the d16 he's suggesting.

[exelsisxax]

You don't get weird distributions because the way he's suggesting you do it is analogous to the way you ordinarily do d%, not 2d10. You get the same distribution as a standard d4, d8 or d16.

The real thing, though, is that it's entirely pointless to do this.

EDIT: To be clear, this (http://anydice.com/program/fab6) is the d16 he's suggesting.

That works, but is certainly not what OP actually describes. Though i'm still confused as to why this would be better unless losing your randomization elements underneath furniture is important for the game.

[Bohandas]

You don't get weird distributions because the way he's suggesting you do it is analogous to the way you ordinarily do d%, not 2d10. You get the same distribution as a standard d4, d8 or d16.

The real thing, though, is that it's entirely pointless to do this.

EDIT: To be clear, this (http://anydice.com/program/fab6) is the d16 he's suggesting.

Precisely this.


Though i'm still confused as to why this would be better unless losing your randomization elements underneath furniture is important for the game.

It's good thematically for rolls in battles with dragons or mammon cultists. Also it allows use of a some die sizes that aren't commonly seen or available (d16 and also d32) which, while not particulatly useful in published adventures using core materials, could potentially be useful for homebrewing and random event tables. Also you can lose dice under the furniture too.

[Jay R]

I've done the decimal equivalent. I've rolled a d10,000,000.000 using these

http://www.eaieducation.com/images/products/506409_L.jpg

and these

http://www.eaieducation.com/images/products/506410_L.jpg

[Cluedrew]

To Jay R: For what?! Why did you need to roll a d10,000,000.000? The closest I have head of that was in the system that should not be named and I hope you weren't playing that.

[Knaight]

You don't get weird distributions because the way he's suggesting you do it is analogous to the way you ordinarily do d%, not 2d10. You get the same distribution as a standard d4, d8 or d16.

In this case it's almost exactly the same as the d100, d1000, and d10000, just in a different base.

[Terrorvein]

I've never considered it because I don't like coin flips personally. I'm rather clumsy and it's difficult enough for me to keep dice on the table let alone catching a coin.

If I have to use a d2, I tend to use another die (d6 or up) and simple call the odd numbers 1 and the even numbers 2. So I'd probably simulate 'binary dice' such as these with normal dice in much the same way.

That said it is an interesting idea.

[Jormengand]

That works, but is certainly not what OP actually describes. Though i'm still confused as to why this would be better unless losing your randomization elements underneath furniture is important for the game.

Yes, it is: he's envisioning using a penny as 1 or 2 (1d2), a nickel as 2 or 0 ((1d2-1)*2), a dime as 4 or 0 ((1d2-1)*4) and it's obvious from context that the quarter is meant to be 0 or 8 ((1d2-1)*8).


You have a penny be 1 or 2 and then have nickel be 2 or 0... Add a further dime (4 or 0) and you have a d8, and finally add a quarter and you have a d16 that's more accessible than any of the traditional die sizes.

[Lord Torath]

I'd avoid it, myself. What you save in money spent on dice, you lose in time spent deciphering your throws. And standard, no-frills dice aren't that expensive.

I usually roll a d6 when something calls for a d2, with an odd result being a "1", and an even result being a "2".

[warty goblin]

And even for a d16, it seems easier to do rejection sampling off of a d20.

[Cicciograna]

By any account, whatever the probability distribution, we all roll dice because dice are BEAUTIFUL. As items theirselves, they're beautiful.

[Cluedrew]

And even for a d16, it seems easier to do rejection sampling off of a d20.Is that the thing where you roll, assign and if you can't assign them evenly than try and create sub-ranges and roll on that? d6->d4 is roll the d6. On a 1-4 that is a result. On a 5 roll a d2 and use that result. On a 6 roll a d2, add to and use that result.

[FreddyNoNose]

I feel like you'd get some really weird probability distributions that way.

All it takes is a little practice to get the bias going.

BTW OP, you can roll dice for binary.

[Sorceress]

Wouldn't it be easier to just roll 1d20 and reroll if you roll higher than a 16?

[warty goblin]

Is that the thing where you roll, assign and if you can't assign them evenly than try and create sub-ranges and roll on that? d6->d4 is roll the d6. On a 1-4 that is a result. On a 5 roll a d2 and use that result. On a 6 roll a d2, add to and use that result.
Not quite. Rejection sampling is -


Wouldn't it be easier to just roll 1d20 and reroll if you roll higher than a 16?

- that.

Basically all you're doing is 'getting rid' of the extra outcomes, which rescales the remaining outcomes to have the correct probabilities. It can be very handy in some cases, i.e. I want to generate points uniformly in a sphere of diameter one. I could do a bunch of math, or I could generate pairs of independent random uniforms between 0 and 1, then ditch the ones that don't fall in the sphere. It's so easy it feels like cheating.

[Nifft]

Wouldn't it be easier to just roll 1d20 and reroll if you roll higher than a 16?

That works fine until you encounter a Dire Black Swan.

[FreddyNoNose]

Wouldn't it be easier to just roll 1d20 and reroll if you roll higher than a 16?

This is what people have done for years. Need a d5, roll a d6 until a non-6 comes up. Trivial non-issue.

[Jormengand]

This is what people have done for years. Need a d5, roll a d6 until a non-6 comes up. Trivial non-issue.

If you want a d5, you can also use a d10 the way you use a d6 for a d3.

[Mystral]

I imagine a lot of people use a coin toss for d2; that's what I've always assumed a d2 was meant to be.

Considering this, and considering how the d% is constructed I was wondering if anyone had ever thought to use coins for higher order dice. Personally the idea just hit me about a week ago but I think it's a good one. You have a penny be 1 or 2 and then have nickel be 2 or 0 and you've got a d4 for cheaper than a d6 (6 cents as opposed to 12.5 cents for a d6 if you buy them in a pack of 8 at Dollar Tree). Add a further dime (4 or 0) and you have a d8, and finally add a quarter and you have a d16 that's more accessible than any of the traditional die sizes.

I think using a bunch of coins as dice would also be good thematically in a fight with a dragon.

Anybody have any thoughts?
The proper way to do this is to assign each coin to a binary number, have heads be 1 and tails be 0 and add 1 to the final result.

[FreddyNoNose]

If you want a d5, you can also use a d10 the way you use a d6 for a d3.

No **** sherlock. Stupid critical reply by you.

[Cluedrew]

That works fine until you encounter a Dire Black Swan.If you roll a 17-20, take your result, subtract 17 (0-3) multiply by 4 (0, 4, 8, 12) and add a d4 roll. If you are also missing a d4 roll another d20, 1-16 use that result, otherwise subtract 16 (1-4) and add it to the four times value from last time. The math is probably slower than re-rolling, but you will always finish in 2 rolls or less.

[kyoryu]

If you have enough different coins, you can do binary, which will be an even distribution.

d2 is just '2 if heads, 1 if tails'. Easy enough.

d4 - the first coin is 2 if heads, 0 if tails. The second is 1 if heads. Add together, add 1 to get 1-4.

d8 - same, but add another coin that's 4 if heads, 0 if tails.

d16 - same, but add another that's 8 if heads, 0 if tails.\

Basically, random binary numbers with the proper length. It'll turn out properly. I'm a programmer, I deal with binary all the time.

[RandomLunatic]

It'll work. The question is, why?

You are more limited in the number of ranges you can generate. There's 6 common die sizes. The sixth power of 2 already takes you up to 64, and I don't see much call for a d64. Your next size down is the d32, which also doesn't seem too handy.

I'd expect it to slow play down too. Even when rolling a lot of dice at once, you can usually grab and roll a good chunk of them at once. Flipping coins is a one-at-a-time thing, and you're going to be doing this multiple times per roll. I'm guessing most people won't count in base 2 as fast as base 10 either.

Finally, there's the human factor. In my experience, there's always that guy at the table who struggles with math in base 10. And you propose to teach him about binary numbers? *Shudder*

[Knaight]

Finally, there's the human factor. In my experience, there's always that guy at the table who struggles with math in base 10. And you propose to teach him about binary numbers? *Shudder*

The human factor seems like the reason to do this - some people really like binary (and/or hexadecimal), and there are certain games where this sort of aesthetic could be fun.

[Nifft]

If you roll a 17-20, take your result, subtract 17 (0-3) multiply by 4 (0, 4, 8, 12) and add a d4 roll. If you are also missing a d4 roll another d20, 1-16 use that result, otherwise subtract 16 (1-4) and add it to the four times value from last time. The math is probably slower than re-rolling, but you will always finish in 2 rolls or less.

That would curtail the tail.

Also it might be slower than the expected single re-roll, but it's probably faster than calculating the coin-flip binary, so your method is a net win for this topic.

[Knaight]

Also it might be slower than the expected single re-roll, but it's probably faster than calculating the coin-flip binary, so your method is a net win for this topic.

It's still a bit slower than compressing the coin flip binary into just a d2 and a d8, using 8(1d2-1)+1d8.

[Jay R]

I've done the decimal equivalent. I've rolled a d10,000,000.000 using these
and these
To Jay R: For what?! Why did you need to roll a d10,000,000.000? The closest I have head of that was in the system that should not be named and I hope you weren't playing that.

https://i.imgflip.com/292pl4.jpg

[Cluedrew]

Oh, OK. That makes sense.