HOLY SHIT

dude, that's like the ULTIMATE beer and pretzels innovation

>>15041438
no dice? no problem!

My world is fucking SHATTERED.

I'm so glad i never realised this during secondary school. The nerdiness that would have overtaken me...

Oh god I can already see the flinging and the breaking and stabbings

HOT DAMN!

Holy shit, don't tell me I was the only one to poke pips in my pencils with a pin in high school.

oh god

>>15041448

is it a good nerdiness?

Huh.

Be right back, performing statistical analysis.

>>15041421
Wow, I realized the use of rolling pencils for odds when a girl in an anime wrote A, B, C, D, E on her pencil and rolled it to randomly get what to answer on each question.
If pencils are 6-sided then that's marvelous though.

THE PATH TO ENLIGHTENMENT HAS BEEN OPENED.

>>15041468

I've known about this since I was a kid. Using dice was a bit more convenient since they were smaller.

Nowadays there are dice rolling apps for that.

What the fuck are you on a-
OH GOD!

http://community.pencils.com/files/Pencil_Dice_White3.jpg

They already exist.

Ok, so after 25 "rolls" here's the results:
side - freq. rolled
1 - 4
2 - 3
3 - 5
4 - 5
5 - 6
6 - 2

Not really enough to be conclusive, but just from a casual perspective, sure it seems fair enough.

....my god.

>This thread

>>15041513
yeah they're called pencils

>mfw I only have mechanical pencils

If I had known this earlier I would never of forsaken the normal kind!

What?!? Explain this sorcery to me, vile magician!

>>15041525
Just get a few of the little golf/bowling pencils.

>>15041516
That isn't enough rolls for good data.

GENTLEMEN!

We must venture forth to /sci/ and figure this out!

I AM OVERCOME

>>15041543
NO. WE WILL DO THIS OURSELVES, AS REAL MEN *SHOULD*!

>>15041543

I know it's not, but I really don't have the patience to sit down and roll and record a thousand times over by hand.

>>15041468
I used to poke marks in the side of my pencils and I'd roll them to see which side came up, but I never thought of it as being dice.

>>15041550
Gentlemen,

We are not savages, we have trained men who like to waste countless years proving absolutely useless things.

Why waste our time when we can have them do this?

This quite literally has never even occurred once to me before.

what is the big deal? I don't get it.

>>15041582

Yeah, me neither. I'm honestly flabbergasted I never thought of it before.

>>15041596
Not many daily use items can be used in a way to produce a statistically random result out of a set number or results, such as a quarter. Dice are items used specifically for this. It has not occurred to the majority of us that a common item like a pencil can also produce this event.

Thus, mindblown

>>15041557
That's what research grants are for, silly.

>This thread

What sorts of games require massive quantities of six-sided dice?

>MFW I never realized this

>>15041715
Shadowrun for starters

>>15041665

Exactly, I'm already getting paid as my job to sit down and painstakingly count every hair on the ass of fruitflies right now.

I'd rather not do something as mind-numbingly tedious in my time off too.

>>15041488
Couldn't that be Kissxsis? I think I just saw that yesterday or something.

>>15041715

you would probably be better off using dice from the dollar store if you are going to need buckets of d6

WHY HAVEN'T I EVER THOUGHT OF THIS?!

>>15041715
W40K.
It's best to buy a small brick of d6's for that game.

>>15041715

Shadowrun.

>>15041715
40k. Hell, you could use a small saw to make several d6 out of only a few pencils.

never thought of it, props to OP being the best.

This is the best thread on teegee right now

>>>/sci/3116395

Gentlemen,

The techpriest have informed us that this new technology shall replace our standard cubic randomization device.

>>15041727
I'm don't think it's the one the anon you're replying to is thinking about, but someone did this exact thing in Baka to Test to Shoukanjuu.

I have a six-sided pencil, a sheet of paper to record, a calculator, and a basic grounding in highscool statistics.

Gentlemen, I'm going to run an experiment and then perform a chi-square test of homogeneity. I should be back in about a half-hour with an initial look at how fair this actually is.

......Pencils as D6? Just who is the OP? it must have taken some kind of tactical genius to come up with-
CREEEEEEEEEEEEEEEEEEEEEEEEEED

/Traditional Games/

Testing the boundaries of the tabletop, and going beyond.

Barrel dice, yo.

Learn the Ways of the Ancients.

You people seriously never realized this before? I thought everyone knew this.

>>15041865
...is it considered fair if you use your test subject to record the results of the test?

>>15041888

the worst part is I already knew barrel dice existed and it still never occurred to me that pencils could be used in this fashion.

>>15041908
Everything is fair in !!SCIENCE!!

>>15041908

I thought of that in advance. I have a separate pencil.

>>15041865
Don't forget to check your degrees of freedom.

>>15041908
Considering that writing wears down the tip and could theoretically throw off the balance, no.

>>15041908

He's right, the pencil is biased in this case.

Use a pen! The fate of your statistical valid results depends on it!

>>15041507
Nnnh ... baby ... so cute ... cheeks ... toothless grin... resolve ... weakening...

>>15041930

I won't. Incidentally, I'm rolling it sixty times for ease of calculation, and I'm halfway through now.

>>15041848
God damn it why did you have to use the Backwards chainsword pic. Now we look like retards in front of the internet scientists.

>>15041961
Make sure each number is opposite of another number which sums up to 7.

ie 1 and 6, 2 and 5

That's the optimum randomization pattern.

rolled 18, 14, 6, 16 = 54

>>15041935
news flash:
so do the holes on the dice.
Its not enough to mess up the game.

pencils

>>15041974
If it's really random the position of the numbers doesn't matter.

An issue is the need for a a VERY flat hard surface, using it on said surface would damage the leas inside making it relatively useless as a pencil.

>>15041865
Godspeed, good sir.

>>15041727
I don't think it's what the anon you're replying to is talking about, but someone did this exact thing in Baka to Test to Shoukanjuu.

>>15041986
I think you mean "uniformly random". Anyways, aside from the nitpicking, I think that the reason they do this is so that not all the high numbers are cluttered next to each other, because if the die IS biased then the expectation value is fucked up.

>>15042036
Surely if you wanted the high numbers not to be clustered together you would put them on opposite sides.

Rather than putting them next to each other. Which is what the 'opposite sides add up to 7' technique does.

>>15041961

ETA on the EXPERIMENT!?

*cue Lightning flash then thunder*

>>15042062
Actually it is a weight issue.

Those little holes affect the results. So by not having all the large numbers on one side the weight is more evenly distributed.

I wouldn't worry too much about the pencils being perfectly random. Due to the plastics dice companies use and other shit most dice aren't perfectly random either. You know that Die that ALWAYS rolls twenties or ones or whatever? There's a reason for that. I think theres a youtube video of some guy at gencon that makes perfectly random dice using vegas standards, but I'm pretty lazy and will only try and find it if people want me too.

anyways, nice job OP. DIY gaming is the shit and you should be commended.

So.... Does the EMPRAH OF MANKIND still play dice with the universe, or is he using pencils?

I was so gobsmacked by this realization that I ascended to a higher plane of existence.

>>15042100
>I think theres a youtube video of some guy at gencon that makes perfectly random dice using vegas standards, but I'm pretty lazy and will only try and find it if people want me too.

9.001 seconds on youtube

http://www.youtube.com/watch?v=bR2fxoNHIuU

http://www.youtube.com/watch?v=PxmkWrDbn34

First, the counts. I rolled:
side - frequency
1 - 8
2 - 9
3 - 10
4 - 8
5 - 12
6 - 13

Because there are six categories, I have five degrees of freedom

H0: The sides are uniform in distribution.
HA: The sides are not uniform in distribution.

With a chi-squared value of 2.2 at 5 degrees of freedom, we get a p-value of .82, leading me to fail to reject the null hypothesis, meaning that we have not yet disproven the supposition that the pencil is fair.

>>15042141
Test with additional pencils and more rolls per pencil?

>>15042150

The chi-square test can never actually prove the null hypothesis. It can only fail to disprove the null hypothesis or disprove the null hypothesis. As it stands, a p-value of .82 is a fairly promising result. It suggests that the pencil is, in fact, fair.

If you want to do more tests with a different kind of pencil and post the counts so that I can crunch the data, please do so.

>>15042150

Yeah, we need replicable results for this to be really valid.

>>15042190
>>15042178
>>15042150
>>15042141

SCIENCE!

>>15042178
WAIT WAIT WAIT.

What probability are you using?

Also we need a comparison here, we are an engineeringish board not a science board.

We do comparisons.

>>15042129
this man is a better person than I will ever be.

>>15042141
I love how you're talking about a chi-squared test and yet you are using a tiny amount of data.

Did you just take Stats for Psych Students or something?

Incidentally, I set the p*-value at .05. This basically means that, should the p-value be less than .05, I'd be forced to reject the null hypothesis.

To explain the jargon more plainly, the p-value of .82 says that if the pencil-rolls were in fact distributed uniformly, an observed chi-square value of 2.2 would occur about 82% of the time, which isn't particularly unusual.

>>15042235

No, I just didn't want to roll the pencil hundreds of times. Like I said, this was a really cursory glance.

>>15042235
>>15042255
Is there two people here with the same name or are you crazy/stupid?

>>15042266

Take an educated guess.

>>15042266
Dosent know hoe to use the three sea shells.

>>15042178

use a Fischer's Exact test. the sample size is too small.

or a G-test. the G-test is just like a chi-squared except easier to calculate and tends to be more conservative. chi-squared can sometimes be prone to type 1 error.

>>15042276
Fukken math. How does it work?

rolled 15, 16, 15, 11 = 57

Ignore me,i suck dicks

>>15042266
There are many Alpharii, some brighter than others.

>>15042276

That's where my knowledge runs out, I'm afraid. I'm not even remotely a statistician. Like I said, my sole statistical grounding is two semesters of highschool stat, so I'm working with what I know.

Perhaps I should bow out and let wiser heads perform more advanced analysis. It's just that no-one was doing anything yet to actually examine the problem, and I figured I might as well put that dusty knowledge to use.

>>15042304
you're lucky we're sticking to frequentist methods here. some one will probably get all Bayesian on our collective asses and then we'll have a huge argument about if the parameter of the population is unknowable and fixed or if it is a variable than can be solved.

i love this shit.

>>15042330
Well we got like half of our buddies from /sci/ in here, we can figure this out.

Does this deserve an archive?

>>15042129
Most of the way through.
I wish I had some calipers on me so that I could see how egg shaped some d20s are.
And egg shaped d20 isn't fair at all.
Any anons with some calipers and some d20 want to give us some relative distances from opposite sides?

>>15042370
I would think so.

>>15042382
I got some and two d20 for checking let me get them.

>>15042400
Okay for the first dice it is about 19.09mm and the second one was worn down too much to give accurate data, it is a 7 year old dice.

>>15042466
is that length different depending on which sides you measure?

>>15042496
Yeah so i measured them all and got an average of 19.08.

It is only a .02 difference between the max and min.

>>15042519
I'd still say that's significant. not enough to be noticeable, but enough to effect the outcome

>>15042535
Well compared to my 7 year old one which i got a difference of a half a mm it isnt going to do much.

>>15042535

affect.

Wait, guys, wait...

Separate, we are nothing.

Together, we are strong.

What if we all rolled our pencils, recorded the results, and posted them, along with the total number of times rolled? Add it all together, and we might have a statistically significant number of rolls, on many pencils, by many people.

Shall we take this to... 1000?

>>15042561
If I had a pencil, I would totally join in on this.

>>15041421

>>15042561

no. the rolls are influenced by the composition/density of the pencil.

>>15042129
while he's right, hes still a scam artist though since he's selling what is essentially an unfinished product for more. Out of the mold his dice are identical to any other companies, he just skips on the painting and then rock tumbler step and then charges you more for the trouble.
Also I find it highly suspect that he managed to find so many dice that where slightly off in exactly the same way for those stacks of his.

>people don't realize the test actually says that pencils are good

My friends retook that fucking course three times and I waltzed through it and the goddamn project alone. They do get me on complex differential equations, though. Weird shit.

>>15042574
I currently have 60 pencils 2ft from me.

>>15042561

>>15042583

>>15042561
Beginning experiment.

Will post results of 50 rolls.

>>15042593
Just found them actually about 200 pencils here but not 6 sided.

Fucking cheap ass pencils.

>>15041516
I got a p value of .7613652679 on a Chi squared goodness of fit test, meaning it's probably that it's a fair "die."

>>15042561
I have three 6-sided pencils, I'll roll 'em each 10 times and then more if I'm intrigued or bored enough.

>>15042561
right good idea
i've managed to find 1 pencil, so i'll roll it a bunch of times and be back with results.

>>15042617
Someone should do a control.
Roll an actual standard cubic randomizing number device and post results

Set of 50 here:

1 - 11
2 - 5
3 - 5
4 - 12
5 - 11
6 - 6

...although actually if we add all these up it's going to displace the bias of individual pencils.

Well whatever.

>>15042620
Results:
1: 4
2: 5
3: 7
4: 6
5: 3
6: 5
Seems pretty even. Pencil 2 rolled 4 three's, but that was the only significantly higher result. And that pencil is kinda old anyways.

So i made a spreadsheet that we can all upload our data too.

Put each set of data into it's own column please to keep the clusterfucking down and label the first row with a unique title

https://spreadsheets.google.com/spreadsheet/ccc?hl=en_US&key=tOQNdgGQtEYJISK1JwNGecA&hl=en_U
S#gid=0

rolled my one pencil 70 times

side frequeny
1 12
2 13
3 11
4 13
5 10
6 11

>>15042690
FOR SCIENCE!

>>15042690
Man, I wish I had non-mechanical pencils. I'd help if I could!

Ok, posting results, 50 rolls:
First pencil:
side freq
1 - 12
2 - 8
3 - 10
4 - 13
5 - 4
6 - 3

Second Pencil
s / f
1 - 9
2 - 4
3 - 11
4 - 5
5 - 9
6 - 12

>>15041421

>back from long day at work
>time to check /tg/ once before sleeps and another day
>see this thread

i love you /tg/ never change

Up to 410 collective rolls.

Hmmm... I'm noticing a disturbing trend in the data so far.

While the grand total of the rolls out of 410 looks pretty good, the pencils themselves can be either good or bad. Some give a good spread, others heavily favor 2 or 3 numbers.

That doesn't look good at all.

The pencil is

>>15042878

Looks about right for goddamn lopsided d6s that everyone rolls.

I'll do a control test. I'll roll an online dice roll an online dice roller 500 times, and record the results. Then a standard cubical randomizer.

>mfw this is distracting me from stats homework

>>15042878

That shouldn't matter too much in the long run. Did a quick chi-square-GOF test with the data collected, and the p-value is at .7614, so we still fail to reject the idea that the pencil is a fair die.

>>15042898

Good idea. This is substandard experiment design, this is, collecting data willy-nilly without setting up a control. /sci/ would be ashamed, if they ever actually did anything /sci/-related.

Wait. Are we talking about pencils with or without erasers?

>>15042941

>>15042941

I rolled with erasers. It was a brand new pencil out-of-the-box, though I did sharpen it to replicate a "real" pencil.

>>15042931
/sci/ is working on this with us right now...

We are of course using the Holy Ticonderoga Lord of all Pencils correct?

(Or anything that isn't an eagle pencil, seriously, fuck eagle)

what are you trying to prove with all this rolling and statistics? I don't understand.

>>15042957

Are they? Excellent! A little inter-board rivalry and friendly ribbing never hurt anyone.

>>15042963
That a hexagonal pencil can be used as an adequate-or-superior d6.

>>15042963

The default premise is that the average six-sided pencil can act as a fair "die", meaning that the probability of any one side rolling face-up should be near-equal to the probability of any other side. The alternative premise is that the pencil is not a fair "die".

By running these tests, we can determine whether or not pencils can be safely substituted for d6s without damaging the random element of play.

>>15042971

what about common sense? it is pretty obvious a pencil would be an okay but not great d6.

>>15042985

but doesn't everyone have different pencils? meaning different results?

We have hit 500 rolls! The results so far...

1's) 86
2's) 68
3's) 83
4's) 90
5's) 92
6's) 91

Looks pretty good.

>>15042989

Common sense is inadequate! We demand more! We demand data! We demand analysis! We demand SCIENCE!

>>15042989
Common sense was, back in the day, that the world was flat. Or that the universe revolved around the earth. Science works.

>>15042998
Hence the community testing.

>>15042998

By rolling many different pencils and pooling the data, we create what is effectively a random sample, thereby accounting for minute variation from pencil to pencil.

Rollin' more pencils...

500 should probably be enough, but an even 1000 would be better.

at this point you guys would do well also rolling a big set of "normal" d6.

you then can ran an Kruskal-Wallis one-way analysis of variance. it beats the parametric ANOVA because you don't assume even distribution, and because if it passes a Kurskal-Wallis it would pass an ANOVA anyway.

this way you can actually compare the performance of the two sets.

i still think someone needs to run a G-test. I'm way to lazy sadly, and R isn't installed on this computer :(

any Bayesians want to give this a try?

>>15043067
I read that and my head is hurting now.

Engineer here.

>>15042898
>>15042931
someone posted about this up here. Not sure if he ever did anything though
>>15042660

>>15043008

Cave Johnson here. Look, sometimes you're going to need to make up a random number. I don't know when, maybe you're choosing lotto numbers or doing algebra because you've got a weird hobby. So we threw some science down and found out an ordinary number 2 pencil can come up with any random number you want. Er, just as long as it's from 1 to 6. We're working on going higher. Maybe we'll make a five-hundred sided pencil. Like you're writing with a polygon or something.

Cave Johnson. We're done here.

>>15042989
>what about common sense?
Humans have terrible intuition when it comes to statistics.
Do not trust your intuitive feelings about probabilities. this is something of a weak point for us.

You guys made my day. /tg/ is crowdsourcing at its finest.

>>15043077
i'm a paleontologist actually, but i wish my undergrad had been in statistics or at least dual major statistics biology. this kind of thing used to hurt my brain all the time too, but recently it is all i think about. funny how things change.

and now this makes almost complete sense to me: http://www.nceas.ucsb.edu/~alroy/JA_commandments.html

Do we have a standarized "1" side? I suggest we use the pencil side with the #2 side as our standard for side "1" and continue with increasing numbers in clockwise order (assuming non-eraser side is facing you)

This should give us a good standard for pencil data control.

>>15043120
Yes that was the common sense part.

We are just basically hoping everyone isnt being a dumbass.

For those interested.

>>>/sci/3116395

>>15043120
actually, we assumed opposite sides added to 7.

>>15043157
Also part of the standardization of common sense.

But that doesnt affect the data too much if you didnt do it because the number would still be random since no side will be favored.

>See this thread a few hours ago
>Cool idea
>Doubt anyone would be willing to sit down and roll out statistical data by hand
>Check /tg/ before getting some sleep
>Everyone has done exactly that

Let it be known that /tg/ still remains the champion of Getting Shit Done.

>>15043202
>Let it be known that /tg/ still remains the champion of Getting Shit Done.

I wasn't aware that this was ever up for dispute.

Patenting d6 numbered pencils and selling them to thinkgeek as we speak

Just rolled 50 dice in an online dice roller, reporting results.

1 - 6
2 - 8
3 - 8
4 - 9
5 - 9
6 - 10

>>15043223

It appears /sci/ found a way to turn pencils into d20s.

>>3117140
What if you took 1 pencil, and labeled its sides with two 0's, two 1's, and two 2's, then another that was labeled either 0 or 5, then another labeled 1-5, with one side being a "reroll?"

>>15043247

Random number generators aren't exactly the same thing as physical dice.

>>15043120
Wait...if it has a 2 on it already, wouldn't that be the 2 side?

Good idea, but wrong. You would be able to roll a 20 or a 29 in that case.

You need 3 0's and 3 1's on the first pencil. That way, you can't generate numbers larger than 20.

>>15043255

doesn't work.

>>15043278
Oh fuck, are we going to wind up standardizing all dice rolls to a series of pencil markings?

>>15043278

This post was supposed to be quoting
>>15043255.

>>15043265

I thought we'd be using an online generator and actual dice separately as controls.

>>15043299
Yeah we need that data in the spreadsheet.

https://spreadsheets.google.com/spreadsheet/ccc?key=0Aqrqn_hAPYCYdE9RTmRnR1F0RVlKSVNLMUp3TkdlY0E&
;hl=en_US#gid=0

>>15043255
looks like he corrected himself.

>>>/sci/3117185
erp, forgot to explain.

The first one is if it is single digit, in the teens, or... wait, no. It should just be split between 0 and 1, shouldn't it?

Okay, let me try again...

Pencil 1 = first digit = split between 0 and 1
Pencil 2 = half of the second digit = split between 0 and 5 (rolls half will be over 5, half under)
Pencil 3 = other half of the second digit = 1-5 and "reroll.

The lowest would be 0-0-1 (1), while the highest would be 1-5-5 (10 + 5 + 5 = 20).

>>15043316

Yeah, that works. Then you have a much less fun to roll 20 sided dice.

>>15043316
Crossposting a better way.

>>>/sci/3117201

>There's an easier way to do it with just two.

>If you roll a six on the first pencil, reroll it. Otherwise subtract one and multiply by four.

>If you roll a five or six on the second, reroll it. Otherwise add the number to the result of the first pencil.

>And you're done.

Got 50 d6 rolls:

1 = 6
2 = 7
3 = 11
4 = 9
5 = 6
6 = 11

>>15043316
But does it have the same probability as a normal d20 though?

>>15043398
Yup. There is exactly a 1 in 20 chance that any particular number (between 1 and 20) will appear. Each number can appear only once.

Alright, with 610 data points, we get a chi-square value of 8.35 and a p-value of .1378. We must still fail to reject the null hypothesis that the pencil is a fair die.

>>15043380
Wouldn't it be easier to just count the six as a zero?

>>15043459
If you also want to be able to roll -3, -2, -1 and 0 then sure.

Otherwise no.

>>15043443
That p-value is descending as we continue isn't it? That seems a bit foreboding.

>>15043443
Is that good or bad?

>>15043430
are you sure about that? if you have 3 pencils, A B and C, where A is 0-0-0-1-1-1, B is 0-1-2-3-4-5, and C is 1-2-3-4-5-reroll, you can only achieve a twenty with A1, B5, C5, a single possibility, but you can roll a 4 with A0, B0, C4 OR A0B1C3 OR A0B2C2 OR A0B3C1, which is 4 separate possibilities. This will end up creating a bell curve on the average, similar to rolling 2d6 rather than 1d12.

Basically, no, multiple hexagonal pencils is an improper substitute for a d20.

>>15043459

No. To emulate a 20 sided dice, you need 20 unique outcomes with equal possibility.

Rolling a 5 sided pencil followed by a 4 sided pencil has 5x4=20 outcomes.

Rolling a 2 sided pencil, followed by a 2 sided pencil, followed by a 5 sided pencil has 2x5x2= 20 outcomes.

Rolling a 6 sided and a 4 sided pencil has 24 outcomes. Even if you assigned them all to one of the 20 numbers, your probabilities would not be equal.

>>15043486

That's actually pretty common with chi-square tests. When you get to really, really high counts it's all but impossible to disprove something.

We're not quite there yet, though, and we can stick with the assumption that the pencil is fair. Hasn't been disproven yet.

>>15043495
You read it wrong. Pencil B isn't 0-1-2-3-4-5, it's 0-0-0-5-5-5.

2 options, not 6.

>>15043486
In theory it could be a fault of non-uniform labelling of pencils. I think totalling each value is not an adequate way to represent the abilities of pencils to be random. Hypothetically, every pencil could be weighted slightly more towards one side, which would throw off the results for that pencil, but if everyone labels their pencils differently, there could easily be an illusion of random rolls.

I think a better approach would be to calculate the randomness of each pencil, then compare the randomness values of all of the pencils to see if there is a trend.

>>15043495
A die with any number of sides can be substituted by hexagonal pencils.

>>15043265
And it's "random" anyways.

>>15043517
okay, thanks for clarifying. that makes much more sense now.

>>15043525

That would be ideal, yes, but it would also require a much greater deal of individual work, work that most fa/tg/uys simply wouldn't be willing to put in. As it stands, this is a decent substitute for getting a starting idea for how fair pencils as a whole are, although I think the idea merits further testing.

>>15043525
That would work better. Aaaand we do have each pencil's data separate from the others.

But I have no idea how to do these maths, so I'm of no help here.

>>15043495

B is not 0-1-2-3-4-5. It's 0 or 5, with 3 results assigned to each.

so, you roll pencil b, and it is either zero OR 5.

So, you get 0 or 1 for the tens digit. 0 or 5 for the first half of the ones digit, and then 1,2,3,4, or 5 for the second half of the one's digit, and add them together.

There are 2x2x5=20 outcomes, the smallest being 0,0,1, or 1, and the largest being 1,5,5, or 20.

>>15043526
7 sided? 11 sided? 13 sided? 17 sided? 19 sided? 23 sided? 29 sided? 31 sided? 37 sided? 41 sided? <more prime numbers> -sided?

>>15043600

As long as you are willing to reroll, you simply generate too many results and discard some of them. rolling 3 six sided dice will generate 216 unique outcomes. If you are willing to put in enough reroll conditions to discard some of them, you can generate procedure to generate a dice roll of any number of sides up to 216 by only rolling 3 6 sided dice a single time. (and possibly rerolling, of course).

>>15043544
so, the probability:
1/2 x 1/2 x 1/5 = 1/20
am I counting this right? Is there a difference in 'randomness' between, say, two dice rolled in succession and two dice at once? because the three-pencil d20 needs three pencils to be thrown in succession, whilst the normal d20 just needs one roll. Does the 'roll three' works like 'roll one' in terms of randomness?

>>15043627
Correct. But as long as the number is divisible by one of 6's non-one factors, you won't need to reroll ever (though side redundancy will occur for any number that isn't 6^x where x is a whole, positive number)

>>15043659
with an entirely, perfectly random die/pencil, it should work fine. but if you have a generator that favours a single result, and you roll that 3 times, it will turn out differently than 3 different generators rolled at once (or even in succession, really).

>>15043600
With y Dx you can simulate a die of any number of sides up to x^y. You just express the number of sides you want in base x and then use the dice for different digits of the number.

It just gets more awkward if you're working with prime numbers because you have to simulate that number +1 and then reroll if you get the largest possible value.

Thread is archived, by the way.

another deep rot and no pencils fuuuuuuuuuuu compulsive sharpening to non existence

>>15043120
Or...you know...make that the standard "2" side.

If nothing else, this creates a wonderful way to punish uncooperative dice. You can burn them!

I feel like the next time I need help with Econometrics or Statistics I should come to you guys.

Holy shit, kids, lookie here!

Lifehacker + OP = WIN

http://lifehacker.com/5804898/use-a-pencil-as-a-6+sided-die

>>150444

Allright, which one of you is 'danjiro2'?

Morning bump

HOLY FUCK SHIT!!! I NEVER KNEW!!! I NEVER KNEW!!!

Is it just me or has all the spreadsheet data disappeared?

Also we need to come up with a standard system for numbering pencils so that you can use them as any kind of dice. There's plenty of space along the sides to make multiple sets of markings.

>>15047931
Yeah, I can't see any of the data either.

>>15047931
Lets see... using multiple pencils (R stands for re-roll):
D3:
Pencil 1 -> 1,1,2,2,3,3

D4:
Pencil 1 -> 1,1,1,2,2,2
Pencil 2 -> 0,0,0,2,2,2

D6:
Pencil 1 -> 1,2,3,4,5,6 (duh)

D8:
Pencil 1 -> 1,1,1,2,2,2
Pencil 2 -> 0,0,0,2,2,2
Pencil 3 -> 0,0,0,4,4,4

D10:
Pencil 1 -> 1,2,3,4,5,R
Pencil 2 -> 0,0,0,5,5,5

D12:
Pencil 1 -> 1,2,3,4,5,6
Pencil 2 -> 0,0,0,6,6,6

D20:
Pencil 1 -> 1,2,3,4,5,R
Pencil 2 -> 0,0,0,5,5,5
Pencil 3 -> 0,0,0,10,10,10

>>15048238
Oh, this is wonderful. The d4 and the d8 are perfect and use most of the same markings too.

>>15041421
OMG YOU GO TO LIFE HACKER TOO?

>>15048238
The flaw in this system is that you get bell-curve distribution, not true random distribution.

>>15048463
Looking at the d4 one at least, it would appear to produce even results.

There's only 4 possibilities:

1 and 0 = 1
2 and 0 = 2
1 and 2 = 3
2 and 2 = 4

Each of these possibilities should have an equal chance of popping up which would make it properly random.

If it's the same with the others, that should be a legitimate way of doing it.

Anyway, here's what your first pencil could look like. The line is the marker to reroll.

>>15048463
Nope, the distributions are uniform.

>>15048534
Using multiple pips would take up too much space. The less space used for the 3 marking/side, the more dice you can get out of a single pencil. Markings that are of equal width would be most efficient (like normal numbers).

Also, suggestion for the distribution of the numbers could be 6, 2, 4, 1, 5, 3. This would even be enough to produce the other sets with nice spacing between. Low numbers vs. High numbers for binary, 12, 34 and 56 for trinary, and reroll = 6 is trivial to remember, but a small bar or pip added to just that number could work.