Double Vision

[ToniXX]

Try it now. Let's see if this is better.

[tristenc]

Fill my eyes... [1d20] = 6,[1d20] = 8,[1d20] = 1,[1d20] = 13,[1d20] = 3
Fill my eyes... [1d20] = 7,[1d20] = 15,[1d20] = 13,[1d20] = 4,[1d20] = 6
Fill my eyes... [1d20] = 19,[1d20] = 5,[1d20] = 2,[1d20] = 8,[1d20] = 18

Much mo-betta. thanks for the quick reply! the service around here is great ^_^

[tristenc]

Hate to resurrect the doubly-dead... but had a couple of 'double vision' deja vu and odd probability moments recently...

Double doubles?
Nis vs firebats [1d20]=14,[1d20]=6,[1d20]=14,[1d20]=6

Crit, fumble and Similar dmg?
t5 claw,claw,bite vs skel 2 and skel1 [1d20]=2, [1d20]=20, [1d20]=1, [1d4+4]=3+4=7, [1d4+4]=1+4=5, [1d8+4]=3+4=7
2 7s and 2 8s in 5 dice?!
t2,3,4,5,6 [1d10]=7,[1d10]=8,[1d10]=7,[1d10]=6,[1d10]=8

[dmw71]

This would be a Greg thing; although, I’m not sure what can be done about this as I’m fairly certain the roller is utilizing results from an external source.

[tristenc]

Yeah, i seem to recall last time they found a way to utilize data from JSON on the randomizer site or some such

[ToniXX]

This would be a Greg thing; although, I’m not sure what can be done about this as I’m fairly certain the roller is utilizing results from an external source.

Yes, it still is getting the values from random.org. I don't know what else I can do to make it 'more random'. Remember, when you roll ONE of any size of die, there is an equal chance of getting any number valid for that die. And even when you roll a bunch of 1d4 or 1d6 or 1d10s in a row, the same applies.

[tristenc]

sure, i know youre supposed to treat each die individually statistically speaking but the probability still seems scarce to have these kinds of duplicates so frequently. But it's all good, just a curiosity.

TLDR: my brain runs away with me sometimes aaaand....

i mean... 2x6 and 2x14 on 4 individual d20s in the same set seems to be pretty high, (though getting any single pair on 4 d20 is still like 27.3-ish%), but for the last 2 to also be a pair makes it far less likely. not sure how to do the second half of that math precisely, since the 27.3-ish is just the probability that Not all 4 dice are different
I.E. (20/20)*(19/20)*(18/20)*(17/20)*=> 5814/8000
=> 72.675 that they Are diff
leaving 27.325 that they are not all diff (2186/8000)

I suppose you could treat the second pair as separate since they don't have to be the same pair as the first set and thus it would be 1/20 (as the first of the 'pair' can be anything and the second is 1/20 likely to match... though technically it should be (19/20)*(1/20) since the 3rd dice of the 4 shouldn't match the original pair... so more like (19/400)*(2186/8000) = > ≈ 1.3%
but this is surely a little off b/c i'm unsure how to account for the lack of a '2 pair' solution in the first half of the equation

but that wouldnt fix the last number being different so it would really be a little off... maybe i'm mixing methods too much to make sense...


Edit: add for the truly probability nerdy:
found a solution for 2 pair in 5 dice from a forum talking stats.. <mmmm... tasty stats.. i miss them sometimes... *cough* sorry>

Warning: I am deep in the weeds now
(6C2)(5C2)(3C2) x (4C1)
6^5

where the (6C2) is the rank, (5C2)*(3C2) are the odds of the first two pairs and the (4C1) is the odds the last die doesnt match the first two pairs, all over the total possibilities (6^5)
(15)(10)(3)*(4)
6*6*6*6*6

=>1800/7776 or 23.15% on 5d6

So in the case of the 5d10 where there were a pair of 7s and 8s and a lone 6 (will ignore the fact the 6 is in sequence) the math would be:
(10C2)rank*(5C2)first pair*(3C2)second pair *(10-2)odds last die doesn't match
all over 10^5

[(45)(10)(3)*8]/100,000
=> 10800/100,000 => 10.8%
not as low as I would have thought! (If I did that right, lol)

... sorry... i can't help it sometimes...

[Rex]

Tonight was our weekly FTF game and 1 round of combat had 10 20's rolled counting all the players and me (the DM). 8 of the 10 rolls were 2, the others were a 3 and a 4. The 2 fighters in the group get 3/2 attacks per round a were both on there 2 attack round and both rolled double 2's. No computers involved, just good old fashioned dice. If you roll enough (computer or dice) you will see some strange things and over the 45 years I have been playing I have seen a lot of strange things. In fact that wasn't the only low chance set of rolls tonight. We do 4d6 vs Int to remember something you should know and I called for a memory check and my wife rolled 24 and another player rolled a 4 at the same time. It got a pretty good laugh since my wife's Warlock has the highest Int in the group and the other character (the thief) had the lowest Int in the group.

[tristenc]

Yeah, i've seen lots of wacky combos over the decades as well. As you say, with enough dice, etc etc, something-something Shakespeare

I just find stats and probability interesting. Ofc, with physical dice there can be bias as well. I used to float my new dice in a suspension to be as sure as I could that they were balanced, but that doesnt do anything for geometric imperfections and wear.
that said, i'm intrigued by computer generated 'random' as well

[Rex]

Yeah, i've seen lots of wacky combos over the decades as well. As you say, with enough dice, etc etc, something-something Shakespeare

I just find stats and probability interesting. Ofc, with physical dice there can be bias as well. I used to float my new dice in a suspension to be as sure as I could that they were balanced, but that doesnt do anything for geometric imperfections and wear.
that said, i'm intrigued by computer generated 'random' as well

Me too, I also find stats and probability interesting. Enough so in fact that I have minors in math (pure) and statistics. Of course it also helps that they are very useful in my job as a chemist too. Most dice are biased to a varying degree and I have even calculated it out before. So aren't computer generated random rolls for that mater. There are things you can do to make them as random as possible but it is actually impossible to make them truly random. But, probably just as or more so then our trusty dice, which is why I don't worry to much about it.

[Scott308]

To be honest, randomness isn't really all that important. I could decide the next ten d20 rolls for a game and let you know what was rolled when asked and it would be just as valid. A low number is just as relevant as a high number since I don't know what the numbers will be needed for. I may know the next number will be a 17, but it could be for a character attack or save where you want a high number, or a dex check where you want low. Or it could be a monster attack roll. The key is that nobody involved in the game knows what the next number will be, and I wouldn't know what roll would need the number so the system can't be gamed.

What matters is there is no way to figure out what the next number will be. Obviously, if the numbers I gave you were 8, 9, 10, 11, and 12, you could extrapolate that 13 might be next. But if they were 8, 13, 4, 19, and 11, you would have no way to figure the next number was going to be another 13. For that matter, if I was providing numbers for enough games, I could even just go 1-20 in order because with a high enough number of calls, you wouldn't get back-to-back numbers very often...which still can happen with randomized rolls anyway. I don't think there are enough games here to make that system work, but if there were hundreds or thousands of games where it would be extremely unlikely that any series of rolls would go to the same game...

[Eris]

Many years ago I was asked to GM a FTF game at a friend's house. I didn't have any dice with me and he didn't have any, other than one d6 scrounged from an old Monopoly game.

I rolled the heck out of that one d6 behind the screen, but seldom used it by itself. What I did was look at the digital clock on the table across the room whenever I needed a random number and used the seconds (with division by a base in my head) as the number. It worked fine for uniform distribution or even simulation of standard distributions when I already knew the probabilities in my head. The players never really caught on to what I was actually doing, and they trusted me to "be fair" so it worked out. Actually, Julian trusted that I was going to try to TPK no matter what, and nothing I could ever do would convince him I wasn't actively trying to kill 'em all! However, he naively believed I'd never "fudge" a roll either.

We, the folks asking for the rolls randomly, simulate something like "Brownian Motion of air molecules in a chamber." So, for our games here, even computer generated 32k INT patterns with a fixed seed number is good enough. With the many, many rolls generated by many different people in many games individual rolls will effectively come at different points in the pattern each time a roll is made. Then if the seed changes every time a roll is requested (based on seconds passed in the day as an INT) that would make it even more pseudo-random.