VDice - Dice Rolling for Proboards v5

[ifoutz]

Hi, I am new here, and I am having a weird issue with the dice plugin. When I enter something like 2d20+7, it works exactly as I would expect. But when I enter 1d20+7 or d20+7, I've tried both, I just get the die roll, no modifier added or total shown. Any idea what I might be doing wrong?

[arek]

This can be worked around using vDice Advanced, but really should be fixed in the plugin.

If you're already using vDice Advanced, I posted the solution a couple of pages back.

--Arek

[ifoutz]

Sorry I missed it before, but I found it this time, and it works, thanks!

[Virgil Sovereign]

*lol* Thanks, arek. Nothing like needing the advanced plugin to un-break the standard one.

I'll take a look at this at the same time I'm looking at Crystal's issue.

[arek]

I did that back when you were busy with other things, Virgil Sovereign. I agree, needing the Advanced plugin to un-break the normal one is seems a bit backwards.

--Arek

[garrus]

Hey,
our forum seems to be having some issues with the RNG, where threes and sixes on the standard 1-6 die appear significantly less than the other numbers. I understand random is random but it looks very suspicious to a lot of us

We're using VDice (Standard Die Edition)Version 1.1.3 Beta

When we roll 100 dice in one block in a single post, the distribution seems what you'd expect.


ones   twos    threes  fours   fives   sixes
15 18 11 15 17 22
14 17 17 18 18 15
11 20 19 18 18 15
19 16 11 15 21 18

59 71 58 66 74 70   /398


But when we've played a game of a long duration (about 12 hours) with only 1-5 dice per post, the distribution looks very suspicious. For example, this is the total number of each die roll rolled by players in a game of Yahtzee, each line being a different player


ones   twos    threes  fours   fives   sixes
29 25 4 24 35 4
25 40 6 20 21 10
29 31 4 22 32 8
19 32 9 24 36 7

102 128 23 90 124 29 /496


and in chart form!




And furthermore, when we roll a block 10-100 dice in a post, the first 10 are again heavily popularised by 1,2,4, and 5s (see here: somethingtodo.freeforums.net/thread/20/dice-conspiracy-thread )


So in summary, it appears to me that dice roll distribution is fairly equal when posting a large amount of dice in a single post, but with small amounts there is a bias against 3s and 6s.

I really do think this isn't just in my head

Our forum is here if you want to take a look
somethingtodo.freeforums.net/

These particular threads should be of interest:
somethingtodo.freeforums.net/thread/20/dice-conspiracy-thread
somethingtodo.freeforums.net/thread/25/return-yahtzee?page=1
somethingtodo.freeforums.net/thread/27/50-dice-yahtzee?page=1
somethingtodo.freeforums.net/thread/10/roll-dice?page=1
somethingtodo.freeforums.net/thread/14/yahtzee?page=1
somethingtodo.freeforums.net/thread/23/krunch?page=1




Thanks for your help, and thanks for all your work on the plugin, we love it a lot

[garrus]

I did some more testing on this, and it is happening on multiple forums for me.

What's more, I increased the range to 1-18 and the same pattern was happening with all multiples of 3 (3,6,9,12,15,18) for a sample of ~200 rolls.



1 15
2 10
3 8
4 8
5 15
6 6
7 18
8 14
9 2
10 13
11 26
12 1
13 14
14 17
15 1
16 15
17 14
18 1



So for now I've got a bit of a temp solution in place.
Using a range of 1-18 for the 6 sided dice, each 1-6 value has two "strong" rolls and one "weak" roll assigned.
ie:
1 is 1,8,15
2 is 2,9,16
3 is 3,10,17
4 is 4,11,18
5 is 5,12,13
6 is 6,7,14

which gives me a more decent distribution:
1 2 3 4 5 6
30 27 35 35 30 41






Anyway, those are my findings It'd be good to know if this is affecting anyone else's forums.

To reprodouce:
roll a small amount of dice (5-10) in a single post at a time until you've got 100 or so rolls to count from.
Rolling 100 in one post will make it appear as even distribution.

[Dakota*]

Hello! I was told to report that vDice is causing problems for guest posting. I had to remove it on my forum as when guests tried posting, for some reason this plugin was causing it to glitch, and no guest was able to post. So hopefully you can look into this. Thanks!

[Virgil Sovereign]

OK, garrus and Dakota*. It looks like some issues are piling up. I'm swamped this week, but I will get around to the various fixes.

[Virgil Sovereign]

garrus: I threw together a quick simulation on jsfiddle for the random number generator.

jsfiddle.net/L6GNu/6/

It appears the problems aren't just limited to the ones you've noticed. For lack of more precise terminology, it appears to be a crappy RNG.

I'll have to come up with something new. If you're interested, you can fork the Fiddle and edit the $V.rrhash function. It should ideally return a non-negative integer uniformly distributed from 0 to 2^32 based on a string argument 's' and an integer argument 'i'. The problem is that the string and the integer both have extremely low entropy. They might only differ by one or two bits between different dice, although they are guaranteed to change from die to die, roll to roll, and post to post.

If you (or anyone else reading this) can come up with a reasonably efficient hash function, you'll have saved me a lot of work.

[garrus]

It's beyond my knowledge sadly

Though that fiddle you gave seems to have weak multiples of 2, whereas it was multiples of 3 being affected for me. The temp solution I had above (giving each face value of the d6 3 values of a d18) has evened it up but I'll still keep an eye on that.

Good luck to you or anyone who tries to fix it, many thanks from us

[Virgil Sovereign]

garrus, I finally put together a decent RNG based on--of all things--two-cycle bit permutations. You can find it at

jsfiddle.net/L6GNu/43/

If you're able to break it, let me know. Thus far it's proven to generate uniformly random values over all permutations of posts, rolls, and dice. It might go down as the least efficient RNG in history, but it does manage to squeeze out reliable randomness from just a few measly bits of entropy.

I'll add it to VDice 1.1.4, along with the fix for the improper roll template being used for single-die rolls, and the fix for VDice interfering with guest posts.

[Virgil Sovereign]

VDice and VDice SDE 1.1.4 are out. They should address all issues/bugs raised by ifoutz, garrus, Dakota*, Crystal, and @littlemissraven. Specifically:

- VDice no longer interferes with guest posting.
- Random number generation is significantly improved.
- Rolls with ranges of the form 1dN+O will now render correctly with the offset and sum showing if the offset is nonzero (and the 'nosum' tweak isn't present).

For users also using VDice Advanced: For the correction to the fix of the 1dN+O syntax to "take", you'll need to update to VDA 1.1.4. However, it's probably just easier to stick with 1.1.3 and make the change manually, which has been the workaround thus far.

The only difference between VDA 1.1.3 and 1.1.4 is this syntax fix. All other issues don't require a VDA update.

[ridicangel]

Hey there. How do I adjust the size of the dice once they're rolled and posted? They're rather large and obnoxious as they are right now, and I'd love to reduce the images to half of their default size.

[Virgil Sovereign]

ridicangel: Under the 'Tweaks' tab, add a width= tweak to override the default image widths. Just enter the desired width in pixels. The VDice SDE version already has this tweak present by default (to reduce the width of the dice from their native resolution of 300x300 to 80x80), hence in that case, just change the 80 to 40.