ben_awesome
Joined: May 11, 2016
|
 Posted:
Aug 18, 2017 - 13:23 |
 
|
Its interesting that you mention the human brain.
Remember the individual players don't see the big picture, they only see the small specific sample in their game. you are talking about (10's of thousands, 100's thousand dice rolls?) they're looking at a sample of what 300-400?
Then there is this:
"Deep 6 just constantly rolls a list of d8 and d6 results to be called upon as the requests come in.
The *order* in which the requests come in for these randomly generated numbers also adds another layer of randomness to the process. eg: Every Kick-Off event requests two sequential d6 results from the table. When these kick-off requests come in is entirely dependant on when the player hits the client's "End Set-Up" button."
So you have a sequence of random numbers, that are then randomly handed out. In the big picture this is fine - in the smaller individual picture the 2nd layer of randomness can affect the 'probabilities' so individuals see a series of rolls that would normally have a very low probability of occurring but as you've effectively squared the randomisation the probability has been adjusted - and no one knows how. |
|
|
Christer
Joined: Aug 02, 2003
|
| c9805222 wrote: | | in the smaller individual picture the 2nd layer of randomness can affect the 'probabilities' so individuals see a series of rolls that would normally have a very low probability of occurring but as you've effectively squared the randomisation the probability has been adjusted - and no one knows how. |
Except that's not how randomness works.
If you have a high-entropy (e.g. random) series of numbers (or rolls), additional "layers" of randomness will not improve or change the fact that the series is random. It doesn't get better, it doesn't get worse. You're simply adding another entropy source to the mix.
However, if your original series of numbers is low entropy, adding more entropy (using more or less whatever means you see fit) will make things better.
In the FUMBBL case, the underlying technology is very very strong (significantly stronger than necessary). Maybe not 100% truly random, but close enough to be more random than any dice you use on the tabletop (even if you're using casino grade dice). Adding more layers will not affect the perceived streaks or how people believe the dice are one-sided and favouring a particular coach. Even if I made use of the TRNG hardware I have (http://www.entropykey.co.uk/ - currently in use to add entropy to SSL connections people use to access the site), rolls in the client wouldn't be noticeable (or measurable). |
|
|
straume
Joined: Dec 01, 2003
|
| Posted:
Aug 18, 2017 - 14:53 |
|
First: Confirmation bias is a thing, no question about it.
The thing about how the game Blood Bowl works is that some bad dice might put you in a difficult position. Which means you will have to roll more dice. Which (often) means more failure and thus a feeling of "streaky dice" against you. Or the other way around: If you take out some players, you will (often) be in a good position to take out more players thanks to the strength and pitch advantage.
Planning for failure and dealing with crappy dice is a big part of getting good at this game. Bad stuff can snowball. Does not mean the dice is not random. |
|
|
DrDiscoStu
Joined: Feb 20, 2006
|
| Posted:
Aug 18, 2017 - 15:10 |
|
A friend of mine made a RNG in high school. A program would take each keystroke and match it against a time log, the hundredth of a second function (0-9) would be stored (tenths and full seconds etc, would be ignored). That would become the list of random numbers.
(cool story)
(not relevant)
(later) |
_________________
Check out my fishing and camping blog.
The Black Pearl Bounty-Board.
GUARD CONQUERS ALL! |
|
Rbthma
Joined: Jan 14, 2009
|
| Posted:
Aug 18, 2017 - 15:28 |
|
I think if you added a dice rolling sound, all these questions would disappear. That soothing friendly dice roll rattle would subdue all fears of dice inequality...or equality - whatever you fear.
apologies for interrupting the quality posts  |
|
|
Thunderfrog
Joined: Jul 25, 2017
|
| Posted:
Aug 19, 2017 - 00:55 |
|
I knew that for awhile the "random" in apple was based off the first track played, and then stuck, so every time you started off with say, "Elenor Rigby", the tracks to follow were static. Or maybe that was my Zune? I don't recall. |
|
|
ben_awesome
Joined: May 11, 2016
|
| Posted:
Aug 19, 2017 - 01:33 |
|
first off, I'm well aware that each dice rolled is independent of the previous dice, but then you talk about serious / good coaches minimising risk, how do they minimise risk if for eaxmple the reroll doesn't account for probability?
humour me please, I'm trying to understand this as much as be a pain in the behind.
according to probability the chance I roll 6 *1' in a row is: 1/46656
however according to the system in place, the rolls are generated randomly (e.g. 1/466656) but then depending on the number of games happening, and when I click end turn or dodge for example I could be assigned a another random number.
to be clear these are the two situations:
I' at my table rolling dice - the chance I roll 6 ones in row is 1/46656
I'm rolling dice on fumbbl, depending on the number of requests for a dice (e.g number of games coaches on lie at any given point is) roll my odds to roll 6 *1 in roll are?
ah yes the 2nd scenario dopes introduce more possibilities but at the same time it actually lessons the skill in dice rolling because it relies on the number of requests tot he dice machine rather than you considering the odds of the dice you roll.
or did I get that wrong (which is possible I get lots wrong) |
|
|
tussock
Joined: May 29, 2011
|
| Posted:
Aug 19, 2017 - 05:23 |
|
The fumbbl random numbers are correct if you just want to study exactly how likely things are to work after many thousands of attempts. 1 in 9 of your 6+ 3+ attempts will work, if you try them many thousands of times, or 13 in 54 with a team RR.
Your tabletop dice are not correct if you just want to study exactly how likely things are to work after many thousands of attempts, those things are all biased, often quite heavily. |
_________________
  |
|
Harad
Joined: May 11, 2014
|
| Posted:
Aug 19, 2017 - 08:41 |
|
c9805222 the odds remain 1/6 for each dice. The probabilities are not affected. Whilst I know quite a bit of maths I am not an expert on RNGs so I could be wrong but I think that the layers of entropy just increased the randomness. I will try for a crude example to
show the difference between randomness and probability (but could stuff it up as probability can be surprisingly confusing and it's early in the morning and I'm stupid). I could have a rubbish RNG which gave you the numbers 1,2,3,4,5,6 in order and repeated: if you didn't know this the odds of what you get are still 1 in 6 for that first number but because it's not very random at all you would soon be able to exploit its lack of randomness. I am not implying all RNG give a perfect 1 in 6 chance (real dice do not for example) just trying to illustrate a difference between randomness and probability.
I think in practise any serious attempt to create a RNG produces something so much more random than my example that it's not going to be exploitable. As Christer says, his is much better than is necessary.
Short answer the probability is as close to the mathematical theory as makes no difference (if there is any) and even if there is an infitessimal difference it's not exploitable because of how random it is. |
|
|
Christer
Joined: Aug 02, 2003
|
| c9805222 wrote: | according to probability the chance I roll 6 *1' in a row is: 1/46656
however according to the system in place, the rolls are generated randomly (e.g. 1/466656) but then depending on the number of games happening, and when I click end turn or dodge for example I could be assigned a another random number. |
The error you are making is to not fully realize how independency works, and expect a RNG to be forcing an equal distribution of results. Let me give you an example:
Let's say you consider two different methods of randomizing one of two numbers:
1. Coin flipping (no sleight of hand tricks to force results)
2. Using a deck of cards, and using the red or black suits as the output of the randomness.
The coin flip is fully random, and results are completely independent. If you flip heads, the probability of heads on the next flip doesn't change. It's still 50%. This is perfectly random (again, assuming to tricks to force the result).
The deck of cards is a much more interesting example. If you were to return the drawn card to the deck and shuffle properly after each of them, the result would still be independent and fully random.
However, if you don't return the drawn card, and keep drawing new ones you will no longer have independent results and your random series is flawed. For example, after 51 drawn cards, you're left with a single one. if you kept track of the first 51 card, you'd be able to "predict" the last card.
Even after the first drawn card, there would be a slight bias towards the opposite colour of card. This latter method of generating randomness is obviously weak.
So, let's go back to the case where you feel it'd be possible to change the results by waiting for someone else to roll first.
Now, I am going to assume you are comfortable with the fact that the RNG used is fully independent. If it wasn't, all bets are off. 
Let's construct a real-world equivalent of this:
You have 100 dice set up on a table, in a line. You start rolling the first few dice one by one in the order they are placed in. After 5 dice rolls you realize all of them have been ones, twos and threes. Knowing the dice are not biased, would it make any difference if you took the 6th die or the 7th one (or the 93rd one for that matter)?
With completely independent (and fair) dice, it obviously wouldn't make any difference. The rolled result wouldn't necessarily be exactly the same, but it would still be independent from the previous rolls, and wouldn't affect your end result from a statistical perspective. In fact, you would never be able to tell the order of dice you picked by simply looking at the rolled numbers.
I hope that's understandable to you. |
|
|
ben_awesome
Joined: May 11, 2016
|
| Posted:
Aug 19, 2017 - 10:49 |
|
Your example is good. I guess I'm confused by the extra layer of randomness. |
|
|
ArrestedDevelopment
Joined: Sep 14, 2015
|
| Posted:
Aug 19, 2017 - 11:31 |
|
Entropy from mouse movement!
*waves mouse frantically across screen while waiting to reroll endzone GFI* |
_________________
 |
|
|
|
| |
Conceding v Goblins/...
Search
