Page 1 of 1

Troyes Reroll

Posted: 13 September 2012, 02:07
by teasel
is there anything wrong with the reroll routine on troyes? i've lost count of how many times i have seen a 1 reroll into a 1,now some people think it's just bad luck but one time i was able to do so 5 times in a row... and when something like that happens it's really hard to point the finger at "luck"

Re: Troyes Reroll

Posted: 15 September 2012, 05:00
by Phelan
Unfortunately, randomness( what you call luck) in computer software is pretty hard to test.
"1 1 1 1 1 1 1 1 1" could be either a fault in the algorithm, or a perfectly valid sequence of rolls, and you'd have no way of knowing.

There was previously a player complaining about the randomness of rolls in "Can't Stop" (you can search for the thread), and from what I remember, BGA said the randomness used was from a standard library function, which means it is used in most software.

So it should be fine.

You can read more about it: http://en.wikipedia.org/wiki/Random_number_generation

Re: Troyes Reroll

Posted: 17 September 2012, 00:20
by beri
I've had the same kind of unpleasant conescutive rerolls at Troyes, with at least four 1 in a row.

So there must be something up there making it go wrong.

Re: Troyes Reroll

Posted: 19 September 2012, 22:05
by Speusippus
It is definitely possible to test this. A single string of five 1's may not mean anything. But an examination of every roll over a large number of games could be used to see if the rolls are close enough to being truly random.

Re: Troyes Reroll

Posted: 02 October 2012, 21:15
by Phelan
beri wrote:I've had the same kind of unpleasant conescutive rerolls at Troyes, with at least four 1 in a row.

So there must be something up there making it go wrong.
In the previous thread( can't stop dice are not random):
sourisdudesert wrote:Hello,

I confirm that dice are totally random. It is using the PHP function "mt_rand(1,6)".

Cheers,
As for
Speusippus wrote:It is definitely possible to test this. A single string of five 1's may not mean anything. But an examination of every roll over a large number of games could be used to see if the rolls are close enough to being truly random.
Note that I didn't say impossible, I said pretty hard. Random.org, a site that uses truly random numbers instead of the pseudo-random numbers that are used in most software, has a pretty interesting section on how to test random numbers. It is pretty hard.
For most purposes, the PHP random function works perfectly well.