Irgy
Joined: Feb 21, 2007

Posted:
May 23, 2024  05:50 

There's a big philsophical debate of "true" randomness vs pseudorandom but it's mostly just that, philosophy. At best. At worst the distinction is just pure nonsense. From a Bayesian (as opposed to "frequentist") perspective, it's random for you if you, personally, don't know the outcome, irrespective of what the underlying mechanism is.
The point of the mouse movements thing is that somebody once claimed they knew the source code, could calculate the seed, and use it to predict the sequence of rolls. It was a theoretically plausible claim at the time. Armed with that knowledge they could indeed chew up 1s that they knew were coming by doing other things first. Whether they were just bluffing we'll never know for sure, but by including the opponent's mouse movements such a claim is no longer even theoretically plausible. But, notice I never needed to use "true" or "pseudo" once in order to describe that actual theoretical problem and solution?
Anything generated by an algorithm will be theoretically predictable if you know the algorithm and the inputs. But that's a long way from saying it has certain properties like 1s being less likely after other 1s. Problems have occurred with patterns in pseudorandom numbers but those patterns are immensely more complex than. Anything anyone would even consider using would have a mathematical proof that simple properties like the probability of a result being statistically independent of recent results holds. 


MattDakka
Joined: Oct 09, 2007

Posted:
May 23, 2024  11:58 

Irgy wrote:  From a Bayesian (as opposed to "frequentist") perspective, it's random for you if you, personally, don't know the outcome, irrespective of what the underlying mechanism is. 
I understand this, but I'd like to know the accurate, scientifical, technical definition of FUMBBL RNG.
True random, semirandom or pseudorandom? And why?
About the fact that, if I can't know the outcome, then it's random: there could be a bag with 100 numbers, with some missing (I don't know which ones).
Numbers are drawn from the bag, now, I can't know the outcome, and that would fit your definition of random, but technically that would not be random. 


Sp00keh
Joined: Dec 06, 2011

Posted:
May 23, 2024  13:31 

it's not a bag of numbers because that implies that things taken out, will influence the remaining options
the relevant info is all in this thread. you don't need any further info 


MattDakka
Joined: Oct 09, 2007

Posted:
May 23, 2024  14:41 

The bag with numbers was just to make an example that not knowing the outcome of something doesn't necessarily mean it can be considered random. I draw one number, I see it, then I put it back in the bag, rinse and repeat. For me that would be random, but not really random, due to some unknown numbers missing.
Ok, but I had no answer to a direct question. Is the RNG true random, pseudorandom, or semitrue random and why?
True random because it uses mouse entropy + algorithm?
Pseudorandom because after 1 million of rolls there will be one roll discrepancy?
I guess that semitrue random is like saying pseudorandom. 


MrCushtie
Joined: Aug 10, 2018

Posted:
May 23, 2024  23:09 

SeriliKirico wrote: 
Personally, I would say they are mental if they truly believe RNG is out to get them. However, when it comes to Cyanide game, which is developed by.. people of questionable skills (what they've proved again and again in the last year), I reluctantly agree with thoughts certain streamer shared before  even if RNG is itself solid, they could still implement the actual usage of it in the game client and/or server badly. Like, one possible scenario is that when client fails to pull a fresh generated value (for example, due to server being unable to handle a spike in traffic), it could just reuse the previous one (would explain occasionally getting too much of snake eyes in one game in different kinds of rolls). It even could be done intentionally by them, to cut costs, if they don't want to pay extras for faster servers to accommodate more users at peak hours. This hypothesis in case of Cyanide looks quite believable. 
Unless we think Cyanide are more careful than a multibilliondollar bank, this feels highly plausible. Citibank (in part 4.27) had a trader fatfinger the wrong number into a form, and in part because their pricing control said "if you can't get the current price of a security, just use $1" things happened... Just recycling random numbers rather than burn electrons communicating back and forth with servers feels like the responsible, environmentally friendly thing for Cyanide to do 
_________________


RDaneel
Joined: Feb 24, 2023

Posted:
May 24, 2024  11:34 

MattDakka wrote:  Irgy wrote:  From a Bayesian (as opposed to "frequentist") perspective, it's random for you if you, personally, don't know the outcome, irrespective of what the underlying mechanism is. 
I understand this, but I'd like to know the accurate, scientifical, technical definition of FUMBBL RNG.
True random, semirandom or pseudorandom? And why?

i find the course of Professor Salil Vadhan of Harvard University, well done, there is also an online book.
A definition of randomness can be found page page 213, chapter 7.1
https://people.seas.harvard.edu/~salil/pseudorandomness/prgs.pdf
for more detail
https://salil.seas.harvard.edu/lecturenotes
https://people.seas.harvard.edu/~salil/pseudorandomness/
In general for who is interested to go deep in detail of Entropy and Cryptography i founded quite interesting these readings
1) the fundamental paper of Shannon about Entropy and in general Mathematical Theory of communication which introduced the concept of Entropy, which is a must every master degree students approaching Probability, Random Variables and Stochastic Processes should have in his library
https://people.math.harvard.edu/~ctm/home/text/others/shannon/entropy/entropy.pdf
2) Another excellent textbook that I recommend for those who want to have a general knowledge on cryptography and personally I founded extremely clear is the following
https://link.springer.com/book/10.1007/9783662690079?source=shoppingads&locale=enit&gad_source=1&gclid=CjwKCAjw9cCyBhBzEiwAJTUWNVj2960GWeOxuUxMNVpLp1Lsk4DiCXLrmvQjanmO34nwKJldsi_BoC6rkQAvD_BwE
3)This is a very common manual among practitioners
Handbook of Applied Cryptography
Alfred J. Menezes, Paul C. van Oorschot, Scott A. Vanstone, CRC Pres 1996
Can be found on Amazon
Have a good study! 
_________________ To judge a man, one must at least know the secret of his thoughts, his misfortunes, his emotions, Balzac 


 