Random Number Generators: Really Random?

[Bill K]

 Hi, guys. It's been awhile since I've been here (although I read the political topics her a lot). Anyway, I play games and obvious random number generators play a massive role. What I am wondering, after talking to a friend studying computer science, if it is possible to create something truly random? I am having a hard time rapping my head around the concept.

 Sincerely, Bill. ʕ•́ᴥ•̀ʔっ

[CarbShark]

Yes.

The way this is proven is the RNG creates sets of millions of numbers millions of times. Then the results are analyzed to determine if any number has a higher (or lower) probability of being generated than any other number.




Sent from my iPhone using Tapatalk

[Alex Simmons]

True randomness requires use of naturally random phenomenon (e.g. atomic quantum fluctuations) as input. It's not possible algorithmically to create true randomness without including such natural phenomenon as part of the generation process. It's possible for an algorithm to create sequence of numbers that demonstrate statistical appearance of randomness, however if you know the starting state the sequence can be repeated exactly.

[Bill K]

True randomness requires use of naturally random phenomenon (e.g. atomic quantum fluctuations) as input. It's not possible algorithmically to create true randomness without including such natural phenomenon as part of the generation process. It's possible for an algorithm to create sequence of numbers that demonstrate statistical appearance of randomness, however if you know the starting state the sequence can be repeated exactly.

Thank you. That is what I thought, but my friend who specializes in computer science (somewhat as you're talking about) believes true randomness can be created in the way you say it cannot.

[The Latinist]

Most computer scientists will call such algorithms ‘pseudorandom number generators’ instead, recognizing that algorithms run on deterministic machines cannot produce truly random numbers. There are lots of techniques for creating pseudo-random numbers and some of them can give good approximations of randomness; but every one of them will, given enough time, begin to repeat the exact same sequence of random numbers.  For SHA1, for instance, that repetition starts after 2^160 random numbers are generated.  This is why, as others said above, when security is on the line we use true random number generators which draw on system information (temperature sensors, etc) as sources of entropy.

[jt512]

True randomness requires use of naturally random phenomenon (e.g. atomic quantum fluctuations) as input. It's not possible algorithmically to create true randomness without including such natural phenomenon as part of the generation process. It's possible for an algorithm to create sequence of numbers that demonstrate statistical appearance of randomness, however if you know the starting state the sequence can be repeated exactly.

Thank you. That is what I thought, but my friend who specializes in computer science (somewhat as you're talking about) believes true randomness can be created in the way you say it cannot.

I think your friend needs to think more deeply about what "true randomness" is.  Computer random number generators, more properly termed "pseudo-random number generators (PRNGs)," are deterministic: given a starting number, or seed, they generate a series of numbers using an algorithm.  Although the sequence of numbers is not "truly random," the algorithms are designed such that the generated sequence cannot be distinguished from a truly random sequence based on a standard set of statistical tests.  Nevertheless, given sufficient time and computing power, it is theoretically possible to analyze the generated sequence and make statistical predictions of future numbers—if the sequence were truly random, this would be impossible even in theory.  But the best PRNG algorithms are so complex that sufficient time and computing power doesn't exist to successfully perform such an analysis.  Hence, for practical purposes, the sequence is as good a truly random sequence.

[arthwollipot]

The online tabletop gaming platform Roll20.net uses what it refers to as a "true random" source of entropy based on the fluctuations in the power of a beam of light to drive its dice rolling engine, which it calls "QuantumRoll".

[Mr. Beagle]

To a true determinist, aren't even atomic fluctuations deterministic? More a philosophical question than a practical one.

When I taught intro programming long ago, my final project would often be to have a student programming contest of pseudo-random number generators, tested against some standard criteria fro randomness. The math students tended to outshine the business students here, though.

[Noisy Rhysling]

Most computer scientists will call such algorithms ‘pseudorandom number generators’ instead, recognizing that algorithms run on deterministic machines cannot produce truly random numbers. There are lots of techniques for creating pseudo-random numbers and some of them can give good approximations of randomness; but every one of them will, given enough time, begin to repeat the exact same sequence of random numbers.  For SHA1, for instance, that repetition starts after 2^160 random numbers are generated.  This is why, as others said above, when security is on the line we use true random number generators which draw on system information (temperature sensors, etc) as sources of entropy.

That's the term I was taught back in the computing neolithic. (Pre-Y2K.)

[CarbShark]

I believe that Apple’s RNG uses the last digit of a series of time fractions to generate the seed.


Sent from my iPhone using Tapatalk

[Noisy Rhysling]

The one Clancy used in Clear and Present Danger sounded like it was a good one, but then Clancy was selling it.

[Soldier of FORTRAN]

my friend who specializes in computer science (somewhat as you're talking about) believes true randomness can be created in the way you say it cannot.

What's his argument?

[jt512]

To a true determinist, aren't even atomic fluctuations deterministic? More a philosophical question than a practical one.

Atoms don't care if you are a true determinist.

[brilligtove]

I can't remember which corp used lava lamps to generate random numbers, but I think I heard about it on Security Now.

Looking...

Security Now Episode 299

Steve: Where we'll set up the problem and get into it, but stop short of going into the solutions which have been devised, which are really interesting. I mean, I think - I'm sure I've spoken on the podcast, for example, that somewhere at Sun Computer there was at one time cameras looking at lava lamps.

Leo: Yes.

Steve: Because lava, the flow of the wax in a lava lamp is a chaotic, unpredictable process. And so they were literally digitizing lava lamp images as a source of chaos to feed into their need for random numbers. So anyway, we've got a great podcast this week - lots of interesting updates and news, and the first half of the issue of random numbers in cryptography and the need for them and the problems of, like, that we have of generating them.

I'm not sure it this counts as "pure" randomness. It's chaotic - like a double pendulum - but not random like the static we used to pick up on the TV, or quantum decay times for a single particle. In practice, sure - utterly unpredictable.

[CarbShark]

How are we defining “true” random and how do we test it to determine if the output is truly random?


Sent from my iPhone using Tapatalk